UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Welding in pyroclastic deposits Quane, Steven Laurance 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2004-902978.pdf [ 17.63MB ]
Metadata
JSON: 831-1.0052759.json
JSON-LD: 831-1.0052759-ld.json
RDF/XML (Pretty): 831-1.0052759-rdf.xml
RDF/JSON: 831-1.0052759-rdf.json
Turtle: 831-1.0052759-turtle.txt
N-Triples: 831-1.0052759-rdf-ntriples.txt
Original Record: 831-1.0052759-source.json
Full Text
831-1.0052759-fulltext.txt
Citation
831-1.0052759.ris

Full Text

W E L D I N G IN P Y R O C L A S T I C DEPOSITS By STEVEN LAURANCE QUANE B.A. (Magna cum laude), University of Colorado, 1997 M.Sc. University of Hawaii, 1999 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Earth and Ocean Sciences) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA April, 2004 © Steven Laurance Quane, 2004 This thesis comprises a collection of papers deriving from my dissertation research and this statement is to acknowledge the contributions of collaborators in accordance with the University of British Columbia. Chapters 2 through 4 represent papers that have been submitted to refereed professional journals. Chapter 2 entitled Ranking Welding Intensity in Pyroclastic Deposits has been accepted pending revision by the Bulletin of Volcanology and was co-authored by J.K. Russell. Russell played an editorial role and helped with data analysis. Anita Grander (Oregon State University) and Martin Streck (Portland State University) were journal reviewers. Chapter 3 is published in the American Mineralogist as: A Low-Load, High-Temperature Deformation Apparatus for Volcanological Studies. The paper was co-authored with J.K. Russell (strain analysis) and L. A . Kennedy (experimental setup). The paper was informally reviewed by Hugh Tuffen (IMPG, Munich) and by journal reviewers Cliff Shaw (University of New Brunswick) and Charles Lesher (University of California at Davis). Chapter 4 entitled Welding: Insights from High-Temperature Analogue Experiments is in review (Journal of Volcanology and Geothermal Research). This paper is co-authored with J.K. Russell. Russell helped with modeling the experimental data and some of the implications. The undersigned hereby acknowledge the roles described aboye are accurate. Steven L. Quane Dr. J.K. Russell (supervisor) Dr. L .A . Kennedy (coauthor) Abstract The process of welding in pyroclastic deposits involves compaction sintering and flattening of hot glassy particles. Pronounced changes in physical properties attend welding; as welding intensifies, for example, primary porosity is reduced, density increases and the deposit becomes progressively more foliated. Consequently, welding intensity in individual deposits varies with stratigraphic depth. This thesis comprises field, laboratory and experimental studies aimed at understanding the conditions necessary for welding, the rheology and mechanisms of welding the prediction of welding intensity and timescales of the welding process. Changes in welding intensity and the accumulation of strain in single pyroclastic flow cooling units are studied using physical property measurements. Combined with petrographic indicators, these measurements are used to develop a classification scheme for welding intensity. The scheme has eight indices, demarcated by specific petrographic features correlated to a range of normalized density values used to calculate strain in welded deposits. The physical mechanisms by which strain accumulates are analyzed through deformation experiments on analogue glass beads and natural pyroclastic materials. The experiments use a new deformation apparatus capable performing high-temperature, low-load deformation experiments and collecting high-resolution rheological data. Total strain is partitioned into axial (porosity loss) and radial (bulging) components. The relative amount of each is dependent on initial porosity, temperature and strain rate. In all cases experimental cores showed a strain-dependent rheology that is more strongly affected by temperature than by load or strain rate. Results from these experiments are used to develop a relationship in which the effective viscosity (r\e) of the experimental cores is predicted by: where r| 0 is melt viscosity, <j) is sample porosity and a is a constant dependent on material properties. This predictive, rheological model provides insight into the relative roles of emplacement temperature, load and glass transition temperature on welding intensity. The model is used to predict strain accumulation with time during welding and the timescales of the welding process. ii Table of Contents A bstract 11 Table of Contents 111 , . , r r . viii List of Figures List of Tables Foreword . _ — x u Acknowledgements_ . __ XIU Chapter 1. Overview —__ 1 Introduction : ——' Previous Work . : ——^ Previous Field Studies '. ^ Previous Experimental Studies •-—^ Previous Models ^ ^ 12 Organization of Thesis References ——^ Chapter 2. Ranking Welding Intensity in Pyroclastic Deposits 19 Abstract ^ 20 Introduction 21 Previous Work 24 Metrics of Welding Intensity ^ 24 Density — Porosity ~* 28 Rock Strength . A O iii Oblateness : : 3 0 Fabric Angle 30 Covariation of Metrics \ 33 Metrics of Strain 38 Objective Classification of Welding 43 Petrographic Ranks 43 Corresponding Physical Properties 49 Application 55 Conclusions__ . 59 References 6 ^  Chapter 3. A Low-Load, High-Temperature Deformation Apparatus for Volcanological Studies _ 65 Abstract 6 5 Introduction 66 Experimental Apparatus 67 Sample Assembly 67 Furnace Assembly 69 Results 7 2 Data Treatment . 72 Experimental Results 74 Application to Welding 78 Conclusions . 80 References 80 iv Chapter 4. Welding: Insights from High-Temperature Analogue Experiments 83 Abstract 83 Introduction 84 Previous Experimental Studies 85 Experimental Device 91 Analogue Experimental Material 96 Experimental Results 99 Constant Displacement Rate .102 Constant Load _105 Data Analysis 107 Mechanisms of Strain 107 Strain Dependent Rheology .113 Implications for Welding __120 Conclusions __123 References ' .123 Chapter 5. Welding Experiments on Natural Pyroclastic Materials 128 Introduction ' 128 Experimental Material 129 Experimental Results 132 Constant Displacement Rate 135 Constant Load • __138 Physical Properties _ 145 Axial (sa) vs Radial (er) Strain 146 v Fabric Angle 149 Strain Dependent Rheology 151 Controls on a 158 Conclusions 166 References 167 Chapter 6. Rheology and Timescales of Welding 169 Introduction .169 Predictive Rheological Model for Welding ; 169 Timescale of Welding .172 Discussion ) 77 Conclusions 80 References . } 81 Chapter 7. Conclusions 182 Future Work ; 184 References_ 184 Appendix 1. Physical Property Measurements _186 Sample Collection 186 Analytical Methods 189 Visual Inspection 189 Bulk Density (p) 192 Porosity (§)_ ___192 Rock Strength (PLST & UCS) 200 Oblateness 201 vi Fabric Angle 202 References 203 Appendix 2. Experimental Data 205 Experimental Data 205 Appendix 3. Derivation of Equation 6.1 207 vii > List of Figures Figure 1.1. Schematic relaxation timescale diagram for welding 3 Figure 1.2. Schematic representation of the welding process_ 7 Figure 1.3. Photomicrographs of welding 8 Figure 2.1. Nomenclature used to describe zones of welding 22 Figure 2.2. Previous welding classification schemes 23 Figure 2.3. Physical properties from the Bandelier Tuff 27 Figure 2.4. Metrics of welding intensity 29 Figure 2.5. Normalized density vs other metrics 35 Figure 2.6. Strain calculated from physical properties 41 Figure 2.7. Macroscropic welding rank characteristics 45 Figure 2.8. Microscropic welding rank characteristics 47 Figure 2.9. Ranks related to physical properties 52 Figure 2.10. Summary of welding indices 54 Figure 2.11. Strain calculated from ranks 57 Figure 3.1. Volcanology deformation rig 68 Figure 3.2. Temperature calibration profiles 71 Figure 3.3. Experimental space of the VDR 73 Figure 3.4. Photomicrographs and photos of experimental end products 76 Figure 3.5. Strain analysis of experimental results 77 Figure 4.1. Previous experimental apparatus 88 Figure 4.2. Volcanology deformation rig 92 Figure 4.3. Experimental space of the VDR 94 viii Figure 4.4. Viscosity-temperature relationships for soda-lime silica glass 98 Figure 4.5. Photomicrograph and photograph of representative starting material 100 Figure 4.6. Experimental results from constant displacement rate experiments 103 Figure 4.7. Photomicrographs and photos of experimental end products 104 Figure 4.8. Experimental results from constant load experiments 106 Figure 4.9. Schematic of strain accommodation mechanisms 109 Figure 4.10. Strain analysis of experimental results (analogure materials) 111 Figure 4.11. Strain analysis of experimental results 114 Figure 4.12. Strain dependent rheology 116 Figure 4.13. Porosity-viscosity relationship for soda lime glass beads 119 Figure 4.14. Application of experimental data to single cooling unit 121 Figure 5.1. Grain size analysis of Rattlesnake Tuff 131 Figure 5.2. Representative starting material 133 Figure 5.3. Pre-experimental correction 136 Figure 5.4.Constant displacement experimental results 137 Figure 5.5. Photomicrographs and photographs of experimental end products 140 Figure 5.6. Correction factor for constant load data 142 Figure 5.7. Constant stress experimental results 144 Figure 5.8. Comparison of axial and radial strain 148 Figure 5.9. Fabric angle vs strain 150 Figure 5.10. Strain dependent rheology 153 Figure 5.11. Porosity-viscosity relationships 156 Figure 5.12. Porosity-viscosity paths 160 ix Figure 5.13. Photomicrographs of representative starting material for different a_164 Figure 6.1. Model prediction of experimental data 171 Figure 6.2. Porosity-viscosity paths 173 Figure 6.3. Model prediction of Bierwirth (1982) data 174 Figure 6.4. Predicted timescale of welding for Bandelier Tuff section 176 Figure 6.5. Calculated cooling times for Bandelier Tuff samples 178 Figure 6.6. Comparison of welding and cooling timescales 179 Figure A1.1. Location map for field study 187 Figure A l .2. Stratigraphic nomenclature for Bandelier tuff 188 Figure A l .3. Stratigraphic sections showing sample collection depths 190 Figure A l .4. Plots to determine connected vs non-connected porosity 199 x List of Tables Table 2.1. Attributes (strength and weakness) of each welding metric 32 Table 2.2. Best-fit relationships between normalized density and metrics 36 Table 2.3. Petrographic characteristics used to define ranks 48 Table 2.4. Parameters used to calculated thickness of Bachelor Mt. Tuff 60 Table 3.1. Experimental conditions for constant displacement rate experiments 75 Table 4.1. Summary of previous experiments 86 Table 4.2. Composition and viscosity of soda lime silica glass 97 Table 4.3. Experimental conditions and properties for analogue materials 101 Table 4.4. Experimental conditions and properties for natural materials 108 Table 5.1. Composition and viscosity of Rattlesnake Tuff ash 130 Table 5.2. Experimental conditions and properties for Rattlesnake Tuff cores 134 Table 5.3. Summary of different starting material properties-l- 161 Table A 1.1. Depths for samples collected from Bandelier Tuff 191 Table A l .2. Physical property data for samples from Bandelier Tuff 193 Table A l .3. Measurements for determination of connected porosity 198 Table A2.1. Sample experimental data worksheet 206 xi Foreword This thesis comprises a collection of papers deriving from my dissertation research and this foreword is to acknowledge the contributions of collaborators in accordance with the University of British Columbia. Chapters 2 through 4 represent papers that have been submitted to refereed professional journals. Chapter 5 is a manuscript that is intended for future publication. Chapter 2 entitled Ranking Welding Intensity in Pyroclastic Deposits has been accepted pending revision by the Bulletin of Volcanology and was co-authored by J.K. Russell. Russell played an editorial role and helped with data analysis. Anita Grunder (Oregon State University) and Martin Streck (Portland State University) were journal reviewers. Chapter 3 is published in the American Mineralogist as: A Low-Load, High-Temperature Deformation Apparatus for Volcanological Studies. The paper was co-authored with J.K. Russell (strain analysis) and L.A. Kennedy (experimental setup). The paper was informally reviewed by Hugh Tuffen (IMPG, Munich) and by journal reviewers Cliff Shaw (University of New Brunswick) and Charles Lesher (University of California at Davis). Chapter 4 entitled Welding: Insights from High-Temperature Analogue Experiments is in review (Journal of Volcanology and Geothermal Research). This paper is co-authored with J.K. Russell. Russell helped with modeling the experimental data and some of the implications. xn Acknowledgements Lets face it; I wouldn't have been able to do all of this by myself! This thesis has benefited from the assistance, interaction and generosity of countless people. I would first like to extend my gratitude to my supervisor Kelly Russell whose unwavering drive, enthusiasm and lateral thinking provided me opportunity to undertake and complete a great project. Kelly, from you I learned much more than just the contents of this thesis. Thanks for continuing to be part of the solution! Getting paid to study volcanoes is the best thing in the world. The Geological Society of America Graduate Student Grant Program (SQ), NSERC Discovery Grants Program (JKR) and UBC Teaching and Learning Enhancement Grant Program (LAK, JKR) provided funding for this project. Support was also provided by the staff of EES-6 at Los Alamos National Laboratory, the staff of IMPG at Ludwig Maximillians University Munich and the machinists at UBC, especially Ray Rodway. I also extend thanks to the people of Canada for letting me live in your wonderful country. Numerous faculty members at UBC and other universities deserve thanks for their assistance. Lori Kennedy, thanks for talking to us volcanologists. Greg Dipple, thanks for reading everything in sight and writing letters. The service of Bob Luth, Paul Hickson and Warren Poole in my examination committee is greatly appreciated Thanks also go to Stuart Sutherland, Oldrich Hungr, James Scoates, Dominique Weis, and Dick Tosdal for letting me interrupt their busy days over the years. Alex Allen, you do a great job. I would also like to acknowledge Anita Grunder for her great attitude and approach to science and Don Dingwell for providing me the opportunity to visit his lab and live in Munich for a while. Thanks to Mike Garcia for the encouragement and countless letters of recommendation. Graduate school is certainly a dynamic, exciting, difficult and strange place to spend a significant amount of time. I was fortunate enough to spend lots of this time with wonderful friends and colleagues. It is great to have friends that can challenge and inspire you in all aspects of life. Thanks Steve Israel, Scott Heffernan, Geoff Bradshaw, Dave Smithson, Mike Henrichsen, Larry Winter and Sarah Gordee for "keepin' it real". My office mate and friend Nick Austin must be acknowledged for surviving a field season and two years sharing an office with me. Steve Piercey, thanks for setting the bar and providing someone to look up to, both personally and professionally. Ken Hickey, you're the man. I would like to thank my family who bore the brunt of the emotional outbursts and financial hardships that accompany a Ph.D. Can't wait to see you more! Finally, I would like to thank my future wife Erika for wanting to share your life with me. xiii Chapter 1 Overview Introduction Welding is a physio-chemical process that involves sintering, compaction and flattening of hot glassy volcanic material (e.g., Smith 1960a; Ross & Smith 1961; Cas & Wright 1987) at temperatures above the material's glass transition temperature (Tg; Giordano et al. 2000; Gottsmann & Dingwell 2001). Welding processes change the constitution of glassy materials and can transform volcanic deposits from unconsolidated porous aggregates to dense glassy solids. In many welded volcanic deposits large (up to -50%) physical property (e.g., porosity, density and strength) variations can occur over centimeter length scales. Volcanic deposits that have undergone welding are geographically and temporally pervasive on Earth (e.g., Cook 1959) and have been recognized in other parts of the solar system (e.g., Mars; Crown & Greeley 1993); the volcanic deposit span the known range of chemical compositions for extursive volcanic rocks (e.g., from basalt to rhyolite and from peralkaline to peraluminous). Furthermore, the presence of welding is pervasive throughout volcanic processes, ranging from the agglutination of low viscosity spatter to form clastogenic lava flows (e.g., Vergniolle & Jaupart 1990) to fracture and rehealing processes on walls of volcanic conduits (e.g., Tuffen et al. 2003) to sintering and densification of both felsic pyroclastic fall (e.g., Sparks & Wright 1979) and flow deposits (e.g., Smith 1960a). The pervasive distribution of welded volcanic deposits in the rock record and the fact that welding occurs over a wide range of deposit types and compositions suggest that it is an interesting and important process in volcanology. This thesis uses field, laboratory and experimental studies to gain an understanding of the conditions necessary for 1 welding. This thesis also seeks to establish the rheology and mechanisms of deformation in hot, porous, glassy aggregates of particulate material (e.g., pyroclastic deposits) and, thus, constrain the timescales of the welding process. In silicate melts, the glass transition is a kinetic divide between liquid and glassy behavior and is approximated by the temperature-dependent curve of melt viscosity (e.g., Dingwell & Webb 1990; Fig. 1.1). This curve tracks the viscosity of the silicate melt as a function of temperature. Dingwell and Webb used the Maxwell (1867) relationship: Ga where r\ is the newtonian viscosity and GO T the elastic shear modulus of the silicate melt to translate the temperature-dependence of viscosity into a map of characteristic relaxation timescales (x) for the melt as a function of temperature (e.g., Dingwell & Webb 1990; Fig. 1.1). Figure 1.1 can now be divided into two fields, a field of "liquid behavior" where viscous flow occurs and a field of "glassy behavior" where brittle deformation occurs. Thus, at a fixed temperature there are timescales (e.g., large values of t; slow processes) where the system will deform viscously (e.g., flow) and timescales (small values oft; fast processes) that cause the same system to deform in a brittle manner (e.g., fragment). This principle allows us to understand single volcanic systems that show both apparently viscous and apparently brittle behavior. For example, in volcanic conduits prior to eruption temperatures are high and strain rates usually moderate, allowing for the melt to behave as a liquid (#1; Fig. 1.1). However, during a high strain rate event (e.g., rapid gas exsolution and vesiculation; Sparks 1978), where strain rates exceed the relaxation timescale (x) of the material, the melt crosses the barrier into the glassy field, resulting in fragmentation (#2; 2 Hi T Low T 1/T (K"1) Figure 1.1. Schematic relaxation timescale diagram for welding in pyroclastic deposits. Solid black line is the rheologic glass transition for a given composition of silicate melt. When the combination of temperature and viscosity, strain rate or relaxation timescale falls below this line the melt behaves as a glass (e.g., fragmentation). When the combination is above the line, viscous deformation (e.g., welding) can occur. Numbers 1-4 represent different points on the strain rate-time path of the eruption, emplacement and welding process followed by most welded ignimbrite deposits. See text for details. 3 Fig. 1.1). Commonly, the pyroclastic material is then transported, cooled and deposited (#3; Fig. 1.1). After deposition, the cooler deposit may compact due to the lithostatic load. The lower strain rates attending the compaction allow time for the material to structurally relax, resulting in viscous flow during welding (#4; Fig. 1.1). These are the requisite conditions for welding. In nature, a wide variety of volcanic deposits representing diverse environmental conditions undergo welding. For example, in high temperature, low viscosity explosive eruptions (e.g., basalt, phonolite; Vergniolle & Jaupart 1990) complete agglutination (or extremely dense welding) of individual spatter clasts can produce clastogenic lava flows having identical characteristics to those produced by effusive eruptions (e.g., Giordano et al. 2000; Gottsmann & Dingwell 2001). A more subtle case in which welding occurs is along the walls of volcanic conduits. High strain rate events (e.g., earthquakes) can rupture hot, viscous, glassy material. After the disturbance ends and strain rates decline, over longer time periods the material will heal and reform a cohesive mass during viscous relaxation (e.g., Tuffen et al 2003). Welding can also occur in pyroclastic fall deposits that are emplaced sufficiently proximal to the eruptive vent that they accumulate at temperatures significantly > T g (e.g., Sparks & Wright 1979). Furthermore, welding occurs in pyroclastic flow deposits ranging from block and ash flows (e.g., Hickson et al. 1999) to ash and pumice lapilli rich ash flows or ignimbrites (e.g., Smith 1960a,b; Ross & Smith 1961; Ragan & Sheridan 1972; Streck & Grunder 1995). The latter are by far the most common and widely studied volcanic deposits that characteristically exhibit welding features. On this basis, the main focus of this thesis is welding of ignimbrite deposits. However, the 4 information gained on the deformation of these hot, porous, glassy aggregates of particulate material is relevant to welding processes in all types of volcanic deposits. Previous Work Previous Field Studies Pyroclastic flow deposits or ignimbrites are poorly sorted, massive volcanic deposits containing variable amounts of ash-sized glass shards, rounded pumice lapilli, and pumiceous blocks (e.g., Sparks et al. 1973). They are the result of deposition from high-speed, hot, surface flows of pyroclastic debris that are transported as high particle concentration gas-solid mixtures (e.g., Wilson 1980, 1984; Freundt 1988). They are generally gravity driven and can be produced by gravitational collapse of Plinian eruption columns or explosive collapse of high viscosity lava domes or flows (e.g., Merapi or Montserrat; Sparks 1976). Pyroclastic flows are topographically-controlled and they fdl valleys and depressions (e.g., Fisher 1979); thicknesses of these deposits vary from several meters to 10's of meters (Fisher & Schminke, 1984) and, in the case of compound ignimbrite sheets, to well over 100 m (Smith, 1960a). Classical field studies of ignimbrites generally result in isopach and isopleth maps and estimates deposit volumes (e.g., Gilbert 1938; DeSilva 1989). The welded aspect of the deposit might only be described qualitatively. Here, I review only studies that have tired to quantify the systematic changes in properties due to welding (e.g., Smith 1960b; Smith & Bailey 1966; Ratte' & Steven 1967; Ragan & Sheridan 1972; Sheridan & Ragan 1976; Streck & Grunder 1995). Under normal conditions of emplacement (e.g., Sparks 1973) a single cooling unit of ignimbrite shows predictable welding zonation. Porosity is highest and density lowest at the top and bottom of the cooling unit and shows a smooth increase to 5 respective maximum and minimum at -40% above the base (Fig. 1.2; e.g., Sheridan & Ragan 1976). This "densely welded zone" is nearly coincident with the depth of maximum heat retention in conductive cooling models for pyroclastic deposits (e.g., Riehle 1973; Miller 1990; Riehle et al 1995). Profdes of welding intensity have been used to map 3-D variations in welding facies to which constrain the location of the volcanic source (e.g., Streck & Grunder 1995). Alternatively, such studies can be used to predict pre-welding pyroclastic deposit thickness (e.g., Ross & Smith 1961; Sheridan & Ragan 1976). Ignimbrite deposits that are welded can show a planar fabric defined by flattened pumice lapilli and glass shards (Fig. 1.2, 1.3). Uniaxial compaction (Ragan and Sheridan 1972) causes juvenile pumice (McBirney 1968) and sometimes unvesiculated juvenile clasts (Gibson and Tazieff 1967) to flatten in a direction perpendicular to the depositional surface. Larger lapilli sized particles will collapse to from, in cross-section, flame-like structures called fiamme (Fig. 1.3; e.g., Ross and Smith 1961). Fiamme appear as dense elongated glassy inclusions in a fine-grained groundmass. One method of semi-quantifying the degree of welding in an ignimbrite is to measure the axial ratio (length/height) of flattened pumice. Peterson (1979) states that deformed pumice may be a more reliable method of measuring the degree of welding in a deposit than porosity, especially in areas of significant post deposition crystallization and diagenesis. However, some care must be taken to avoid situations of secondary flattening of pumice lapilli by diagenetic compaction (e.g., McPhie et al 1993). Flattening ratios in densely welded tuffs usually measure between 5 and 10 (e.g., Ragan and Sheridan 1972; Bierwirth 1982). However, flattening ratios of up to 200 have been noted in ignimbrite deposits (Schminke and Swanson, 1967). 6 Figure 1.2. Schematic representation of the welding process, (a) An undeformed pyroclastic flow deposit immediately after deposition comprising Y-shaped ash shards and subspherical pumice lapilli (black), (b) The same deposit after undergoing 30% compaction due to welding. Pumice and shards are flattened, (c) Change in porosity % and density with depth in the welded deposit. Low porosity and high density correspond with the greatest flattening in (b) and typically occurs 35-40% up from the bottom of the deposit (e.g., Ragan & Sheridan, 1972). (d) Schematic strain ellipses showing the effects of welding on shape of pumice lapilli. Flattening ratios (length/width) are another means of mapping welding intensity. 7 A — 0 5 V _ _> ss ,9 —• — CJ E 3 CO — cd C3 -a. TJ o -_ cj -c 5 52 S3 "O _E rt — CJ a. rt — . co o 00 -O cj -O J5 3 O - D tt? rt 3 .3 c -s § ^ U "5 cu . £ S 6 1 1 > go § .£ _ 'o IS CJ — c ID E rt QQ CJ CJ cu CJ TJ © t N £ V £ - E g GO Si J3 CO c. S3 M "S 2 .£p i •_ 0, « A >? 3 GO co — GO O E £= • - co I g3 GO O co __ ,s3 u cn T J _J - 2 -8 > E CJ > S3 % <u — 1) S3 >-> _ E "St - E _ CJ GO co O - g T3 O E ^ - E E D . co O * S3 I f g o 2 ~o -S -5 E EQ - E —-» -— r<M si —! GO « .£ 60 O PH "5 3 — E rt E o -a E rt CJ 32 •CJ CJ CO E CJ c E "« « £ •55 S3 • — S3 8 Flattening ratios of magnitude higher than about 10 are thought to be representative of rheomorphic flow (Wolff and Wright, 1981). Walker and Swanson (1968) propose that laminar flowage is the result of either lower than normal glass viscosity, greater than normal load pressures, deposition and compaction on steep slopes, or a combination of these factors. Elston and Smith (1970) use the stretching lineatjons of pumice to infer flow direction of the rheomorphic tuff. Hence, it seems that apparently over-flattened pumice are the result of post-emplacement flow of the ignimbrite. The amount of strain accumulated during welding has been estimated using techniques developed to measure tectonic strain (e.g., Ramsay 1967; Dunnett 1969; Elliot 1970; Ragan & Sheridan 1972; Sheridan & Ragan 1976; Sparks & Wright 1979; Wolff & Wright 1981). Compactional strain in the form of strain ratio is calculated using the deformation induced geometrical changes of ash shards with assumed original angles of 120°, originally spherical bubble shards and originally spherical pumice lapilli (Fig. 1.3; e.g., Ragan & Sheridan 1972). Results indicate that flattening ratios (x/y) of pumice lapilli of up to 25 can be attributed to strain solely by porosity loss (coaxial strain), whereas ratios >25 must be attributed to constant volume deformation resulting from non-coaxial strain or rheomorphic flow (e.g., Ragan & Sheridan 1972; Wolff & Wright 1981). Previous Experimental Studies Welding of pyroclastic deposits is a process where laboratory experimentation can exactly match the natural conditions (e.g., temperature, load pressure, time scale; e.g., Friedman et al. 1963; Bierwirth 1982). Previous experimental studies of welding in pyroclastic deposits fall into two main categories: those meant to establish a minimum temperature of welding (Boyd & Kennedy 1951; Taneda 1957; Boyd 1961; Yagi 1966; 9 Akelaitis 1999; Mossing 2003; Grunder et al in review), and those investigating the rates of welding deformation (Smith et al 1958; Friedman et al 1963; Bierwirth 1982). The experiments designed to determine minimum welding temperatures were mainly performed on small samples (<50 mm ) in hydrothermal bombs by varying the parameters chemical composition, temperature, fluid pressure, confining pressure and time. Results indicate that more silicic compositions under increased confining pressure have the lowest minimum welding temperatures. Moderate H2O pressures lowered minimum welding temperatures by lowering the viscosity of the glass, however, excess H2O pressures created pore fluid pressures that inhibited welding compaction and effectively raised the minimum temperature for efficient welding. Very few studies published photomicrographs and, in those that did, most of the deformation was either unpredictable or non-directional (e.g., Boyd 1961) and did not replicate welding textures found in nature (e.g., Fig. 1.2; deformed Y-shaped ash shards; parallel alignment of grains). Furthermore, the degree of welding reported in these studies was entirely qualitative. Smith (1958), Friedman et al (1963) and Bierwirth (1982) developed apparatus capable of running deformation experiments at variable temperature, fluid pressures and constant compactional loads equivalent to those found in naturally occurring tuffs. These studies observe the time-dependent compaction of ash mixtures under different environmental conditions. In these studies, metal jackets laterally confine the samples; therefore, all deformation is perpendicular to the imposed stress and translated into porosity loss. Results indicate that the rate of porosity loss is greater at higher temperature and increased load. Addition of fluid pressure results in slower rates of compaction (e.g., 10 Friedman et al. 1963) due to the presence of pore fluid pressure. Details of specific experimental techniques, conditions and results are reviewed in Chapter 4. Previous Models Two models for cooling and welding compaction of pyroclastic flow deposits have been developed based on experimental data. Riehle (1973) and Riehle et al. (1995) propose that strain rate (e°) is proportional to the uniaxial load (crz) through a compaction coefficient az=j3(s") (1.1) Riehle (1973) expresses P as a function of temperature, volatile (water) pressure, and porosity using Friedman et al. (1963) compaction rate data. Riehle (1973) assumes a strain model equivalent to purely viscous deformation governs the compaction of porous ash. Model results produce density profiles equivalent to those found in natural welded ignimbrite sheets (Riehle 1973; Miller 1990; Riehle et al 1995). Bierwirth (1982) uses a different approach to modeling the welding of ignimbrite sheets. The study disagrees with the approach of Riehle (1973) stating that due to the effects of pore geometry, the compaction of porous ash is not equivalent to viscous deformation. Bierwirth uses a power law equation to predict strain rate (s°) in porous material: s" =K-f(p)-P" -e-Q/RT (1.2) where K is a strain rate constant, n is a stress factor, R is the universal gas constant, Q is activation energy P is load pressure and f(p) is a density function based on pore geometry (Caristan 1981). Perfectly viscous fluids (e.g., volcanic glass), exhibit n = 1 and n values between 2 and 5 are expected for crystalline materials. From the experimental data Bierwirth (1982) determines n = 1.7. This is likely due to heterogeneity (e.g., crystals, lithic 11 fragments) in the ash mixture. Bierwirth (1982) then converted strain into density and integrated equation 1.2 to derive an equation to define compaction rates for anhydrous Bandelier Tuff ash that is useful for samples with porosity >10% (e.g., Cas & Wright 1987). Model results indicate dense welding for an ignimbrite emplaced at 850°C would occur within one week (Bierwirth 1982). Organization of Thesis The main objectives of this thesis are to: a) explore the physical properties of natural welded deposits for ways to quantify the amount of strain accumulated during welding, b) determine, through experiment, the rheology and mechanisms of deformation for hot, porous aggregates of glassy material and c) develop a rheological model for deformation of pyroclastic material capable of predicting the timescale of strain accumulation under environmental conditions consistent with natural welding processes. This thesis represents a collection of research manuscripts (chapters) at various stages of publication aimed at reaching these objectives. Some of the chapters have already been published (Chapter 3) in peer-reviewed international journals or accepted for publication (Chapters 2 & 4) while others have undergone internal review intended for future publication in international journals (Chapters 5 & 6). Each chapter of this thesis can be viewed as a stand-alone document. Efforts have been made to minimize repetition, however, some overlap between chapters is unavoidable. Chapter 2 comprises the results of quantitative stratigraphic mapping of welding facies from four drill cores of Bandelier Tuff. This chapter develops an objective means of ranking welding intensity that combines petrographic observations with physical property measurements. The classification system uses measurements of the physical properties 12 density, porosity, point load strength, uniaxial compressive strength, pumice lapilli oblateness and micro-fabric orientation. Eight ranks of welding intensity are defined on the basis of specific changes in macroscopic or microscopic textures and are assigned specific values of each physical property. This ranking scheme can be used to quantify strain as a result of welding deformation in diverse pyroclastic deposits. Chapter 3 describes a new experimental apparatus designed to perform high-temperature, low-load deformation experiments on pyroclastic materials. Included in this chapter are the specifications and calibration of an apparatus that can run constant load and constant displacement rate experiments at temperatures up to 1100°C with continuous collection of precise rheologic data. The application of this rig to the welding process is shown by a series of constant displacement rate experiments on analogue glass beads. The apparatus provides the means for deforming glassy, particulate materials at controlled temperatures, stresses and strain rates consistent with welding. Chapter 4 represents results from constant displacement rate and constant load experiments on analogue glass beads. The rheology of these porous mixtures is determined to be strain dependent. Furthermore, the individual spherical glass beads are used as strain markers and constrain the mechanisms of strain during welding. Physical property measurements on pre and post experiment run products are used in conjunction with rheological data to develop a constitutive relationship in which the effective viscosity of porous, glassy, particulate aggregates is predicted as a function of melt viscosity and porosity. Chapter 5 represents the application of concepts developed in earlier chapters to welding in natural deposits. A series of constant displacement rate and constant load 13 experiments on sintered cores of rhyolite ash are used to develop a parallel model (e.g., Chapter 4) that describes the deformation of natural pyroclastic materials. This model is constrained by physical properties measured on the experimental products. Chapter 6 tests a rheological model based on results from Chapters 4 & 5 developed external to this thesis (Russell & Quane in review; Appendix 3) capable of predicting the accumulation of strain during welding under different environmental conditions using experimental data from Chapter 5 and from Bierwirth (1982). The model is modified to predict the timescale of welding deformation for a single cooling unit of Bandelier Tuff (Chapter 2) relative to the timescale of conductive cooling for the same deposit. Chapter 7 provides a summary of the major findings in this thesis and outlines recommended avenues for further research. References Akelaitis, C , 1999. Characterization of strain during welding of pyroclastic flow deposits: Devine Canyon, USA and Mt. Meager, Canada. B.Sc. Honours Thesis, University of British Columbia, Vancouver, B.C., 31 pp. Bierwirth, P.N., 1982. Experimental welding of volcanic ash. B.Sc. Honours Thesis, Monash University. Boyd, F.R., 1961. Welded tuffs and flows in the rhyolite plateau of Yellowstone Park, Wyoming. Geological Society of America Bulletin, 72(3): 387-426. Boyd, F.R. and Kennedy, G . C , 1951. Some experiments and calculations relating to the origin of welded tuffs. Transactions - American Geophysical Union, 32]: 327-328. Cook, E.F., 1959. Ignimbrite bibliography. Idaho Bureau of Mines and Geology Info Circular 4, 30pp. Caristan, Y., Harpin, R.J. and Evans, B. 1981. Deformation of porous aggregates of calcite using the isostatic hot-pressing technique. Tectonophysics, 78: 629-650. Cas, R.A.F. and Wright, J.V., 1987. Volcanic successions, modern and ancient; a geological approach to processes, products and successions. Allen & Unwin, London, United Kingdom, 528 pp. 14 Crown, D.A. and Greeley, R., 1993. Volcanic geology of Hadriaca-Patera and the eastern Helles region of Mars. Journal of Geophysical Research-Planets, 98 (E2) 3431-3451. de Silva, S.L., 1989. Geochronology and stratigraphy of the ignimbrites from the 21° 30'S to 23° 30' S portion of the Central Andes of N . Chile. Journal of Volcanology and Geothermal Research 37 (2): 93-191. Dingwell, D.B. and Webb, S.L., 1990. Relaxation in silicate melts. European Journal of Mineralogy, 2(4): 427-449. Dunnet, D., 1969. A technique of finite strain analysis using elliptical particles. Tectonophysics, 7(2): 117-136. Elliot, D. 1970. Determination of finite strain and initial shape from deformed elliptical objects. Geological Society of America Bulletin, 81: 2221-22236. Elston, W.E., Smith, E.I., 1970. Determination of flow direction of rhyolite ash-flow tuffs from fluidal textures. Geologic Society of America Bulletin, 81: 3393-406. Fisher, R.V., 1979. Models for pyroclastic surges and pyroclastic flows. Journal of Volcanology and Geothermal Research, 6(3-4): 305-318. Fisher, R.V. and Schminke., H.U., 1984. Pyroclastic Rocks. Springer-Verlag, Berlin, 472 pp. Freundt, A. 1998. The formation of high-grade ignimbrites; I, Experiments on high- and low-concentration transport systems containing sticky particles. Bulletin of Volcanology Volcanol 59:414-435. Friedman, I., Long, W. and Smith, R.L., 1963. Viscosity and water content of rhyolite glass. Journal of Geophysical Research, 68(24): 6523-6535. Gibson, I.L. and Tazieff, H., 1967. Additional theory on the origin of fiamme in ignimbrites. Nature, 215: 1473-4. Gilbert, C M . , 1938. Welded tuff in eastern California. Geological Society of America Bulletin, 49(12, Part 1): 1829-1862. Giordano, D., Dingwell, D.B., and Romano, C. 2000. Viscosity of a Teide phonolite in the welding interval. In: J. Marti and J.A. Wolff (Editors), The geology and geophysics of Tenerife. Elsevier. Amsterdam, Netherlands. Gottsmann, J. and Dingwell, D.B., 2001. Cooling dynamics of spatter-fed phonolite obsidian flows on Teneriefe, Canary Islands. Journal of Volcanology and Geothermal Research 105(4): 323-342. 15 Guest, J.E., 1967. The sintering of glass and its relationship to welding in ignimbrites. Proceeding of the Geological Society of London. 1641: 174-177. Hickson, C.J., Russell, J.K. and Stasiuk M.V. , 1999 Volcanology of the 2350 B.P. eruption of Mount Meager volcanic complex, British Columbia, Canada; implications for hazards from eruptions in topographically complex terrain. Bull Volcanol 60(7): 489-507. Maxwell, J.C. 1867 On the dynamical theory of gases. Phil. Trans. Roy. Soc. A157. 49-88. McBirney, A.R., 1968. Second additional theory of origin of fiamme in ignimbrites. Nature (London), 217(5132): 938. McPhie, J., Doyle, M . and Allen, R. 1993. Volcanic textures; a guide to the interpretation of textures in volcanic rocks. University of Tasmania, Centre for Ore Deposit and Exploration Studies, Launceston, TAS, Australia, 196 pp. Miller, T.F., 1990. A numerical model of volatile behavior in nonwelded cooling pyroclastic deposits. Journal of Geophysical Research B, 95: 19,349-19,364. Mossing, M . 2003. The role of temperature, stress, strain and porosity on pumice from Mount Meager, Canada. B.Sc. Honours Thesis, University of British Columbia, Vancouver, B.C., 25 pp. Peterson, D.W., 1979. Significance of the flattening of pumice fragments in ash-flow tuffs. In: C E . Chapin and W.E. Elston (Editors), Ash-flow tuffs. Special Paper - Geological Society of America. Geological Society of America (GSA), Boulder, CO, United States, pp. 195-204. Ragan, D.M. and Sheridan, M.F., 1972. Compaction of the Bishop Tuff, California. Geological Society of America Bulletin, 83(1): 95-106. Ramsay, J.G., 1967. Folding and fracturing of rocks. New York: McGraw Hill . Ratte, J.C. and Steven, T.A., 1967. Ash flows and related volcanic rocks associated with the Creede Caldera, San Juan Mountains, Colorado.. Riehle, J.R., 1973. Calculated Compaction Profiles of Rhyolitic Ash-Flow Tuffs. Geological Society of America Bulletin, 84(7): 2193-2216. Riehle, J.R., Miller, T.F. and Bailey, R.A., 1995. Cooling, degassing and compaction of rhyolitic ash flow tuffs; a computational model. Bulletin of Volcanology, 57(5): 319-336. Ross, C.S., and Smith, R.L., 1961. Ash-flow tuffs-their origin, geologic relations, and identification. U.S. Geological Survey Professional Paper, 366: 81. 16 Rust A.C. and Russell, J.K. 2000. Detection of welding in pyroclastic flows with ground penetrating radar; insights from field and forward modeling data. Journal of Volcanology and Geothermal Research, 95(1-4): 23-34. Schmincke, H.U. and Swanson, D.A., 1967. Laminar viscous flowage structures in ash-flow tuffs from Gran Canaria, Canary islands. Journal of Geology, 75(6): 641-664. Sheridan, M.F., 1971. Particle-Size Characteristics of Pyroclastic Tuffs. Jpurnal of Geophysical Research, 76(23): 5627-5634. Sheridan, M.F. and Ragan, D.M., 1976. Compaction of Ash Flow Tuffs. In: G.V.C.a.K.H. Wolf (Editor), Developments in Sedimentology. Elsevier, Amsterdam, pp. 677-713. Smith, R.L., 1960a. Ash Flows. Geological Society of America Bulletin, 71: 795-842. Smith, R.L., 1960b. Zones and zonal variations in welded ash-flows. U.S. Geological Survey Professional Paper no 354-F. Smith, R.L. and Bailey, R.A., 1966. The Bandelier Tuff; a study of ash-flow eruption cycles from zoned magma chambers. Bulletin of Volcanology, 29: 83-103. Smith, R.L., Friedman, I.I. and Long, W.D., 1958. Welded tuffs, Expt. 1. Transactions -American Geophysical Union, 39(3): 532-533. Sparks, R.S.J., Self, S. and Walker, G.P.L., 1973. Products of Ignimbrite Eruptions. Geology (Boulder), 1(3): 115-118. Sparks, R.S.J, and Wilson, L., 1976. A model for the formation of ignimbrite by gravitational column collapse. Journal of the Geological Society of London, 132(Part 4): 441-451. Sparks, R.S.J. 1978. The dynamics of bubble formation and growth in magmas: a review and analysis. Journal of Volcanology and Geothermal Research, 3: 1-37. Sparks, R.S.J, and Wright, J.V., 1979. Welded air-fall tuffs. In: C E . Chapin and W.E. Elston (Editors), Ash-flow tuffs. Special Paper Geological Society of America, Boulder, CO, United States, pp. 155-166. Streck M.J. and Grunder A.L. , 1995. Crystallization and welding variations in a widespread ignimbrite sheet; the Rattlesnake Tuff, eastern Oregon, USA. Bulletin of Volcanology, 57, 151-169. Taneda, S., 1957. Geological and petrological studies of the "Shirasu" in south Kyushu, Japan; part II, Preliminary note (2). Memoirs of the Faculty of Science, Kyushu University, Series D: Geology, 6(2): 91-105. 17 Tuffen, H., Dingwell, D.B. and Pinkerton, H. 2003. Repeated fracture and healing of silicic magma generate flow banding and earthquakes? Geology 31(12): 1089-1092. Vergniolle, S., & Jaupart, C. 1990. The dynamics of degassing at Kilauea volcano, Hawaii. Journal of Geophysical Research 95(B3): 2793-2809. Walker, G.W. & Swanson, D.A., 1968. Laminar flowage in a Pliocene soda rhyolite ash-flow tuff, Lake and Harney counties, Or. Geol. Surv. Research 1968, Chap. B. U. S. Geological Survey Professional Paper. U. S. Geological Survey, Reston, V A , United States, pp. B37-B47. Wilson, C.J.N., 1980. The role of fluidisation in the emplacement of pyroclastic flows: An experimental approach. Journal of Volcanology and Geothermal Research, 8: 231-49. Wilson, C.J.N., 1984. The role of fluidization in the emplacement of pyroclastic flows; 2, Experimental results and their interpretation. Journal of Volcanology and Geothermal Research, 20(1-2): 55-84. Wolff, J.A. and Wright, J.V., 1981. Rheomorphism of welded tuffs. Journal of Volcanology and Geothermal Research, 10: 13-34. Yagi, K., 1966. Experimental study on pumice and obsidian. Bulletin of Volcanology, 29: 559-572. 18 Chapter 2 Ranking Welding Intensity in Pyroclastic Deposits Abstract Welding of pyroclastic deposits involves flattening of glassy pyroclasts under a compactional load at temperatures above the glass transition temperature and progressive welding is recorded by changes in the petrography (e.g., fabric) and physical properties (e.g., density) of the deposits. Mapping the intensity of welding can be integral to studies of pyroclastic deposits, but making systematic comparisons between deposits can be problematical. Here I develop an objective means of ranking welding intensity which combines petrographic observations with physical property measurements. I develop the classification using physical property data from welded ignimbrites including density, porosity, point load strength, uniaxial compressive strength, pumice lapilli oblateness and micro-fabric orientation. My dataset comprises measurements on 100 samples of a single cooling unit of Bandelier Tuff as well as measurements on samples from more densely welded deposits. Eight ranks of welding intensity are defined on the basis of specific changes in macroscopic or microscopic textures. The petrographic markers identifying the onset of ranks I-VII all have corresponding values of normalized density and this allows for mapping of more subtle variations in welding intensity. Rank VIII encompasses all welded samples that have undergone rheomorphic flow and deformation. I use the ranking scheme, in association with published data, to reconstruct the pre-welding thickness of two pyroclastic flows. A version of this chapter has been accepted for publication pending moderate revisions. Quane, S.L. & Russell, J.K. Ranking welding intensity in pyroclastic deposits. Bulletin of Volcanology. [accepted pending revisions 12/03]. 19 Introduction Pyroclastic deposits emplaced at high temperatures and having sufficient thickness become welded. The welding process involves sintering, compaction and flattening of hot glassy pyroclastic material (e.g., Smith 1979; Ross & Smith 1980; Cas & Wright 1987). Pronounced changes in physical properties attend welding; as welding intensifies, for example, primary porosity is reduced, density increases (e.g., Ragan & Sheridan 1972; Streck & Grunder 1995; Rust & Russell 2000) and the deposit becomes progressively more foliated (e.g., Smith 1960a; Ragan & Sheridan 1972; Sheridan & Ragan 1976; Peterson 1979). In general, the intensity of welding reflects the aggregate effects of the load of the overlying column and time of residence at temperatures above the material's glass transition temperature (Tg) (Riehle et al. 1995; Gottsmann & Dingwell 1999; Giordano et al. 2000). Consequently, welding intensity in individual deposits generally varies with stratigraphic depth. Profiles of welding intensity have been used to: map spatial variations in welding facies (e.g., Streck & Grunder 1995), predict pre-welding pyroclastic deposit thickness (e.g., Sheridan & Ragan 1976), estimate displacements across faults (e.g., Peterson 1979), estimate hydrologic variability in tuffs for waste disposal assessment (e.g., Istok et al. 1994), and predict strength variations for engineering applications (e.g., Price & Bauer 1985). At present there is no systematic means of tracking and comparing welding intensity in and between pyroclastic deposits, which prevents quantitative comparisons of welding intensity. The purpose of this paper is three-fold. Firstly, I review the past practices of mapping welding facies variations. Secondly, I present a comprehensive dataset of metrics of welding intensity including: porosity ((()), density (p), point load strength (PLST), uniaxial compressive strength (UCS), pumice lapilli oblateness and micro-fabric orientation. These 20 data are used to evaluate the extent to which each metric records welding intensity and total strain in welded pyroclastic deposits. Lastly, I introduce a classification system for welding intensity that comprises a set of indices (or ranks) tied explicitly to deformation-induced textural changes. These indices provide an objective means of recording welding intensity and can be used universally on welded pyroclastic deposits. Previous Work Historically, the nomenclature used to describe welding intensity in pyroclastic deposits (Fig. 2.1) has been diverse and arbitrary rather than strictly defined. Many of the terms have overlapping or conflicting usage. In general, these classification schemes are qualitative and based on visual comparisons of texture (e.g., eutaxitic texture) rather than based on specific changes in physical properties (e.g., porosity). At best, such schemes can track welding variations within a single cooling unit, but they do not facilitate comparisons between deposits. Unit 4 of the Tshirege member of the Bandelier Tuff has been characterized fully by many workers and serves as an example (e.g., Smith & Bailey 1966; Vaniman & Wohletz 1990, 1991; Goff 1995). The unit is interpreted to represent a single cooling unit of ignimbrite and has a base clearly defined by a sandy surge deposit (Krier et al. 1998). The same section of the deposit has been described, by different workers, as nonwelded to partially welded (Broxton & Reneau 1995) or as nonwelded to densely welded (Krier et al. 1998). Each study is internally consistent but appears to describe different ranges in welding intensity. More rigorous schemes used to describe welding intensity are summarized in Fig. 2.2 (Smith 1960b; Smith & Bailey 1966; Sheridan & Ragan 1976; Peterson 1979; Streck & 21 High Frequency of term use unwelded incipiently welded moderately welded densely welded moderately welded incipiently welded unwelded I fused ~1 sintered highly welded Low slightly welded extremely welded slightly welded Figure 2.1. Comparison of nomenclature used to describe zones of welding. Schematic column on left represents textural variations in ash shards (open) and pumice lapilli (solid). Terms are listed in order of increasing frequency of use in the literature (left to right) and boxes denote corresponding ranges of welding intensities. 22 (a) (b) (c) (d) (e) Figure 2.2. Previous schemes for classifying zones of equal welding intensity in a single cooling unit of pyroclastic material based on changes in physical properties (e.g., p, ())) and shard shape (columns a-e): a) Smith (1960b; petrographic observations on Bandelier Tuff) b) Smith & Bailey (1966; porosity of Bandelier Tuff), c) Sheridan & Ragan (1976; density of Bishop Tuff), d) Peterson (1979; fiamme elongation of Apache Leaf Tuff) e) Streck & Grunder (1995; density of Rattlesnake Tuff). Abbreviations used include: NW: non-welded, Up: upper, Mid: middle, Lo: lower, T: transitional, DW: densely welded, PW: partially welded, IW: incipiently welded, PWP: partially welded with pumice, and PWF: partially welded with fiamme. 23 Grunder 1995). Smith (1960b) proposed six zones of welding based on petrographic traits of lapilli and ash sized fragments (Fig. 2.2a). Smith & Bailey (1966) created six zones based on the range of estimated porosities in multiple sections of Bishop Tuff (Fig. 2.2b). Sheridan & Ragan (1976) and Peterson (1979) used measurements of bulk density and pumice flattening ratios, respectively, to identify three grades of welding (Figs. 2.2c & d). Streck & Grunder (1995) developed the most complete scheme; they distinguished five zones based on measurements of bulk density and petrographic features (Fig. 2.2e). Metrics of Welding Intensity I present measurements of p, <j>, PLST, UCS, oblateness and fabric angle for samples of Bandelier Tuff, New Mexico (e.g., Smith & Bailey 1966). The suite includes 100 samples collected at a depth spacing of < lm from four drill cores (SCC-1, SCC-2, SCC-4 and NISC-2) of Unit 4 in the Tshirege member of the Bandelier Tuff (Broxton & Reneau 1995). At this location, Unit 4 is an ~20 m thick unit with a base marked by the presence of a crystal-rich, sandy surge deposit (Krier et al. 1998). These data are used to evaluate the ability of each metric to record welding intensity and to establish the nature of covariation between properties. Density Bulk densities for all 100 samples were determined using the hydrostatic weighing technique at 25 °C. Samples were rendered impermeable to H2O by spraying them with a negligible volume and weight of Krylon® Crystal Clear aerosolcoating. Densities (p-r) were calculated from: PT=p f[W,/(Wi-W 2)] (2.1) 24 where pf is the density of the fluid, Wi is the weight of the samples in air and W 2 is the weight of the sample immersed in H_0. Specific details on the procedure can be found in Hutchison (1974) and Muller (1977). Accuracy of the technique was established by measurements of multiple samples of pure quartz (±0.01 g/cm3). An estimate of precision was determined by replicate measurements on two of the unknowns. The measurements have a 1 a uncertainty of <1%. In four sections of the Bandelier Tuff, density steadily increases down-section to maxima between 13 and 17 m depth followed by a steady decrease towards the base (Fig. 2.3). Porosity Porosities for all 100 samples were measured using helium pycnometry. Samples were weighed and then flooded with helium to determine the volume of framework material (i.e., the skeletal volume) and, assuming all porosity is connected, skeletal density (ps) is determined. Total porosity is calculated using the equation: fT = £JLZ£J_ ( 2 . 2 ) Ps where pj is the bulk density of the material. If all the pores are connected then <|)T is a true estimate of sample porosity. In order to determine if samples of Bandelier Tuff have any unconnected (isolated) porosity, samples were powdered and the same measurement made, (e.g., Rust et al. 1999). A l l porosity is connected in samples investigated by this study. One sigma uncertainties were determined by replicate measurements to be - 1 % of the total. Porosity steadily decreases in all four sections of the Bandelier Tuff to minima between 13 and 17 m depth followed by a steady increase to the base of the unit (Fig. 2.3). 25 Figure 2.3. Physical properties of samples from drill core sections from Unit 4 of the Tshirege member of the Bandelier Tuff are plotted as a function of depth (m), including density (g/cm3), fractional porosity, point-load strength (PLST) (Mpa), and oblateness of pumice lapilli. Samples are from 4 different drill hole sections (SCC-1, SCC-2, SCC-4 and NISC-2). A l l metrics vary systematically with depth and define maxima or minima between 13 and 17 m depth. Error bars denote 1 s uncertainties on PLST and oblateness measurements; corresponding 2a uncertainties on density and porosity are smaller than symbols (see text for discussion of methods used to measure each metric). 26 PLST (MPa) i • i • i • i 0. Oblateness ' O.VoVo's'o.Voj 0.8 " 0.3 OA 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 0.8 0.3 0.4 0.5 0.6 0.7 O.f 27 Rock Strength The point load strength test (PLST) provides an indirect measure of the tensile strength of a rock specimen (i.e., Quane and Russell 2003). PLST measurements were made on forty-one samples from two of the drill cores of Bandelier Tuff (SCC-1 and SCC-4; Fig. 2.3). The test applies a vertical concentrated load that induces tensile stresses that cause failure parallel to the loading direction. Al l measurements were made on multiple samples with the platen direction perpendicular to the fabric orientation. Al l reported data are size corrected to equivalence with a 50 mm core diameter and the relative precision of the measurement (lo) is -10% (e.g., Quane & Russell in press). In the two measured sections, PLST increases to maxima between 14 and 17 m depth followed by a decrease to the base. Uniaxial or unconfined compressive strength (UCS) determines the breaking strength of a material under axial loading. Measurements were performed on right circular cylinders with a length (L) at least 2 times the width (W) having a minimum diameter (D) of 50 mm. Reported UCS values are the mean of the peak load at failure from multiple measurements (-5) from each sample and average standard deviation (la) is approximately 15%). Measured values of UCS define an apparent maxima at -16.8 m depth (Fig. 2.4). Uniaxial compressive strength measurements require large samples that can be machined to exact specifications. Often pyroclastic rocks are not suited to this preparation. Only nine samples from SCC-1 were large enough for UCS measurement. Alternatively, UCS can be predicted from measured values of PLST using the equation: UCS = 3.86 • PLST2 + 5.65 • PLST (2.3) (Quane & Russell in press). Calculated UCS values increase slowly down the core and then jump to a maximum at 16.8 m depth followed by a steady decrease to the base (Fig. 2.4). 28 Density (g/cm) CL 16 20 0 5 10 15 20 25 30 UCS* (MPa) i—1—i—'—r 0.3 0.4 0.5 Porosity o •0 (d ) o-o to 4 -o • o -• o • o • o o 8-o _o 12 -• . o o 16-o ° o o 20 -1 1 1 1 1 1 1 1 1 1 1 0.3 0.4 0.5 0.6 0.7 0.8 Oblateness i 1 i 1 i 1 i 0 0.4 0.8 1.2 1.6 2 PLST (MPa) OH 12 16H 20 (f) >—e—i i-e-i 1 - 6 - 1 i - e - H i-e-i i-e-< i-e-i i-e-n i O ' H-e-i i-e-i i-e-< i-e-i i-e-< i-e-i 20 30 40 50 Fabric Angle (°) Figure 2.4. Metrics used to track welding intensity in section SCC-1 of Unit 4 of the Tshirege member of the Bandelier Tuff, including density, porosity, PLST (MPa), calculated (open circles; Quane & Russell in press) and measured (solid circles) uniaxial compressive strength (UCS*), fabric angle, and oblateness. Metrics show maxima or minima at ~17 m depth. Error bars denote l a uncertainties on UCS and fabric angle measurements. Experimental uncertainties of other properties are treated as in Fig. 2.3. 29 Oblateness During progressive welding of ignimbrites, both the ash matrix and pumice lapilli deform (e.g., Ragan & Sheridan 1972). The pumice lapilli deform to form flattened ellipsoids having equatorial axes a and b and polar axis c. Ragan & Sheridan (1972) demonstrated, from measurements on cube shaped, oriented samples of Bishop Tuff and Aso 4 Tuff, that the equatorial axes (a and b) change equally as the vertical axis (c) shortens. Hence, when measured perpendicular to flattening direction, the height (c) and the length (a) of pumice lapilli fully describe the extent of deformation. Commonly, the ratio of axial length to height (a/c) is used to describe "flattened" lapilli (e.g., fiamme; Sheridan & Ragan 1976; Peterson 1979). However, to better describe deformation of pumice lapilli, I use the mathematical function for oblateness: which assumes that the equatorial axes of the ellipse (a and b) are equal. I measured oblateness for 78 of the samples collected; some samples were too poor in pumice lapilli for obalteness to be measured accurately. Reported values of oblateness are the means of measurements on > 20 pumice lapilli per sample. Each lapilli had to have a cross sectional area of at least 5mm2 and the average uncertainty (la) is ~7%. Oblateness increases to define a maxima between 13 and 17 m in all four measured sections followed by a decrease to the base of the section (Fig. 2.3). Fabric Angle the direction of flattening (e.g., Smith 1960a). The orientations of individual glass shards (relative to horizontal) were determined through thin section analysis. Digital (2.4) During progressive welding ash shards align to form a fabric perpendicular to 30 photomicrographs of oriented thin sections were taken for 19 samples from section SCC-1. I traced a minimum of 100 particles from each photomicrograph using the pencil tool in Adobe Illustrator®. Near the edges of mineral grains (e.g., phenocrysts), shards commonly show enhanced alignment or deformation. Consequently, all fabric measurements were made well away (e.g., > 1.5 times the diamter of the mineral grain) from the edges of any crystal. Using the Scion® (NIH) image analysis program, I fit an ellipse to each particle trace to best represent its orientation. The mean of each population is reported as the fabric angle: the angle of deviation from the horizontal. The most intensely welded (e.g., deformed and aligned) samples will have the lowest fabric angle. The average measurement uncertainty (la) is 3% and this was estimated by replicate analysis (e.g., five separate populations of particles) on a single thin section. Fabric angle decreases, in SCC-1, to a minimum at -13.5 m depth and stays at the same value to -16.8 m depth before increasing to the base (Fig. 2.4). Al l of the metrics collected for a single drill core section (SCC-1) are plotted in Figure 2.4. To a first order, when plotted versus depth, all metrics show significant and systematic variation throughout the core and all appear to record similar variations in welding intensity. Furthermore, the metrics show similar patterns indicating a maximum in welding at -17 m depth (Fig. 2.4). There are inherent strengths and weaknesses in the measurement and application of each metric (Table 2.1). For example, while density is relatively easy to measure and seems to provide a direct and very accurate record of subtle changes in the welding process, it can be significantly affected by post-welding crystallization (e.g., lithophysae), or the presence of dense lithic materials and abundant crystals, rendering it relatively useless in tracking welding. Normalizing density (bulk 31 < o _l 0-0. I 2. a, Ct -a c a o E LU s LU CH Z3 V) < LU c 1? l l < c c o o ~ ~ ~ *-S «= V) (0 (/) (0 ro _ o _> o o <D <D CO CO CO CO £ 0 0 0 0 ro ra o) oi xi n ro x: x: c c cz ro ro ro o o o & 0 0 cu O T= = S 3 T= CU XI XI X! JD i t 3 3 3 3 (0 (/>(/) (A (0 0) CO W CD 0 O CT) CT) ^ rod) ro x o 0 XI 3 co co J£ o o a) 3 c c o o ro ro 2i £ ro " r o XI XI "CT "O 0 CD XI CO CO o ^ ^ 0) t t I t o o 0 CD ro co .!= 3 ro ro c c 0 o <a-CNJ CO ^ ~ i 1 < C _ ro c ro 3 ro ^ X "O c .E? c ro J -.- 0 _P 0 O .— (— r— « 3 _ £ J= <-> _ CT) CO o x: E > , . . . . Q. CD CD 0 > > cq TJ o 3 3 3 CO "O c ro "co CO CO CO 0 0 0 CT) o E -S "S J5 O 3 T3 O t_ £L 0 u_ X CT) XI O 3 "CT O Q. c 0 0 ' E CT) 0 i— 3 CO II ro „ "^  -s 0 ro 0 . 1 o 6 £ c 0 E 0 i_ 3 CO ro 0 E X) ro o _5 0 co CO 0 0 > O CH . 1 -LU S CL —' CO CO < density/matrix density) can combat some of these effects. Porosity tends not to be measured as precisely as density but has the same capacity to track subtle changes in welding intensity. Porosity is also affected by post-welding alteration (e.g., secondary porosity) but is less susceptible to the affects of crystal and lithic concentrations because both bulk and skeletal (or matrix) density are used in its calculation. Rock strength offers a more indirect measure of welding intensity. PLST measurements can be made quickly in the laboratory or field on machined or irregular samples (e.g., Quane & Russell in press), but tests are destructive and require a significant amount of material. PLST can be used to calculate UCS, which is a universally accepted metric of rock strength (e.g., Quane & Russell in press). The uncertainties on rock strength measurements of welded pyroclastic deposits are relatively high compared to the other metrics (Fig. 2.4). Oblateness and fabric angle are direct measures of welding intensity that are less sensitive to the effects of post-welding alteration. Oblateness is very easily measured in the field or laboratory, however, some subtle changes in welding intensity are lost due to high variability of pumice flattening, perhaps due to differences in original porosity. It is however, a direct measure of strain accumulated by pumice lapilli (Fig. 2.4). Fabric angle measurements can be labor intensive and also do not seem to track subtle changes in welding as well as the other metrics (Fig. 2.4). Covariation of Metrics The purpose of this section is to explore the relationships, or covariation, between the individual metrics used for welding intensity. Below, density is used as the standard for tracking welding intensity. Density has commonly been used for this purpose (e.g., Sheridan 33 & Ragan 1976; Streck & Grunder 1995) and it has a low experimental uncertainty. Furthermore, by normalizing values of density to the corresponding matrix density, it can be used to compare different welded deposits. The most densely welded samples of Bandelier Tuff in my dataset have a density of ~1.8 g/cm3 and porosity of -30%. We have supplemented the dataset with measurements on more densely welded material from the Rattlesnake Tuff (e.g., Streck & Grunder 1995), Devine Canyon Tuff (e.g., Greene 1973) Walcott Tuff (e.g., Carr & Trimble 1963) and Bishop Tuff (e.g., Ragan & Sheridan 1972). Covariation between physical properties is illustrated by plotting each metric against normalized density (pn) (Fig. 2.5). Each dataset (e.g., p nvs. <j>) is fit by least squares optimization (Table 2.2) to the generalized function: y = a*Pn +c (2-5) The form of this relationship is purely descriptive and not meant to reflect physical processes controlling variations between the metrics. Porosity shows a near perfect 1:1 correlation with p n (Fig. 2.5a; Table 2.2). Density and <|) are linearly correlated through both measurement and process. This strong linear relationship indicates that densification during welding is all accommodated by a loss in porosity. If another process other than porosity loss were driving densification (e.g., secondary crystallization) then the ideal 1:1 relationship would not be maintained. The relationship between rock strength (PLST and UCS) and p n is non-linear (Fig. 2.5b & 5c; Table 2.2). For normalized densities up to -0.65, rock strength shows a moderate and near linear increase but, above this value, rock strength increases dramatically for relatively small changes in p n. At higher values of p n rock strength is less precise and, as seen in previous studies (e.g., Price & Bauer 1985; Quane & Russell in press), the 34 0.4 0.5 0.6 0.7 ).8 0.9 1 Pn Figure 2.5. Plots of normalized values of density (pn) vs. measured values of a) porosity (<))), b) PLST, c) UCS, d) oblateness, and e) average fabric angle for all samples in this study. Suite includes samples from: the Bandelier Tuff, Devine Canyon Tuff, Walcott Tuff and Bishop Tuff. Solid lines represent best-fit approximations between properties (Table 2.1). Error bars are as discussed in Figs. 2.3 & 2.4. Table 2.2 Best fit relationships and parameters between nomalized density and each metric used in this study (Fig 5). METRIC a b c R 2 . PLST UCS OB FA -0.98 8.78 117.79 -0.16 16.39 1.04 5.12 6.22 -1.82 -1.34 0.984 0 0 1 0 1 0.86 0.961 0.901 0.934 <|> = porosity; PLST = point load strength; UCS = uniaxial compressive strength; OB = oblateness; FA = average fabric angle 36 measurements are only applicable to pnup to -0.9. Unfortunately, because adequate material was not available for testing, UCS measurements were not made on supplemental material of higher p n than the Bandelier Tuff samples. However, Price & Bauer (1985) indicate a similar non-linear trend between UCS and (j) for Yucca Mt. Tuff over a very large range in strength and (|>. It is apparent that after a certain amount of densification due to welding, the mechanical strength of pyroclastic flows drastically increases. This increase results from the sintering of glassy particles as the contact area of material increases with densification. The relationship between p n and oblateness is highly non-linear (Fig. 2.5d; Table 2.2). There is a rapid increase in oblateness during the first third of densification. Above values of -0.6 p n oblateness increases only slightly and is almost constant at p n > 0.9. Sheridan & Ragan (1976) suggest that, during welding, pumice deform more rapidly than the surrounding ash matrix. Pumice lapilli show a complete loss of porosity when the matrix still has -10% porosity. This apparent decoupling of deformation processes during welding (e.g., lapilli vs. matrix) could explain the non-linear relationship between pn and oblateness (Fig. 2.5d). Fabric angle also shows non-linear variation with p n (Fig. 2.5e; Table 2.2). As the samples densify, the decrease in fabric angle with welding intensity progressively lessens. Changes in fabric angle are presumably related to rotation of individual shards into the plane of the eutaxitic texture during progressive welding. In the initial stages of welding (e.g., high low pn), there is significant room for ash shards to rotate. However, as densification proceeds (e.g., pn > -0.7) there is little opportunity for rotation of ash shards to fill void space. 37 Metrics of Strain The welding process can be thought of as the accumulation of strain in pyroclastic materials and therefore metrics that track welding are also indicators of total strain. The nature of strain induced by welding can be conceptualized by two end-member processes. Under plane strain, strain is induced by pure shear involving homogeneous flattening and requiring volume conservation. This requires flow in the plane of flattening. Conversely under volume strain, strain (e.g., flattening) is accumulated strictly by volume reduction (e.g., porosity loss). Most studies suggest that volume strain is most important during welding (e.g., Sheridan & Ragan 1976) until the onset of rheomorphic flow. If strain in pyroclastic deposits is accommodated entirely by porosity loss then <j> and p n serve as direct measurements of accumulated strain. Strain strictly accumulated from porosity loss can be calculated directly from porosity measurements s(<j>) via: where <j>o is the initial porosity of the deposit (assumed here to be 60%) and <j>i is the porosity of the strained sample. A similar estimate of total strain based on p n is given by: £(P„)=P"1~P"° (2-7) where pno is the initial normalized density of the deposit (assumed here to be 0.4) and p ni is the normalized density of the strained sample. Oblateness is, by convention, a measure of strain in pyroclastic rocks. However, it must be noted that oblateness is tracking strain in only the pumice lapilli while s(pn) and s((|>) are tracking strain in the bulk deposit. 38 The calculated values of strain based on measurements of p n, <j> and oblateness are plotted for samples from section SCC-1 (Fig. 2.6a & c). The full extended dataset is plotted in Figure 2.6b & 6d. Where plotted as s(pn) vs. e(<t>), the data essentially define a 1:1 relationship (Fig. 2.6) which is expected where the two physical properties (pn and (|>) are completely coupled (e.g., densification is driven only by porosity loss). As discussed earlier, pumice lapilli begin to deform at lower degrees of welding than does the ash matrix (e.g., Sheridan & Ragan 1976) and, ultimately, pumice lapilli may give up all porosity when the surrounding ash matrix still has -10% residual porosity. From this point onward, any further strain involving pumice lapilli (fiamme) must be on a constant volume basis (pure shear). Plots of lapilli oblateness vs. e(pn) elucidate the nature of strain in the samples of welded ignimbrite. Pumice lapilli that deform in a manner consistent with volume strain (e.g., strain accommodated porosity loss) will describe a linear relationship with a zero intercept (solid lines; Fig. 2.6c & d). Conversely, if lapilli deformation occurred on a constant volume basis (pure shear and no porosity loss) measurements of oblateness and £(pn) will have a nonlinear relationship (dashed curves Fig. 2.6c & d). The majority of oblateness and e(pn) data from SCC-1 exhibit a linear trend. However, unlike the ideal model, the observed trend does not pass through the origin. The most likely explanation is the pumice lapilli had an original oblateness prior to any welding deformation. The best approximation for data from SCC-1 is an original oblateness of 0.24, a reasonable value when compared to observations made on measurements of pumice fiamme elongation ratios from the Aso 4 and Bishop Tuffs (e.g., Sheridan & Ragan 1976; -0.35). Still some data deviate and fall below the model line for homogeneous strain with an original oblateness of 39 Figure 2.6. Values of strain calculated from physical properties measured in this study. Strain calculated from porosity, e ((j))and density, e (p) for a) SCC-1 samples and b) all samples in this study. As expected, data fall very close to model line with unit slope and zero intercept (solid line). Oblateness of pumice fiamme vs. £ (p) for c) SCC-1 samples and d) all samples in this study. Heavy solid line denotes model relationships between oblateness and strain accumulated strictly from porosity loss. The Y-intercept of the different model lines (solid and light) represent the original oblateness of the pumice lapilli. Dashed lines represent model relationship between oblateness and constant volume strain. In the SCC-1 section (c) all samples of welded ignimbrite plot closest to a model line for homogenous strain with an original oblateness value of 0.24. The majority of data collected in this study (Bandelier Tuff (open circles); Bishop Tuff (gray circles) and Devine Canyon Tuff (solid circle)) plot in the region defined by homogenous strain model lines with original oblateness values between 0.2 ±0.12. Where samples deviate from these model curves (top (T) and bottom (B) of sections), they record a greater loss in porosity than fiamme shape would suggest. This is a clear indication that, at the top and bottom of welded units, flattening of pumice lapilli can be decoupled from the progressive loss of porosity. 40 41 0.24. Such samples record a greater strain in terms of densification than suggested by measurements of oblateness. In virtually every instance, these samples were collected from the top and bottom of the flow unit where cooling is the fastest (e.g., Riehle 1973; Riehle 1995). This suggests that, at the top and bottom of welded pyroclastic flow deposits, the processes behind flattening of pumice lapilli and bulk porosity loss become decoupled. The decoupling may indicate that these processes occur at substantially different time scales. For example, early densification of the ashy matrix may result from a largely mechanical compaction of particles that operates on a short time-scale, relative to the viscous deformation driving the collapse of pumice lapilli. The "quenched" tops and bases of ignimbrites, therefore, record the points on the welding path where compaction processes dominated over processes limited by viscous flow. The entire dataset is shown in Fig. 2.6d and, although there is considerably more scatter, the pattern is similar. Nearly all data fit into the region covered by model lines allowing for an original oblateness of 0.24 ±0.12. Again, samples from the Bandelier Tuff that fall below the model line are from the top and bottom of their respective flow units. The Bishop Tuff samples (gray circles; Ragan & Sheridan, 1976) span a much larger welding range than the Bandelier Tuff samples (open circles). However, they appear to follow a linear model trend, further suggesting that, within the uncertainty of the method used, most strain in the welding interval is accomodated by porosity loss. I am well aware that some welded tuffs (e.g., Schmincke & Swanson 1967; Walker & Swanson 1968; Elston & Smith 1970; Wolff & Wright 1981; Branney & Kokelaar 1992) contain textures (e.g., highly flattened and attenuated fiamme) indicative of non-homogenous deformation. In 42 these situations, the use of physical properties and textures (e.g., density, porosity and oblateness) as strain indicators is substantially more complicated. Objective classification of welding Welding in pyroclastic deposits is accompanied by observable macro and microscopic textural changes (e.g., Smith 1960b) and, as seen above, significant changes in physical properties. Below, I present a scheme for ranking the intensity of welding in pyroclastic deposits. My classification scheme comprises 8 ranks (I-VIII) that are defined by specific changes in macroscopic or microscopic textures. Petrographic Ranks The following are criteria used to demarcate individual rank divisions (Table 2.3; Figs. 2.7 & 2.8). Rank I consists of undeformed and randomly oriented pumice blocks and lapilli in an unconsolidated matrix of ash (Fig. 2.7a) within which can still be found undeformed bubble and Y-shards (Fig. 2.8a). In Rank II the matrix and pumice lapilli remain undeformed and randomly oriented but the deposit is now consolidated (Figs. 2.7b & 2.8b). Individual pumice lapilli can be plucked from the deposit and fractures pass around, rather than through pumice lapilli (e.g., Smith 1960b). The matrix often shows incipient adhesion of glass shards but no deformation of bubble or Y-shards. By Rank III the textural effects of welding are apparent (e.g., Sheridan & Ragan 1976); pumice lapilli begin to show slight alignment (Fig. 2.7c) and bubble shards and the limbs of Y-shards show the first signs of being deformed (Fig. 2.8c). Rank IV is indicated by the first deformation of glass shards around rigid inclusions (e.g., phenocrysts; Fig. 2.8d) but the remainder of the matrix comprises mainly randomly-oriented particles that show some deformation (e.g., bubble shards and limbs of Y-shards). Pumice lapilli are clearly flattened and define a pervasive 43 Figure 2.7. Images containing some of the macroscopic petrographic characteristics used to delineate welding ranks (I-VIII) in this study, a) Photograph of pyroclastic flow deposit from Mt. Meager (Hickson et al., 1999; rank I). The deposit is random and unconsolidated with randomly oriented pumice lapilli in a loose ash matrix. Deposit is easily disturbed by hand, b) Photograph of pyroclastic flow deposit from the Bandelier Tuff (rank II). Pumice lapilli are randomly oriented in a matrix of consolidated ash (camera lens cap in upper right hand corner for scale). Deposit forms solid wall but individual pumice lapilli can be plucked with hammer and fractures go around pumice lapilli. c) Scan of hand sample slab from the Bandelier Tuff (rank III). Pumice lapilli are incipiently flattened and crudely aligned into eutaxitic texture (arrows). Fractures go through, rather than around, pumice lapilli. d) Photograph of hand sample from Devine Canyon Tuff (rank IV). Fiamme are showing a progressive alignment into a eutaxitic texture which is now easily recognizable in hand sample. Most fiamme are moderately collapsed and a few exhibit "flame-like" ends (arrows), e) Outcrop photograph from the Rattlesnake Tuff (rank V). Pumice lapilli are now significantly alignment into a strongly defined eutaxitic texture. Remnant porosity in pumice lenticules often appears as wisps stretched parallel to the direction of flattening (arrows), f) Scan of slabbed section of the Bishop Tuff (rank VI). Pumice lapilli are fully collapsed to vitrophyric fiamme with virtually zero porosity (arrows). However, the matrix still retains some porosity, g) Photograph of hand sample from the Walcott Tuff (rank VII). Sample appears as massive vitrophyre with virtually zero porosity. However, individual shards can be seen with a hand lens on cut or broken faces, h) Outcrop photograph from the Rattlesnake Tuff (rank VIII). Pumice fiamme exhibit a deformed eutaxitic texture indicative of syn or post-welding non-coaxial rheomorphic flow. 44 45 Figure 2.8. Microscopic petrographic characteristics used to delineate welding ranks (I-VIII) in this study (all thin sections are oriented perpendicular to flattening direction), a) Photomicrograph of unconsolidated glass shards from the Rattlesnake Tuff (rank I). Arrows point to selected bubble and Y-shards although most shards are tabular in form. None of the glass shards are deformed of adhered to one another b) Photomicrograph from the Devine Canyon Tuff (rank II). Undeformed bubble and Y-shards as well as an undeformed pumice lapilli (lower left) are shown with arrows. Incipient adhesion of ash shards can often be seen c) Photomicrograph of Devine Canyon Tuff (rank III). Arrows point to incipient but consistently deformed in the direction of flattening bubble and Y-shards. The remaining matrix is still randomly oriented, d) Photomicrograph of Devine Canyon Tuff (rank IV). Ash shards are wrapped around a phenocryst of quartz (bottom) while remaining mostly random elsewhere. Continuing deformation of bubble and Y-shards is seen on the right of the image, e) Photomicrograph of Devine Canyon Tuff (rank V). A weak but apparent eutaxitic texture is developed in the ash shard matrix. In this rank pumice lapilli more clearly define the alignment into eutaxitic texture (arrows), f) Photomicrograph of Devine Canyon Tuff (rank VI). The eutaxitic texture is moderately developed in the ash shard matrix with significant deformation of bubble shaped shards (arrows). In addition, pumice lapilli are strongly collapsed to fiamme (arrow), g) Photomicrograph of the Walcot Tuff (rank VII). Ash shards are strongly aligned into a eutaxitic texture and all bubble and Y-shards are flattened, h) Photomicrograph of Grey's Landing Ignimbrite (courtesy of Graham Andrews; rank VIII). Ash shards are strongly flattened and non-coaxially deformed, indicative of syn or post-welding rheomorphic flow. Further details of the petrographic characteristics used in this study can be found in text and Table 2.2. 46 47 I© V r--o (N d o A &0 1Z> — £ O « o >> ' to - D -a .1 -3 T3 £3 •o c to X ° S <U to (L) c •a § s ~° 3 C O cO _ i-'•2 s T3 C X 3 X 3 C O C >-cj 2 « CJ •I a 5 o c o M U c2 X M g ° -X u S i S - x .a o -a c -a _ _ C <0 CO « X r = O C &o —. cd o cj *j > s CJ CO o "Hi c o CJ O T3 C 3 O C O 6 c .SP "co ~o c CO c C X CJ s s cS X CO G _ 3 0 .5 -o -fi CJ CO ^ s _ _ cd . - -O •_ S o 0 3 H C X CJ >• " „ 1 •_ « T3 C CJ « o ,o -o . E CO CJU 3 X CJ 3 O >. cj c aa o Q. Q. 3 0 0 c X a. o -a x CO o CJ I c o c 48 foliation (eutaxitic texture). Some pumice lapilli are deformed to form fiamme with "flame-like" terminations (Fig. 2.7d). Rank V rocks are eutaxitic at all scales. The pumice lapilli are strongly flattened and aligned (Fig. 2.7e) and any residual porosity is stretched perpendicular to flattening. The eutaxitic texture in the ash-rich matrix is weak but pervasive (Fig. 2.8e). Rank VI clearly illustrates the different rates of strain governing deformation of pumice lapilli versus the ash matrix (e.g., Sheridan & Ragan 1976). Pumice lapilli are fully collapsed to vitrophyric fiamme and have virtually no porosity (Fig. 2.7f) whereas the matrix shows a moderate eutaxitic texture and has remnant porosity (e.g., Streck & Grunder 1995) (Fig. 2.8f). In Rank VII hand samples often appear as massive vitrophyre (Fig. 2.7g) however, individual ash shards can be detected on broken or cut surfaces. Fully collapsed pumice lapilli are often difficult to distinguish from the ash matrix. Ash shards show a strong eutaxitic texture (Fig. 2.8g). Rank VIII comprises samples that exhibit textures associated with post-welding flow or rheomorphism. Such features include overly elongated pumice fragments, pull-apart features and tension cracks, shadows around rigid inclusions, folds (Fig. 2.8h), pumice lapilli imbrication, and ramp structures (Fig. 2.7h). For an explanation of these features and other rheomorphic characteristics refer to Schmincke & Swanson (1967), Walker & Swanson (1968), Wolff & Wright (1981) and Branney & Kokelaar(1992). Corresponding Physical Properties The changes in petrographic character discussed above directly track the progression of welding in pyroclastic deposits and can commonly be used even where there has been substantial post-welding alteration (e.g., Streck & Grunder 1995). However, physical property measurements' (e.g., pn) provide for a more quantitative, precise and continuous 49 tracking of variations in welding intensity. For example, p n measurements can be used to measure the down-section gradients in welding intensity (Figs. 2.3, 2.4) or to provide estimates of accumulated strain (Fig. 2.6). Each petrographic marker has a corresponding value of p n and, therefore, values of pn can also be used to indicate rank (Table 2.3). The ranges of the other physical properties (e.g., <j>) corresponding to each rank are calculated from p n using the functions listed in Table 2.2 (Fig. 2.5). When combined, the petrographic rank divisions and corresponding ranges of welding metric values comprise an objective system for classifying welding intensity. The divisions of each rank are shown graphically in Fig. 2.9 and summarized in Fig. 2.10. The range of p n values for each rank is approximately equal (Fig. 2.9a; Fig. 2.10) suggesting a nearly exact correlation between progressive observable deformation of the pyroclastic material and densification. The range in porosity values for each rank is roughly equal because of its strong correlation with p n (Fig. 2.9b). The range of values of rock strength (UCS & PLST) in each division significantly increases with welding degree because of the non-linear relationship between these properties and p n (Fig. 2.9c & d). Conversely, the range of values for oblateness and fabric angle decrease with increasing degree of welding (Fig. 2.9e & f). It should be noted that Rank VIII does not have a corresponding set of metric values because rheomorphic textures can be evident across almost the entire welding range from as low as Rank III or IV (e.g., Streck & Grunder 1995) to most commonly Rank VII (e.g., Schmincke & Swanson 1967; Walker & Swanson 1968). Figure 2.10 compares my ranking scheme to other approaches in the literature. My system has more divisions, which allows for recognition of more subtle changes in welding. 1 use slightly different petrographic characteristics than those of Smith (1960b) to delineate 50 Figure 2.9. Ranks 1-VII for welding intensity are related to physical properties, (a) Each petrographically discriminated rank (Table 2.2; Figs. 2.7 & 2.8) is assigned upper and lower boundaries of normalized density (pn). Corresponding values of pn are then used to set index boundaries for other metrics (b-f). Rank boundaries are picked by intersection of critical values of rn with best-fit curves to individual datasets (Table 2.1; Fig. 2.5) on covariation plots of each metric vs. rn. Uncertainties are treated as in Fig. 2.3. 51 52 Figure 2.10. Rank of welding intensity (this study; I-VIII) is summarized as critical ranges of values of metrics commonly used to track degree of welding: normalized density ( p n o r m a i i z e d ) , porosity (())), PLST, UCS, oblateness and fabric angle (FA) as determined from Fig. 2.9. Bottom row shows values of strain corresponding to each rank for an ignimbrite with an original normalized density of 0.4. Ranking scheme from this study is compared to semi-quantitative welding schemes proposed in the literature, a) Smith (1960b; petrographic observations on Bandelier Tuff) b) Smith & Bailey (1966; porosity of Bandelier Tuff), c) Sheridan & Ragan (1976; density of Bishop Tuff), d) Peterson (1979; fiamme elongation of Apache Leaf Tuff) e) Streck & Grunder (1995; density of Rattlesnake Tuff) and f) Rock Mass Rating based on UCS and PLST values from geotechnical engineering practice from Heok & Brown (1980) as used by Quane & Russell (in press). Abbreviations are the same as in Fig. 2.2. 53 X CM Q D O N O S « ^ A CD A CM cn CD CO O ^ h- co °9 ^ ^ cri O 0 0 CO co o C D ° ° -f~~ Z< cr> T — co co in co i : N CM 9 rS CM cr> co r— CM o 1 0 ^ CD • O CM CM • i > to in m CD CO O C D CO cd CO CD in CO CO CO o o LO CO CO o CD CD CD o CO CM 0 0 LO CD LO CO ^ o ^ ° ^ LO o O LO CD LO CM O LO CD LO CD Q LO CM V A CD V V o V 73 .N «= — - J o O . - 0 - Q - -t> CO CO U l — I O CD A ^— CO f"— CO T — LO CD ' CD i r~- i CO CD CM N CM W CD CD OO N CM h d ^ O CM CD LO co CD 1 i LO 5 ^  O co r— N (M CD • CD • CD • CD cd CO CD CM CO CD CD cn CO T J - T -LO CO CO CO CD CM CD LO •5-A CM CD V 5 CO CD SZ TD CD t_ CD CL 14— o CD £Z o N CL.I i CNJ im imil ro _Q o CD CM o 1 TD =3 "oo CO Id I— 54 different zones. My petrographic-based divisions are substantially different from the zones of Smith & Bailey (1966) defined by variations in porosity, those of Sheridan & Ragan (1976) based on density variations, and Peterson's (1979) based on fiamme elongation. The most similar scheme to mine is that of Streck & Grunder (1995) which also uses petrographic features. The rock mass rating scheme of Hoek & Brown (1980) based on UCS and PLST values is more or less insensitive to variations in welding intensity in the middle of the welding range. The ranking scheme I have developed provides improved sensitivity to welding variation when compared to all of these schemes, especially in the beginning and middle of the welding range. In addition, because rank divisions in my system are based on petrographic observations that are common to most welded pyroclastic deposits, it can be applied universally. Application Welded pyroclastic deposits are the final products of a complex sequence of volcanic processes involving eruption, flow and emplacement of pyroclastic materials followed by degassing, compaction, annealing and flow of glassy material (e.g., Smith 1960a; Ross & Smith 1961; Freundt 1998). Two of the most sought after and important pieces of information contained in a welded ignimbrite deposit are the thickness of the deposit at the time of deposition and the corresponding dense rock equivalent amount of erupted material. Ragan & Sheridan (1972) created a single welding profile for the Bishop Tuff on the basis of numerous density measurements on a composite section. I have converted their reported bulk densities into p n and divided the section into welding ranks (Fig. 2.1 la). As indicated earlier, assuming all welding compaction is due to the removal of pore space, pn values can be used as a proxy for strain in welded pyroclastic deposits. Strain is calculated 55 Figure 2.11. Application of ranking system developed in this study to the Bishop Tuff and Bachelor Mt. Tuff, a) Schematic representation of welding rank distribution in single cooling unit of Bishop Tuff determined from analysis of data from Sheridan & Ragan (1976). Welding rank divisions are based on normalized density values from Fig. 2.10. b) Strain with depth for each data point calculated from reported density values, c) Integrated strain with depth for the same section of Bishop Tuff, d) Schematic representation of welding rank division in a multi-flow single cooling unit of Bachelor Mt. Tuff as described by Ratte' & Steven (1967). e) Average strain accumulated in each rank, f) Integrated strain with depth for each rank. Such diagrams allow for analysis of post-welding distribution of physical properties and are a basis for calculation of initial deposit thickness and distribution. 56 Degree of Welding Low High Low High Low High 1 0 0.2 0.4 0.6 0 0.1 0.2 0.3 0.4 0.5 Density Strain Integrated Strain Strain Integrated Strain 57 for each sample for which density has been reported (Fig. 2.1 lb). The s(p)-depth data were fitted to a polynomial expression which was then integrated to provide estimates of average total strain as a function of depth (Fig. 2.1 lc). The average strain in the section increases down section to a maximum at -130 m depth and then decreases slightly as the curve enters the less welded (Rank I-III) base (Fig. 2.1 lc). The value at the base (e(p) = 0.429) represents the average total strain over the entire section. This value can then be used to compute the original thickness of the section (Lo) from: = (2-8) where Li is the current thickness (-165 m). My estimate is 289 m, which is in close agreement to estimates from similar treatment by Sheridan & Ragan (1976). This calculation of original thickness for a welded ignimbrite requires accurate measurements of both bulk and matrix density and an assumption of the average pre-welding porosity of the deposit. Unfortunately, not all welded sections lend themselves to such measurements. In some instances density or porosity data may not be available. Alternatively, significant amounts of post welding crystallization (e.g., lithophysae) can change the bulk density and porosity of part or all of a welded section, rendering this method inaccurate at best. Therefore, an accurate prediction of original deposit thickness based on something other than density or porosity measurements could prove quite useful, especially in older, more altered welding facies. Petrographic features are a viable alternative, tending to be more resilient to post-welding processes and remaining useful even in significantly altered rocks (e.g., McPhie et al. 1993). Here I calculate the pre-welding thickness of a multi-flow, simple cooling unit section of the Bachelor.Mountain Tuff (Ratte' & Steven 1967) using the ranking system 58 proposed in this study. Ratte' & Steven (1967) indicate that the Bachelor Mountain Tuff is pervasively altered by crystallization during cooling of the ash flows, hence physical property measurements do not provide a reliable estimate of welding-induced changes. I have divided the Bachelor Mountain Tuff stratigraphic column into ranks (Fig. 2.1 Id) using petrographic characteristics recovered from published photographs and photomicrographs and stratigraphic descriptions from a thick section of variably welded ignimbrite (Ratte' & Steven 1967). The approximate vertical extent and rank assigned to each stratigraphic interval of the Bachelor Mountain tuff is summarized in Table 2.4. The average amount of strain recorded by each interval (sa) is calculated in the same fashion as above using the mid-value of p n consistent with the assigned rank (Table 2.4; Fig. 2.1 le). The total strain achieved in the deposit (sj) is determined using the equation: vii E,.=Yj%hr-sa (2.9) i=l where h r is thickness (%) of material constituting each individual rank in the deposit and ea is the average strain for each rank division (Table 2.4; Fig. 2.11 f)- The total calculated strain is 0.445. Using the equation above I estimate an original thickness of 2196 m for the Bachelor Mt. Tuff. This calculation demonstrates the usefulness of having an objective system for ranking welding intensity in pyroclastic deposits. Conclusions I have explored the abilities of different physical properties to map variations in welding intensity using data collected from a single cooling unit of Bandelier Tuff. I have used these data to develop a scheme for ranking or classifying the intensity of welding in pyroclastic deposits. My ranking scheme comprises 8 divisions defined in terms of specific 59 Table 2.4. Parameters used to calculate original thickness of Bachelor Mt. Tuff. Welding Index Bachelor Mt. Tuff Rank Pn ea span (m) %hr er I 0.45 0.111 59.5 4.9 0.005 II 0.54 0.259 148.6 12.2 0.032 III 0.62 0.355 118.9 9.8 0.035 IV 0.705 0.4236 237.8 19.5 0.083 V 0.795 0.4965 208.1 17.1 0.085 VI 0.88 0.545 237.8 19.5 0.106 VII 0.96 0.583 208.1 17.1 0.100 Total — 1218.8 100 0.445 pn = average normalized density; ea = average strain represented in each rank; span (m) = the thickness in meters of deposit each metric covers; %h r = the amount of deposit comprising each metric; er = total strain accumulated within the span of each rank 60 textural changes recording progressive deformation. I have also tied Ranks I-VII to characteristic ranges in physical properties (e.g., normalized density). The attributes of this proposed classification are two-fold. Firstly, it facilitates semi-quantitative comparisons of welding processes within different deposits or bodies. Secondly, by coupling the petrographic-based ranking scheme to physical properties the individual ranks represent an approximate measure of strain. References Branney MJ, Kokelaar P (1992) A reappraisal of ignimbrite emplacement; progressive aggradation and changes from particulate to non-particulate flow during emplacement of high-grade ignimbrite. Bull Volcanol, 54: 504-520. Broxton DE, Reneau SL (1995) Stratigraphic nomenclature of the Bandelier tuff for the environmental restoration project at Los Alamos National Laboratory. Los Alamos Nat Lab Rep, LA-13010-MS: 21 pp. Carr W, Trimble DE (1963) Geology of the American Falls Quadrangle, Idaho. USGS Bull. ppGl-G44. Cas, RAF, Wright JV (1987) Volcanic successions, modern and ancient; a geological approach to processes, products and successions. Allen & Unwin, London, United Kingdom, pp 1-528. Elston WE; Smith EI (1970) Determination of flow direction of rhyolitic ash-flow tuffs from fluidal textures. Geol Soc Amer Bull 81: 3393-3406. Freundt A (1998) The formation of high-grade ignimbrites; I, Experiments on high- and low-concentration transport systems containing sticky particles. Bull Volcanol 59: 414-435. Giordano D, Dingwell DB, Romano C (2000) Viscosity of a Teide phonolite in the welding interval. In: J. Marti and J.A. Wolff (Editors), The geology and geophysics of Tenerife. Elsevier. Amsterdam, Netherlands. Goff F (1995) Geologic map of technical area 21, in D.E. Broxton & P.G. Eller, eds., Earth science investigations for enviromental restoration—Los Alamos Laboratory Technical Area 21, Los Alamos National Laboratory report LA-12934-MS. Gottsmann J, Dingwell DB (1999) The cooling of welded fall-out deposits; calorimetric geospeedometry applied to Montana Blanca phonolites on Tenerife, Canary Islands. In: 61 Anonymous (Editor), A G U 1999 fall meeting. Eos, Transactions, American Geophysical Union. American Geophysical Union, Washington, DC, United States, pp. 1189. Greene RC (1973) Petrology of the welded tuff of Devine Canyon, southeastern Oregon. USGS Prof. Pap. USGS, Reston, VA, United States, 26 pp. Hickson CJ, Russell JK, Stasiuk M V (1999) Volcanology of the 2350 B.P. eruption of Mount Meager volcanic complex, British Columbia, Canada; implications for hazards from eruptions in topographically complex terrain. Bull Volcanol 60(7): 489-507. Hoek E, Brown ET (1980) Underground excavations in rock. Inst. Min. and Metall., London, United Kingdom, 527 pp. Hutchison CS (1974) Laboratory Handbook of Petrographic Techniques. New York: John Wiley & Sons 257 pp Istok, JD, Rautman CA, Flint LE, Flint A L (1994) Spatial variability in hydrologic properties of a volcanic tuff. Ground Water, 32(5): 751-760. Krier D, Caporuscio F, Lavine A, Gardner J (1998) Stratigraphy and geologic structure at the SCC and NISC building sites, technical area 3, Los Alamos National Laboratory, New Mexico, Los Alamos National Laboratory technical report LA-13507-MS. McPhie J, Doyle M , Allen, R (1993) Volcanic textures; a guide to the interpretation of textures in volcanic rocks. University of Tasmania, Centre for Ore Deposit and Exploration Studies, Launceston, TAS, Australia, 196 pp. Muller LD (1977) Density determination in Zussmann J (ed) Physical methods in determinative mineralogy London, Academic Press 663-673. Peterson DW (1979) Significance of the flattening of pumice fragments in ash-flow tuffs. In: C E . Chapin and W.E. Elston (Editors), Ash-flow tuffs. Special Paper - Geological Society of America, Boulder, CO, United States, pp. 195-204. Price RH, Bauer SJ (1985) Analysis of the elastic and strength properties of Yucca Mountain Tuff, Nevada. In: E. Ashworth (Editor), Research and engineering applications in rock masses. Proceedings - Symposium on Rock Mechanics. A .A. Balkema, United States, pp. 89-96. Quane SL, Russell JK (2003) Rock Strength as a Metric of Welding Intensity in Pyroclastic Deposits. Eur Jour Mineral 15: 855-864. Ragan DM, Sheridan MF (1972) Compaction of the Bishop Tuff, California. Geol Soc Amer Bull, 83(1): 95-106. 62 Ratte' JC, Steven TA (1967) Ash flows and related volcanic rocks associated with the Creede Caldera, San Juan Mountains, Colorado. USGS Prof Pap 524-H, H1-H58 . Riehle JR (1973) Calculated Compaction Profiles of Rhyolitic Ash-Flow Tuffs. Geol Soc Amer Bull, 84(7): 2193-2216. Riehle JR, Miller TF, Bailey RA (1995) Cooling, degassing and compaction of rhyolitic ash flow tuffs; a computational model. Bull Volcanol, 57(5): 319-336. Ross CS, Smith RL (1961) Ash-flow tuffs; their origin, geologic relations, and identification. USGS Prof Pap 366 Reston, VA, United States, 81 pp. Ross CS, Smith RL (1980) Ash-flow tuffs; their origin, geologic relations and identification and zones and zonal variations in welded ash flows. Special Publication - New Mexico Geological Society, 9. New Mexico Geological Society, Socorro, N M , United States. Rust AC, Russell JK (2000) Detection of welding in pyroclastic flows with ground penetrating radar; insights from field and forward modeling data. Jour Volcan Geotherm Res, 95(1-4): 23-34. Rust AC, Russell JK, Knight RJ (1999) Dielectric constant as a predictor of porosity in dry volcanic rocks. Jour Volcanol Geotherm Res, 91(1): 79-96. Schmincke HU, Swanson DA (1967). Laminar viscous flowage structures in ash-flow tuffs from Gran Canaria, Canary islands. Jour Geol, 75(6): 641-664. Sheridan MF, Ragan D M (1976) Compaction of ash-flow tuffs. In: G.V. Chilingarian and K.H. Wolf (Editors), Compaction of coarse-grained sediments, II. Elsevier Sci. Publ. Co., Amsterdam, Netherlands, pp. 677-717. Smith RL (1960a) Ash flows. Geol Soc Amer Bull, 71(6): 795-841. Smith RL (1960b) Zones and zonal variations in welded ash flows USGS Prof Pap 354-F, 149-159. Smith RL (1979) Ash-flow magmatism. In: C E . Chapin and W.E. Elston (Editors), Ash-flow tuffs. Special Paper - Geological Society of America. Geological Society of America (GSA), Boulder, CO, United States, pp. 5-27. Smith RL, Bailey RA (1966) The Bandelier Tuff; a study of ash-flow eruption cycles from zoned magma chambers. Bull Volcanol, 29: 83-103. Streck MJ, Grunder A L (1995) Crystallization and welding variations in a widespread ignimbrite sheet; the Rattlesnake Tuff, eastern Oregon, USA. Bull Volcanol, 57(3): 151-169. 63 Vaniman D, Wohletz K (1990) Results of geological mapping/fractures studies, TA-55 area, Los Alamos National Laboratory Seismic Hazards MemoEESI -SH90-17. Vaniman K, Wohletz K (1991) Revisions to report EES1-SH90-17, Los Alamos National Laboratory Seismic Hazards Memo EES1-SH91-12. Walker GW, Swanson DA (1968) Laminar flowage in a Pliocene soda rhyolite ash-flow tuff, Lake and Harney counties, Oregon, Geological Survey research 1968, Chap. B. USGS Prof Pap Reston, V A , United States, pp. B37-B47. Wolff JA, Wright JV (1981) Rheomorphism of welded tuffs. Jour Volcanol Geotherm Res, 10(1-3): 13-34. 64 Chapter 3 A Low-Load, High-Temperature Deformation Apparatus for Volcanological Studies Abstract I describe a new experimental apparatus designed to perform high-temperature, low-load (< 1136 kg) deformation experiments relevant to the volcanological sciences. The apparatus accommodates samples that are up to 7.5 cm in diameter and 10 cm long and can be used to run constant displacement rate and constant load experiments. The rig is ideal for volcanological studies because it uses experimental conditions that closely match those found in volcanic processes: temperature (25 to 1100°C), stress (0 to >50 MPa), strain rates (10"6 to 10"2 s"'), and total strains of 0 to > 100%. I present experimental data that show how total strain (ej) is distributed in pyroclastic material during welding. My experiments use cores of analogue (glass beads) and natural (ash and pumice) materials. Coaxial deformation of the glass beads involves equal amounts of axial (ea; volume strain) and radial (er; pure shear strain) strain until 40% strain where porosity is reduced to less than 10%. Radial strain dominates at this point. Natural materials show a different pattern because both the matrix and clasts are porous. High ratios of ea to er are maintained until all porosity is lost (ej « 80%). The implication is that welding in pumiceous pyroclastic deposits proceeds mainly by volume strain; in natural materials pure shear strain is minimal except in special circumstances. A version of this chapter has been accepted for publication. Quane, S.L., Russell, J.K. & Kennedy, L.A. (2004) A Low-Load, High-Temperature Deformation Apparatus for Volcanological Studies. American Mineralogist vol. 89 (in press). 65 Introduction Experimental methods for studying deformation processes in rocks at elevated temperatures (T) and high confining pressures (P) are well-established (e.g., Handin et al. 1972; Tullis & Tullis 1986). In particular, rock deformation presses have provided an effective means of studying mechanisms of flow in both crustal (e.g., Rutter 1993) and mantle (e.g., Karato et al 1998) environments. Despite the importance of rheology to volcanic processes (e.g., Bagdassarov et al. 1994; Dingwell 1998), with few exceptions, these experimental techniques have not been exploited by the volcanological sciences. High-T deformation experiments can supply data pertinent to the formation and collapse of lava flows and domes (e.g., Spieler 2003), the transport and fragmentation of magma in conduits (Tuffen in press), or the high-T rheology of pyroclastic material (e.g., Boyd 1961; Friedman et al. 1963; Yagi 1966; Bierwirth 1982; Quane et al. 2002, 2003; Grunder et al 2003). The purpose of this paper is three-fold. First, I introduce the volcanology deformation rig (VDR) which is designed for high-T, unconfined, low-load deformation experiments (Quane et al. 2002, 2003). The range of experimental conditions available to this apparatus make it ideal for replicating conditions associated with volcanic processes (e.g., T, load, strain rate). Second, I demonstrate its capacity for reproducible experiments that return real material properties. Lastly, I demonstrate the data that can be collected using a series of constant displacement rate experiments on cores of glass beads, volcanic ash and pumice. Welding of pyroclastic deposits is an ideal process for experimental study because the conditions attending welding in natural systems, including timescales, stresses, strain 66 rates, total strain, and temperature, are all experimentally attainable. Consequently, the experimental results can be applied directly to the natural process. Experimental Apparatus The base unit of the VDR (Fig. 3.1) is a LoadTrac II® loadframe manufactured by Geocomp Corporation. The unit performs both constant displacement rate and constant load tests. Displacement is achieved by controlling the position of the bottom platen using 6 2 an electronic stepper motor with a displacement speed range from 5x10" to 2.5 x 10" cm s" 1 and measured using a built-in linear variable differential transformer (LVDT) displacement transducer with a 7.6 cm travel range and 0.00013 cm resolution. Load is measured using an S-type load cell attached to a fixed crossarm (Fig. 3.1). Samples can be loaded at rates from zero to 1.9 kg/s and the maximum attainable load is 1136 kg with 0.086 kg resolution. The unit is factory calibrated for apparatus distortion during loading. An internal processor applies calibration factors for displacement (determined using a gauge block) and load (determined using a proving ring) thereby converting raw data into corrected output. Experimental output (e.g., measurements) can be collected every 0.01 s throughout the experiment. This stand-alone base unit has been modified to allow for running deformational experiments at magmatic temperatures on both consolidated and unconsolidated material. Sample Assembly The piston and sample assembly have been constructed to facilitate high-T experiments (e.g., >500°C up to 1100°C; Fig 1) on large samples. The VDR accommodates cylindrical samples having maximum diameters and lengths of 4.95 cm and 10 cm, respectively (Fig. 3.1). The upper piston is machined from Rescor 960 alumina (diameter 67 Figure 3.1. Schematic representation of the Volcanology Deformation Rig (VDR). Base unit is a Geocomp® Corporation Load Trac II load frame with LVDT displacement transducer (1), and load cell with an 1136 kg limit (2). Basic load frame is modified for high-T experiments by adding thermocouple (3), fiber insulated furnace (4), and temperature controller (5). The sample assembly comprises a steel spacer (a) attached to the load cell and two high temperature ceramic pistons (b) located above and below the sample (c). 68 4.95 cm; length 17.75 cm) and is suspended from the load cell by a 2.5 cm length stainless steel threaded spacer. The lower piston is identical but shorter (10.16 cm in length). The alumina is insulating and retains its high compressive strengths (415 MPa) at temperatures up to 1650°C. A tapered hole was machined through the center of the lower piston to accommodate a type-K stainless steel sheathed thermocouple. The lower piston is seated on a 15.25 cm diameter, 10 cm high, stainless steel cylinder that has a notch for the thermocouple wire. The experiments in this study are peformed on non-jacketed cores of coherent material (e.g., pumice blocks) however, the VDR can be used to experiment on unconsolidated materials, such as unsintered volcanic ash. Unconsolidated material is wrapped tightly in a jacket of 0.25 mm steel foil. Additionally, to ensure only volume strain during the experiment (e.g., no bulging) the sample can be inserted into a graphite-lubricated type L copper tube. Furnace Assembly High temperatures are required if experiments are to be pertinent to volcanic processes. Temperature was attained by adding a factory build Zircar® -type FIH fiber insulated heater tube furnace to the VDR. The furnace has helically wound Fe-Cr-Al alloy resistance wire elements embedded in a rigid body of high temperature refractory fiber, is 30.5 cm long and has inner and outer diameters of 7.6 cm and 15.25 cm, respectively. It is seated on the steel base and surrounds the lower piston, the sample and most of the upper piston (Fig. 3.1). The furnace generates temperatures up to 1100°C. A K-type thermocouple connected to a Fuji PXZ-4 PID temperature controller is used to monitor and control temperature during the experiment. The basal steel cylinder and upper steel spacer are 69 wrapped in 0.635 cm copper tubing through which tap water is run continuously to keep the load cell and LVDT at the recommended working temperatures (< 60°C). One attribute of this experimental design is that I can run experiments on relatively large samples (e.g., 750 cm3). Using large samples, however, raises the issue of temperature gradients within the sample during the experiment. To address this issue I prepared a set of pumice cores with center holes drilled throughout their length. In addition to the regular VDR monitoring thermocouple (at the base of the core), a calibration thermocouple attached to an external temperature-monitoring device was inserted into the center hole. The calibration thermocouple was then moved up the center hole to map the temperature profile from the bottom to the top of the core (Fig. 3.2). This procedure was used to measure temperature profiles (sampled at 0.5 cm intervals) for a variety of experimental configurations and as a function of pre-experiment dwell time (0.5, 1.0, 2.0 and 3.0 hrs; Fig. 3.2). Without insulation, convection of air causes a 40-50°C gradient across a 6 cm core (Fig. 3.2; curve i). I dampened this effect by wrapping Cotronics Rescor® blanket insulation around the sample assembly to completely fill the space between the assembly and inner furnace wall (Fig. 3.2). Temperature profiles are shown in Figure 3.2 for five different baffle geometries (T = 600°C). By wrapping insulation only around the upper piston (Fig. 3.2a; v) I produced a symmetric temperature profile that reaches steady-state after a 1 hour dwell time, and shows a total variation of ~12°C (Fig. 3.2b). 70 Figure 3.2. Temperature (T°C) profiles in experimental cell as a function of: a) different insulation strategies (i-v; shown below), and b) dwell time (see labels in hours) for insulation model v. Shading highlights the temperature profile used at the start of experiments after a dwell time of 1 hour; total temperature variation from top to bottom of sample is 12°C. c) Schematic of furnace and piston assembly to show range of insulation strategies (i-v; gray shading). 71 Results Data Treatment The VDR can perform isothermal experiments under the conditions of constant displacement rate or constant load. Each experiment generates a nearly continuous set of raw measurements including: time (s), load (kg), and displacement (cm) from which I compute stress (MPa), strain and strain rate (s"1). Stress (a) is calculated from: a = load I nr2 (3.1) where r is the radius of the sample core. This relationship between core geometry and stress can be exploited to extend the upper and lower ranges of the VDR even though the load cell has an upper limit of 1136 kg. In natural systems welding typically operates under stresses < 5 Mpa, however, the VDR can achieve stresses as high as -150 MPa by using 1 cm diameter cores (Fig. 3.3a). I mainly used 4.4 cm diameter cores because they provide abundant material for post-experiment analysis. Strain is calculated as: e = Alll (3.2) where Al is the experimental displacement and / is the original length of the sample. The VDR allows a total displacement of 7.6 cm; this means that a wide variety of sample lengths can be used for relatively high strain experiments (Fig. 3.3b). A 10 cm long core, for example, can undergo 75% shortening. Most of my experiments used 6 cm length cores. Cumulative (or incremental) strain rates are calculated from values of time and displacement: e°=(Alll)IAt (3.3) where At is the measurement time interval. 72 Load (kg) Displacement (cm) Figure 3.3. Ranges of stress and strain that are achievable with the VDR using different core geometries, (a) Load (kg) vs. stress (MPa) relationships showing stress limits for cores that are 1 to 7.5 cm in diameter (see labels) given the maximum load of apparatus (1136 kg), (b) Displacement (cm) vs. strain (%) realtionships showing corresponding limits on strain for cores between 2 and 10 cm in length (see labels). The rig has a maximum displacement of 7.5 cm. 73 Experimental Results Below, I explore the rheological behavior of pyroclastic materials at elevated temperatures using a set of isothermal (600°C), constant displacement rate (5 x 10"4cm/s) experiments. The experiments use fabricated (e.g., sintered) cores of 2 mm soda lime glass beads (see Table 3.1). Figure 3.4 shows cores and thin section images of the starting material (Fig. 3.4a) and run products that sustained different amounts of strain (Fig. 3.4b, 3.4c). In these experiments on sintered glass beads, strain is manifest in several ways, including: a) reduction of core length (e.g., shortening), b) change of core geometry (e.g., barrelling), c) reduction of primary porosity (e.g., volume loss), and d) deformation of original spherical beads (e.g., flattening). These indicators of strain represent the combined effects of axial strain (ea) accommodated by porosity loss (volume strain) and radial strain (er) which conserves volume and requires geometric changes (pure shear strain). One attribute of these experiments is that independent post-experiment measurements (<j>, Al, r) can be used to show how strain is being accommodated (e.g., volume vs pure shear). In Figure 3.5a I have compared the total strain implied by machine displacement (sxm) against the total strain based on shortening of the sample (STS)- Here, over the full range of strain explored by these experiments Sim is essentially equivalent to 8TS (Fig. 3.5a). The shortening of the sample records the total strain accumulated in the sample, however, I express this deformation in terms of two separate components. These components are calculated independently; axial strain is calculated from: 74 Table 3.1. Summary of experimental conditions used for constant displacement rate experiments illustrated in Fig. 3.4, including temperature, displacement rate (AL/At), and total strain (sT). Other parameters include the initial (<|>o) and final ({J)F) fractional porosity, as well as, the computed axial (sa) and radial (sr) strain. No. T(°C) "Ami v°a; l u u l " 1 AL/At (cm s"1) <t>0 <t>F sq0804b 600 0 0.311 0.311 0 0 sq0717a 600 5 x 10"4 0.377 0.331 0.123 0.22 0.21 sq0627a 600 5 x 10"4 0.625 0.343 0.053 0.27 0.48 75 Figure 3.4. Photographs and photomicrographs of run products from constant displacement rate experiments, (a) Starting materials are cores made by sintering of 2 mm soda-lime glass beads. Run products are shown at b) 40% and c) 60% strain. Strain is accommodated during experiment by porosity loss (axial strain) and changes in core radius (radial strain). Photomicrographs show how originally spherical beads (B) progressively track strain by deforming (coaxially) to form to ellipses. Porosity, represented by colored epoxy (E) is also reduced simultaneously 76 Figure 3.5. Analysis of strain in high-T deformation experiments of cores of glass beads, pumice, and volcanic ash. a) Comparison strain computed from machine displacement (£Tm) to strain recorded by samples as shortening (ETS). Data are distributed evenly along 1:1 line for full range of strains (see text), b) Comparison of axial strain (£a) and radial strain (er) for experiments involving soda lime glass beads (open circles) including the 2 experiments summarized in Table 3.1 (solid circles). Axial strain is calculated from the reduction of porosity and represents volume strain. Radial strain is calculated from the geometry of the core (barreling) at the end of the experiment and represents pure shear strain. Dashed lines are iso-strain contours, c) Comparison of axial strain (£a) and radial strain (£r) for experiments on natural materials (solid circles) and compared to data from glass beads (e.g., b). See text for discussion. 77 where <(>o' is the initial sample porosity and <b i the porosity of the run product determined by image analysis (Table 3.1). Radial strain is given by: r \ where ro is the radius of the core before experiment and v\ the mean radius of the run product. Calculated values of ea and er for 32 experiments on glass beads performed at different experimental conditions are plotted in Figure 3.5b. For the analogue experiments there is a 1:1 relationship between ea and er until -40-45% strain is attained. At this point porosity is reduced to -10% and sr becomes dominant. Application to Welding The glass beads serve as an excellent analogue material for investigating aspects of welding processes in volcanology. The material is effective because the beads are compositionally homogeneous, they have a uniform, well-defined glass transition temperature (Tg; 440-460°C) and they have a regular (e.g., spherical) geometry thereby providing another strain marker. As shown in Fig. 3.4, the spherical beads accommodate strain by flattening to form oblate spheroids. Indeed, measured values of oblateness for these strain markers fall on the theoretical curvilinear relationship for constant volume coaxial strain (Quane & Russell in review). In this regard, the analogue experiments do not capture the full essence of welding in pyroclastic deposits. In natural systems, starting porosities are higher (-50-75% e.g., Smith 1960; Ross & Smith 1961; Smith & Bailey 1966; Sheridan & Ragan 1976) and comprise matrix porosity as well as the porosity in individual pyroclasts. As strain accumulates clasts will deform (flatten) but their deformation is not limited by a constant 78 volume constraint (e.g., Sheridan & Ragan 1976; Quane & Russell in review) as is the deformation of glass beads. The consequence of this difference is shown by experiments on cores of pumiceous rhyodacite (N=7) and sintered cores of rhyolite ash (N=6). These data are plotted in Fig. 3.5c. In every case ea (e.g., volume strain) dominates over sr until very high values (~ 80%) of total strain. At this point, presumably all pore space has been lost and, thus, further strain is dominantly radial (er; Fig. 3.5c). One implication is that welding of pyroclastic deposits is controlled by volume strain processes and, therefore, welding in pyroclastic deposits is most likely limited by porosity loss. My welding experiments indicate that in natural systems pure shear strain (sr) contributes only a minor (<10%) component to the total strain during welding as long as there is significant porosity. I suggest that pure shear strain is, in fact, not important to welding processes in most pyroclastic deposits except under extraordinary circumstances. Pure shear strain may be important in situations where: a) the deposit's residence time in the welding window (e.g., T > T g) is great relative to the average timescale of deformation, or b) the deposit is on a substantial slope. The former situation might be realized in very thick deposits or in deposits emplaced at T » T g . Higher strain rates would result from increased load (thickness) or because of the very short relaxation timescales associated with the high emplacement temperatures (e.g., Dingwell & Webb 1990). The higher strain rates allow for attainment of near zero porosity before the system is thermally quenched. In the remaining deformation interval, pure shear strain would dominate. In the latter situation, where the pyroclastic deposit is on a substantial slope the overburden load may be resolved into shear stresses that induce non-coaxial strain that is not 79 necessarily coupled to porosity loss (e.g., pure or simple shear). Therefore the sr contribution to total strain could be substantial (e.g., rheomorphism; Wolff & Wright 1981). Conclusions The experimental device presented here is capable of exploring a unique portion of experimental space (e.g., T, stress and strain rate). The VDR provides a vehicle for exploring relationships between temperature, load and strain and the extent of compaction, sintering and flattening of hot pyroclastic mixtures (e.g., ash and lapilli). Such experimental results are required to: a) constrain the mechanisms of deformation controlling welding processes, and to b) develop constitutive relationships for the rheology of pyroclastic materials. This basic science is needed to address volcanological issues such as predicting the distribution of welding in ignimbrites, or establishing timescales of cooling, welding, and compaction of ignimbrites, or estimating the paleothickness of partly eroded ignimbrites. References Bagdassarov, N . , Dingwell, D.B., & Webb, S.L. (1994). Viscoelasticity of crystal-and bubble-bearing melts. Physics of the Earth and Planetary Science Interiors. 83, 83-99. Bierwirth, P.N. (1982) Experimental welding of volcanic ash. Bachelors Thesis, Monash University. Boyd, F.R. (1961). Welded tuffs and flows in the Rhyolite Plateau of Yellowstone Park, Wyoming. Geological Society of America Bulletin, 72, 387-426. Dingwell, D.B. (1998): Recent experimental progress in the physical description of silicic magma relevant to explosive volcanism. In: The Physics of Explosive Volcanic Eruptions (Eds. J.S. Gilbert & R.S.J. Sparks). Geological Society, London, Special Publications, 145, 9-26. Dingwell, D.B. & Webb, S.L. (1990) Structural relaxation in silicate melts. European Journal of Mineralogy 2, 427-449. 80 Friedman, I., Long, W. and Smith, R.L. (1963) Viscosity and water content of rhyolite glass. Journal of Geophysical Research 68, 6523-6535. Grunder, A. L., Laporte, D. & Druitt, T. H. (2003) Experimental constraints on welding in rhyolitic ignimbrite. AGU-EUG-EGS Joint assembly, Nice, France 2003. Handin, J., Friedman, M . , Logan, J.M., Pattison, L.J., & Swolfs, H.S. (1972). Experimental folding of rocks under confining pressure; Buckling of single-layer rock beams. In: Flow and Fracture of Rocks, Geophysical Monograph, American Geophysical Union, 16, 1-28. Karato, S., Zhang S., Zimmerman, M.E., Daines, M.J., & Kohlstedt, D.L. (1998) Experimental studies of shear deformation of mantle materials; towards structural geology of the mantle. Pure and Applied Geophysics. 151, 589-603. Quane & Russell (in review) Ranking welding intensity in pyroclastic deposits. Bulletin of Volcanology. Quane, S.L., Russell, J.K. & Kennedy, L.A. (2002) The rheology of welding: experimental deformation studies. Experimental Mineralogy Petrology and Geochemistry Conference IX, 7, 86. Quane, S.L., Russell, J.K. & Kennedy, L.A. (2003) Rheology of welding: experimental constraints. AGU-EUG-EGS Joint assembly, Nice, France 2003. Ross, R.S. & Smith, R.L., (1961) Ash-flow tuffs; their origin, geologic relations and identification. United States Geological Survey Professional Paper 366. Rutter, E.H., Experimental rock deformation: techniques, results and applications to tectonics. Geology Today, 9, 2, 61-65. 1993. Sheridan M.F. and Ragan D.M. (1976) Compaction of ash-flow tuffs. In: G.V. Chilingarian and K.H. Wolf (Editors), Compaction of coarse-grained sediments, II. Elsevier Sci. Publ. Co., Amsterdam, Netherlands, pp. 677-717. Smith, R.L., (1960). Ash flows. Geological Society of America Bulletin, 71, 795-842. Smith, R.L. and Bailey, R.A.(1966). The Bandelier Tuff, a study of ash-flow eruption cycles from zoned magma chambers. Bulletin of Volcanology, 29, 83-104. Spieler, O., Dingwell, D.B., & Alidibirov, M . (2003) Magma fragmentation speed: an experimental determination. Journal of Volcanology and Geothermal Research, 129, 109-123. Tuffen.H., McGarvie, D.W., Pinkerton, H., Gilbert, J.S. & Brooker, R. (in press) Simultaneous collapse and fragmentation of a partially-disaggregated rhyolite foam erupted beneath a glacier at Dalakvisl, Torfajokull, Iceland. Geology. 81 Tullis, T.E. & Tullis, J. (1986) Experimental rock deformation techniques. In: Hobbs-B-E & Heard-H-C (eds.) Mineral and Rock Deformation; laboratory studies; the Patterson Volume. Geophysical Monograph 36, 297-324. Wolff J.A. & Wright, J.V. (1981) Rheomorphism of welded tuffs. Journal of Volcanology and Geothermal Research. 10, 13-34. Yagi, K., (1966) Experimental study on pumice and obsidian. Bulletin of Volcanology 29, 559-572. 82 Chapter 4 Welding: Insights from High-Temperature Analogue Experiments Abstract The rheological behavior of pyroclastic deposits during welding is incompletely understood and is based on a surprisingly small number of experimental studies. Here I present results from a new experimental apparatus comprising an automated uniaxial compression load frame that can run constant load (up to 1150 kg) or constant displacement rate (10~6 to 0.25 cm/s) tests at elevated temperatures (< 1100°C). Deformation experiments were performed on pre-fabricated cylinders (4.5 cm diameter, ~6 cm length) of soda lime silica glass beads (N=32), sintered rhyolite ash (N=7) and cores of pumiceous rhyodacite (N=6). Experimental runs used strain rates from 10*5 to 10"3 s"1 and stresses of ~0 to 5.24 MPa. Temperatures varied from 535 to 650°C for experiments on soda lime silica glass beads and 825 to 950°C for natural materials. In all cases experimental cores showed a strain-dependent rheology that is more strongly effected by temperature than by load or strain rate. Results from these experiments are used to develop a relationship in which the effective viscosity (r|e) of the experimental cores is predicted as by: where r\o is melt viscosity and is sample porosity. This rheological model provides insight into the relative roles of emplacement temperature, load and glass transition temperature on welding intensity. A version of this chapter has been accepted for publication pending minor revisions. Quane, S.L. & Russell, J.K. Welding: Insights from High-Temperature Analogue Experiments. Journal of Volcanology and Geothermal Research [accepted 4/04]. Ve = Vo e x P 83 Introduction Welding of pyroclastic deposits involves flattening of glassy pyroclasts under a compactional load at temperatures above the glass transition temperature (Tg; e.g., Smith 1960; Ross & Smith 1961; Ragan & Sheridan 1972). Progressive welding is attended by pronounced changes in the physical character of the deposits (e.g., porosity loss, density increase, increased foliation; Smith & Bailey 1966; Peterson 1979; Streck & Grunder 1995; Quane & Russell in review; Chapter 4). The factors governing the intensity of welding in a pyroclastic deposit ultimately involve emplacement temperature, load pressure, particle size and distribution and amount of time the deposit is held at temperatures > T g (e.g., Smith 1960; Guest 1967; Riehle 1973). Welding in pyroclastic deposits is a process where laboratory experimentation can match the natural conditions (e.g., temperature, load pressure, time scale; e.g., Friedman et al. 1963; Bierwirth 1982). Experiments on natural ash-lapilli-crystal mixtures, for example, can provide direct estimates of the minimum temperatures required for welding (e.g., Boyd 1961; Friedman 1963; Bierwirth 1982). However, the complexities of the natural materials (e.g., crystal content, glass composition, variable grain size) can obscure the fundamental aspects of welding. My approach is to use both natural and analogue materials to explore the mechanisms and rheology of welding processes in pyroclastic rocks. Specifically, I have performed a series of high-temperature deformation experiments at both constant load and constant strain rate. My results on porous materials show a strain-dependent rheology. This implies that welding intensity is non-linearly related to parameters such as load and temperature. Qualitatively, my experiments show that the effects of increasing temperature are at least an order of magnitude greater than the effects of increasing load. Lastly, my 84 experimental data are used to establish a constitutive relationship between melt viscosity, porosity and effective viscosity of the material being welded. Previous Experimental Studies Previous experimental studies of welding in pyroclastic deposits fall into two main categories (Table 4.1). Several were designed to establish a minimum temperature of welding (Boyd & Kennedy 1951; Taneda 1957; Boyd 1961; Yagi 1966; Akelaitis 1999; Mossing 2003; Grunder et al in review). Studies of Smith et al (1958), Friedman et al (1963) and Bierwirth (1982) also investigated the rates of welding deformation. Boyd and Kennedy (1951) were the first to experimentally investigate welding in natural materials. They showed that dry rhyolite ash from Mono Craters, California welded at temperatures between 775 and 900°C depending on load pressures. Boyd (1961) expanded the study to include experiments on rhyolite ash (95%) containing 0 - 0.4% H2O. Small samples were sealed in platinum capsules and placed in cold seal bombs under a fixed pressure (argon). He showed that the extent and minimum temperature of welding were strongly affected by the water content of the glass, confining pressure, and the experimental dwell time. Boyd (1961) concluded a minimum temperature of welding for the Yellowstone tuff on the order of 600°C. The study includes a single photomicrograph of the experimentally welded ash and Boyd (1961) noted that the sample lacked the common traits associated with welding in pyroclastic material (e.g., eutaxitic alignment and deformation of bubble and Y-shaped ash shards). This result was ascribed to the fine-grained nature of the sample (no bubble-rich or Y-shaped shards) and the presence of a hydrostatic pressure. Nevertheless, these studies clearly demonstrate the effects of water content and confining pressure on welding processes. 85 o o o3 C t-—i O !2 (D X CM X <u o H-» cu c .2 c o u T3 C 03 cu <H-H o 0 3 X 03 H s ^ 3 <J 03 a. Q Q so © in CN so p in ro' r-l oo in © CN I ? s >3 o CN OS o o CN o o SO O <N SO in o IT) U e 3 OH O >^ X U. m ON 03 •O CU C 03 H X ! OO 03 O >> X ! u CO SO 5 D D os in oo in so os T3 o CQ CN ro d cu a o a o IT) SO in © ro >^ o © o in -a-o p CN 10-1 so ro i CN © 1/0 o CN O in in ( N o 00 SO 00 © o © o O 00 m in wo so CN ro CN CN O in CN T—H CN X X X X CN m ro Os "3b X <U ro SO Os oo in 03 ^ o x •S a <U -^v .•S cu I—I +-> >s <u •s ^  cz> 03 SO SO Os M 03 O o Os CN >^'G X o3 CN OO O S X Ul CU s ro © O o o © © sO O in o © u o3 T3 O >s X Ul Os Os Os o o in in o ro 00 sO m in m i *—' oo V V Z> 2 © o Q Q SO 00 in CN o in Os o in oo o ro O CN o X & © i—i D 2 © CN O © CN I O o sO in ro X X o3 <u o X ro O o w c ro O o CN CD -a c 3 O m CN O © in SO in co m co ro © in 03 <U X 60 •a 3 X H X cci c u O cu o u S c > -C C cu O o o< \S <U X o3 l a w g <U n U, "Jj -^v 4-j p CO C O ^ 1 -3 K.S o fit u w 03 Ui X 7 3 oo CU Ui 3 cu b —^» 3 C 5 °* CU & ,CU c .5 60 OH OT 2 S § ^ •2 t x c x? y ^ "r x g s ? .5 S p •a -S 3 CU OH Xi co 23 03 O • 6 CO 3 cu 3 03 CO H-> fU cu CU OH S CU O H X § XT' l-l / - i . "2 7 3 "o 3 oo cu OO 03 UH C O 2 — co a CU o p b <u 5 5 u, 3 =-(—i <U c/j H-) oo cu C -S T3 CD *—' a | £ ^ UIS •>> OH o3 « d j : H-I , r> c a IS u c CJ o iS O OH C oo P T3 86 Taneda (1957) varied temperature and load pressure in unconfined experiments to produce welded material. He used these results to make stratigraphic correlations between unwelded and welded facies of the same ignimbrite but also constrained welding to conditions below 920°C at 0.5 MPa load pressure (-30 m depth). Smith (1960) suggested that the minimum welding temperatures could be even lower because Taneda (1957) samples would have been fully dehydrated at the temperatures used in the experiments. In similar constant load experiments, Akelaitis (1999) showed that dry rhyodacite material deformed at average strain rates of 4 x 10"7, 1 x 10"6 and 3 x 10"6 s"1 at temperatures of 800, 900 and 1000°C, respectively under a stress of 0.003 MPa. Yagi (1966) explored the influence of chemical composition on welding efficiency by working with three different materials. The compositions ranged from rhyolite (Shiritaki obsidian and Toya pumice) to andesite (Usu pumice). The samples were crushed and sieved to < 0.2 mm and sealed in silver tubes with 10-20 cm3 of H2O and heated. Again, the hydrostatic pressure in the experimental runs prevented the development of compactional textures typical of natural welded tuffs. The study did, however, show the compositional dependence of welding. At a temperature of 622°C (under water-saturated conditions) the more siliceous Shiritaki obsidian was welded whereas the andesitic Usu pumice fall failed to weld. Smith (1958) and Friedman et al (1963) aimed to constrain the actual rates and paths of welding. Friedman et al (1963) used a deformation apparatus capable of constant load experiments at variable T and fluid pressure (Fig. 4.1a). Experiments were performed on samples of unwelded ash from the Bandelier Tuff having an initial porosity of-50%. Material was shaken and compacted to -1 cm thickness in a stainless-steel-jacketed copper 87 (a) potentiometer "O" ring -water jacket copper gasket to water pump furnace D sample weights (b) J___L to chart recorder to Argon gas supply ft weights holding screw , displacement transducer thermocouple sample I 30 cm i to power source Figure 4.1. Apparatus used in previous experimental studies of welding, a) Friedman etal (1963). b) Bierwirth (1982). .88 vessel with a platinum liner and cover. A series of pistons were seated on top of the sample. The bomb was sealed with a cooled rubber O-ring and weights were placed on a lever connecting the steel piston to load the sample. Fluid (H2O) pressure in the bomb was controlled with a hand pump. Displacement (e.g., compaction or shortening) was measured as a function of time by a potentiometer and recorded on a strip chart (Fig. 4.1a). The samples were confined laterally and, therefore, all shortening was assumed to be due to porosity loss. Thus, Friedman et al. (1963) was the first to analyze data in terms of porosity reduction with time. Experiments were run at temperatures between 485 and 900°C and at load pressures up to 3.6 MPa (equivalent thickness of-120 m). These data have been the basis for a number of computational models for simulating compaction and welding of pyroclastic material (e.g., Riehle 1973; Riehle et al 1995). There were no photomicrographs of the experimental run products, so the mechanisms of deformation in the experiments were not directly investigated. Bierwirth (1982) also investigated the rates of compaction on ash from the Bandelier Tuff using a constant load apparatus (Fig. 4.1b). The starting material (71% ash and 29% crystals and lithics by weight) was sieved for the ash fraction (<2mm) and packed into a small copper vessel to a thickness of 1 cm (Fig. 4.1b). Starting porosities were -50%. The vessel was fitted on top with a copper disc and placed in a metal holder. A metal plunger was joined to a steel piston by a heat resistant alumina rod designed so that it had free travel into the sample vessel (Fig. 4.1b). Samples were flooded with argon during the experiment to prevent oxidation of the glass. Weights were placed on top of the piston assembly to deliver a specified load. Load was imposed instantaneously by release of a retaining screw (Fig. 4.1b). A displacement transducer connected to the central piston measured shortening 89 as a function of time (Fig. 4.1b) and strain was assumed to represent porosity loss. Bierwirth (1982) showed several photomicrographs of the resulting ash mixture. Normal textures associated with progressive welding facies are present (e.g., bubble and Y-shaped shard deformation). Furthermore, deformed pumice have flattening ratios up to ~7; consistent with observations made on densely welded natural samples (e.g., Ragan & Sheridan 1972; Quane & Russell in review). The experiments of Friedman et al (1963) and Bierwirth (1982) used the same material, however, they produced different results. For example, under conditions of atmospheric confining pressure and no H2O pressure a sample deformed by Friedman et al (1963) at 635°C under a load pressure of 3.63 MPa showed complete loss of porosity in -16 hrs. Bierwirth (1982) experimented under a similar load (2.89 MPa) and at a higher temperature (650°C) but produced virtually no porosity loss over the same dwell time. The different results can be attributed to different experimental techniques. Specifically, the sample vessel of Bierwirth (1982) is not sealed, thus, pore pressure is always equal to atmospheric pressure and, upon heating the sample, H2O escapes from the glass, raising its effective viscosity. Conversely, the Friedman et al (1963) vessel was sealed and the glass was able to retain its H2O. However, as the fluid pressure in the system is increased the data from Friedman et al (1963) show slower welding compaction rates. This counterintuitive response results from the effects of pore fluid pressure, which operates against load pressure (Bierwirth 1982). Potentially, higher H2O pressures would cause the glasses to absorb H2O (Friedman et al 1963; Bierwirth 1982) thereby lowering its viscosity (e.g., Sparks et al 1999). This would increase the potential for welding. However, it is unlikely all the water in these experiments was dissolved into the glass, in which case, the remaining H2O pressure 90 must provide pore fluid pressure that negates the effects of the compactional load. This issue needs to be addressed in future studies. When combined, these pioneering studies elucidate the effects of composition, load and confining pressure on the nature and degree of welding in pyroclastic deposits. Furthermore, they provide minimum welding temperatures and a first look at the rates of welding for a variety of different environmental conditions. However, these data are neither sufficiently comprehensive nor coherent enough to fully describe the rheology of pyroclastic mixtures. In addition, the small sample sizes used in these studies do not allow for detailed examination of the microstructural and geometric changes associated with welding compaction. Inspired by these studies, we have designed an apparatus capable of running controlled deformation experiments on large samples at elevated temperatures. Experimental Device The base unit of the Volcanology Deformation Rig (VDR; Fig. 4.2) is a LoadTrac II® loadframe manufactured by Geocomp Corporation. Specifics on the device and its calibration can be found in Chapter 3. The unit performs both constant displacement rate and constant load tests. Displacement is achieved by controlling the position of the bottom piston using an electronic stepper motor with a displacement speed range from 5 x 10"6 to 2.5 x 10"2cm s"'. The displacement is measured using a built-in LVDT displacement transducer having a 7.6 cm travel range and 0.00013 cm resolution. Load is measured using an S-type load cell attached to a fixed crossarm (Fig. 4.2). Samples can be loaded at rates from zero to 1.9 kg/s and the maximum attainable load is 1136 kg with 0.086 kg resolution. The unit is factory-calibrated for apparatus distortion during loading. An internal processor applies calibration factors for displacement (determined using a gauge block) and load 91 Figure 4.2. Schematic representation of the Volcanology Deformation Rig (VDR). Base unit is a Geocomp® Corporation Load Trac II load frame with L V D T displacement transducer (1), and load cell with an 1136 kg limit (2). Basic load frame is modified for high-T experiments by adding thermocouple (3), fiber insulated furnace (4), and temperature controller (5). The sample assembly comprises a steel spacer (a) attached to the load cell and two high temperature ceramic pistons (b) located above and below the sample (c). 92 (determined using a proving ring) thereby converting raw data into corrected output. Experimental output (e.g., measurements) can be collected every 0.01 s throughout the experiment. The VDR accommodates cylindrical samples having maximum diameters and lengths of 4.95 cm and 10 cm, respectively (Fig. 4.2). The experiments in this study are performed on non-jacketed cores of coherent material (e.g., pumice cores) however, the VDR can be used to experiment on unconsolidated materials (e.g., volcanic ash). The piston and sample assembly have been constructed to facilitate high-T experiments (e.g., >500°C; Fig 2) on large samples (e.g., 750 cm3). I have ensured a minimum temperature gradient within the sample during the experiment by wrapping Cotronics Rescor® blanket insulation around the upper piston. My measurements show a symmetric temperature profile that reaches steady-state after a 1 hour dwell time, and shows a total variation of ~12°C (Chapter 3). The VDR is designed to explore a range of experimental parameters that is unique within the rock deformation community (Fig. 4.3a). Many volcanic processes, including welding, are characterized by conditions of low stress (Fig. 4.3b) and conventional triaxial rock presses have poor resolution at such low loads (e.g., Handin et al 1972; Austin 2003; Fig. 4.3a). Furthermore, welding occurs at strain rates substantially slower than those explored during decompression experiments (e.g., Spieler et al 2003; Mangan et al 2003; Fig. 4.3a & b). The VDR allows for isothermal experiments under the conditions of constant displacement rate or constant load. Each experiment generates a nearly continuous high-resolution set of raw measurements including: time (s), load (kg), and displacement (cm) from which I compute stress (MPa) and strain. Stress (a) is calculated from: 93 Strain Rate (£) ( s 1 ) Strain Rate (g) ( s 1 ) Load (kg) Displacement (cm) Figure 4.3. Range of experimental parameter space available to VDR (gray shaded box) compared to: a) stress-strain rate available to: i) conventional triaxial rock press (e.g., Handin et al 1972; Austin et al in review) ii , iii) decompression experiments simulating fragmentation (Spieler et al 2003) or vesiculation and fragmentation (Mangan et al 2003), respectively, b) Average stresses and strain rates attending common volcanic events. Stress for each event is calculated from loads at the base of the deposits. Strain rates are implied by average extrusion rates, flow rates or compaction rates depending on the process. Explosive events (phreatomagmatic and plinian eruptions) occur at a variety of stresses and strain rates which are higher than shown on this schematic, c) & d) Relationships between machine controlled parameters (load and displacement) and rheological variables stress and strain for the VDR. c) Load (kg) vs. stress (MPa) diagram with lines showing corresponding stresses for different experimental core diameters used on the VDR. Maximum load of apparatus is 1136 kg. Thick line represents core diameter used in this study, d) Displacement (cm) vs. strain (%) with lines showing corresponding strain for different core lengths. Maximum allowable displacement of the apparatus is 7.5 cm. Thick line represents approximate core length used in this study. 94 a-load I nr2 (4.1) where r is the radius of the sample core. Although the load cell has an upper limit of 1136 kg, this relationship between core geometry and stress can be exploited to extend the upper and lower ranges of the VDR. In natural systems, welding typically operates under stresses < 5 Mpa (Fig. 4.3b), however, the VDR can achieve stresses as high as -150 MPa by using 1 cm diameter cores (Fig. 4.3c). I principally use 4.5 cm diameter cores because they provide abundant material for post-experiment analysis. The VDR allows a total displacement of 7.6 cm which means that a range of core lengths can be used for relatively high strain experiments (Fig. 4.3d). My experiments use -6 cm length cores. Strain (s) is calculated as: e = Al/lc • (4.2) where Al is the experimental displacement and lc is the original length of the sample. The core length is corrected for pre-deformational shortening which occurs over the heating (30 min) and equilibration dwell time (60 min) that precedes each experiment. At temperatures >Tg the glass can deform under its own weight causing changes in l c and the sample porosity. A linear calibration was used to correct for this pre-experiment deformation by quenching the experiments without deformation and re-measuring the core length and porosity. The corrections are 0, 1.35, 4.5, and 9% for temperatures 535, 550, 600 and 650°C, respectively. This shortening is accommodated exclusively by porosity loss (e.g., no core radius change). After applying this temperature-dependent correction, displacements are converted to strain. 95 Analogue Experimental Material Analogue materials can provide a vehicle for simplifying experimental models. For example, they are used in experiments that simulate lava flow dynamics (e.g., waxes; Fink & Griffiths 1990; Griffiths 2000; Soule & Cashman 2003), caldera collapse processes (e.g., sand rubber bladders and silicone; e.g., Roche et al 2000; Acocella et al 2003; Lavalee et al 2003) or fragmentation of magma (e.g., Hill & Sturtevant 1990; Mader et al 1994; Phillips et al 1995; Mourtada-Bonnefoi & Mader 2003). Here, 1 have used soda lime silica glass beads as an analogue material for exploring the mechanisms and timescales of welding in pyroclastic deposits. The glass beads have several important attributes. First, the individual beads have known physical properties. This allows us to use the density of the glass (2.49 g/cm3) as a known for calculating porosity (<))) in my pre- and post-run products from: t=P*~Pr (4.3) Pm where p m is the matrix (bead) density and pr is the bulk density of the sample. Second, the beads are spherical in shape having a maximum variation in sphericity of ± 10%. Hence, when deformed under known conditions, each individual bead acts as a perfect strain marker. Third, soda lime silica glass has a uniform composition (Table 4.2) and a relatively low glass transition temperature (T g = ~490-550°C). This allows deformation experiments to be run at lower temperatures (e.g., <600°C). I have calculated the viscosity of the soda lime glass beads as a function of temperature (Lakatos 1976; Fig. 4.4a; Table 4.2) from which I calculate the characteristic relaxation timescales (x) for the soda lime silica liquid: r = rilGK (4.4) 96 CO o o u S CO tj f o OH CD CD > 73 C 03 C CD CD X CD 73 CD O s 1 s CD t o CD w 73 co C O 0 3 .2 03 X , <L> 1 - 1 C ^ 5 <+H 03 CD 73 73 03 fD CD X) CD o3 73 O O a, e o U CN X3 03 H DH 03 4 ^ 3 O iff M CD CD PH 5 03 > > " C D 73 O CO 73 O 73 C o3 "o3 C 3 DM O u, O 03 S3 P" 00 o -ON O CN i n CN CN 1 SO Os C N Os C N SO CO d CN i—i CO oo o m sO sO Os f-p Os 00 SO oo r-- t~~ <-* so os r - r-~ os co t~- CN Os OO in C N r-Os i r i ^ - T t m n N c N N r t H t-» oo r-~ r~ so os o o i> os ro r~; CN a h *d « ^ \t co m Os —< co C N C N in co oo ^ ^ oo CN —< d d CN CO Os Os Os m CN I/O o o ° , o < N S U Z W PQ Vi OS Tt oo in CN Os CO CO CO C N C N CN 00 in —i in oo SO i n oo C N CO o oo oo Os SO o o p o o o o o o o o m O i n o i n o i n o m m i n s O s o r ^ r ^ o o o o o s O s o o o o o o in o i n o O O -^1 r-l C N in * d ° l "oi CO o o SO CN o U U o o in oo CO "3" r-- m M 5p| .5 -2 £ "c3 " CD <C c o c co o3 97 16 14 12 10 co CO CL 8 6 C D _o 4 2 0 I I I I I I I ! Soda-Lime Silica Melts j IN J 1 /"Glass"-" "Liquid" y / 1+—550 °C | \<—535 °C 650 °C—• i i i i i 11 i H I i i 4 b 2 C D S -4 -6 -8 -10. 8 9 10 11 12 13 14 15 10000/T(K) 1 day 1 1 • i . - 1 hour 1 1 / _ 1 minute / t l 1 1 A r 1 I "brittle 1 second _ 1 1 deformation" 1 i "viscous / | 1 i i i deformation" X | i 1 i -/ i 1 i \<—550 °C ' i 600 0 C ^ > ' 1 i ' L « — 5 3 5 °C 1 i I I 650 °C—>\ 1 1 i 1 i i 1 i i i 6 7 8 9 10 11 12 13 14 15 10000/T(K) Figure 4.4. a) Viscosity (Pas)-temperature (K) relationship for soda lime silica glass. VTF fit equation (Table 4.2) is calculated from nominal composition using procedure of Lakatos (1976). b) Relaxation timescale (x)-temperature relationship for soda lime silica glass determined using the Maxwell relationship. Vertical dashed lines represent experimental temperatures from this study. Gray horizontal bars show relaxation timescales at specific time intervals used as a guide to select experimental conditions for each temperature. 98 where r\ is the viscosity of the liquid and GO T is the bulk shear modulus of silicate melts (10 l 0 ± 0 5 Pa; Maxwell 1867; Burcato & Dardy 1974; Bansal & Doremus 1986; Dingwell & Webb 1990). The characteristic relaxation timescales of the soda lime glass liquid range from Is to ~ 1 hour over the temperature interval used in this study (Fig. 4.4b). At a given temperature the experimental deformation rate must be slower than the relaxation timescale to support viscous deformation. My experiments were performed on cores of sintered 1 or 2 mm soda lime beads. The cores were fabricated using an 8 cm long stainless steel cylinder that was split lengthwise. The inner surface of the steel cylinder was coated with molycote® and dry flake graphite lubricant to avoid the glass wetting the steel and the cylinder was filled with beads to a height of 6-7 cm. The halves of the cylinder were fastened by stainless steel hose clamps and the beads were capped by a ceramic spacer (4 cm diameter; 2 cm length) to provide a flat upper surface to the core. The sample was held at 600°C for 75 minutes and then cooled to room temperature over 12 hours to avoid thermal cracking of the glass. The core was removed, washed and dried for 24 hours at ~100°C. Resulting cores (Fig. 4.5) have a diameter of 4.5 cm and lengths of ~6 cm. Porosity of the starting material ranges from -30 to 36% (Table 4.3) compared to 36.3% which is the theoretical minimum porosity for random packing of solid spheres (Scott & Kilgour, 1969). Experimental Results My results derive from thirty-two experiments on sintered cores of soda lime silica glass beads performed under a constant displacement rate or constant load constraint (Table 4.3). Displacement rates varied from 2.5 x 10"4 cm/s to 1 x 10"3 cm/s; loads ranged from ~5 to 50 kg (o = 0.03 to -0.35 MPa). Experiments were at constant temperatures of 535, 550, 99 Figure 4.5. Representative starting material used in this experimental study, a) Right circular cylinder comprising sintered 2 mm soda lime silica glass beads. Porosities of starting materials range from -30-36%. Method used to fabricate cores is described in text, b) Thin section photomicrograph of the same core after impregnation with dyed epoxy (E). The individual beads (B) remain essentially undeformed during sintering. Cracks in individual glass beads (see arrows) result from thin section preparation. 100 C/D -a cd u Xl 0 s „ C/3 CD u. cd •4—> T3 O oi) ai o -a <o o O O on tan T J c/i c rme o CJ UH £ O s— ,—s CO D. T3 j c -a CO p 3 cd i-< i . & c X co CO e C CU o o cd ma ispl o -a 4-1 co c TD 5 Oi CA U: C 3 o cd u CO co :r mp 3 co i C CO JS E 'JS eri 4H o X w e o '•3 •o o c 3 o T3 u O of D. mmary s of run Su rtie CO co ft o ble a. ble -o c H C<5 C3\C^ONOONONC>CX3000000000C)OOOOOOOr^OOOOC^C3NOOO^ O O O ^ O O O O ^ O O O O O O O O O O O O O O O O — < ^ - < 0 0 0 0 0 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + U W W W W W W B B W W B W W U W W W W W B W B C q W W W U W W W W ( N i n o o m m N v o o - r t m ^ M n i f l ' O ^ c n h O N h o o o o o o o o o o N ' H n ^ N O ^ l m » l O l n c < ) O N ^ r N ^ ( ) 0 0 \ ' H ^ l r n m ^ • - ^ M t , ^ l ^ l 0 ^ ^ l l A N O O O O O O ' O O ' o o ' o O O O O O O ' O O O ' o m m co vo t-~ m d o o o o o o o o o o c e o ^ ^ c o ^ m a c « N ^ O i n c < ) K i O ( N O O i n c o o o r n r r i r r i r H ( \ | i / i ^ o i n O \ O oo ON ON r~ >0 ^ °i ^ " ^ "I O O O O O O O O O O O O O O O O O O O O O N ' H ( N i M c M « q « ' - ; q - | ' - | ( S N ' o o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o o > n o N r ^ c 3 > r n o o o o N > o ^ ^ o o o o ^ r ^ r ^ w o ^ r n ^ a N ^ ^ c ^ o \ ^ ' - ^ t ^ - H O \ r ^ O O O C N O O O O O O O O O O O O O^c^oorNcOrtcnr^r^r^fNOOVOONVOrHooiritSC^OvlvcncJi ^CNoocNv- ) ina \c>c>o^Ou-><jo0^a \cN^aNCN^OOsoor - - -rtrtO^rHrtrHrtNmrtrtO^^^NcsNNrNrHqqiN o o ' o o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o o ' o ' o ' o o ' o ' o ' o ' i o ^ r ^ o o o o \ o c o ^ r H o o o ^ c n ^ o t n o \ o \ ^ N , ! t O \ c x j i r i ' t o \ c n - H ^ c o ^ - o o » * N N ^ o o o ^ ^ n o ^ i • o ^ ' ^ o ^ ^ s N r ^ f ^ l « o ^ O ' O O o o ^ • ( N l ( ^ v l ^ l c o ^ M ^ c n r o n n c n r n c n r N c n N f n N n o l c o c n r n n n N t n m r n o o o o o o o o ' o o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o o ' o ' o ' o o c > O N O O c ^ O O O O ^ O O > o O i n o > o O ^ O ^ O ^ O N ^ ^ o r ^ c c r o O > n o ' o ' o ' o ' o ' o o ' o o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o ' o o o ' o ' o ' o ' o ' o ' o o O m O O O O w i m O h M O i n O o o h O l O O N C O l O C N O O 0 > n O O O O O O O O O O m O O O O m > o O i n O O O o i c N — ' i / i O O O c o o o o o t - ^ i n m o ^ r - ^ c N m c N i n o o O O u ~ i r ^ ^ ^ f \ O c n oo O O O O O O O O * O O O O O O O O O OU-)'—'W-i^-cfN^OOO r ^ - H - - 1 COf) N M >o >c o ' o ' o ' o ' o ' o ' o ' o ' o ' o o o o o o o o o o o o W W W W W W W W ^ W W i n q q q q q i n q q q q * * * * * © W p vi * * © p W W * o o O O O 0 O 0 0 O O 0 0 O O O 0 O 0 O O O 0 0 O O O » n i / - > > / - > > / - i < / - > O 0 O O O O ^ O O O ^ ^ O O O O O O O O O O O O w o O ^ w n c o c n r o r o r o O p O st J3 n XI cd cd X •3- »/"> >0 t~~ CN CN CN CN C*) © O I I I I I I I vo t-~ r-~ O O O O O O o c d X ) c d . O c d X ) c d c d . n o c d . O I I I I I I I I I I I k I I ' o o o o o o o o o o o o c d X c d c d c d X X c d X c d c d ^ ^ c x j ^ ^ ^ r ^ o o o o c y i — ' © o d © o, o o, - H , <-, o. • 1 1 1 1 1 1 1 1 .J 1 ~J r ^ r ^ r ^ o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o 101 600 and 650°C and used cores comprising 1 mm, 2 mm and 1 and 2 mm beads. To a first order, there were no apparent differences in the rheology that could be attributed to grain size. Constant Displacement Rate Results are presented in Figure 4.6 for constant displacement rate experiments at rates of 5 x 10"4 cm/s, 2.5 x 10"4 cm/s and 1 x 10"3 cm/s and temperatures of 550 and 600°C. For a 6 cm length core these displacement rates translate into strain rates of 4.16 x 10~5 s'1, 8.3 x 10"5 s"1 and 1.6 x 10"4 s"1 respectively. These strain rates allow for 50% strain over timescales of 12,000 to 3000 seconds. In constant displacement rate experiments the bottom platen of the VDR moves upward at a specified rate resulting in a linear accumulation of strain with time. Increasing strain is accompanied by a relatively smooth non-linear accumulation of stress on the sample (Fig. 4.6) and is manifest in the cores by a loss of porosity and by flattening of individual beads perpendicular to the displacement direction (Fig. 4.7). Changing displacement rate has a noticeable effect on the stress-strain relationship (Fig. 4.6a). Faster rates of displacement result in larger stresses at lower amounts of strain. Lowering temperature has a similar effect on the rheology; lower temperatures result in a more rapid accumulation of stress (Fig. 4.6b). Temperature has a . much larger effect on rheology than displacement rate. A reduction in temperature from 600 to 550°C (~10%>) results in a -10 times increase in stress at a strain of 0.2, whereas a 100% increase in displacement rate (5 x 10"4 to 1 x 10"3 cm/s; Fig. 4.6) causes only an -2 times increase in stress. Relatively smooth curvilinear stress-strain relationships are shown by 550 and 600°C experiments up until a stress of -0.04 MPa is reached (Fig. 4.6a & b). A stress of-0.04 102 Strain Strain Figure 4.6. Experimental results from constant displacement rate experiments on cores of glass beads. Data are plotted as stress (MPa) vs strain, a) 600°C isothermal experiments performed at 3 different strain rates (see labels). Experimental output shows smooth non-linear increase in stress until -0.04 MPa where the stress/strain ratio increases markedly. Shaded area is magnified in c). b) Constant displacement rate experiments illustrating the effect of temperature (550, 600, 650°C) on rheological behavior of glass bead cores. The same step in stress-strain curves is seen at -0.04 MPa. Shaded area is enlarged in d). 103 Figure 4.7. Photographs and corresponding thin section photomicrographs for experimental end products observed in this study, a) Representative starting material sample sq-08_04b with an initial porosity of 31%. Starting materials in this study had initial porosities from -30-36%. Virtually undeformed glass beads (B) can be differentiated from colored epoxy (E) in thin section, b) Sample sq-06_25b; 2mm beads, 55.8% strain a strain rate of 8.3 x 10-5 s-1. Final porosity is 20.5%>. c) Sample sq-06_27a; 2mm beads, 63.6% strain at a strain rate of 8.3 x 10-5 s-1. Final porosity of 5.3%). d) Sample sq-08_06b; 1mm beads, 15.5%) strain at a constant stress of 0.035 MPa. Final porosity is 29%. e) Sample sq-08_06a; 1 mm beads, 28.9%> strain at a constant stress of 0.11 MPa. Final porosity is 14.6%. f) Sample sq-07_28a; 1mm beads, 39.7% strain at a constant stress of 0.17 MPa. Final porosity is 15.6%. 104 MPa seems to separate two different regimes. Below -0.04 MPa increasing strain causes a smooth increase in stress. At -0.04 MPa there is a discontinuous jump in stress after which the stress-strain curve steepens markedly. The position of this rheologic discontinuity, in terms of strain, is controlled by both displacement rate and temperature. At the same temperature, increasing displacement rate causes the discontinuity to shift to lower values of total strain (Fig. 4.6a). Again, temperature has a stronger control than displacement rate on this process. At the same displacement rate lowering temperature causes the discontinuity to occur at much lower values of strain (Fig. 4.6a & b). When explored in detail the discontinuity has the same form, however, the lower temperature experiment shows a much higher jump in stress (Fig. 4.6c & d). Constant Load Here, I present data from four experiments performed at constant loads of 5.2, 15.95 and 24.95 kg and temperatures of 550 and 600°C. These loads were selected from different points on the stress-strain relationships illustrated in Figure 4.6b (i-iv). For a 4.5 cm diameter core, these, loads equal stresses of 0.036, 0.11 and 0.17 MPa, respectively and are roughly equivalent to depths of -2, 6 and 10 m depth in a pyroclastic deposit. In a constant load experiment, the bottom platen of the VDR is raised at a specified rate until the present load is reached. From this point onwards, the platen only moves upward in order to maintain the specified load. Strain increases in the sample non-linearly with time (Fig. 4.8). As strain accumulates, pore space is lost and individual beads become progressively more deformed (Fig. 4.7). I have calculated apparent strain rates from the derivatives of the curves fitted to the strain-time data sets (Fig. 4.8a, c). The strain rates are plotted against time in Figure 4.8b & d. The strain rates and the ultimate strain achieved are 105 0 2000 4000 6000 8000 0 0.1 0.2 0.3 Time (s) Strain Figure 4.8. Experimental results from constant stress experiments on cores of glass beads, a) Strain vs time plot for experiments on 1 mm beads at 550°C and at different imposed stresses (i, ii, iii). Stresses derive from points on 550 C rheological curve (Fig. 4.6b). Larger stresses result in greater amounts of strain for a given time, b) Evolution of strain rate with time for experiments shown in (a). Higher stress experiments support higher strain rates but converge to a common value, c) Effect of temperature on constant load deformation paths. Higher temperature experiment (iv; 600°C) deforms much more rapidly than lower temperature (i, 550°C) at the same imposed stress, d) Strain rate vs. strain relationship for constant load experiments. The higher temperature sample (iv) remains at a much higher strain rate than the lower temperature samples (i, ii, iii). 106 dependent on the imposed stress and temperature (Fig. 4.8). Logically, higher loads (e.g., stress) result in higher initial strain rates and more total strain for a given amount of time (Fig. 4.8). As shown by constant displacement rate experiments, the rheology of the material is strongly temperature dependent. Decreasing the temperature from 600 to 550°C (-10%) results in a 10 times reduction in accumulated strain at a time of 750 s (Fig. 4.8c). Conversely, a 300% increase in stress (0.036 MPa to 0.11 MPa) merely triples the accumulated strain (Fig. 4.8a). These relationships have important implications for the efficiency of welding processes in nature. For example, relatively thin deposits of pyroclastic material can achieve significant welding (e.g., Sparks & Wright 1979) if they are emplaced at temperatures much greater than the material's T g. Data Analysis Mechanisms of Strain The above results describe the rheology of welding in an analogue system. To what extent can these results be applied to welding processes in natural systems (e.g., ignimbrites)? To address this issue, I compare how deformation is accommodated by glass beads to a preliminary dataset from experiments on cores of pumiceous rhyodacite (N=6) and sintered cores of rhyolite ash (N=7; Table 4.4). Strain accumulation in the experimental cores is manifest in several ways, including: a) reduction of core length (e.g., shortening), b) change of core geometry (e.g., barreling), c) reduction of primary porosity (e.g., volume loss), and d) deformation of original spherical particles (e.g., flattening). These indicators of strain represent the combined effects of axial strain (sa) accommodated by porosity loss (volume strain) and radial strain (sr), which conserves volume and requires geometric changes (pure shear strain) (Fig. 4.9). 107 Cu <R •a 13 e © 3 & c -a i2 C co cd C •« 8 8 T3-w ' <i) I o o o •o o o + + + + + + + w w w W w w w * * * * * * SO Os so m m 00 oo o so *—< Os CN SO ro ro m m CN ro _ CO Os CO o <—i <—i ' — i ' — i ' — i CN -st- I—i *—i CN in d d d d d © d d d d d d d co o 00 <N •n oo CN CN oo CO ro CN in SO SO in SO d d d d d d d d d d d d d e cu e cu cj Cd el, I CO •3 I i CO c O o <u co 3 CO C 1> E 'C <u X W o 3 T3 o 3 e e <u o -e-3 hi ro co CN CN CN^HOOOOCOr-iOSO-rfCN m m N c N t N ^ c N ^ c N i r i N ( N t s i ( s i ' - < ' - < ^ ' H ' - ' ' - < , — i SO OO in so SO d d d d d in r-^  m oo r- r~ CN CN ^i-d d d so in co r-ro ro ro o d d _, O CN CO O OO c^f" oo^ocNcNin-Hin ^ d> d> d> d> d> d somoor--c?sco^i-r^socNin^-c3s ro^rj-^J-^cNcosor— msosor^ o in O o in m o oo in OS CN so in oo CN CN CN CN — 1 o oo * * * * * CN o o o o so in O xt-oo so OS CN ^ d in •3-o O o p p O o o WWW W W W # W W o O o m o o o o in m in CN p p in in CN CN oi 1 1 wo in oi oi o m in o o o O o o o CO in m in m m m oo oo oo 00 oo 00 00 oo Os cd cd X) cd cd X Os CN CO in m SO CN i CN 1 CN CN CN CN i CN 1 1 co 1 CO oo 00 00 00 00 i 00 1 co CU (—• u o o o o o o O B B o o o o I I CO GO GO Cu CU CU I I I co I 1 108 fe 1 - c / a ~ 0 d - c / a - 0 . 4 • 1 - c / a - 0 . 6 Figure 4.9. Schematic illustration of mechanisms for accommodating strain during experiments on glass bead cores, a) Starting material showing original porosity (())o), sample length (lo) and core radius (r„). Particles are essentially spherical, b) & c) Progressive deformation causes changes in porosity (cj)o reduced to cj)i) length (Al) particle shape (oblateness) and radius (bulging). Physical property measurements on experimental end products are used to determine the proportion of axial (ga) and radial (£r) strain for each sample (see text). 109 I measured the following physical properties for each experimental core: a) initial porosity (<j)0), b) final porosity (<)>i), c) initial radius (r0), d) final radius (n) and e) average oblateness of the individual beads in each core (Fig. 4.9; Table 4.3). Values of § 0 derive from measurements of density (Eq. 4.3) whereas <j)| is based on image analysis of thin sections (Table 4.2). Values of r\ and r 0 represent the average of ~10 caliper measurements taken at regular intervals along the length of the pre and post-experiment cores, respectively. I determined particle oblateness in each end product using: \-cla (4.5) where c is the axial (height) and a the radial (length) dimension of each individual bead. To determine the axial and radial dimensions of individual beads we made an image of each thin section using a standard microfiche reader and hand traced bead outlines (-500 for 2mm bead section and -1000 for 1 mm bead sections). Traces were scanned and individual grains were fit to an ellipse using Scion® image analysis software. The fit provided values of c and a for each particle. The reported values for oblateness (Table 4.3) are the mean of all traced beads in each section. Average values for oblateness plot on the theoretical curvilinear relationship for pure shear strain (i.e., constant volume strain; Fig. 4.10a). This also shows all deformation in these experiments to be purely coaxial. However, because these experiments are not laterally confined, strain has two components; axial strain (sa) accommodated by porosity loss and radial strain (sr) accommodated by bulging of the sample. These components are calculated independently from: (4.6) 110 Figure 4.10. Analysis of strain in high-T deformation experiments of cores of glass beads, a) Comparison of average particle oblateness from each deformation experiment and total strain. Data plot on or near solid line which is the theoretical relationship between particle oblateness and strain for pure shear strain. Dashed line is theoretical relationship for volumetric strain, b) Comparison of axial strain (ea) and radial strain (Er) for experiments involving soda lime glass beads. Heavy dashed lines are iso-strain contours. Dotted line represents 1:1 relationship between 8a and £r. Schematic in upper right corner shows the physical manifestation of the different types of strain accumulation. Experiments are coded for experimental conditions: lower T and high (gray); higher T, low a and (circles with crosses) and intermediate (open circles). Radial strain is more dominant in strain paths exhibited by the lower T and high experiments. I l l where <j)0 is the initial sample porosity and <j)i the porosity of the run product (Table 4.2) and from: 2 e r = \ - \ (4.7) where ro is the radius of the core before experiment and n the mean radius of the run product (Table 4.2). Calculated values of ea and er are plotted in Figure 4.10b. For experiments with moderate T (e.g., 600°C) and deformation rate 5 x 10"4 cm/s there is a 1:1 relationship between sa and sr until -40% total strain is attained (Fig. 4.10b). At this point porosity is reduced to ~10%> and sr becomes dominant. The datum plotted as a cross (Fig. 4.10b) represents a constant displacement rate experiment (600°C) that was stopped at the rheologic discontinuity shown in Figure 4.6. Because this experiment plots essentially on the 1:1 line, it appears that, up to the discontinuity, strain is accommodated equally through porosity loss (ea) and barreling (sr). This point demarcates the inception of er as the dominant method of strain accommodation. The paths for experiments under other conditions are different and appear to be controlled by T and st . For example, in lower T experiments (gray circles; Fig. 4.10b) radial strain (sr) dominates the early portion of the deformation path. Conversely, higher T and lower sj experiments produce a path where sa is dominant (circles with X ; Fig. 4.10b). In natural systems, starting porosities can be higher (-50-75%) e.g., Smith 1960; Ross & Smith 1961; Smith & Bailey 1966; Sheridan & Ragan 1976) and include both matrix porosity and the porosity of the individual pyroclasts. As strain accumulates, clasts deform (flatten) but, in contrast to the glass beads, the deformation of the pyroclasts is not limited 112 by a constant volume constraint (e.g., Sheridan & Ragan 1976; Chapter 2). The consequences of this difference are expressed in two ways. Firstly, values of oblateness from natural samples of the Bandelier Tuff (Quane & Russell in review), after being corrected for original oblateness of the pumice lapilli, deform solely by volumetric strain in contrast to the glass beads, which demonstrate pure shear strain (Fig. 4.1 la). Secondly, in natural materials s a (e.g., volume strain) dominates over sr until very high values (~ 80%) of total strain. At this point, presumably all pore space has been lost and, thus, further strain is dominantly radial (sr; Fig. 4.1 lb). Regardless of these differences, my analysis indicates that, in analogue and especially high porosity natural materials, porosity distributions ultimately control the mechanisms and extent of welding. Strain Dependent Rheology Results from both constant displacement rate and constant load experiments indicate that the glass bead mixtures have a strain dependent rheology. In Figure 4.12a I plot experimental data from 600°C experiments at three different strain rates (x-axis). The results are plotted as stress (observed variable) and strain rate (fixed experimental condition) for equal increments of strain (e.g., 0.05 to 0.5). The ratio of stress to strain rate describes the effective viscosity (r)e) of the cores during deformation. The data in Figure 4.9a clearly show that as strain increases the r\e (slope) changes drastically (from ~107 to ~109 Pas). The •ne at a fixed amount of strain is somewhat dependent on strain rate at low values of strain, whereas at higher strains the rheology is independent of strain rate (e.g., linear trend). The upper limit to this is reached where porosity goes to zero and the glass bead core approaches the viscosity of the soda lime glass (10 9 7 9 Pas at 600°C). 113 Figure 4.11. Analysis of strain in high-T deformation experiments of cores of glass beads, pumice, and volcanic ash. a) Comparison of average particle oblateness for experiments on soda lime glass beads (open circles) and deformed pumice lapilli (black squares) from the Bandelier Tuff (Chapter 2). Natural materials plot on the straight-line relationship representing volumetric strain (<() loss). Natural pyroclasts lose porosity during deformation unlike the solid glass beads used in this study, b) Comparison of axial strain (8a) and radial strain (&-) for experiments involving soda lime glass beads (open circles) and natural pumice and ash (solid circles). £a is dominant in natural materials up to -0.8 total strain. This relationship is due to the high inherent porosity of natural pyroclasts. 114 Figure 4.12. Summary of constant displacement and constant load experiments on cores of glass beads, a) Experimental stress is plotted against imposed strain rate for fixed increments of strain. Solid lines indicate equal strain and map relative differences in effective Newtonian viscosity. The differences in solid lines show the evolution in rheological properties with increasing strain. Dotted lines denote fixed viscosities of 107, 8 Q 10 and 10 Pas. b) Results of constant load experiments are summarized as calculated effective viscosity (r|e) vs. strain for five experiments at two temperatures: i) 550°C, 0.035 MPa (triangles), ii) 550°C, 0.11 MPa (squares) and iii) 550°C, 0.17 MPa (diamonds) are the same as in Figure 4.6. Crosses are data from 600°C experiment at 0.05 MPa and circles from 0.033 Mpa stress. T|e shows an increase with strain that is linear at 600°C and non-linear at 550°C and dependent on the imposed stress. 115 0 5.0x10"5 1.0X104 1.5X104 2.0x104 Strain rate (1/s) 10 i 1 i 1 i 1 r 0 0.1 0.2 0.3 Strain 116 For the constant load experiments I have calculated r\e at each time step using the expression: where £, is the instantaneous strain rate for each time step in the experiment calculated from: s , ^ (4.9) At where As is the incremental increase in strain for each time step and At is the duration of each time step (s). These data are plotted in Figure 4.12b and show how t|e increases with increasing strain. For example, the 600°C experiments show a moderate and steady increase in r| e with strain. Similarly, the 550°C datasets show a strong, rapid increase in r\e with increasing strain. Together, the 550 and 600°C datasets show that r\s and its rate of change are controlled by temperature. The magnitude of stress can also affect the rheological evolution of the samples. For the 550°C experiments, increasing load delays the increase in n e relative to lower load deformation paths. During the deformation experiments, strain is partially accommodated in the cores by porosity loss (e.g., Fig. 4.7), which, in turn, causes changes in the r)e (e.g, Ducamp & Raj 1989; Sura & Panda 1990). The relationship between n,e(Eq. 4.8) and porosity (<j>) is explored by plotting c}» vs r\e (Fig 4.13a). Experimental data are grouped by temperature and fitted to: (4.8) log,0 7, =l°g 1 077 0 - « (4.10) 117 Figure 4.13. Summary of porosity-viscosity relationship for experimental results on soda lime beads, a) Porosity vs log T| data for each experimental temperature 650°C (squares), 600°C (circles), 550°C (crosses) and 535°C (triangles) show the increase in tie as porosity decreases. Best-fit curves for each set of experiments have a Y-intercept value that defines the viscosity of soda lime liquid (T|0) at the experimental T. Inset shows agreement between values of T | 0 calculated from Lakatos (1976; Table 4.2) and the values of T|o derived from fitting these experimental data, b) The viscosity of the mixture T) E is normalized by T ) 0 and plotted against §. The data have an exponential relationship described by the equation in upper right hand corner. 118 119 where a is an adjustable parameter, r\e is the observed value of viscosity at <j>i (Table 4.3) and t|o is the viscosity at zero porosity. Ideally, by fitting Equation 4.10 to the data I obtain an estimate of the T-dependent viscosity (r)o) of the soda lime silica melt. Indeed, there is good agreement between the expected values of t|o (e.g., Lakatos 1976; Table 4.2) and the values derived from fitting data from these deformation experiments (Fig. 4.13a). My fit to the experimental data (Fig. 4.13b) returns a value for a of 5.3 and allows for the prediction of r|e as a function of porosity and melt viscosity (no): A similar relationship using a values between 2.4 and 4.0 was found for experiments run on dry glass powder (Ducamp & Raj 1989). This relationship is a first step in developing a constitutive relationship for welding in pyroclastic deposits comprising the parameters rj, 8, si and <j). Implications for Welding Within a sheet of pyroclastic material, stress increases linearly with depth (Fig. 4.14a). Furthermore, each position in a pyroclastic deposit has a characteristic residence time in the welding interval (e.g., T>Tg) that can be predicted by simple conductive cooling models (e.g., Riehle 1973; Miller, 1990). The top and base cool quickly (e.g., short residence times) whereas the interior of the deposit is insulated and has substantially longer residence times (e.g., Riehle 1973; Miller 1990; Fig. 4.14b). I represent these times for each sample using Dt* which is calculated by normalizing the time a specific sample has within the welding interval to the maximum time seen by any sample. The data from constant load deformation experiments run at 535°C (Fig. 4.14c) are used to represent samples at 2.2, 10 (4.11) 120 I ' I 1 I 1 I 1 I i—r -o <=> £ « CM P*-m O £ ID £ O (LU) ifldea (ui) indea 3 O 4= c 60 C O S © •S ,P CO 03 " CU c^  o o i -a w o c _cu cu *u _ . cu • o, a E x cn cu cu X .t. + J cu O . T 3 CU o > 03 C OH — o C N C 03 E o o C N c o co co cd cn 3 <-> s O 03 cd ^ s s 03 w O cu <-> T 3 H2 C 03 <U T3 cu £ CU E -S" h CO ft* o CN CN CD cd _c co ° E ^ 3 •S E <u 'x C 03 •<= s cu <u x -S cu cj E «) 'C .5 o-o X g cu ° c x o -5 CU T 3 >- E cd 03 3 •a T3 CU H-» 03 o _o (U •— o3 CU 5 S - E 03 o .22 C U <D T 3 CJ c 3 ,3 (SO ^ !HP 03 co co • — o3 2 £ - H CU a- -o * DO bb « c £ c o CU • — •a J5 o 03 <U CU T 3 . VJ TD *-03 « CJ CU T 3 CU '3 a. cu T3 u C 03 X -*-» H—* 03 3 * W 03 O 2 '5 r-X _ — <+H c +-> cT^ O O C y 03 T 3 -—' " CD co Ji 3 E  o cj -3 « Cu CU co T3 T 3 <D *S N JS x 3 co H-3 .3 X (U C+H O o es . .2 cj — u • - o3 U °-cu u < 03 P c E <u (U 03 CU —• u O 3 O •HP £, uu o. cd _cd E <U o Pi co X) cd 2 J CD OH E CD E 3 3 C •§ E 8 -o •o c o H—» X cd O OH X c o >s 03 a - E E © cd ,Z X T 3 O X o -CS w ttJ o ^ T3 r-,—s 03 g CD CJ /-s °-c -E CO O cu CU T3 CU -a c 3 CU H- ^ "5 'co x <u ±3 X X> S cE 03 ^ co w >. 03 g o3 c/] +J cj H 5 ^ 3 E u c tH VH C o ,o - 5 CU T 3 CD _3 +3 > >s X 121 and 20 m depth in a 20 m thick pyroclastic deposit (Fig. 4.14a). The experimental data are plotted in Figure 4.14c; each sample describes a unique deformation path (e.g., strain accumulation with time). Conceptually, each sample is arrested along its potential deformation path when it passes out of its welding window (t> Dt*; Fig. 4.14c) and is "quenched" by reaching the T g isotherm. The resulting strain in each sample (Fig. 4.14d) is a product of both the load (stress) and time of residence within the welding window. The deformation paths can be very different for each sample. For example, samples associated with large Dt* values (e.g., middle of deposit) can undergo significant strain under lower loads (Fig. 4.14d). Conversely, samples with low values of Dt* require high strain rates to attain equivalent strain. For example, the sample at 20 m depth accumulates more strain than the one at 2.2 m depth even though it is quenched earlier because the high load provides higher strain rates (Fig. 4.14d). Not all pyroclastic deposits follow this simple model for cooling and strain accumulation. For example, thin pyroclastic deposits have been observed to undergo significant welding (e.g., Sparks & Wright 1979). Thin packages of pyroclastic material feature low loads and cool relatively quickly (e.g., Riehle 1973; Miller 1990) thereby shortening the residence time in the welding interval. However, my constant load experiments show that, even at low values of imposed stress, if T » T g the resulting si can be surprisingly high. Samples lost >50% of their porosity over a timescale of 1000's of seconds (Fig. 4.8c). Therefore, under conditions of high emplacement temperatures (relative to T g) even thin packages (low a) of pyroclastic material can undergo significant welding. This reflects the fact that where T e » T g the relaxation timescale of the melt is very short 122 (e.g., Dingwell & Webb 1990) so significant amounts of strain can accumulate in a relatively short amount of time (e.g., Fig. 4.8c). Conclusions I have explored the viability of soda lime glass beads as an analogue material for welding deformation in pyroclastic deposits. The predictable viscosity-temperature relationship, relatively low glass transition temperature and regular geometry of the glass bead analogue materials facilitate study of the rheology of glassy porous mixtures. First, I learned that temperature has a much stronger influence on the rheology of these materials than load or strain rate. The implication is that the difference between emplacement temperature and glass transition temperature (T e-Tg) is the most important variable in welding. Second, in nature, volume strain by porosity loss dominates during welding except under extraordinary conditions. These conditions can be met in situations where emplacement temperatures are extremely high or materials are deposited on a significant slope (e.g., Wolff & Wright 1981). Lastly, the rheology of porous mixtures is strain dependent. The effective Viscosity (r)e) of the material increases exponentially with decreasing porosity resulting in non-linear e-t paths. References Acocella, V., Funiciello, R., Marotta, E., Orsi, G., de Vita, S., 2003. The role of extensional structures on experimental calderas and resurgence. J. Volcanol. Geotherm. Res. 129, 199-217. Akelaitis, C , 1999. Characterization of strain during welding of pyroclastic flow deposits: Devine Canyon, USA and Mt. Meager, Canada. B.Sc. Honours Thesis, University of British Columbia, Vancouver, B.C., 31 pp. Alidibirov, M. , Dingwell, D.B., 1996. Magma fragmentation by rapid decompression. Nature 380, 146-148. 123 Austin, N.J., 2003. An experimental investigation of textural controls on the brittle deformation of dolomite. Masters thesis. University of British Columbia 95pp. Bansal, N.P., Doremus, R.H. 1986. Handbook of glass properties. Academic Press, New York. pp. 680. Bierwirth, P.N. 1982. Experimental welding of volcanic ash. Bachelors Thesis, Monash University. Boyd, F.R., 1961. Welded tuffs and flows in the rhyolite plateau of Yellowstone Park, Wyoming. Geol. Soc. Am. Bull., 72, 387-426. Boyd, F.R. and Kennedy, G .C , 1951. Some experiments and calculations relating to the origin of welded tuffs. Trans. Am. Geophys. Un., 32, 327-328. Burcaro, J.A., Dardy, H.D., 1974. High-temperature Brillouin scattering in fused quartz. J. Appl. Phys. 45, 5324-5329. Dingwell, D.B., Webb, S.L. 1990 Relaxation in silicate melts. Eur. J. Mineral. 2, 427-449. Ducamp, V.C., Raj, R. 1989. Shear and densification of glass powder compacts. J. Am. Ceram. Soc, 72, 798-804. Fink, J., Griffiths, R.W., 1990. Radial spreading of viscous gravity currents with solidifying crust. J. Fluid Mech. 221, 485-509. Friedman, I., Long, W., Smith, R.L., 1963. Viscosity and water content of rhyolite glass. J. of Geophys. Res. 68, 6523-6535. Griffiths, R.W., 2000. The dynamics of lava flows. Ann. Rev. Fluid Mech. 32, 477-518. Grunder, A. L.; Laporte, D.; Druitt, T. H. in review Experimental constraints on welding in rhyolitic ignimbrite. Jour. Vole. Geotherm. Res. Guest, J.E., 1967. The sintering of glass and its relationship to welding in ignimbrites. Proc. Geol. Soc. Lon. 1641, 174-177. Handin, J., Friedman, M . , Logan, J.M., Pattison, L.J., Swolfs, H.S., 1972. Experimental folding of rocks under confining pressure; Buckling of single-layer rock beams. In: Flow and Fracture of Rocks, Geophysical Monograph, Am. Geophys.Un., 16, 1-28 Hill, I.G., Sturtevant, B., 1990. An experimental study of evaporation waves in superheated liquid. In: Meier, G.E.A., Thompson, P.A. (Eds.), Adiabatic Waves in Liquid-Vapor Systems. IUTAM Symp. Goettingen, 1989, pp. 25-37. 124 Lakatos, T., 1976. Viscosity-Temperature relations in glass composed of Si02-Al203-Na20-K 20-Li20-CaO-MgO-BaO-PbO-B 203. Glasteknisk tidskrift, 31, 51-54. Lavallee, Y., Stix, J., Kennedy, B., Richer, M. , Longpre, M.-A., 2003. Caldera subsidence in areas of variable topographic relief: Results from analogue modelling. J. Volcanol. Geotherm. Res. 129,219-236. Mader, H.M., Zhang, Y., Phillips, J.C., Sparks, R.S.J., Sturtevant, B., Stolper, E., 1994. Experimental simulations of explosive degassing of magma. Nature 372, 85-88. Mangan, M. , Mastin, L., Sisson, T., 2003. Gas evolution in eruptive conduits: Combining insights from high temperature and pressure decompression experiments with steadystate flow modeling. J. Volcanol. Geotherm. Res. 129, 23-36. Maxwell, J.C. 1867 On the dynamical theory of gases. Phil. Trans. Roy. Soc. A157. 49-88. Miller, T.F., 1990. A numerical model of volatile behavior in nonwelded cooling pyroclastic deposits. J. Geophys. Res. B, 95, 19,349-19,364. Mossing, M . 2003. The role of temperature, stress, strain and porosity on pumice from Mount Meager, Canada. B.Sc. Honours Thesis, University of British Columbia, Vancouver, B.C., 25 pp. Mourtada-Bonnefoi, C . C , Mader, H.M., 2003. Experimental observations of the elect of crystals and pre-existing bubbles on the dynamics and fragmentation of vesiculating flows. J. Volcanol. Geotherm. Res. 129, 83-97. Peterson, D.W., 1979. Significance of the flattening of pumice fragments in ash-flow tuffs. In: C E . Chapin and W.E. Elston (Editors), Ash-flow tuffs. Spec. Pap. Geol. Soc. Am. (GSA), Boulder, CO, United States, pp. 195-204. Phillips, J .C, Lane, S.J., Lejeune, A . -M. , Hilton, M . , 1995. Gum-rosin-acetone system as an analogue to the degassing behaviour of hydrated magmas. Bull. Volcanol. 57, 263-268. Quane, S.L., Russell J.K., in press. A low load high temperature deformation apparatus for volcanological studies. Am. Mineral. Quane, S.L., Russell J.K., in review. Ranking welding intensity in pyroclastic deposits. Bull. Volcanol. Ragan, D.M., Sheridan, M.F., 1972. Compaction of the Bishop Tuff, California. Geol. Soc. Am. Bull., 83, 95-106. Riehle, J.R., 1973. Calculated Compaction Profiles of Rhyolitic Ash-Flow Tuffs. Geol. Soc. Am. Bull. 84, 2193-2216. 125 Riehle, J.R., Miller, T.F. and Bailey, R.A., 1995. Cooling, degassing and compaction of rhyolitic ash flow tuffs; a computational model. Bull. Volcanol., 57, 319-336. Roche, O., Druitt, T.H., Merle, O., 2000. Experimental study of caldera formation. J. Geophys. Res. 105, 395-416. Ross, C.S., and Smith, R.L., 1961. Ash-flow tuffs their origin, geologic relations, and identification. U.S. Geol. Surv. Prof. Pap., 366, 81pp. Scott, G.D., Kilgour, D. M . , 1969 The density of random close packing of spheres. J. Appl. Phys. 2, 863. Sheridan, M.F., Ragan, D.M., 1976. Compaction of Ash Flow Tuffs. In: G.V.C.a.K.H. Wolf (Editor), Developments in Sedimentology. Elsevier, Amsterdam, pp. 677-713. Smith, R.L., 1960. Ash Flows. Geol. Soc. Am. Bull., 71, 795-842. Smith, R.L., Friedman, 1.1., Long, W.D., 1958. Welded tuffs, Expt. 1. Trans. Am.Geophys. Un. 39, 532-533. Smith, R.L., 1960 Zones and zonal variations in welded ash flows. USGS Prof Pap 354-F, 149-159. Smith, R.L., Bailey, R.A., 1966. The Bandelier Tuff; a study of ash-flow eruption cycles from zoned magma chambers. Bull. Volcanol., 29, 83-103. Soule, S.A., Cashman, K.V. , 2003. The mechanical properties of solidified polyethylene glycol 600, an analog for lava crust. J. Volcanol. Geotherm. Res. 129, 139-153. Sparks, R.S.J., Wright, J.V., 1979. Welded air-fall tuffs. In: C E . Chapin and W.E. Elston (Editors), Ash-flow tuffs. Special Paper Geol. Soc. Am. (GSA), Boulder, CO, United States, pp. 155-166. Sparks, R.S.J., Tait, S.R., Yanev, Y., 1999. Dense welding caused by volatile resorption. J. Geol. Soc. Lon. 156, 217-225. Spieler, O., Dingwell, D.B., Alidibirov, M . , 2003. Magma fragmentation. J. Volcanol. Geotherm. Res. 129, 109-123. Streck M T , Grunder A.L. , 1995. Crystallization and welding variations in a widespread ignimbrite sheet; the Rattlesnake Tuff, eastern Oregon, USA. Bull Volcanol, 57, 151-169. Sura, V., Panda, P., 1990. Viscosity of porous glasses. J.Amer. Ceram. Soc. 73, 2697-2701. 126 Taneda, S., 1957. Geological and petrological studies of the "Shirasu" in south Kyushu, Japan; part II, Preliminary note (2). Memoirs of the Faculty of Science, Kyushu University, Series D: Geology, 6(2): 91-105. Wolff JA, Wright JV 1981 Rheomorphism of welded tuffs. Jour Volcanol Geotherm Res, 10, 13-34. Yagi, K., 1966. Experimental study on pumice and obsidian. Bull. Volcanol., 29, 559-572. 127 Chapter 5 Welding Experiments on Natural Pyroclastic Materials Introduction The purpose of this chapter is to investigate the rheology and deformation mechanisms of natural pyroclastic material at temperatures and loads consistent with welding. To achieve this goal I apply the techniques used and knowledge gained from high-temperature deformation experiments on analogue glass beads (Chapter 4). Here, I present results from 8 constant displacement rate and 6 constant load deformation experiments on fabricated cores of natural ash collected from the Rattlesnake Tuff (e.g., Streck & Grunder 1995, 1997). Cores of natural ash material are different than cores of glass beads in two main ways. Firstly, the individual clasts are irregular in shape (rather than spherical) and vesicular (rather than solid). Second, the natural ash cores are substantially more porous. My results indicate that the rheology of sintered cores of natural ash is similar to cores of analogue glass beads. In both cases, rheology is strain dependent and changes in temperature have a greater effect on rheology than changes in load. However, the high porosities of the natural materials allow for substantially greater amounts of strain due to porosity loss relative to the glass bead cores. Furthermore, deformation is attended by changes in average fabric angle that are directly proportional to porosity loss. These differences translate into a different constitutive relationship between melt viscosity, porosity and effective viscosity of the material being welded than was established for glass beads (Chapter 4; Fig. 4.13) and glass powder compacts (e.g., Venkatachari & Raj 1986; Rahaman et al 1987; Ducamp & Raj 1989; Sura & Panda 1990). 128 Experimental Material Pyroclastic flow deposits comprise predominantly volcanic ash and variable amounts of pumice lapilli, lithic fragments and crystal inclusions (e.g., Sparks et al 1973). Each component has a unique effect on the composite rheology of a pyroclastic flow deposit and the mechanisms of welding deformation. For example, lithic fragments and crystal inclusions behave as rigid lumps in a viscous matrix of glass (Ragan & Sheridan 1972) and pumice lapilli commonly have higher porosity and lower viscosity than the surrounding ash matrix (Sheridan 1970; Sheridan & Ragan 1976). However, the rheology and deformation mechanisms attending welding are likely dominated by the ubiquitous grain-to-grain interaction of ash-sized glassy particles. Therefore, to best elucidate the first-order rheology and mechanisms of strain during welding in ash-dominated ignimbrites, I have collected material from the distal nonwelded base of the Rattlesnake Tuff for use in deformation experiments. The 7.1 Ma Rattlesnake Tuff rhyolite is nearly aphyric (<2 wt% crystals) and of known composition (Table 5.1; Streck & Grunder 1995, 1997; Grunder et al in press). Th facies of the Rattlesnake Tuff is relatively well sorted for a pyroclastic deposit (Fig. 5.1) and testing in immersion oils indicates all material is entirely vitric and undisturbed by post-emplacement alteration. Experiments were performed on cores of sintered Rattlesnake Tuff ash. Prior to sintering, the natural material was sieved to isolate the size fraction 0.6 to 2 mm (-1 to 0.73 cj>) (Fig. 1). Sieving to this fraction ensures all material is true ash grade (e.g., Cas & Wright 1987), eliminates interstitial dust and maximizes the amount of ash-sized material that can be used for post-experiment textural analysis. The sieved material was fabricated as cores in an 8 cm long, 4.3 cm diameter stainless steel cylinder split lengthwise. The halves were 129 Table 5.1. Normalized, anhydrous XRF major element composition of Rattlesnake Tuff ash for starting material, sintered material and post-experiment material used in experiments and values of viscosity computed after Shaw (1972) and viscosity of the melt implied by Condition starting sintered post-experiment T(°C) Oxide Wt% Wt% Wt% logn caic logri f i t Si0 2 77.40 77.64 77.24 800 9.95 11.44 Ti0 2 0.16 0.17 0.16 835 9.57 11.30 Al 2 0 3 12.68 12.48 12.82 850 9.22 11.10 Fe 20 3 0.39 0.91 0.90 875 8.88 10.76 FeO 0.90 0.39 0.39 900 8.55 10.70 MnO 0.07 0.07 0.07 MgO 0.04 0.01 0.01 CaO 0.30 0.31 0.30 Na20 3.41 3.38 3.44 K 20 4.62 4.62 4.66 P2O5 0.02 0.01 0.01 H 2 0+ 3.30 0.15 0.24 + determined by infrared spectroscopy 130 Grain diameter (mm) -4 -2 0 2 4 Grain diameter((|>) Figure 5.1. Grain size analysis of unwelded Rattlesnake Tuff starting material. Results show a poorly sorted grain size distribution typical of pyroclastic flow deposits. Grain sizes between 0.6 mm (0.73<j)) and 2 mm (-1<|)) were selected for use as experimental material (gray shaded box). 131 fastened by stainless steel hose clamps and filled with ash to a height of 6-7 cm. The ash was capped with a graphite spacer (4 cm diameter; 2 cm length) in an attempt to limit oxidation of the glass. The sample was held at 900°C for 20 minutes, removed and cooled at room temperature. At these temperatures and times, the ash showed signs of pervasive volatile exsolution (e.g., small bubbles forming on ash shards) during heating and became slightly oxidized, changing from whitish-gray to tan in color. These physical changes are accompanied by chemical changes (Table 5.1). Major element and H 2 O analyses of starting material prior to and after sintering indicate exsolution of H 2 O (e.g., from 3.3 wt% to ~0.2 wt%) and oxidation of the glass (Fe 2 + changes from 0.9 to 0.39; Table 5.1) during sintering. There are, however, no significant changes in geochemistry during deformation experiments, indicating all volatile exsolution and oxidation occurs during the sintering process (Table 5.1). The sintered cores of ash were machined to desired length by cutting both ends with a fine-toothed handsaw using the steel cylinder as a guide to ensure flat ends. Resulting cores (Fig. 5.2) have diameters of 4.3 cm and a lengths of -5.5 cm. Porosity of the starting material ranges from -78 to 80% (Fig. 5.2; Table 5.2). Experimental Results My results derive from fourteen experiments on sintered cores of Rattlesnake Tuff ash performed under constant displacement rate or constant load constraint (Table 5.2) on the Volcanology Deformation Rig (VDR; Chapter 3; Quane & Russell 2004). Displacement rates varied from 1.25 x 10"4 to 5.0 x 10"4 cm/s; loads ranged from -22.5 to 90 kg (a = 0.17 to 0.64 MPa). Experiments used constant temperatures of 800, 835, 850, 875 and 900°C, a heating time of 45 minutes and a dwell time of 1 hour. These conditions result in some pre-experiment shortening and subsequent loss of porosity. To correct for pre-experiment 132 Figure 5.2. Representative starting material used in this experimental study, a) Right circular cylinder comprising sintered Rattlesnake Tuff ash. Porosities of starting materials range from -78-80%. Method used to fabricate cores is described in text, b) Thin section photomicrograph of the same core after impregnation with dyed epoxy. The individual Y-shaped cuspate shards and bubble shaped shards remain essentially undeformed during sintering. 133 CN CM LO LO CD CN CO o o o o o o o o o o o o o o CN CO CJ) CD CN CO CO CO CO o o o o o "-3 cm ^ • 0 5 c O O S C O S O ) f O O ) C O O C D C M CN] 0 ) L O r - - C O ' < - a ) C I ) L D C N I - C O T - C N C O C N C N I C O T -CD CM CO 00 CD T - T - O O 0 ) ( D C O L O C O ( \ | l f ) ^ N COOinCNCN'^CN'r-Cp o o o o o o o o o t t m o (M CO CO O) If) N N CO t - 0) TJ" CO O CN T f T - •<- CN LO o ci d ci d o CO N S (M CO i -LO CD CD LO CD ci ci ci ci ci ci CO r- O CD T CN COCOCOCOCOCOCNCOCNCNCOCNCNCO I CN^rCNT-<NLO C>i(NCNCNCNCNCNCNCNCNCNCNC\icM CD O O CO CO co CD o LO o> CD CO CO CD I CO CO o ^ r~- T J - co CD co N LO N (D CO CO O f O LO CO r - f - ^ i -o o o o o o o o o o o o IT) U) N 00 00 N r~- CN CN ci ci ci CD LO c o h -CO CO CO o d d CM CD CO h- CO 05 00 0) 00 0 r~- r-- r - oo O O O O S O T f f O N O O ' t n O O (M CO O CO ^ C O C O O O O h - O T O O C O C O 0 ) O C N C N L O i - L O N N N C O S N N N N ^ ^ ^ ^ ^ f"": ^ c i c i c i c i c i c i c i c i c i ci ci ci ci ci ci d * n * ( O C D N ' - ' _ N t C O O C M ' - 0 c o ^ T r ^ r T i ' T f C N i c D c o c o L o c o c N r ^ d d c i c i c i c i d c i c i c D c i c i c i c i CO N- CD CN LO CT) O (D N m CD CD s ci ci ci ci ci ci ci OOLOLOOOLOCOLOCOOCDOCO O N t N w oo g L O L O C N T - L O T - C N L O N - L O C O C n C M C n LOLOLOOCMLO'J • ( N 0 ) 0 1 0 ) S S O 0 ) ^ C N ( p i n i N U ) r; q in CD N s in CO CNCNCNCNCNLO^COCO'CNCO^'^ 0 0 0 0 0 0 0 co co N r o CN CO CO CO 1 - CO ci ci ci ci ci ci CD LO o CO CD O) N ^ d iri "3- T o o o o o o T f o LU UJ LU LU LU LU LU o o O LO LO LO LO CN o. o o o C N C N C N C N - < - L O L O L O LU LU o o LO LO CN CM O L O O L O O O O O O O O O O O OCOLOh-LOLOLOLOLOLOLOLOLOO C O O O O O O O O O O O O O O O O O O O O O C O O O C T ) 0 0 0 0 0 3 0 0 0 ) 0 ) 0 ) O O O O O O O LO LO LO LO LO LO LO ro ro co xi co CD CN CO CO '<t CN CN CN CM. CN, CN ro .a ro xi ro LO LO CO CD TT xi ro xi o •Tf LO LO LO CD T - CN CO LO 00 CD r v w v v v i™r r r r i' i ' i O O O O O O C O O O O O O O T - C O T - T - T - T - T - T -O O O O O O O O O O O O O O O to (/) CL CL U) to' CO CL CL CL E E E •a ca CS 3 o U l o a . s u ,<u .—. cu c/5 CD em s s o o CUD 3 's o . 3 OD 3 CO CO a> 00 ca U i cu > CO s .5 'co 2 o -—-I—I <!. a o 3 "S cu cu T3 U i 15 CO a. T3 CO U i •a 3 CO CO cu 3 CJ CJ cu s -a s CO oo CO cu cu 3 T3 s _o ' s cu T3 3 cu E * U i cu cu X cu cu s cu cu "O CD - « CO 2 « 1 CJ S '2 cn C L , jo 3 a. £ z « re e r> O S 3 C cu •-< 3 x e s cu E 2 o <J 3 3 .5 134 deformation, a linear correction factor was developed. The factor was determined by measuring physical property changes in an 850°C experiment (Fig. 5.3; sq-01_16 in Table 5.2) that was quenched prior to the start of experimental deformation. During the heating and dwell time the length of the sample shortened by 2.75% (Fig. 5.3). By assuming pre-experiment shortening is negligible at the lowest experimental temperature (800°C) and fitting a line between the two points, the amount of shortening for temperatures greater than 800°C is determined to be 0.055%) per °C. The resulting correction factors are 0, 1.9, 2.75, 4.1 and 5.5% for temperatures of 800, 835, 850, 875 and 900°C, respectively. This shortening is accommodated entirely by porosity loss (e.g., no core radius change). After applying this temperature-dependent correction, displacements are converted to strain using the equation: e = Al/l* (5.1) where Al is the displacement and 1* is the corrected original length of the sample. Constant Displacement Rate Results are presented in Figure 5.3 for constant displacement rate experiments at rates of 1.25 x 10"4, 2.5 x 10"4 and 5.0 x 10"4cm/s and temperatures of 800, 835, 850 and 875°C. The effective strain rates calculated from experimental displacement rates on a core with a starting length of 5.5 cm are 1.25 x 10"5, 4.5 x 10"5 and 9.0 x 10"5 s"', respectively. These strain rates allow for 50%> strain over the timescales of 5,555 and 40,000 s. During each experiment the bottom platen of the VDR moves upward at the constant, specified rate. Total strain accumulates linearly with time. A relatively smooth, non-linear increase in stress accompanies the increase in strain (Fig. 5.4). The effects of increased strain are manifest in the experimental end products by loss of porosity and alignment and 135 0.06 0.04 h-0.02 h-800 825 850 875 Temperature (°C) 900 Figure 5.3. Experimental determination of temperature-dependent, pre-experiment correction factor for cores of Rattlesnake Tuff ash. Deformation (AL/L) during pre-experiment heating and dwell time was determined at 850°C (solid circle; Table 5.2), see text for details. To determine pre-experiment displacement for other experimental temperatures (open circles) a linear fit was applied assuming negligible pre-experiment deformation at 800°C. 136 0 0.1 0.2 0.3 0.4 0.5 S t r a i n Figure 5.4. Experimental results from constant displacement rate experiments on cores of Rattlesnake Tuff ash. Data are plotted as stress (MPa) vs strain, a) 850°C isothermal experiments performed at 3 different strain rates (see labels). Experimental output shows a relatively smooth non-linear increase in stress, b) Constant displacement rate experiments illustrating the effect of temperature (800, 835, 850 & 875°C) on rheological behavior of Rattlesnake Tuff ash. 137 deformation of individual shards (Fig. 5.5). Both increasing strain rate and lowering temperature increase the stress for a given amount of strain (Fig. 5.4). However, changing experimental temperature has a much greater effect on the stress-time path than does changing strain rate. For example, a decrease in temperature from 875 to 800°C (-10%) results in a 10 times increase in stress for a given strain of 0.2, whereas a 100% increase in strain rate (2.25 x 10"5 to 4.5 x 10"5 s"1) merely doubles the increase in stress. Constant Load In this section, data is presented from four experiments performed at constant loads of 22.5, 45 and 90 kg and temperatures of 850 and 900°C. For a 4.3 cm diameter core these loads equal stresses of-0.17, 0.33 and 0.64 MPa respectively, and are approximately equivalent to depths of 10, 20 and 40 m in a pyroclastic flow deposit. In these constant load experiments the lower platen of the VDR was programmed to increase load at a rate of 0.4 kg/s until the pre-set load is reached. At this point, the platen continues to move upward to maintain the set load as the sample deforms during the experiment. The optimal loading rate was established on the basis of several experiments run under different loading rates. High rates are limited by causing brittle fracturing of the experimental core because effective strain rates exceed the relaxation timescale of the material (e.g., Dingwell & Webb 1990). Low rates of loading are limited operationally because the material deforms substantially before the pre-set load condition is met. In all constant load experiments the instrumental strain recorded by the VDR (ETM) is systematically lower than the strain implied by shortening of the core (STS) (Fig. 5.6; open circles). The deviation of constant load experimental data from a model 1:1 relationship between £TM and STS is shown in Figure 5.6. This situation is problematic for further analysis 138 Figure 5.5. Photographs and corresponding thin section photomicrographs for experimental end products observed in this study, a) Representative starting material with an initial porosity of 79%. b) Sample sq-08_19a, 34% strain at a deformation rate of 2.5 x 10"4cm/s and temperature of 800°C. Final porosity is 73.6%. c) Sample sq-01_13a, 61% strain at a strain rate of 2.5 x 10"4 and a temperature of 850°C. Final porosity is 54.3%. d) Sample sq-0115b, 21% strain at a constant stress of 0.17 MPa and a temperature of 850°C. Final porosity is 75%. e) Sample sq-01_14b, 50%> strain at a constant stress of 0.64 MPa and a temperature of 850°C. Final porosity is 63.4%>. f) Sample sq-01_15c, 70% strain at a constant stress of 0.33 MPa and a temperature of 900°C. Final porosity is 43.4%. 139 140 141 Figure 5.6. Determination of correction factor for constant load experimental data. Strain determined from machine displacement (ETM) VS strain determined from sample shortening (&rs). Constant displacement rate experiment results (triangles) plot close to model 1:1 correlation line (dashed). Uncorrected constant load experiment results (open circles) fall off model line and have a relationship with slope 0.73 (solid line). The slope is used as a correction factor for constant load experiments. Corrected data (solid circles) plot near model 1:1 line. 142 of constant load experimental data. However, the discrepancy between 6 T M and S T M is systematic; the data deviate linearly from the model relationship with a slope of 0.73, indicating 8TM = 0.73eis (Fig. 5.6). Constant displacement rate experiment data fall on or near the model 1:1 relationship between E T M and 8TS (Fig. 5.6; triangles) This suggests the discrepancy between 8 T M and STS is limited to constant load experiments and in these experiments the VDR is recording less strain than is actually accumulating. Accepting the values of ETS; £ T M is recalculated as: e * =£™_ (5.2) ! M 0.73 where the correction factor of 0.73 is the slope of the line fitted to the constant load experiment data in Fig. 5.6. The corrected constant load experiment data are shown in Fig. 5.6 (solid circles). The correction allows for systematic comparison of constant load and constant displacement rate (triangles) experiments. Al l results discussed below are corrected. During constant load experiments strain increases in the sample non-linearly with time (Fig. 5.7a,c). As strain accumulates, pore space is lost, individual glass shards deform and samples become progressively more foliated (Fig. 5.5d-f). Al l e-t curves were fitted to a polynomial. Strain rates were calculated by taking the derivatives of the fitted curves. The strain rates are plotted against time in Figure 5.7b & d. The calculated strain rates and ultimate strain achieved are dependent on the imposed stress and temperature (Fig. 5.7). Locally, higher loads (e.g., stress) result in higher initial strain rates and more total strain for a specified amount of time (Fig. 5.7). As shown by constant displacement rate experiments, the rheology of the material is strongly temperature dependent. Increasing the temperature 143 Time (s) Strain Figure 5.7. Experimental results from constant stress experiments on cores of sintered Rattlesnake Tuff ash. a) Strain vs time plot for experiments at 850°C and at different imposed stresses (i,ii,iii). Larger stresses result in greater amounts of strain for a given time, b) Evolution of strain rate with time for experiments shown in (a). Higher stress experiments support higher strain rates but converge to a near common value, c) Effect of temperature on constant load deformation paths. Higher temperature experiment (iv) deforms much more rapidly than lower temperature (i) at the same imposed stress, d) Strain rate vs strain relationship for constant load experiments. The higher temperature sample (iv) remains at a much higher strain rate than the lower temperature samples. 144 from 850 to 900°C (-5%) results in a 220% increase in accumulated strain at a time of 5000 s (Fig. 5.7c). Conversely, doubling the stress from 0.17 to 0.33 MPa results in only a 40% increase in strain at 5000 s (Fig. 5.7a). These results are similar to those from experiments on analogue glass bead cores (Chapter 4) and imply that temperature has a much greater effect on welding than changes in load. Physical Properties The accumulation of strain during welding of pyroclastic deposits is manifest by measurable changes in physical properties (e.g., Ragan & Sheridan 1972; Sheridan & Ragan 1976; Streck & Grunder 1995; Chapter 2). Strain accumulation in experimental cores is attended by several physical property changes, including: a) reduction in core length (e.g., shortening; Al) b) change of core geometry (e.g., barreling) c) reduction in porosity and d) progressive alignment of particles (e.g., fabric angle). These indicators of strain represent the combined effects of axial strain (ea) accommodated by porosity loss (volume strain) and radial strain (sr), which conserves volume and requires geometric changes (pure shear strain). Here, physical property measurements on experimental cores of Rattlesnake Tuff ash are used to elucidate the mechanisms of strain accumulation during welding in natural pyroclastic deposits. The following physical properties were measured on each experimental core: a) initial length (10), b) final length (If), c) initial radius (r0), d) final radius (rf), e) pre-experiment mass (m0), f) post-experiment mass (trif), g) matrix density (ps) and h) fabric angle: the average orientation of individual ash shards from the horizontal (angle) (Table 5.2). Initial porosity was calculated using bulk (pt) and solid or matrix density (pm): 145 + (5.3) Pm Bulk density was determined using the following equation: mass P,=—. (5-4) volume where caliper measurements of 10 and r 0 were used to calculate initial core volume. Values of m 0 and mf were measured on a Mettler® 3000 balance. The p m of the Rattlesnake Tuff is 2.387 g/cm3 as measured by helium pycnometry on a finely crushed sample fraction. Values of <j)f were determined in the same manner as ty0 using measurements of If, rf and mf. Fabric angle for each experimental end product was measured using a minimum of 100 individual shards traced from thin section images of each core using the pencil tool in Adobe Illustrator®. Individual traces were fit to an ellipse using Scion® image analysis software. The fit provided an orientation angle (0-90°) for each particle. A vertically oriented shard has an orientation angle of 90° and a horizontally oriented shard 0°. The reported values for fabric angle (Table 5.2) are mean values for each population of ash shards. Analytical uncertainty was determined by replicate analysis of a single sample by 5 different individuals. Average lrj uncertainty is -3%. Axial (€q) vs Radial (sr) Strain Here, I investigate the components of strain attending deformation experiments on natural pyroclastic materials. Measurements on samples of Bandelier Tuff of pumice lapilli oblateness show that welding deformation in natural samples is almost entirely accommodated by porosity loss (e.g., Fig. 2.6; Fig. 4.7a). In previous experimental studies, samples were confined laterally to ensure this relationship (e.g., Friedman et al. 1963; Bierwirth 1982). In my experimental setup, however, cores are unconfined during 146 deformation and have the potential to strain by both porosity loss (ea) and bulging (sr). These components are calculated independently from physical property measurements on pre-and post-experiment cores: l-<(>, and: (5-6) (Table 5.2). Results indicate that nearly all strain in experiments on Rattlesnake Tuff ash is accommodated by porosity loss (e.g., ea > sr)- Values of ea and sr are roughly equal up to a strain of ~0.2, after which virtually all strain is accommodated through porosity loss (sr = 0; Fig. 5.8). The trend of data in Figure 5.8 implies that, although unconfined, the natural ash cores accumulate strain more efficiently through porosity loss (volume strain) than by bulging (pure shear strain). These data are consistent with pumice oblateness measurements on natural samples (Fig. 2.6; Fig. 4.7) and in direct contrast to the data derived from deformation experiments on cores of glass beads (Chapter 4). The accumulation of strain solely by porosity loss terminates when samples have undergone large amounts of total strain relative to their initial porosity. In experimentally deformed cores of pumice from Mt. Meager (solid circles; Fig. 5.8; Table 5.2; Mossing 2003) the proportion of 8r increases at high values of total strain (e.g., low porosity). The implication is that at this point strain is more efficiently accommodated by sample bulging than porosity loss. This deviates from behavior seen in pumice oblateness measurements from the Bandelier Tuff (Fig. 2.6; Fig. 4.7) and suggests that at high values of strain and low 147 Figure 5.8. Comparison of axial (&.) and radial (e,) strain for experiments involving Rattlesnake Tuff ash (solid black circles), Mt. Meager pumice (solid gray circles) and soda lime glass beads (white circles; Chapter 4). £a is dominant up until -0.8 total strain. At this point because most of the porosity has been driven out of the samples, the material must bulge (i.e., increase e,) to continue deformation. Dashed lines represent iso-strain contours (total strain). Schematics of deformed cores in upper right represent idealized geometric response of experimental end products to different proportions of £aand £<. 148 values of porosity (i.e., significant welding) natural samples must be confined to deform solely by porosity loss. Fabric Angle Non-welded, uncompacted pyroclastic deposits feature a completely random orientation of shards. This translates into a fabric angle of 45° relative to the horizontal (e.g., 0°). As welding proceeds and a eutaxitic texture develops perpendicular to the loading direction (e.g., Fig. 5.5; e.g., Smith 1960; Ragan & Sheridan 1972), fabric angle will decrease. Here, I plot fabric angle against s a to test the degree to which individual ash shards track the welding process. For all samples deformed in this study, fabric angle values systematically decrease as welding intensifies (e.g., ea increases; Fig. 5.9a). Normalizing the fabric angle values makes this plot universal for all pyroclastic deposits and allows us to test whether shard rotation is recording all accumulated strain. Fabric angle is normalized by dividing by 45°, the theoretical orientation of an undeformed sample and ea is normalized by dividing by the original porosity of the pyroclastic material. For experimental cores of Rattlesnake Tuff ash, the starting porosity is -0.8 (Table 5.2). Normalized values for the experimental cores plot on a model line representing a l.T relationship between change in fabric angle orientation and change in ea (Fig. 5.9b). This indicates that in my experimental cores strain from porosity loss is directly communicated into changes in fabric orientation. These results indicate that, in a well-controlled experimental environment, on relatively homogenous material, fabric angle robustly tracks progressive welding. Do these experimental conclusions hold true in more heterogeneous natural systems? To address this question 1 compare my experimental results to those collected on samples of Bandelier Tuff (Quane & Russell in review; Chapter 2). Fabric angles for the Bandelier Tuff samples range 149 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 1 Normalized £ a £ a Normalized £ a Figure 5.9. Analysis of strain in high-temperature experiments on Rattlesnake Tuff ash. a) Measured fabric angle vs calculated axial strain (£a) for experiments on Rattlesnake Tuff ash. Data indicate linear decrease in fabric angle as strain increases, b) Normalized fabric angle plotted vs normalized axial strain. Data plot on or near model line indicating individual ash shards adequately track strain in the experimental samples (see text for details), c) Measured fabric angle vs calculated axial strain (£») for experiments on Rattlesnake Tuff ash (circles) and Bandelier Tuff samples collected in the field (triangles), d) Normalized fabric angle plotted vs normalized axial strain for experimental samples (circles) and Bandelier Tuff field samples (triangles). Data from samples collected in the field fall show the same relationship between fabric angle and axial strain. Error bars indicate 2 a 150 from -20 to -45°. The initial porosity in this section of the Bandelier Tuff is estimated to be 0.51 using non-welded material. The Bandelier Tuff samples show roughly the same systematic decrease in fabric angle with increasing e a(Fig. 5.9c). When normalized, the natural samples overlap the experimental samples and plot close to the mode! line (Fig. 5.9d). Deviation from the model line is likely due to inherent complexities associated with natural materials that can effect the communication between strain accumulation and ash shard orientation (e.g., rigid crystal inclusions, pumice lapilli inclusions, post-emplacement crystallization). The significant overlap of microstructural data from natural and experimental samples suggests my experimental simulations are accurately replicating the natural welding process. Strain Dependent Rheology Results from both constant displacement rate and constant load experiments indicate that the rheology of the Rattlesnake Tuff ash cores is strain dependent. In Figure 5.10a I plot experimental data from 850°C experiments at three different strain rates (x-axis). The results are plotted as stress and strain rate as iso-strain contours (e.g., 0.05 to 0.45). The ratio of stress to strain describes the effective viscosity (r|e) of the material during deformation. The data clearly show r\e increasing as strain increases (Fig. 5.10a) and porosity is lost (e.g., Fig. 5.5). The r\e is dependent on strain rate at low amounts of strain (e.g., nonlinear iso-strain contours) and becomes independent at -0.35 strain (e.g., linear iso-strain contours). Values of n e for these still porous materials fall in the range of 109 to 101 0 Pas. The r\e values will continue to increase until all porosity is lost and approaches the viscosity of the melt (r\0). 151 Figure 5.10. Summary of constant displacement and constant load experiments on sintered cores of Rattlesnake Tuff ash. a) Experimental stress is plotted against imposed strain rate for fixed increments of strain. Solid lines indicate equal strain and map relative differences in effective Newtonian viscosity. The differences in solid lines show the evolution in rheological properties with increasing strain. Dotted lines denote fixed viscosities of 109, 101 0 Pas. b) Results of constant load experiments are summarized as calculated effective viscosity (r|e) vs. strain for four experiments at two temperatures: i) 850°C, 0.64 MPa (circles), ii) 850°C, 0.33 MPa (crosses) and iii) 850°C, 0.17 MPa (triangles) are the same as in Figure 6. Diamonds are data from 900°C experiment at 0.33 MPa. r|e shows a non-linear increase with strain. 152 0 2x10"4 4x1 CT4 6x10"' Strain rate (1/s) 1.0x1011 -i 0 0.2 0.4 0.6 0.8 Strain 153 For the constant load experiments I have calculated r)e at each time step using the expression: r)e=^7 (5.7) where instantaneous strain rates sj for these calculations are determined by taking derivatives of curves fit to strain-time experimental data (e.g., Fig. 5.7). These data are plotted in Figure 5.10b and show the dependence of r\e on strain. Values of n e increase non-linearly with strain and are influenced by temperature; higher values of temperature result in lower values of n e . The effect of load on n e is not as pronounced as the effect of temperature (Fig. 5.10b). Load has essentially no effect on n e except at low values of strain after which values of r)e for experiments with different loads converge (e.g., i,ii,iii; Fig. 5.10b). Most strain in deformation experiments on Rattlesnake Tuff ash is accommodated by porosity loss (e.g., Fig. 5.8) resulting in corresponding changes in the r|e (e.g., Fig. 5.10). Here, I use the empirical relationship between t|e and porosity: log10*7e = l o g 1 0 » 7 o - a — L ~ (5.8) (Chapter 4; Ducamp & Raj 1989; Sura & Panda 1990) to develop a predictive model for the rheology of porous natural ash. Deformation experiments are grouped by temperature and fit to Equation 5.8. Values of T|0 are determined for each experimental temperature using the Y-intercepts of the fitted curves in Figure 5.11a (Table 1). The fit values of r\0 are 1.5 to 2 log units higher than values for r\Q calculated for a nominal composition of Rattlesnake Tuff rhyolite (Streck & Grunder 1997; Table 1) using 154 Figure 5.11. Summary of porosity-viscosity relationship for experimental results on Rattlesnake Tuff ash. a) Porosity vs log r\ data for each experimental temperature 800°C (squares), 835°C (solid circle), 850°C (open circles), 875°C (cross) and 900°C (triangle) show the increase in r)e as porosity decreases. Best-fit curves for each set of experiments have a Y-intercept value that defines the viscosity of the rhyolite liquid (r\0) at the experimental T (Table 1). These fit values of r\0 are plotted against calculated r\0 values using Shaw (1972) using a nominal composition of Rattlesnake Tuff rhyolite. Dotted line indicates model 1:1 relationship between calculated and fit values. Data fall well below model line and indicate experimental fit values for liquid viscosity are considerably higher than calculated values. For comparison, experimentally determined viscosity values of selected high silica rhyolites BL6 (black diamonds), EDF (gray diamonds) and LGB1 (white diamonds); Stevenson et al. 1995) are plotted vs calculated values using Shaw (1972). b) The viscosity of the mixture tie is normalized by t|0 and plotted against (j). The data have an exponential relationship described by the equation in upper right hand corner. 155 1 1 .5 j — 1 1 1 1 1 — i r (|> 156 Shaw (1972) (Fig. 5.1 la). There are several possible explanations for why the fit values for r\0 are higher than predicted. Firstly, the nominal chemical composition used in calculating r| 0 (Table 1) might not accurately represent the composition of the experimental material. Furthermore, volatile exsolution and roasting of the glass observed during pre-experiment higher viscosity experimental product. Secondly, calculated values for viscosity using Shaw (1972) do not always accurately predict experimentally determined viscosities. For example, experimentally measured melt viscosities for three high-silica rhyolites are plotted against values calculated from Shaw (1972) using reported chemical compositions (Stevenson et al. 1995). For one of the samples, the method of Shaw (1972) accurately predicts the measured melt viscosity (black diamonds; Fig. 5.11a; Stevenson et al. 1995). However, for the two other samples (gray diamonds and white diamonds; Fig. 5.11a) calculations from Shaw (1972) are up to 0.75 log units lower than experimentally determined viscosites (Fig. 5.1 la; Stevenson et al. 1995). Lastly, pyroclastic materials (e.g., individual ash shards, crystals and pumice inclusions) annealed into a zero-porosity mass are likely to have different material properties and higher effective viscosities than a homogeneous melt of the same composition (e.g., Bagdassarov et al. 1994). My fit to the experimental data returns a value for a of 0.63 and allows for prediction of r|e as a function of § and r i 0 (Fig. 5.11b) using the equation: The a value of 0.63 is lower than a determined for deformation of soda lime glass bead cores (5.3; Fig. 4.13) and a values determined for isothermal sintering experiments on 157 heating may have induced changes in chemical composition (e.g., alkali loss) resulting in a (5.9) dry glass powder compacts (2.4 to 4.0; Ducamp & Raj 1989). These different a values result in different predicted CD-TJ paths for materials during densification. The different paths and a reference path for a = 1 are compared in Figure 5.12a. Furthermore, the rate of change of r| e with respect to change in <j) is determined by taking the derivative of the curves in Figure 5.12a. Rates derived for each set of experiments plotted in Figure 5.12a are compared in Figure 5.12b. Higher a values (e.g., >5) lead to concave (jj-n paths (Fig. 5.12a). In this scenario, § is lost without significant increase in r\e during the initial stages of deformation (e.g., at high <j>; Fig. 5.12b). However, as deformation continues and <j> loss increases, the concomitant increase in r| e is significantly greater and continues to rapidly increase until all porosity is lost (Fig. 5.12b). As a decreases, the §-r\ paths become progressively less concave (Fig. 5.12a) and the rate of increase in r\ with respect to increase in <j> evolves to have its highest values at progressively higher porosities (Fig. 5.12b). Values of a < 1 (e.g., 0.63) produce convex <j)-r) deformation paths in which the effects of <j) loss on r\e increase are greatest early in the deformation path (e.g., <j) = -0.7) and slowly decrease until all (j) is eliminated (Fig. 5.12). Controls on a There are three fundamental differences between experimental starting materials that likely contribute to the different a values for pressed glass powder compacts (e.g., Rahaman 1987; Ducamp & Raj 1989; Sura & Panda 1990), soda lime glass beads (Chapter 4) and Rattlesnake Tuff ash. These differences are: a) starting porosity and subsequent packing of the starting materials, b) geometry and character of individual glassy clasts and c) ability of individual clasts to rearrange or rotate during deformation (Table 5.3). 158 Figure 5.12. Comparison of porosity-viscosity paths determined from results of deformation experiments and physical property measurements on experimental end products, a) The path of increasing normalized effective viscosity (r\Jr\0) w ' t n decreasing porosity ((j)) for cores soda lime glass beads (dotted), pressed glass powder compacts (gray shaded region; Ducamp & Raj 1989), reference line of a = 1 (dashed) and cores Rattlesnake Tuff ash (solid) comprise significantly different a-factors and shapes and are dependent on material properties, b) Slopes of curves for all data plotted in (a) using same symbols. 159 160 c CD I T3 C cz CO CD co CD ra| ro o ro E £ Z! CO cri iri n co I -CU • =! O ^ CO •5 ® CO -S cj> ,co O O I— O ) O CT> CD CO X J c CO C L CN J D n co CO q: Q. E CO o Q cn T J * C N o <" % % E •£ CD fO _cc - E . 1= a) >J w Q- -"> >. .21 •o o J _ " Q — CO" S o "5 cl co in o "? oo I « O ) 1^-( CO o T J ( 0 CO w iii CO ra _co CD CD ra-j§ E "o c = cu in m o o) ™ CO " S <D CO co D . cc: Representative starting porosities (<j)0) for materials from the three different experimental types are listed in Table 5.3 and photomicrographs of each material are shown in Figure 5.13. Cores of sintered soda lime glass beads are closely packed (Fig. 13a) and have <j)0 values similar to but less than cubic close packing of spheres (e.g., 36.4%; Scott & Kilgour 1969). Interfaces between particles are well established and regular (Fig. 5.13a). Pressed glass powder compacts (Fig. 5.13b) have (j>0 values of ~49-55%> (e.g., Rahaman et al 1987; Ducamp & Raj 1989; Sura & Panda 1990). Interfaces between tightly packed, irregular particles are well established and semi-regular because the material is uniaxially cold-pressed at stresses up to -20 MPa prior to experiment (e.g., Rahaman et al 1987). The materials are closely packed but have high angularity and low sphericity resulting is high minimum porosity values (e.g., Latham et al 2002). Cores of sintered Rattlesnake Tuff ash have (|>o of ~78%> (Table 5.2; Fig. 13c). Contacts between individual particles are relatively small and disperse due to random orientation and low overall volume percent of ash shards (Fig. 5.13c). Representative clast geometries for each experimental starting material are listed in Table 5.3. Individual soda lime glass beads are solid spheres with zero porosity (Fig. 5.13a). Crushed glass powder particles are blocky and semi-equant in shape with zero porosity (Fig. 5.13b). Individual particles of Rattlesnake Tuff ash are highly irregular in shape (e.g., Y-shaped shards, remnant bubble shards) and some particles are porous (e.g., micro-pumice inclusions; Fig. 5.13c). The geometry and packing arrangement of experimental starting materials dictates the amount of contact surface area between particles. The amount of contact surface area determines the ability of individual shards to rearrange and rotate during deformation. 162 Figure 5.13. Photomicrographs of representative starting materials and deformed end products from experiments discussed in Figure 5.12. a) Representative starting material of sintered soda lime glass beads used in deformation experiments (Chapter 4). Sintered cores have a starting porosity of-30%. Beads are solid spheres with consistent geometry and regular packing, b) Unpressed starting material from Rahaman et al. (1987). Shards are solid, semi-equant and blocky. Starting porosity of pressed powder compacts are -49-55%). c) Representative starting material of sintered Rattlesnake Tuff ash. Sintered core has a starting porosity of -80%. Shards are very irregular and occasionally porous, d) Experimental end product from experiment sq-07_07b (Chapter 4). 1 mm-sized glass beads are consistently deformed perpendicular to loading direction (from top). Dark outlines of individual beads are visible. Sample has a porosity of 22.8% (Table 4.3) e) Experimental end product of glass deformed during isothermal sintering experiment (Sura & Panda 1990). Porosity is represented by black areas and was measured at 24%. Glass particles have no preferred orientation and boundaries between grains have largely been destroyed, f) Experimental end product from experiment sq-01_15c (Table 5.2). Individual ash shards are deformed parallel and rotated perpendicular to loading direction (from top). Final porosity of material is 43.4%. 163 164 Cores of glass beads have high contact surface areas and low potential for individual shard rotation (Fig. 5.13a, d; Table 5.3). Particles deform by becoming oblate perpendicular to the loading direction in response to imposed stress (Fig. 5.13d; e.g., Fig. 4.10). Pressed glass powder compacts also have abundant contact surface areas due to pre-experimental cold-pressing (Fig. 5.13b; e.g., Rahaman et al. 1987; Sura & Panda 1990). The close packing of these materials does not permit significant rotation of individual grains (Fig. 5.13e). Conversely, the particles appear to fuse along existing contact interfaces (Fig. 5.13e). Cores of Rattlesnake Tuff ash show significant rearrangement and rotation of individual clasts during deformation (Fig. 5.13c,f). Starting materials have very few anddispersed contact surfaces allowing the materials to freely rearrange and rotate perpendicular to the imposed stress (e.g., Fig. 5.9; Fig. 5.13f). The a factors for different experimental materials appear to be controlled by a combination of the characteristics discussed above (Table 5.3). High a values are present in materials with relatively low <j)0 where contact surface areas between closely packed particles of regular geometry are high (e.g., glass beads). Individual particles in these materials are not permitted to rotate during deformation. As a result, as porosity is lost, the relative increase in viscosity is very high (e.g., high slope values; Fig. 5.12b). As particles become more angular and <j)0 values increase, a factors decrease. Pressed powder compacts (e.g., Rahaman et al 1987; Sura & Panda 1990) have abundant contact surfaces. However, due to the angular nature of the individual particles, the contact distributions and contact surfaces are irregular (e.g., Latham 2002). Furthermore, materials with moderate a values appear to comprise particles that have limited ability to rearrange and rotate. In these materials, the increase in viscosity relative to decrease in porosity is less dramatic and 165 begins at higher porosities than systems with higher a values (e.g., Fig. 5.12b). The lowest a values are present in materials with high (j)0 and very irregular, widely dispersed particles that are free to rotate during deformation. In these systems, viscosity increases most dramatically early in the deformation process (high <j>0; Fig. 5.12b). During these early stages of deformation, the individual particles easily establish new surface contacts. The a value of 0.63 determined here for sintered cores of Rattlesnake ash is likely a minimum for natural pyroclastic flow deposits.' Most likely, the presence of crystals, pumice lapilli and interstitial dust create lower <(>0 and the increase in grain contact areas will increase a values for ignimbrites with all constitutive components (e.g., Sparks et al. 1973). The individual particles, however, remain free to rotate (e.g., Fig. 5.8c,d) thus keeping a values relatively low. Conclusions In this chapter, deformation experiments are used to investigate the rheology and deformation mechanisms of naturalpyroclastic materials under conditions of welding. The rheology of Rattlesnake Tuff cores, like cores of glass beads, is strain dependent and more sensitive to changes in temperature than changes in load. The Rattlesnake ash materials, however, have significantly different mechanisms for accommodating strain. Firstly, though unconfined, the cores of natural ash deform predominantly by porosity loss and, unlike glass bead cores, show limited amounts of radial bulging. Second, individual particles in the cores of Rattlesnake tuff ash are able to freely rotate and their orientation angle is directly proportional to the amount of accumulated axial strain. Both these processes replicate those in naturally deformed ignimbrites. Lastly, experiments performed in this chapter indicate that the different physical properties and mechanisms of deformation in natural materials 166 compared to glass bead cores and pressed powder compacts translate into a different constitutive relationships between melt viscosity, porosity and effective viscosity of the material being welded. This new relationship can be used in future models to predict £-t paths for welding deformation. References Bagdassarov, N.S., Dingwell, D.B., Webb, S.L. 1994. Viscoelasticity of crystal- and bubble-bearing rhyolite melts. Phys. Earth Plan. Int., 83(2): 83-89. Bierwirth, P.N. 1982. Experimental welding of volcanic ash. Bachelors Thesis, Monash University. Cas, R.A.F., Wright, J.V.,1987. Volcanic successions, modern and ancient; a geological approach to processes, products and successions. Allen & Unwin, London, United Kingdom, pp 1-528. Dingwell, D.B., Webb, S.L. 1990 Relaxation in silicate melts. Eur. J. Mineral. 2, 427-449. Ducamp, V.C., Raj, R. 1989. Shear and densification of glass powder compacts. J. Am. Ceram. Soc, 72, 798-804. Friedman, I., Long, W., Smith, R.L., 1963. Viscosity and water content of rhyolite glass. J. of Geophys. Res. 68, 6523-6535. Grunder, A. L.; Laporte, D.; Druitt, T. H. in review Experimental constraints on welding in rhyolitic ignimbrite. Jour. Vole. Geotherm. Res. Latham, J.P., Munjiza, A., Lu, Y. 2002. On the prediction of void porosity and packing of rock particulates. Pow. Tech. 125:10-27. Mossing, M . 2003. The role of temperature, stress, strain and porosity on pumice from Mount Meager, Canada. B.Sc. Honours Thesis, University of British Columbia, Vancouver, B.C., 25 pp. Quane, S.L., Russell J.K., in press. A low load high temperature deformation apparatus for volcanological studies. Am. Mineral. Quane, S.L., Russell J.K., in review. Ranking welding intensity in pyroclastic deposits. Bull. Volcanol. Quane, S.L., Russell J.K., in review. Welding: insights from high-temperature analogue experiments. J. Volcanol. Geotherm. Res. 167 Ragan, D.M., Sheridan, M.F., 1972. Compaction of the Bishop Tuff, California. Geol. Soc. Am. Bull., 83, 95-106. Rahaman, M.N., Dejonghe, L.C., Sherer, G.W., Brook R.J., 1987. Creep and densification during sintering of glass powder compacts. J. Am. Ceram. Soc. 70(10): 766-774. Scott, G.D., Kilgour, D. M . , 1969 The density of random close packing of spheres. J. Appl. Phys. 2, 863. Shaw, H.R., 1972. Viscosities of magmatic silicate liquids: an empirical method of prediction. Amer. J. Sci., 272, 870-893. Sheridan, M.F., 1970. Fumarolic mounds and ridges of the Bishop Tuff, California. Bull. Geol. Soc. Am., 81: 851-868. Sheridan, M.F., Ragan, D.M., 1976. Compaction of Ash Flow Tuffs. In: G.V.C.a.K.H. Wolf (Editor), Developments in Sedimentology. Elsevier, Amsterdam, pp. 677-713. Smith, R.L., 1960. Ash Flows. Geol. Soc. Am. Bull., 71, 795-842 Sparks, R.S.J., Self. S., Walker, G.P.L., 1973. Products of ignimbrite eruptions. Geology. 1(3), 115-118. Stevenson, R.J., Dingwell, D.B., Webb, S.L., Bagdassarov, N.S., 1995. The equivalence of enthalpy and shear stress relaxation in rhyolitic obsidians and quantification of the liquid-glass transition in volcanic processes. Jour. Volcanol. Geotherm. Res. 68: 297-306. Streck, M.J., Grunder, A.L. , 1997. Compositional gradients and gaps in high silica rhyolites of the Rattlesnake Tuff, Oregon. J. Petrol. 38(1), 133-154. Streck, M.J., Grunder, A.L. , 1995. Crystallization and welding variations in a widespread ignimbrite sheet; the Rattlesnake Tuff, eastern Oregon, USA. Bull Volcanol, 57, 151-169. Sura, V . M . , Panda, P.C., 1990. Viscosity of porous glasses. J. Am. Ceram. Soc, 73(9), 2697-2701. Venkatachari, K.R., Raj, R., 1986. Shear deformation and densification of powder compacts. J. Am. Ceram. Soc 69(6): 499-506. 168 Chapter 6 Rheology and Timescales of Welding Introduction The focus of this thesis is to understand and quantify the mechanisms and timescales of welding. The experimental data on the high-temperature deformation of analogue and natural materials produced here (e.g., Chapters 4 & 5) have been used to create a constitutive relationship that relates strain (e.g., deformation) to stress, melt viscosity and porosity as a function of time. Here, I test this relationship against the experimental dataset from Chapter 5 and the few experimental datasets from the literature. After this validation, 1 use the constitutive relationship to explore the implications of my experimental program for the timescales of welding in natural deposits using data from samples of Bandelier Tuff rhyolite (Chapter 2). Predictive Rheological Model for Welding Insights gained from field, laboratory and experimental work in this thesis were used to develop a rheological model for deformation during welding of natural pyroclastic materials (Chapters 4 & 5). This constitutive relationship between strain (e), stress (cr), time (t), porosity (tj)) and melt viscosity (r\0) was developed by Russell & Quane (in review) to predict the accumulation of strain in pyroclastic deposits under different environmental conditions as a function of time: A full derivation of Equation 6.1 can be found in Appendix 3. In this relationship, a is an empirically determined physical constant (Chapter 5). Values of r\0 and <j>0 are universal 169 (6.1) traits of the deposit whereas rj is specific to individual samples (e.g., depth) in a deposit. This relationship predicts the strain-time path for pyroclastic material at elevated temperature under uniaxial loads. Below, this relationship is tested against two sets of experimental data. During deformation experiments on Rattlesnake Tuff ash (Chapter 5) the material was unconfined. Therefore, total strain (ST) was partitioned into the components of axial (ea porosity loss) and radial strain (sr; bulging of the sample): eT =sa+er (6.2). The model (Eq. 6.1), however, predicts strain solely by porosity loss. Therefore, if uncorrected, predicted tj values will be significantly lower than experimental ST- TO account for sr, a correction factor was developed by fitting a polynomial with zero intercept relating the components of er and sa for experiments on Rattlesnake Tuff ash cores (Fig. 5.8): e, =-0.59s2a +0.6lsa (6.3). This relationship was substituted into Eq. 6.2 yielding: eT = -0.59e'a +\.6lea (6.4) where s a represents the strain predicted from Eq. 6.1. Here, experimental data are compared to predicted model results for the Rattlesnake Tuff ash (Chapter 5). Figure 6.1 shows strain-time model results for Rattlesnake Tuff ash under the conditions of 850°C and 0.17, 0.33 and 0.64 MPa stress using an a of 0.63 and § 0 of 79% (e.g., Table 5.2). The value for logi 0r| o at 850°C is 11.4 Pas (Table 5.1). Model results show good agreement with experimental data for all three values of stress. 170 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Time (hrs) Figure 6.1. Model prediction of experimental data. The strain-time data from three 850°C constant load experiments on cores of Rattlesnake Tuff ash are compared to strain-time values predicted values using Equation 6.2. Experiments were performed at 0.17 (triangles), 0.33 (crosses) and 0.64 MPa (circles); these values were used as input parameters in Equation 6.2. 171 The model is further tested against the experimental data from Bierwirth (1982). Bierwirth (1982) performed anhydrous deformation experiments on Bandelier Tuff ash at 800°C using an experimental setup in which samples were confined and starting porosity was -52%. Both these factors will have significant effect on resulting a values (e.g., Chapter 5). Therefore, a new value for a must be determined. Figure 6.2 compares the a values determined for cores of Rattlesnake Tuff ash (Chapter 5), cores of soda lime glass beads (Chapter 4) and unconsolidated Bandelier Tuff ash (Bierwirth 1982). Not surprisingly, due to confinement, the a value (6.7) for data deriving from Bierwirth (1982) experiments is significantly higher than the Rattlesnake Tuff ash (0.63) (Fig. 6.2). Using an a of 6.7, strain-time paths are predicted for Bandelier Tuff ash at 800°C and stresses of 3.62. 2.89 and 2.17 MPa using Eq. 6.1. Due to the similar composition of Bandelier Tuff to Rattlesnake Tuff, the logioTio value of 11.4 Pas determined for the Rattlesnake Tuff in Chapter 5 (Table 5.1) was used in the model. A starting porosity of 52% was used (Bierwirth 1982). Predicted results again show good agreement with experimental data (Fig. 6.3) confirming that Eq. 6.1 is a robust tool for predicting strain accumulation in pyroclastic material. Timescale of Welding A main objective of this thesis is to determine the time required to reproduce the observed strain in welded volcanic deposits (i.e., the timescale of welding). Previous workers (e.g., Riehle 1973; Miller 1990; Riehle et al. 1995) suggest that the timescale of welding is equivalent to the time the pyroclastic material is above its T g. Therefore, the extent of welding is dictated by the cooling rate of the deposit. In this section, I test their hypothesis by predicting the timescale of welding for twenty samples from a single cooling 172 Figure 6.2. Comparison of porosity-viscosity paths determined from results of deformation experiments and physical property measurements on experimental end products. The path of increasing normalized effective viscosity (T|e/r|o) with decreasing porosity ((j)) for cores soda lime glass beads (dotted), cores Rattlesnake Tuff ash (dashed) and charges of Bandelier Tuff ash (solid; Bierwirth 1982) comprise different a-factors and are dependent on material starting properties and experimental methodology. Circles represent data used to determine a for Bandelier Tuff ash (Bierwirth 1982). 173 0.6 0.1 Bierwirth (1982) fit - this study (a = 6.7) 1 • 40000 80000 120000 160000 Time (s) Figure 6.3. Comparison of predicted and experimentally determined strain-time paths for Bandelier Tuff Rhyolite. Strain is predicted as a function of time (s) using Equation 6.2 at 800°C and stresses of 3.6, 2. and 2.2 MPa using an a of 6.37. Predicted strain (solid) shows good agreement with experimental data (dotted; Bierwirth 1982). 174 unit of Bandelier Tuff (Chapter 2) using the validated rheological model developed above (Eq. 6.1). In essence, I compare my timescales of welding predicted by Equation 6.1 to the expected cooling times for each sample using simple conductive cooling models. The timescale of welding for a given sample can be determined by rearranging Equation 6.1 to isolate time: A; exp <'-'•> - exp0"*0 ( l - < 0 -V o (6.5). aa Equation 6.5 has the power to predict the timescale of welding for any sample where the values of a, a, s, <j)0 and t|0are known. To predict the timescale of welding for samples collected from section SCC-1 of Bandelier Tuff (Fig. 6.4) I use the value of a determined empirically for unconfined natural pyroclastic deposits (0.63), a <j>0 value of 0.55 and the r\0 value determined experimentally at 800°C for the Rattlesnake Tuff rhyolite (Chapter 5; Table 5.1). In section SCC-1 of the Bandelier Tuff, each sample has a unique value for a corresponding to its depth in the deposit (Fig. 6.4b). Values of s for each sample are calculated from physical property measurements (e.g., (j>, p; Chapter 2; Fig. 6.4c) assuming <j)0 is 0.55. Results indicate each sample in this section of the Bandelier Tuff under these conditions requires a finite time of at least 6 days to up to 10 days to accumulate strain (Fig. 6.4d). The times required for each sample from section SCC-1 to cool from an emplacement temperature (Te) of 800°C to a T g of 650°C are predicted using a simple conductive cooling model using methods similar to Riehle (1973), Miller (1990) and Riehle et al. (1995). Details of the procedure can be found in Russell & Quane (in review). Predicted cooling times range from near zero values at the top and base of the deposit where 175 +++++ + + ++++ + + + + - 00 CD CN CO CM CD O CM C/) >^  CO ; o E C O ro m • • • • • • • • • • I C E 1 1 o 1 1 1 CO 1 1 CM 1 CD 1 O CN (LU) LjldGQ 03 CD C N ;3 •-1 rt cu T3 3 B- « _ CQ ;3 o •e -a 3 rt H o J> C -a s= 5 ca „ m ° _ / — s cu ° « cn J3 CU cj 3 6 8 C/5 as O o - t; 3 <u s rt 73 O CD s o £p .s on TO 0, cn 03 O rt 3 '3 cn £ 3 a a as — J3 Q O a s -s ^ 3 £ -o CO <u 03o -o C N <0 as o o ^ rt o CU <-> 73 cn O CU £ a. ao S 3 "3 • S cn co O cu .2 > • 3 aS L -rt _G cn CU c CU .3 T3 03 J—< +-« > C/> cn <u S cn T3 cu to S 3 -3 .2 o as cn cu ^ ,o ri rt ^ .3 CN CU T3 3 oS CQ cu J3 O 3 # o o cu cn cn IS _3 cu "a, 6 cO cn x: CJ 03 CU CU rt CU T3 as > -3 3 as 3 cu .£Pl3 (j- GO cn ^ -a o <u J-I—I CJ rt • T3 ^ <U '3 ^ Er W g ao X c « 'cn w 3 <u __ 3 7 3 3 <U H .S 3 03 f— u rt u / - v cu CN -a 176 cooling is fastest due to interaction with the relatively cool air and ground, respectively, to a maximum value of-550 days slightly below the middle of the deposit (Fig. 6.5). The shape of the cooling profile in Figure 6.5 agrees with previously calculated cooling profiles for 20 m thick ignimbrite deposits (e.g., Riehle 1973; Miller 1990; Riehle et al. 1995). Discussion For all samples except those at the very top and bottom of the deposit, the timescales of welding predicted by Eq. 6.5 for this section of the Bandelier Tuff are substantially faster than the calculated timescales of cooling (Fig. 6.6). In some cases, where cooling times are the longest (e.g., middle of deposit), the timescale of cooling is > 1.5 orders of magnitude larger than the timescale for deformation (Fig. 6.6). Therefore, a sample may reach its welding potential (e.g., maximum attainable strain) while it is still at temperatures » Tg. However, despite being at a temperature where it is able to deform, because the rheology of the material is strain dependent (e.g., Chapter 5), each sample reaches a finite amount of strain at which its rate of deformation becomes sufficiently slow that on the timescale of cooling it is effectively no longer deforming. This suggests that, unlike previously thought (e.g., Riehle 1973; Miller 1990; Riehle et al. 1995), cooling of the deposit does not directly control the extent of viscous deformation during welding. For example, in the scenario presented here for the Bandelier Tuff (Fig. 6.6), the welding and cooling history of the samples are only weakly coupled. However, in this model (Eq. 6.5) the predicted welding timescales are very sensitive to r|o (e.g.,Te). Effectively, every log unit change in r| 0 corresponds to a factor of 10 change in timescale. This fact emphasizes the importance of T e relative to T g . The scenario presented above describes a relatively thin (-20 m) deposit in which T e is high relative to T g . 177 0 - F 0 200 400 600 Time (days) Figure 6.5. Calculated cooling times for samples collected from drill core SCC-1 of the Bandelier Tuff (e.g., Fig. 6.4a). Crosses represent time required to cool from an emplacement temperature of 800°C to a glass transition temperature of 650 C using a simple conductive cooling model (e.g., Riehle 1973; Miller 1990; Riehle etal 1995). 178 CO T3 E 8 Q_ Q 1 2 16 20 c o E cc 0 o o+ o + 9> ++ o + o + o + o + o + o + - o + o + o + - o + — o + o - o + o + i i m i l l I 1 10 100 1000 T i m e ( d a y s ) Figure 6.6. Comparison of timescale of welding deformation (open circles 6.4) and conductive cooling (crosses; Fig. 6.5) for a 20 m thick section of Bandelier Tuff ash (Chapter 2). 179 In this case, the pyroclastic material has sufficiently low viscosity and deforms quickly under relatively low stresseses. Therefore, it appears that as T e increases relative to T g for a deposit of any thickness the timescales of welding and cooling become increasingly decoupled. An extreme example of the decoupling between the timescales of welding and cooling is the production of clastogenic lava flows (Vergniolle & Jaupart 1990) in which complete welding of low viscosity material occurs almost instantaneously upon deposition (e.g., very short welding timescale) to form a lava flow that has a significantly longer timescale of cooling. Alternatively, for deposits emplaced at T e «T g the timescale of welding and cooling become increasingly more coupled. In this scenario, due to low values of T e relative to T g, values of r\0 are effectively greater limiting the ability of the material to viscously deform and weld. In this scenario, the thickness of the deposit becomes increasingly important because sufficient stresses (e.g., thick deposits) are needed to drive deformation at high viscosities (e.g., Chapter 5). The deformation timescale of these high viscosity materials may be long enough that samples, especially under low uniaxial stress, are interrupted during deformation because the timescale of cooling is sufficiently fast and the deposit reaches T g. In this case, the timescale of welding and cooling are strongly coupled. Conclusions In this chapter, the rheological model to predict strain accumulation during welding developed by Russell & Quane (in review; Appendix 3) is validated using experimental data from this study (Chapter 5) and Bierwirth (1982). Timescales of welding predicted for 20 samples from a single cooling unit of Bandelier Tuff rhyolite range between 6 and 10 days and, for most samples, the timescale of welding is at least an order of magnitude faster than 180 the timescale for cooling. This suggests that cooling does not necessarily control the extent of welding in pyroclastic deposits, as previously thought, and indicates that, for welding in most volcanic deposits, coupling of the two timescales is directly proportional to the magnitude of T e/T g . References Bierwirth, P.N. 1982. Experimental welding of volcanic ash. Bachelors Thesis, Monash University. Friedman, I., Long, W., Smith, R.L., 1963. Viscosity and water content of rhyolite glass. J. of Geophys. Res. 68, 6523-6535. Miller, T.F., 1990. A numerical model of volatile behavior in nonwelded cooling pyroclastic deposits. J. Geophys. Res. B, 95, 19,349-19,364. Riehle, J.R., 1973. Calculated Compaction Profiles of Rhyolitic Ash-Flow Tuffs. Geol. Soc. Am. Bull. 84, 2193-2216. Riehle, J.R., Miller, T.F., Bailey, R.A., 1995. Cooling, degassing and compaction of rhyolitic ash flow tuffs; a computational model. Bull. Volcanol., 57, 319-336. Russell, J.K., Quane S.L., in review. Rheology of welding: field constraints. Jour. Volcanol. Geotherm. Res. Sparks, R.S.J., Tait, S.R., Yanev, Y., 1999. Dense welding caused by volatile resorption. J. Geol. Soc. Lon. 156, 217-225. Vergniolle, S., & Jaupart, C. 1990. The dynamics of degassing at Kilauea volcano, Hawaii. Journal of Geophysical Research 95(B3): 2793-2809. 181 Chapter 7 Conclusions This thesis combines field-based studies with experimental work to gain a greater insight into welding processes in volcanology. The field work quantifies the distribution of strain within a single cooling unit of welded pyroclastic material using physical property measurements and provides a semi-quantitative ranking index for welding intensity across pyroclastic deposits. The deformation of analogue and natural pyroclastic materials via a new experimental deformation apparatus elucidates the rheology of hot, porous aggregates of glassy material. Together, these studies are used to produce a new constitutive relationship between strain accumulated due to welding and time, stress, porosity and melt viscosity. Outlined below are the key findings of this thesis. The semi-quantitative index scheme for ranking welding intensity developed in Chapter 2 supplants previous schemes to describe the intensity of welding. The indices provide a straightforward terminology that allows for direct comparison between different welded pyroclastic deposits. The index also provides a method to relate macro and microscopic petrographic changes observed in the field to laboratory quantitative physical property measurements. Ultimately, the index serves as a means to quantitatively map the accumulation of strain in pyroclastic deposits. Chapter three details the design, calibration and operation of the Volcanology Deformation Rig (VDR) a high-temperature, low-load deformation apparatus for use in volcanological studies. The apparatus has the capability to run deformation experiments at constant loads and constant rates of displacement up to temperatures of 1100°C on relatively large samples. The high sensitivity of the load cell and displacement transducer make the 182 rig capable of producing high-precision rheological data that can be used to investigate the rheology of hot, porous aggregates of glassy material at temperatures, stresses and strain rates consistent with natural volcanic processes. Experimental deformation of analogue glass beads in Chapter 4 provides insight into the rheology and deformation mechanisms of welding in pyroclastic deposits. Results indicate that the rheology of porous, particulate mixtures is strain dependent and strongly controlled by temperature. It is determined that total strain during deformation of particulate materials has two components: radial (constant volume) and axial (porosity loss) strain. Combined with physical property measurements, the experimental data are used to develop a constitutive model to predict the effective viscosity (ne) of materials at different stages of deformation as a function of porosity ((()) and melt viscosity (r)0) The rheology of natural pyroclastic materials under environmental conditions of welding is developed in Chapter 5. Experimental results indicate that the rheology of these materials is strain dependent and during deformation the most efficient way in which natural materials accumulate strain is by porosity loss. Furthermore, experimental results support the idea that the deformation mechanisms of different particulate mixtures are controlled by clast geometry and packing, original porosity and the ability of individual clasts to rotate during deformation. The main conclusion of Chapter 5 is that that the evolution of r| e during deformation of natural pyroclastic materials, like that of analogue glass beads (Chapter 4), can be approximated using equation 7.1. (7.1). 183 A new rheological model developed by Russell & Quane (in review; Appendix 3) capable of predicting the accumulation of strain during welding as a function of time, stress, viscosity and porosity is tested in Chapter 6. The model, which is based on experimental results from Chapters 4 & 5 and Equation 7.1 accurately predicts strain-time paths for experimental results from Chapter 5 and from Bierwirth (1982). The model is also used to predict the timescale of strain accumulation for 20 samples from a drill core into Unit 4 of the Bandelier Tuff (Chapter 2). Results indicate that the timescale of strain accumulation in these deposits is sufficiently faster than and strongly decoupled from cooling of the deposit. Future Work The model presented here proves to be useful for predicting strain accumulation and timescales of welding in anhydrous pyroclastic materials. However, natural systems comprise numerous variables that may cause the rheology and deformation mechanisms of pyroclastic materials to deviate from those determined in this thesis. The most important of these is likely H 2 0 . (e.g., Friedman et al. 1963; Bierwirth 1982; Sparks et al. 1999). H 2 0 effectively lowers the viscosity of the glass, however, excess H 2 0 can produce pore fluid pressure that acts against the effects of load and inhibits welding. Both of these processes have the potential to profoundly affect the relationship of stress and viscosity during welding. Further variables that must be considered are the effects of pumice lapilli inclusions, crystal inclusions and the interstitial dust on the rheology and mechanisms of welding. Therefore, further experimental work should be aimed at determining the effects these heterogeneities on the welding process in pyroclastic deposits. References Bierwirth, P.N., 1982. Experimental welding of volcanic ash. B.Sc. Honours Thesis, Monash University. 184 Friedman, I., Long, W. and Smith, R.L., 1963. Viscosity and water content of rhyolite glass. Journal of Geophysical Research, 68(24): 6523-6535. Russell, J.K., Quane S.L., in review. Rheology of welding: field constraints. Jour. Volcanol. Geotherm. Res. Sparks, R.S.J., Tait, S.R., Yanev, Y. , 1999. Dense welding caused by volatile resorption. J. Geol. Soc. Lon. 156,217-225. 185 Appendix 1 Physical Property Measurements Sample collection A series of ten holes were drilled through Unit 4 of the Tshirege member of the Bandelier Tuff in New Mexico, USA to assess seismic hazards at Los Alamos National Laboratories. The Bandelier Tuff is a large collection of pyroclastic deposits from large caldera collapsing eruptions of the Valles Caldera associated with the Rio Grande rift. The Bandelier tuff includes, in decreasing age, the Guaje pumice fall, the Otowi pyroclastic flows and the Tshirege pyroclastic flows (e.g., Smith & Bailey 1966; Self et al. 1986; Broxton & Reneau 1995). The holes were located at the corners of proposed building footprints for the Strategic Computing Center (SCC) and the Nonproliferation and International Security Center (NISC) (Fig. A 1.1). Cores were originally examined to determine any offset due to faulting in the immediate vicinity (e.g., Krier et al. 1998). The amount of offset was determined by the contact between Unit 4 and underlying Unit 3 of the Tshirege member pyroclastic flows (Fig. A l .2). This contact was determined both physically (presence of base surges) and chemically (Unit 3 lower in TiCb). No apparent offset was found in the cores. The holes drilled were 3.5 inch core diameter using a hollow stem auger with a split-spoon barrel and wireline retrieval system. Due to the shallowness of the holes, (<100 ft) high core recovery was attained (-96%; Krier et al. 1998). However, because of drilling action coupled with the lack of cohesiveness of the rock material many of the samples were 186 -03 c o r/J <U oo C D. 3 3 X) X J " J 1> 1 .s to «« .<2 § -n * c er o 5 o B . O " » - r , ° -cu X J J3 « to £ 2j e 03 3 O tu c_ 3 3 '+-> o u 3a ^ — C3 " CQ r3 U 5 u o T3 c<_ "3 *S « CD cu .Ss c o '43 co 03 o ° s — C 03 CQ o c w c O .s Ji -H ~3 u XI to ed — >. T3 u C cu '—1 co —; o < d g © 3 Q. 00 CS E E — > C S c3 to Q c 5 .2 S. 03 S C J CO 0 187 Figure A l .2. Stratigraphic nomenclature for the Bandelier Tuff. Unit 4 is a single cooling unit located at the top of the Tshirege member (after Broxton & Reneau 1995). 188 destroyed during handling. The samples were described, archived and stored in the Core Storage Facility at Los Alamos National Laboratories. This study used the core logs produced in the seismic study to reinvestigate the cores for welding variation. Each of the ten cores (SCC1-SCC5 and NISC1-NISC5) were examined. The cores with the best preservation of the welding character were selected: NISC2, SCC1, SCC2 and SCC4. Each core was sampled at the closest possible spacing (-0.75 m interval). A total of 100 samples of material ranging from 1 foot sections of competent core to disked core were collected (Fig. A1.3; Table A l . l ) . Analytical Methods Visual Inspection The density, porosity and macroscopic degree of pumice lapilli flattening all vary in a systematic way within each section suggesting each is a single cooling unit (e.g., Sheridan & Ragan 1976). In addition, each of the cores has a 1-2 ft. thick, crystal rich (-80 vol% quartz and feldspar), fine-grained (-20 vol% ash) layer directly above the underlying Unit 3. One core exhibits laminar and low angle cross beds. Most of the others are poorly indurated sand. This unit is inferred to be a co-ignimbrite surge, a common unit in typical pyroclastic flow deposits (Sparks et al. 1973). Unit 4 is known to have many intermittent base surge deposits throughout its stratigraphy implying many pyroclastic flow pulses (e.g., Broxton & Reneau 1995; Fig. A1.2). Intercalated surges were observed elsewhere around Los Alamos in Unit 4 outcrops. However, in the drill cores there is no evidence for these surges, hence this area can be inferred to be a single flow lobe with intact simple cooling unit stratigraphy. 189 sec 1 SCC 2 SCC 4 NISC 2 21 H t Unit 3 mm? Unit 3 r— 0 L 21 24 Figure A1.3. Stratigraphic sections showing depths and spatial distribution of samples collected from Unit 4 of the Tshirege member of the Bandelier tuff. Sandy surge deposit defines physical and chemical boundary between Unit 4 and underlying Unit 3. 190 Table A1:1. Depths for samples collected from four drill cores in Unit 4 of the Tshirege member of the Bandelier tuff. Sample Depth (m) Sample Depth (m) Sample Depth (m) Sample Depth (m) SCC1-1 0.00 SCC2-1 0.00 SCC4-1 0.00 NISC2-1 0.00 SCC1-2 1.30 SCC2-2 0.46 SCC4-2 0.38 NISC2-2 0.23 SCC1-3 2.36 SCC2-3 1.22 SCC4-3 1.68 NISC2-3 1.37 SCC1-4 3.51 SCC2-4 1.98 SCC4-4 2.59 N1SC2-4 1.75 SCC1-5 3.89 SCC2-5 3.43 SCC4-5 3.51 NISC2-5 2.44 SCC1-6 4.73 SCC2-6 4.42 SCC4-6 3.89 NISC2-6 3.05 SCC1-7 5.64 SCC2-7 5.18 SCC4-7 4.42 NISC2-7 3.43 SCC1-8 6.55 SCC2-8 6.40 SCC4-8 4.95 NISC2-8 3.96 SCC1-9 7.62 SCC2-9 7.70 SCC4-9 5.95 NISC2-9 4.65 SCC1-10 8.23 SCC2-10 9.07 SCC4-10 6.40 NISC2-10 4.99 SCC1-11 8.99 SCC2-11 10.37 SCC4-11 7.09 NISC2-11 5.49 SCC1-12 9.91 SCC2-12 11.81 SCC4-12 8.23 NISC2-12 6.17 SCC1-13 11.81 SCC2-13 12.84 SCC4-13 8.77 NISC2-13 6.63 SCC1-14 12.42 SCC2-14 13.57 SCC4-14 9.45 NISC2-14 7.09 SCC1-15 13.57 SCC2-15 14.33 SCC4-15 10.02 NISC2-15 7.47 SCC1-16 16.92 SCC2-16 14.94 SCC4-16 10.52 NISC2-16 8.08 SCC1-17 17.45 SCC2-17 15.63 SCC4-17 11.39 NISC2-17 8.69 SCC1-18 18.06 SCC2-18 16.54 SCC4-18 11.93 NISC2-18 9.26 SCC1-19 18.75 SCC2-19 18.06 SCC4-19 12.65 NISC2-19 9.60 SCC1-20 19.44 SCC2-20 19.28 SCC4-20 13.26 NISC2-20 10.14 SCC4-21 13.87 NISC2-21 10.90 SCC4-22 14.63 NISC2-22 11.66 SCC4-23 15.09 NISC2-23 12.42 SCC4-24 15.85 NISC2-24 13.11 SCC4-25 16.46 NISC2-25 13.95 SCC4-26 16.92 NISC2-26 14.71 SCC4-27 18.06 NISC2-27 15.40 SCC4-28 18.67 NISC2-28 16.16 NISC2-29 17.38 NISC2-30 19.89 191 Bulk Density (p) Bulk densities for all 100 samples were determined using the hydrostatic weighing technique. The weight of the samples were first measured dry and then weighed submerged in room temperature H2O using a Mettler® 3000 precision balance. In order to render the samples impermeable to H2O the samples were sprayed with a negligible volume and weight Krylon® Crystal Clear aerosol coating. Densities were calculated using the following equation: ' Pr=P/[W1/(Wl-W2)] (A l . l ) where pf is the density of the fluid, W i is the weight of the samples in air and W2 is the weight of the sample immersed in H2O. Details on the procedure can be found in Duncan (1999). Analytical uncertainty for the technique was determined by replicating the same density measurements (10 times) on a pure quartz crystal of known density. In addition, measurement uncertainty was determined by replicating (5 times) the measurement on one of the samples. Results and reported 2rj measurement uncertainty are in Table A l .2. Porosity (</>) Porosity was measured using helium pycnometry, a technique based on the ideal gas law. Samples were weighed and then flooded with helium to determine the skeletal volume (e.g., volume of framework material). Skeletal density was then calculated. Total porosity is calculated using the equation: ^^BLlBl. (A1.2) Ps where pT is the bulk density of the material, and ps is the skeletal density. If all the pores are connected in the sample then </>T is equal to </>conn (connected porosity). In order to 192 o o o o o o o o o o o o o o o p o o o o d d d d d d d d d d d d d d d d d d d d O N N ^ f ^ r N i T - i n N C T J c o c o T - o c o o i i n ^ c N i n d d d d d d d d d d d d d d d d d d d d r o c o c o i n N ^ ^ o o T f ^ t C O L O ^ T - r - C O c q o m o t N i n i n ^ t D T -a ) T - ( D c o i c c o w o ) o o d d a J - ^ o d r ^ ^ c M c o ^ o o c o r o c o c o o o r o n •c CD CL O L_ CL "co o CO n N c o o i i D c O T - c o c o i n CNICDCOLO^J-LD - t -CDCOCO 0) i -CT)T-CN^ l 'C3 )a ) l^ -Tt -^ f r O N r - O i n N T - O N d d c N ^ C N C N l O C N l d o d COCOCOCOCMCMCMCOCMCM CD in o CO o in T -m co C D co C M -<a-d d d t CO c o s c o j CO CM CM CO CM c o ^ i n O) CO 1- 0O f--CD ^ ! CM CM C D C O C M t C O L O O O C O C D O O C O O C M O O C M C O m T - O O ' ^ -O O O O O O O O O O O T - O O O O O T - O C D d d d d d d d d d d d d d d d d d d d d C O M f f l O I N ' - C B O C O ^ f t N C O O O ' - ' - O a i C M C D c N C M T - T - c M c o c o ^ i n i n i n r ^ i n r ^ c n c D ^ c o c o i n d d d d d d d d d d d d d d d T - T - i - d d ( D c D ( D C D ( 0 ( D m i r ) i n i f l i o m u ) * ^ ' * ^ t t f i i n o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o p o o o o d d d d d d d d d d d d d d d d d d d d OTOTOOh-COiniOCOCVlT-OiOJCD^- O T - CO oo oo i n \ f ' f ' t ' < f ' f ^ ^ ^ T f ^ c o n c o c o c O F ) c i c o c o d d d d d d d d d d d d d d d d d d d d CMCMCNCMCNCMCMCOCOCOCOCOCOCOCOCOCOCOCO—, O O O O O O O O O O O O O O O O O O O 0 O O O O O O O O O O O O O O O O O O O d d d d d d d d d d d d d d d d d d d r-- C M CM CO ^ • ( D C O O ' f i n O C N r O N O l ^ O O C O t N O C D o o i O ' - c n n ^ i n c M c o c B i -O C O C O m c O N C D i n C D N O l O ) O T - C N i c o ' c o ^ i r i d N c o c d o ) T - f M S C M L O C O i n ^ f T ^ c M c o d i ^ c b o o a i C M C O ^ f i n C D N O O C J ) c M n ^ m i o s c o o i o T - T - T - T - T - T - T - 1 - C M O O O O O O O O O O O O O O O O O O O O o o o o o o o o o o o o o o o o o o o o 193 CO 10 co CO LO LO 00 00 00 CM "3- LO o o o o o ! °. o o o o o : q | O o o d d d d d o d d d d d o d d d o LO LO CO CN o o LO o CO oo 00 "3- LO : ^ LO CO co co co ! ^ CD d d d d d o d d d d d o d d d Q. IQ CL| E ns ICO CD O o CD o o CD O O CD o o CD o o CD o o LO O o LO o o LO o o LO o o LO o o o o o o o o o o N" o o o o o o LO o o LO o o d d d d d d d d d d d d d d d d d d d d LO CM LO o LO 00 00 LO LO o 00 oo 00 oo CD oo LO oo CM CO CM CO o CO O) CM CM oo LO LO d d d d d d d d d d d d d d d d d d d d CM o o CM o o CM o o CM O O CM o o CM o o CM o o oo o o 00 o o CO o o CO o o CO o o CO o o CO o o CO o o CO o o CO o o CO o o CM o o CO o o d d d d d d d d d d d d d d d d d d d d oo CM oo CM oo 00 CD oo CD oo o "3-oo LO CM LO 00 CD oo CO LO CD CM r-CM 00 00 00 00 LO "3" o o CD CM CM 00 CO CM 00 o o N- o d d T — T~~" CO LO cd d r - T - T j - r ^ c o M - c o ^ c D o o O O C O C O L q c O O T C D L O O C N O T - ' c s i c o ^ t ' s r L o d o o o S o T~ CM CO LO CD 00 a> o CM CO LO CD 00 a CM I CN CM CM CM CM CM CN CM CM CM CM CM CM CM CN CN CM CM CM CM o O o o o a o o o o o o o o o o o o o o O o o o o a o o O o o o o o o o o o O o CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO 194 x t CO 00 x t x t x t CNI CO x t CO CO LO CO CO CM CO CO x t LO x t o ! ° ! ° ! <=> ! °. o o O o o o o o o o o o o o o O c i d o o o O d d d d d d d d d d d d d d d oo T m 00 CD CNJ CD _^ CO CO x t I s - CT) CT) o CN| o CNI CNI O) CO 1 i n ! : ^ : LO LO CD CD : <°. CD co CD CD CD I s - CD CD CD CD X t o ' o o o o d d d d o d d d d d d d d d d d 81 15 o d o d •Nt CN1 o o o o o o o o o o o o O 00 CD i n x t x t d d d o o o o o o o o 0 0 0 0 T - m o x -CO CO x t x f 00 00 CT) T - OO CO CD LO LO 00 O O x — CNI CO CO CNI CO x t 4 LO CD x t x t x t LO i n LO oo I s - oo I s - oo o o CNI x t oo CD CM O 1 o o o o o j o o x — x — CNI x — x — x — o o o O 1 i d d d d d d d d d d d d d d d d d d LO o CO CD CT) CN| o o CO I s - CNJ CD o i n CT) CO 1 CO CO CNJ CO I s - j I s - CD I s - co CT) o 00 00 CO CM O j 1 1 d d d d d d d d d d d d x ~ d d d d d LO LO i n LO LO i n LO LO x t LO x t x t x t x t x t x t x t LO m CD CD CD o O o o o o o o o o o o o o O o o o o o o o o o o o o o o o o o o o o o O o o o o o o o d d d d d d d d d d d d d d d d d d d d d d LO CD CO CO CNJ o CT) CT) I s - CT) CD x t CNI CNI CD I s - CD x t co o 00 x t x t x t x t x t x t CO CO CO CO 00 CO CO oo CO CO co CO x t x t i n x j -d d d d d d d d d d d d d d d d d d d d d d CN| CNI CO CO CO CO CO CO CO CO CO 00 CO CO CO oo CO CO CO CM CNI CM o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o d d d d d d d d d d d d d d d d d d d d d d CNJ CJ) CT) CT) oo CT) CT) CO T — I s - CO CT) CO oo CO CT) m CM 00 CD x t x t x t x t x t i n LO LO CD CD CD I s - I s - I s - I s - CD CO LO x t x t 00 CO CN| CNI CT) 00 LO CD I s - CO C7> LO CD CM CD I s -CNJ x t LO OO m CT) o x t CT) O CO CNI I s - LO x t o LO CO CT) CD CNJ 00 CD o oo x t CT) o cp x t x t i r i CD CO 00 d d d CM CO CO 4 LO LO CD CD 00 oci o x — CNI CO x t m CD I s - CO CT) o CNI CO x t i n CD I s - 00 I s - 00 CT) CNJ CNI CNJ CNI CM CM CM CM CM 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 O o O o o o O o O O o o o o O O o o O o o o O o O o O o O o O O o O O O O O o o O o o o CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO CO 195 •c CU D-in o o m o LO : ° o ! ° o o CO o LO o CO : P CM O CO q ! CO | o CO : q i o ' 1 . 1 CO | o LO o d d d d d o d d d d o d o d o o d d CD CO o o in o CT> CT) 1 in o co CO LO CM CD LO 00 ! CD CM ! 1 0 ! : q ! ^ CM i h» i i . i 1 <° CT) d d d d d d d d d d o d d o o O d d CD CD CD CO CD CD CO CO in LO LO LO LO LO o O O o O O o o o o o o o o o o O o O O o o o o o o o o d d d d d d d d d d d d d d CM CO CM oo 00 r-- CD CD in CO CO o LO LO LO "3-d d d d d d d d d d d d d d CM CM CM CM CM CN CM CM CM CM CO CO ro ro o o O o o o O O o o o o o o o o O o o o O O o q o o q o d d d d d d d d d d d d d d CO CO 00 CO CD CO CO o CO CD CD LO CM CM CM CO CO ro CO CO •<t in o CO LO LO CO CD m CT> CD CO cn o CM co o cn CD O) •3- CD o d d , ~ C\i CO CO CO LO CD CD o „_ CM CO T- CM 00 LO CD oo O) CM CM CM CM CM CM CM CM CM CM CM CM CM CM O O O O o o o u O u o O o O CO CO CO CO CO CO CO CO CO CO CO CO CO CO 2 Z z Z Z z z z z z z z z z o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o d d O O O O O O O O o ci ci ci ci C B C O O N C O ^ C O I O N O c o c o c o c o c o c o c o c o c o ^ r O d d c i d O O O O o CO o d d r~ o LO d d o o o o o o o o o o o o CO CO CO CO CM CM o o o o o o o o o o o o o o o o o o o o o o o o o d d O CD C M CD • e - OO CO LO 00 00 O < XI CO h -© Q-l E ro CO S CO CD CO O <Cf O CD N CO cd oci oi ai i^- o C D CM in T-i - CD CO f O) S d d T— C M co ro O CO 00 O) •<r T- co co LO cb ai in co r» oo cn o T- CM co CM CM CM CM in C D i— co cj) C N CM CM CM CM CM CO CMCMCNCMCNCNCNCMCMCMCMCM z z z z z z z z z z z z z z z z z 196 determine if this is true samples of material were crushed eliminating the possibility of trapped air in pore space inaccessible to the helium, and the same measurement made. Samples with the lowest and highest densities were measured from each core to test for connected porosity throughout the entire range of present porosity (Table A l .3). Results indicate (within uncertainty) that all measured porosity in the samples is connected (Fig. A1.4a). The relationship between porosity and density is approximated by two best-fit lines (Fig. A l .4b) one with a set Y-intercept of the measured matrix density of the material 2.576 g/cm3 (dashed) and one best-fit where the Y-intercept is allowed to find the optimum value (2.578 g/cm3; Fig. A1.4b). The slight discrepancies between the preferred model line (solid; Fig. A l .4b) and the Bandelier tuff datasets (Table A l .2) are non-systematic and have several explanations (see arrows in Fig. A1.4b). In situations where the measured porosity actually represents total porosity, variations in matrix density (e.g., variations in % crystals; e.g., Walker 1972) can cause data to drift above (higher pm) or below (lower pm) the model line. Conversely, if some porosity is isolated then the data will plot to the left of the model line (Fig. A l .4b). Scatter to the right of the model line (higher porosity) is most easily ascribed to higher values of pm . Based on the similarity of the model lines in Figure A l .4b, it is safe to consider all porosity in the Bandelier tuff cores connected, however, caution must be taken when dealing with more densely welded material as pores are more likely to become more isolated as porosity decreases and pathways are closed. Reported porosities (Table A1.2) are calculated using the skeletal density measured for each sample and reported measurement uncertainty (2c) is based on replicate measurement (5 times) of both a relatively low density and relatively high-density sample. 197-Table A1.3. Measurements used to identify connected vs isolated porosity. Sample P(matrix) n 2 P(skelelal) <t>3 ^connected) SCC1-5 2.595 2.602 0.469 0.470 SCC1-16 2.590 2.569 0.304 0.299 SCC2-5 2.568 2.617 0.471 0.479 SCC2-13 2.572 2.652 0.332 0.352 SCC2-14 2.566 2.599 0.311 0.319 SCC4-7 2.573 2.601 0.447 0.452 SCC4-23 2.571 2.584 0.365 0.369 NISC2-6 2.567 2.571 0.469 0.470 NISC2-24 2.566 2.577 0.332 0.335 1 density determined on powder 2density determined using bulk sample 3porosity calculated using 1 "porosity calculated using 2 198 Total porosity 0.25 0.3 0.35 0.4 0.45 0.5 0.55 Porosity Figure A1.4. Plots to investigate the nature of porosity in Bandelier tuff samples, a) Connected porosity (skeletal density) vs Total porosity (matrix density) plot indicates all porosity in Bandelier tuff samples is connected, b) Measured values of porosity compared to model lines for a constant pm of 2.576 (measured; dashed) and 2.578 (solid; fitted). Arrows show effects of variations in pm or isolated porosity (see text). 199 Rock Strength (PLST & UCS) Rock strength was measured using both the point load strength index test (PLST) and uniaxial compressive strength (UCS). Detailed methods for both of these tests can be found in Quane & Russell (2003). The point load strength test is an indirect measure of the tensile strength of a rock specimen. The test applies a vertical concentrated load ( a i ) , which induces tensile stresses that cause failure parallel to the loading direction. However, the stresses are not entirely tensile because of the significant compressive component induced by the loading device. Hence, PLST is an index test rather than a direct measure of either compressive or absolute tensile strength (Broch and Franklin 1972). The point load testing machine consists of a portable loading system (load frame, conical platens, and pump), a system for measuring the load (P) required to break the sample (10,000 lb. load cell with digital readout), and a system for measuring the distance between the platens (D). The loading platens are 60° spherically truncated cones with a radius of 5 mm. The two platens meet tangentially and are made of hardened steel ensuring they remain undeformed during testing (ISRM 1985). Al l PLST measurements are made at ambient temperature and moisture-laden samples should be avoided. In this study all samples were dried at ~100°C to remove adsorbed H2O which could lower values of rock strength (e.g., Ghweir 1995). The sample is inserted between the conical platens and the distance between the platens (D) is measured. The load is increased gradually and systematically so that the sample breaks within 10-60 seconds. For a PLST measurement to be considered valid, the fracture surface must pass through both loading points. Al l measurements in this study were made with the platen direction perpendicular to the fabric orientation. Al l reported data are size corrected to 200 equivalent 50 mm core diameter. Each reported value of PLST is the average of several (-7-10) measurements and measurement uncertainty is reported as (la) standard deviation (Table A 1.2). Uniaxial or unconfined compressive strength (UCS) determines the breaking strength of a material under axial loading (al). It serves as a universal rating for rock strength. For example, several rock strength classifications are based on UCS measurements (e.g., Broch and Franklin 1972; Hoek and Brown 1980) and it is used to predict other rock properties (e.g., triaxial rock strength; Hoek and Brown 1980). The experiment is performed on unconfined material (e.g., o2 = rj3 = 0). Al l experiments are performed on right circular cylinders with a length (L) at least 2 times the width (W). Our samples were dried at 100°C for 24 hours and measurements were performed on specimens having a minimum diameter (D) of 50 mm. Experiments on coarse grained rocks need to use core with D > 10 times the largest grain size (e.g., Brown 1981). The sample core is compressed between two fiat platens of hardened steel that match or exceed the diameter of the sample core. The sample is compressed at a constant rate of 0.5 to 1.0 MPa/s until the load exceeds the compressive strength of the rock causing fracture. The peak load at failure (P) is recorded. Each rock sample was cored and measured multiple times (-5). Reported values and uncertainty (la) are the average and standard deviation, respectively. Uniaxial compressive strength measurements require large amounts of sample material that can be machined to exact specifications. Often pyroclastic rocks are not suited to this preparation. Hence, we determined a conversion from the more efficient PLST to UCS. Calculated values of UCS from PLST measurements are determined using the equation: 201 UCS = 3.86 • PLST2 +5.65 • PLST (A 1.3) (Quane & Russell 2003). Oblateness (OB) During progressive welding of ignimbrites, both the ash matrix and pumice lapilli deform (e.g., Ragan & Sheridan 1972). The pumice lapilli deform to form flattened ellipsoids having equatorial axes a and b and polar axis c. Ragan & Sheridan (1972) demonstrated, from measurements on cube shaped, oriented samples of Bishop Tuff and Aso 4 Tuff, that the equatorial axes (a and b) change equally as the vertical axis (c) shortens. Hence, when measured perpendicular to flattening direction, the height (c) and the length (a) of pumice lapilli fully describe the extent of deformation. Commonly, the ratio of axial length to height (a/c) is used to describe "flattened" lapilli (e.g., fiamme; Sheridan & Ragan 1976; Peterson 1979). However, to better describe deformation of pumice lapilli, we use the mathematical function for oblateness: which assumes that the equatorial axes of the ellipse (a and b) are equal. I measured oblateness for 78 of the samples collected; some samples were too poor in pumice lapilli for oblateness to be measured accurately. Reported values of oblateness are the means of measurements on > 20 pumice lapilli per sample. Axes a and c were measured using a standard metric ruler. Each lapilli had to have a cross sectional area of at least 5mm2. Mean and standard deviation (lo") results are reported in Table A l .2. Fabric Angle (Shard) During progressive welding ash shards align to form a fabric perpendicular to the direction of flattening (e.g., Smith 1960a). The orientations of individual glass shards (A1.4) 202 (relative to horizontal) were determined through thin section analysis. Digital photomicrographs of oriented thin sections were taken for 19 samples from section SCC-1. Individual glass shards are difficult to recognize in the SCC-1 samples due to devitrification. To better recognize particles the thin sections, I used the Nomarski method (e.g., Pearce & Clark 1989), which involves treating the samples with boric acid before analysis to better differentiate the glassy shards from the interstitial material. I traced a minimum of 100 particles from each photomicrograph using the pencil tool in Adobe Illustrator®. Near the edges of mineral grains (e.g., phenocrysts), shards commonly show enhanced alignment or deformation. Consequently, all fabric measurements were made well away (e.g., > 1.5 times the diameter of the mineral grain) from the edges of any crystal. Using the Scion® (NIH) image analysis program, I fit an ellipse to each particle trace to best represent its orientation. The mean of each population is reported as the fabric angle: the angle of deviation from the horizontal. The most intensely welded (e.g., deformed and aligned) samples will have the lowest fabric angle. The average measurement uncertainty (la) is 3% and this was estimated by replicate analysis (e.g., five separate populations of particles) on a single thin section by different individuals. References Broch, E. & Franklin, J.A., (1972). The point-load strength test. Int. J. Rock Mech. Min. Sci., 9: 669-693. Brown, E.T., (1981). "Rock characterization testing & monitoring ISRM suggested methods." Pergamon Press, Oxford, United Kingdom, 221 pp. Broxton, D.E. & Reneau, S.L., (1995). Stratigraphic nomenclature of the Bandelier Tuff for the environmental restoration project at Los Alamos National Laboratory. Los Alamos Nat. Lab Rep., LA-13010-MS: 21 pp. Duncan, R.A. (1999) Physical and chemical zonation in the Emerald Lake pluton, Yukon Territory. Master's Thesis. University of British Columbia. 203 Ghweir, M . , (1995). Point load strength index of Yucca Mountain tuff at elevated temperatures. Master's Thesis, New Mexico Institute of Mining & Technology., Socorro, N M , 78 pp. Hoek, E. & Brown, E.T., (1980): Underground excavations in rock. Inst. Min. & Metali, London, United Kingdom, 527 pp. International Society for Rock Mechanics, C.o.T.M., United States (1985): Suggested method for determining point load strength. Int. J. RockMech. & Min. Sci. & Geomech. Abs., 22: 51-60. Krier, D., Caporuscio, F., Lavine, A. & Gardner, J., (1998): Stratigraphy & geologic structure at the SCC & NISC building sites, technical area 3, Los Alamos National Laboratory, New Mexico, Los Alamos National Laboratory technical report LA-13507-MS. Pearce, T.H., Clark, A .H. (1989): Nomarski interference contrast observations of textural details in volcanic rocks. Geology 17(8): 757-759. Peterson, D.W., (1979): Significance of the flattening of pumice fragments in ash-flow tuffs. In: "Ash-flow tuffs. Special Paper - Geological Society of America". C E . Chapin & W.E. Elston (Editors), Geological Society of America (GSA), Boulder, CO, United States, pp. 195-204. Quane SL, Russell JK (2003): Rock Strength as a Metric of Welding Intensity in Pyroclastic Deposits. Eur Jour MineralT 5: 855-864. Ragan, D.M. & Sheridan, M.F., (1972): Compaction of the Bishop Tuff, California. Geol. Soc. Am. Bull., 83: 95-106. Self, S., Goff, F., Gardner, J.N., Wright, J.V. & Kite, W.M. (1986). Explosive rhyolitic volcanism in the Jemex Mountains: vent locations, caldera development and relation to geological structure. Journal of Geophysical ResearchB2: 1779-1798. Sheridan, M.F. & Ragan, D.M. , (1976). Compaction of ash-flow tuffs. In: "Compaction of coarse-grained sediments, II". G.V. Chilingarian & K.H. Wolf (Editors), Elsevier Sci. Publ. Co., Amsterdam, Netherlands, pp. 677-717. Smith, R.L. & Bailey, R.A., (1966): The Bandelier Tuff; a study of ash-flow eruption cycles from zoned magma chambers. Bull. Vole, 29: 83-103. Sparks, R.S.J., Self, S. and Walker, G.P.L., 1973. Products of Ignimbrite Eruptions. Geology (Boulder), 1(3): 115-118. Walker, G.P.L., 1972. Crystal Concentration in Ignimbrites. Contributions to Mineralogy and Petrology, 36(2): 135-145. 204 Appendix 2 Experimental Data Experimental data collected for this thesis is presented on an attached cd. The data is in a single Microsoft Excel fde entitled data.xls. Each experiment has an individual worksheet within the single fde. Worksheets are labeled using sample numbers. Table A2.1 is an example of the worksheets on the cd. Raw data collected from the VDR for each experiment is in columns labeled Time (s), Load (lbs) and Displacement (in). Area (cm3) is the computed cross sectional area for the experimental cores. Stress (MPa) is calculated at each time step using the values for load corrected to 0 for the start of the experiment and computed areas. L 0 (cm) is the original length of the core and L c o r r is the original length corrected for pre-experiment heating. Strain is calculated using L c o r r . 205 CD CD in 'ro I S 0. (0 w I co co c CD E| CD CD ef' CO X LU CN < JD XI CO O) CD T t LO CO CO O T - 00 LO CO CN _ CD c o CD o t o i n O i -o o o o o o o o o 1- T- LO N CD m r- oo m T- cn co I O CN CN OO CO CN CN CO CO O O O O O ~ ~ o o o cb d d LO i n CD CN CD T -CO CN r-~ T-o CD CD T-T - h -h - T -o CN CD •<t m o o q ° d d t o ^ ¥t LO s s §§ §1 o d CM OO 00 CO r- CM M- •>-CN T- CD o o o o d d cn oo CD CO c o CO CN o CN CO CN h -00 00 o o o o d d c o T m i - O CD O CO 4 O OO CD CN LO CO CO CO CD CD O O O T-o o o o d d CD CD cn o LO h-_ i - o 5; cn 00 _ - 00 CD LO LO CD CO LO O CD 1- CM CM CD CD O T — O d 1- CO LO O O ) t O CO s 1- 00 in N O CM S CN CO CO m o o d d o o d d CO CO CO 00 CO 00 CO CO CO 00 CO 00 CO CO CO CO CO CO CO CD CO CD CO CD CD CD CD CD CD CD CD CD r-- CD CD CD CO CD M- CO CO CO CN CO CN CO CN CM CN CN CN CO r- CO 0 CO 0 0 r-- 1^  N- 1"- T — T — T — m T — m T — LO LO in LO LO CD 00 CD cn CD cn CD 00 00 00 00 00 00 C0 00 CO CO CO CO h- 00 00 c- r-_ r~- CO r~- m h- m in CO CO CO T — CO CO T — T — cn cn CD CD cn r~- N- N-O fSJ CO CN in CN in LO 00 00 T — 00 CO CO T — T — CO CO CO CD CO CD CO CD CD CD CD CD 0 0 0 0 O 0 0 0 0 T — 0 ^— 0 0 T — T — T — T — T — T — T — T — T — T — x— t — •*— T — 0 0 0 0 O 0 0 0 0 O 0 O 0 0 O O O O O 0 O O O O 0 O O 0 0 0 0 0 O 0 0 0 0 O 0 O 0 0 O O O O O 0 O O O O 0 O O 0 d d d d d d d d d d d d d d d d d d d d d d d d d d d d o ^ r-~ in m T — cn N" CM CN CD CN 0 T — CO CD CM CD 0 00 CN LO T — CN LO CD OO CD O O 0 0 O O O O O 0 0 O O O d d d d d d d CD CO CD CD CD CN LOOOCDCOCDCO'3-OOOOOOCOCD'^-CD * C O C N L O c o " ( 0 L O f 0 0 ) N N O * 0 0 r , J N Y w I N « J N A , « , n f. m N W M - ( N K O CN CO O O d d N (N CO CO •5}- CO 1- T J -C O S 0 1 0 C M n ^ C O N f f l O i - C < ) T f T - T - T - C N C N C N C N C N C N C N C O C O C O C O O O O O O O O O O O O O O O 0 0 LO g CO CD u y N _ nr» O O O O O O O O O O O r-- 00 c o c o o o i n CO i n cn LO cn cn CO CO r~- CO CO CO N- I-- ,_ c o ^— CD ,- CD CD CD CD CD CO CN CO CO >- CN CN CD CN CD CN CN CD CD m T — i n i n LO 10 LO i n CD 0 CD cn CD cn cn 0 0 0 0 0 0 0 0 0 CD h - CD r~- CD CD CD CD i n CD i n CD CD LO m CO CO CO CO CO CO CO ob od 00 00 06 06 00 00 CO CO CO 00 06 06 06 CO 00 00 06 00 06 CO cd CO 00 od od od • e - , - m L o r M r n r o r o c ^ i ^ ^ . r - m ^ C N C D C O C O O r ^ ^ t ^ C O i n c \ | C D C O 206 Appendix 3 Derivation of Equation 6.1 Equation 6.1 is derived by Russell & Quane (in review) to predict strain accumulation in pyroclastic materials as a function of viscous deformation using the following methodology. Viscous deformation can be described by: a = tjs (A3.1) where CT is the stress due to lithostatic loading, e is the strain rate and n is the viscosity of the pyroclastic deposit which implies: dsv=-dt (A3.2) 1 The effective viscosity of porous, glassy mixtures during high-temperature deformation can be modeled as: TJ = T]„ exp -a •</> (A3.3) (Rahaman et al. 1987; Ducamp & Raj 1989; Sura & Panda 1990) where r\0 is the viscosity of the pyroclastic material at no porosity and a is an adjustable parameter dependent on porosity, clast geometry and ability to rotate during deformation. The best-fit a value for deformation of porous pyroclastic material is 0.63 (Chapter 5). Substitution of Eg. A3.3 into Eq. A3.2 yields: dsv - — exp lo (A3.4) 207 It was demonstrated in Chapters 2, 4 and 5 that changes in porosity are equivalent to volume strain (sa). Therefore, Eq. A3.4 can be written as: asv exp V dt (A3.5). Separation of variables provides the following integral expressions f— eft = [exp ds„ (A3.6). The variable r| 0 is treated as constant. Integration of Eq. A3.6 provides an expression relating stress (e.g., load), strain, and time during viscous deformation of a porous medium: a-At _{\-^) exp^ + C (A3.7). The integration constant C is solved for by requiring the viscous component of accumulated strain to be zero under the condition At = zero: C = ^ A ) e x p ^ (A3.8) a yielding: aa • At f«-(gv-<0 exp^ ' - exp (A3.9). Rearranging Eq. A3.9 provides a constitutive relationship that relates ea to original porosity, framework viscosity, stress and time: a -a-<f>0 a-a At (jiT) + expv *°' (A3.10). 208 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0052759/manifest

Comment

Related Items