UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Durability of buried galvanized steel structures in British Columbia Galtung-Døsvig, Tom 1995

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

Item Metadata


831-ubc_1995-0253.pdf [ 12.78MB ]
JSON: 831-1.0050395.json
JSON-LD: 831-1.0050395-ld.json
RDF/XML (Pretty): 831-1.0050395-rdf.xml
RDF/JSON: 831-1.0050395-rdf.json
Turtle: 831-1.0050395-turtle.txt
N-Triples: 831-1.0050395-rdf-ntriples.txt
Original Record: 831-1.0050395-source.json
Full Text

Full Text

DURABILITY OF BURIED GALVANIZED STEEL STRUCTURES IN BRITISH COLUMBIA by TOM GALTUNG D0SVIG B.A.Sc, U.B.C, 1993 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTERS OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES Department of Civil Engineering We accept this thesis as conforming to the required standard  UNIVERSITY OF BRITISH COLUMBIA APRIL 1995 © Tom Galtung Dosvig, 1995  In  presenting this  degree at the  thesis  in  University of  partial  fulfilment  of  of  department  this or  thesis for by  his  or  scholarly purposes may be her  representatives.  permission.  Department The University of British Columbia Vancouver, Canada  for  an advanced  Library shall make  it  agree that permission for extensive  It  publication of this thesis for financial gain shall not  DE-6 (2/88)  requirements  British Columbia, I agree that the  freely available for reference and study. I further copying  the  is  granted  by the  understood  that  head of copying  my or  be allowed without my written  ABSTRACT The local goal of this thesis was to collect information regarding the determination of the service life of galvanized steel plates, which are used in the construction of soil-steel structures. British Columbia's Ministry of Transportation and Highways (MoTH) detected durability problems in some of its water carrying soil/steel structures, and requested that research be done to adopt design life specifications for the future installations of structures. This work summarizes the processes of degradation which affect the durability of soil/steel structures and discusses the design life practices currently employed by many European countries and the U . S. Based on a survey of 35 galvanized steel drainage structures and underpasses conducted by the  author  on Vancouver Island during the  summer  of 1994,  recommendations are given regarding the durability of these particular structures.  A  method was devised to evaluate design service life based on the new design life philosophy, specifically fitted to environmental conditions encountered in Coastal British Columbia.  ii  TABLE OF CONTENTS ABSTRACT LIST OF FIGURES LIST OF TABLES ACKNOWLEDGMENTS INTRODUCTION CHAPTER 1 - LITERATURE REVIEW 1.1 Basic corrosion theory 1.2 Factors affecting corrosion in soil 1.2.1 General 1.2.2 Soil resistivity 1.2.3 SoilpH 1.2.4 Soluble salts 1.2.5 Moisture content 1.2.6 Aeration 1.2.7 Redox potential 1.2.8 Soil-related bacterial corrosion 1.2.9 Climatic conditions 1.3 Water chemistry and water flow characteristics affecting corrosion 1.3.1 General 1.3.2 Corrosive salts 1.3.3 Protective salts 1.3.4 Water pH 1.3.5 Water related bacterial corrosion 1.3.6 Velocity, abrasion, aeration, and temperature 1.4 Atmospheric corrosion 1.5 Zinc as a protective coating 1.5.1 General 1.5.2 Hot dip galvanizing process 1.5.3 Coating thickness 1.5.4 Protection offered by zinc 1.5.5 Service life of zinc coating 1.5.6 Wet storage stain 1.6 Corrosion of steel 1.7 Design Service life considerations 1.7.1 General 1.7.2 California Method of determining service life 1.7.3 Available corrosion rate data 1.7.4 The new design life philosophy British design requirements American design requirements Design life comparisons CHAPTER 2 - MoTH'S STRUCTURES STUDY 1994 2.1 Location selection 2.2 Field work 2.2.1 Core sample extraction 2.2.2 Coating thickness 2.2.3 pH collection 2.2.4 Water and soil sample extraction 2.2.5 Water speed 2.2.6 Visual inspection iii  ii v vii viii 1 3 3 6 6 6 8 10 12 14 15 15 16 18 ...18 18 18 20 20 21 23 26 26 26 27 28 29 30 31 32 32 32 35 40 41 44 47 49 49 51 51 56 56 57 57 58  2.2.7 Photo documentation 2.3 Laboratory work 2.3.1 Soil moisture content 2.3.2 Soil resistivity 2.3.3 Water resistivity 2.3.4 pH 2.3.5 Coupon evaluation CHAPTER 3 - RESULTS AND ANALYTICAL PROCEDURES 3.1 General 3.2 Corrosion parameters 3.2.1 pH 3.2.2 Soil moisture content 3.2.3 Resistivity 3.2.4 Microbiological corrosion 3.3 Structural condition 3.3.1 Inside condition - above the high water mark 3.3.2 Soil side condition - above the high water mark 3.3.3 Invert inside (water side) condition - Unprotected inverts 3.3.4 Invert inside(water side) condition - Concrete protected inverts 3.3.5 Invert soil side condition 3.4 Corrosion rates 3.4.1 Establishing corrosion rates above high water mark 3.4.2 Establishing invert corrosion rates 3.5 Observed corrosion rates vs. American and British design loss rates 3.6 Time to perforation: California Method vs. Extrapolation in B.C 3.7 Corrosion rates/corrosion parameter correlation 3.8 Conclusions CHAPTER 4 - RECOMMENDED PRACTICE 4.1 General 4.2 Backfill requirements 4.3 Corrosivityclassification 4.4 Concrete invert protection 4.5 Sacrificial thickness requirements 4.6 Design life 4.7 Deicingsalt 4.8 Further testing 4.9 Future developments 4.10 Other design features 4.11 Measurement techniques BIBLIOGRAPHY  iv  58 58 58 59 59 60 60 62 62 63 63 64 65 68 70 70 71 72 78 87 87 87 92 97 99 100 105 109 109 109 Ill 112 116 119 119 119 121 121 125 127  LIST OF FIGURES Figure 1.1.1 Basic corrosion cell Figure 1.1.2 Uniform and localized corrosion Figure 1.2.1 Effect of pH on Corrosion of zinc Figure 1.2.2 Effect of chloride and sulfate concentrations on soil resistivity Figure 1.2.3 Resistivity and corrosion rate as functions of water content in soil Figure 1.2.4 Soil resistivity vs. average annual rainfall Figure 1.3.1 Rate of wastage vs. severity of abrasiveflowon metal culverts Figure 1.4.1 Rusting of steel with increasing relative humidity due to various impurities Figure 1.4.2 Life of protection vs. Thickness of zinc and type of atmosphere Figure 1.5.1 Hot dip galvanized coating on steel Figure 1.5.2 Corrosion plug resulting in corrosion resistance Figure 1.7.1 Chart for estimating average life of plain galvanized culverts Figure 2.1.1 Frequency distribution of surveyed structures' ages Figure 3.2.1 pH frequency distribution for (a) soil and (b) water samples obtained in survey Figure 3.2.2 Distribution of resistivities obtained in B.C Figure 3.2.3 Water and soil resistivity distribution according to corrosiveness Figure 3.2.4 Invert of Eagles Nest culvert displays nodules of rust throughout the invert Figure 3.2.5 Drainage area of Eagles Nest culvert Figure 3.3.1 Frequency distribution of inside zinc coating thickness Figure 3.3.2 High water mark indicated by steel/zinc transition Figure 3.3.3 Extreme bolt damage from exposure to large boulders Figure 3.3.4 Boulders up to one meter in diameter were observed in Island HW + 38 km culvert Figure 3.3.5 Pitting corrosion on down-stream side of corrugations Figure 3.3.6 Heavy perforations in Roy Creek culvert Figure 3.3.7 Abrasion caused by improperly laid concrete invert lining in Mile 61 culvert Figure 3.3.8 McKay pipe showed zinc loss at the concrete/steel interface after one year in service Figure 3.3.9 McKay culvert shows a limited amount of concrete lining in the invert Figure 3.3.10 Invert concrete lining in Schley pipe Figure 3.3.11 Invert concrete lining in Hydro Hill West pipe Figure 3.3.12 Invert concrete lining in Hirsch pipe Figure 3.3.13 Accumulation of debris along the concrete/steel interface in Schley pipe Figure 3.3.14 Extreme abrasion observed in Island HW + 34 km culvert Figure 3.4.1 Preliminary zinc loss rate distribution above high water mark Figure 3.4.2 Distribution of average soil side zinc thickness above the high water mark Figure 3.4.3 Soil side zinc loss rates calculated above high water mark Figure 3.4.4 Upper and lower bound invert steel loss rate distribution Figure 3.4.5 Mean invert steel loss rate distribution Figure 3.4.6 Upper and lower bound invert pitting rate distribution Figure 3.4.7 Ranking of pitting observed with accompanying mean rates Figure 3.4.8 Estimated average years to perforation for surveyed structures Figure 3.5.1 Comparison of design zinc loss rates with observed B.C. rates Figure 3.5.2 Comparison of specified steel loss rates with observed B.C. rates Figure 3.6.1 Comparison of time to perforation; California Chart v.s. B.C. extrapolation Figure 3.7.1 Soil resistivity v.s. preliminary zinc loss rates above high water mark Figure 3.7.2 Mean invert loss rate distribution at different soil resistivities Figure 3.7.3 Mean invert loss distribution at different culvert slopes Figure 3.7.4 Mean invert loss rate v.s. structural age Figure 3.7.5 Mean invert loss rate distribution at different water resistivities Figure 4.4.1 Gradual transition from steel to concrete prevents buildup of debris Figure 4.4.2 Grate installed to avoid contact between large boulders and culvert Figure 4.5.1 Comparison of specified mean steel loss rates with observed rates  v  3 4 9 11 12 17 22 24 25 27 29 34 51 63 65 67 69 69 71 74 75 75 76 77 80 81 82 83 83 84 85 86 89 90 92 94 95 95 96 96 97 98 100 101 101 102 103 103 113 115 118  Figure 4.10.1 Dip in concrete causes retention of organic material at Seymour Ave. underpass Figure 4.10.2 Accelerated deterioration at Seymour Ave. underpass due to the hostile environment present Figure 4.10.3 Design feature at Paradise Arch allows for moisture buildup at the concrete/steel interface Figure 4.10.4 Rusting noticed due to moisture retention in Paradise Arch  vi  122 123 124 124  LIST OF TABLES Table 1.1.1 Table 1.2.1 Table 1.2.2 Table 1.4.1 Table 1.5.1  Galvanic series of metal Corrosion classification according to resistivity Corrosiveness of different types of soil Effect of sea-salt on atmospheric corrosion Comparison of corrosion rates of uncoated steel and zinc-coated steel in various moist environments Table 1.7.1 Corrosivity classification test for existing soil in G.B Table 1.7.2 Corrosivity Classification for existing soil in the G.B Table 1.7.3 G.B. specification requirements for backfills to corrugated steel buried structures Table 1.7.4 Corrosivity classification of water or effluent Table 1.7.5 G.B.'s specified rates of deterioration Table 1.7.6 U.S.'s select granular backfill gradation requirements Table 1.7.7 Quality requirements for select granular backfill in the U.S Table 1.7.8 Electrochemical requirements for select granular backfill in the U.S Table 1.7.9 Comparison of electrochemical limits for select backfills used in various countries Table 1.7.10 Comparison of required design life for soil/steel structures in various countries Table 1.7.11 Comparison of total required sacrificial thickness for galvanized steel reinforcement in select backfill intended to last through the required design life Table 2.1.1. Location of structures examined through summer, 1994 Table 3.4.1. Zinc loss rates by division of sample coupons into age groups Table 3.4.2 Armco's specified total thicknesses for structural plate Table 4.2.1 Recommended electrochemical requirements for select granular backfill Table 4.2.2 Recommended quality requirements for select granular backfill Table 4.2.3 Recommended select granular backfill gradation requirements  vii  5 7 14 23 26 42 42 43 43 44 46 46 46 47 48 48 50 91 94 110 110 110  ACKNOWLEDGMENTS The author would like to thank his supervisor Dr. Sigfried Stiemer for his financial support, patience, and the introduction to this thesis topic. Furthermore, warm thanks should be expressed to the Ministry of Transportation and Highways for their financial support, and in particular, mention should be made of Kang Ho, Peter Brett, Orlando Tissot, and Robert Dick for initiating this project.  The author is also grateful for the  assistance offered by John Harrison during the field work period of the project.  viii  INTRODUCTION Galvanized steel manufacturers, suppliers, and consultants often have different opinions on service life calculations applied for mechanically stabilized earth structures constructed with galvanized steel as metallic reinforcement. One method widely used by different U.S. State Transportation Agencies for durability considerations is the California Method of computing service life. This procedure for estimating the service life of metal culverts is based on a survey of over 12,000 culverts. It was developed using data from a wide range of site conditions. The California data include non-corrosive, non-abrasive sites, corrosive sites, abrasive sites, and abrasive-corrosive sites. However, the method correlates only the lowest values of the soil or effluent p H and resistivity to estimate years to first perforation. Some average level of abrasion is therefore implicitly included. The American Association of State Highway and Transportation  Officials  (AASHTO) has recently introduced its design service life requirements which are based on the present general design philosophy already in use in countries such as Germany, France, and Great Britain (G.B.). The design practice requires that the galvanized reinforcement be placed in a select granular soil whose electrochemical properties exceed certain mandated limits associated with mildly corrosive regimes. Required design life is obtained by adding a sacrificial thickness to the structural thickness, which corresponds to the amount of metal presumed to be affected by corrosion during the structure's intended life. The final objective of this study was to devise a method of evaluating design service life of soil-steel structures based on the new design life approach fitted to environmental conditions encountered in British Columbia. An appropriate method of determining a structure's resistance to chemical and mechanical wear is to evaluate the actual performance in its service environment.  Site  conditions have a significant effect on how long a facility will last, and laboratory corrosion testing rarely gives a proper indication of the varied in-situ conditions. With this in mind, investigation of existing in-service structures and their surrounding environment 1  seemed to be the obvious approach for the collection of data for determining the service life of multi-plate steel structures in British Columbia. During the course of the summer of 1994, a survey of metal loss of existing soil steel structures on Vancouver Island was conducted. Thirty-two multi-plate culverts, 1 pipe culvert, and 2 multi-plate underpasses were sampled from locations across the island, with emphasis on Highways #4, #14, and #19. Ages of the structures were fairly evenly distributed, ranging from 2 to 44 years. The sites were inspected in the months of June and July 1994, in order to take advantage of the low water levels, permitting a thorough examination of the multi-plates. Metal core samples were extracted from every site in order to access possible soil side corrosion and measure the overall thickness of the plate. Furthermore, water and soil samples were obtained from each location for laboratory analysis. Environmental variables, which have been proven to affect corrosion, such as resistivity, pH, and moisture content, were measured. The data sets obtained from the field investigation were analyzed to obtain sets of corrosion rates which, used  in accordance  with the  recommended  design life  specifications, should assure required design lives of future installations of soil/steel structures in British Columbia.  2  CHAPTER 1 LITERATURE REVIEW 1.1 Basic corrosion theory Corrosion is defined as the surface wastage that occurs when metals are exposed to reactive environments.  The chemical compounds or corrosion products which  constitute such wastage are similar to the mineral rocks that are found in the earth's crust. In other words, corrosion is reversion of metal to its original ore. For the corrosion process to occur, four items must be present. They are: an anode, a cathode, an electrolyte, and a metallic path connecting the anode to the cathode. The place at which the metal corrodes, or is oxidized, is the anode. Here, positive metal ions are released into the electrolyte, which is an ionic conducting substance such as water or soil, in which the anode and cathode are immersed. Electrons, released at the anode, find their way through the metallic path to the cathode where reduction of some constituent, e.g. oxygen or hydrogen, from the surrounding electrolyte is the principal reaction. Figure 1.1.1 shows a schematic of a basic corrosion cell. It should be noted that in the absence of any of the aforementioned components, the current will not flow, resulting in no corrosion (Smith, 1989) , (Uhlig, 1963) . 1  2  Figure 1.1.1 Basic corrosion cell (Smith, 1989). E l e c t r o n F l o w (e ) Metallic ' Connection  3  The driving force for corrosion is a difference in electrical potential between the anode and cathode; this causes the electrons to flow.  Potential differences exist at  different points or areas on the surface of the metal. Potential differences may be brought about by local variations in both the metal and the electrolyte. On a microscopic scale, this potential difference can be equally distributed throughout the surface of a single metal due to different crystal orientation and chemical composition of the metallic grains. If the electrolyte in contact is sufficiently homogeneous, it will lead to uniform corrosion of the metal surface (Figure 1.1.2 a). On a macroscopic scale, different oxygen concentrations can exist on different parts on the metal due to a heterogeneous electrolyte; the formation of corrosion products will also cause a potential difference usually leading to a more localized attack of the metal (Figure 1.1.2 b); in this case the oxygen-starved area acts as the corroding anode (EFCP, 1990) . 3  Figure 1.1.2  a) Uniform corrosion. (EFCP, 1990) b) Localized corrosion.  Two different metals coupled together cause preferential corrosion of that metal which has the more negative potential or which requires the most energy to convert ore to  4  a metallic state. A galvanic series of metals has been established (Table 1.1.1) which lists the metal with reference to its relative corrosion potential. The series was established based on each of the metals potential with reference to copper/copper sulfate reference electrode. Galvanic corrosion occurs as a result of one metal being in contact with another in a conducting, corrosive environment.  The corrosion is stimulated by the potential  difference that exists between the two metals. The more noble metal acts as a cathode where some oxidizing species is reduced, the more active metal, which corrodes, acts as an anode. From the figure, it can be observed that zinc will corrode preferentially to iron if coupled together. Also, their relative proximity in the Galvanic series prevents excess corrosion of the zinc when coupled with iron and placed in an electrolyte. Zinc coating the steel therefore creates an effective protector for the steel substrate. However, it is important to note that as long as the zinc coating is continuous and unbroken, the coating will corrode like solid zinc. Not until the base steel becomes exposed through some break or corroded zone in the zinc coating will the galvanic protection properties come into effect. Table 1.1.1 Galvanic series of metals (Turner, 1988). Most energy required potassium (more electro-positive) magnesium aluminum zinc chromium iron nickel tin copper silver Least energy required platinum (more electro-negative) gold The rate at which corrosion occurs, is proportional to the amount of current passing through the electrolyte from the anode to the cathode; this is due to the fact that  5  metal ions are dissolved into the electrolyte at the same rate as electrons are released. It follows that this current flow, or the corrosion of a given metal i.e. the rate of corrosion, is dependent upon the precise environmental conditions to which the metal will be exposed. In addition, its own physical properties, those of the metal to which it may be coupled, and the relative potential between the two metals, all determine the rate of corrosion of the more active or the metal of most negative potential (Turner, 1988) . 4  1.2 Factors affecting corrosion in soil 1.2.1  General Modern design of buried steel structures is based on measurements of certain key  index parameters of the soil reinforcement backfill, which governs the corrosivity. Several parameters determine soil corrosivity, and these are generally interrelated but can be measured independently. N o direct connection between any one soil parameter and a quantitative corrosion relationship has been proven to exist, but certain parameters have been shown to be more accurate indicators of corrosion potential than others.  The  parameters most commonly used today in identifying the potential corrosiveness of soils are resistivity, pH, and presence of aggressive soluble salt (chlorides and sulfates). Other factors that affect the corrosion of metal in soil, but can be difficult to quantify, are moisture content, aeration, and temperature. 1.2.2 Soil resistivity A general consensus has emerged from publications on studies of buried metals that states that resistivity, the inverse of conductivity, is the one parameter that best indicates a soil's aggressiveness. When a piece of metal is buried in the soil, a difference of potential exists between two points of the buried metal. Electrons start to flow from the anode to the cathode through the metal, and back to the anode through the soil, thereby completing the circuit. The corrosivity of the metal is proportional to the current passing through it, which 6  follows Ohm's law, I=V/R (where I = current, V = voltage, and R = resistance).  The  resistivity of the soil, through which the current must pass, is thus inversely proportional to the corrosion rate of the metal. Soil resistivity is a measure of how easily a soil will allow an electric current to flow through it, and thus a measure of how effective the soil is as an electrolyte.  The lower the resistivity of a soil, the better it will behave as an  electrolyte and the more likely it is to promote corrosion. Measured in ohm/cm, resistivity can vary from 30 Ohm/cm in sea water to in excess of 100,000 ohm/cm in dry sand and gravel. Table 1.2.1 lists relationships between soil resistivity and soil corrosivity that are generally accepted (Elias, 1990) . 5  The electrical resistivity of the soil is a function of the water content and the nature and amount of soluble salts in the water solution. The greater the salt content, the lower the resistivity.  This is not a one-to-one ratio, since the various ions available have  different weights associated with one charge. Similarly, a high water content in soil can Table 1.2.1  Corrosion classification according to resistivity (Elias, 1990).  Aggressiveness  Resistivitv in Ohm/cm  Very corrosive Corrosive Moderately corrosive Mildly corrosive Non-corrosive  <700 700 - 2,000 2,000 - 5,000 5,000 - 10,000 > 10,000  produce low-resistivity values since there are large areas of water through which the current can flow and more complete hydration of the ions can occur. Reducing the water content can increase the resistivity of the soil by several orders of magnitude greater than the value determined at the saturated state, e.g. minimum resistivity.  The resistivity test most commonly used, measures soil resistivity at a  saturated water content, thus producing a minimum resistivity measurement not  7  necessarily corresponding to the true resistivity observed in the field. Only if the structure were continuously exposed to saturated soil would the minimum resistivity value be representative of the in situ conditions.  On the other hand, if the in situ soil only  experienced a certain degree of rainfall and groundwater seepage, the minimum resistivity would compare conservatively to the long-term in situ resistivity, resulting in the erroneous classification of a fairly non-corrosive soil as corrosive. Due to changing water contents in the soil over the course of the year, in situ soil resistivity cannot be used to predict specific corrosion rates. The only resistivity values useful in this prediction are minimum resistivity of the soil known to be continuously saturated. For soils of varying moisture content, the minimum resistivity can only be a guide to possible corrosion severity (Palmer, 1989) . 6  1.2.3  SoilpH Soil pH defines the acidity or alkalinity of the soil media, and represents the  hydrogen ion concentration in solution. There exist today mixed opinions about how accurate a soil's p H can serve to determine susceptibility for corrosion. Some agencies have found that locally, p H is a good indicator of corrosivity. Other studies has shown that for bare steel, within the pH range of about 4 to 10, the corrosion rate is virtually independent of pH, but depends on how rapidly oxygen diffuses on the metal surface (Meacham, Hurd, Shisler, 1982) , 7  (Worley, 1971) . 8  The American Galvanizers Association reports that p H is a primary factor governing the  corrosion behavior  of the galvanized coating in liquid chemical  environments. However, galvanizing performs well in solutions of pH above 4 and below 13 (Figure 1.2.1). Within these limits, a protective film forms on the zinc surface and the galvanized coating protects the steel by slowing corrosion to a very low rate. The exact chemical composition of the protective barrier and its adherence is somewhat dependent  8  upon the specific chemical environment.  Below p H of 4, the acidity has a tendency to  dissolve the corrosion product which act as a protective film. In very high alkaline soils, the soil's corrosivity increase because of high alkalinity usually creates an environment of high electrical conductivity (American Galvanizers Association, 1990) . 9  Natural soils generally have a pH between 4 and 8. Usually when the soil p H goes beyond these limits, deposits from other sources might have been added. A low p H can indicate use of ammonium sulfate as fertilizer or pollution by industrial waste, which often show signs of acidity. A resistivity of about 200 ohm/cm and a pH of about 11 probably shows that liquid manure has been added to the soil. Values of soil p H only represent the hydrogen activity in the soil, and not the total acidity of the soil. Thus, p H values as a measure of corrosion may be misleading in certain soils because organic acids, which are  Figure 1.2.1 Effect of pH on Corrosion of zinc (American Galvanizers Association, 1990) 14 12  10 8  1 1<D  DC  t :  :  6  4 2 • i L i 3 1 2 3 4  i i i i i r 1 , , 5 6 7 8 9 10 11 12 13 14 15 16 pH of Contact Material  not accounted for by the p H scale, might be present (Elias .1990) . 10  These acids might  facilitate dissolution of surface films as well, therefore being relatively corrosive to zinc and steel. In this case, the total acidity measured by titration will be a more relevant index in determining the corrosivity of the soil. Use of inorganic soil is preferred as backfill for  9  soil steel structures in order to eliminate any detrimental effects on corrosion brought about by organic acids produced by microbial growth. 1.2.4 Soluble salts The amount of soluble salts in a soil solution is directly proportional to the conductivity of that solution. Thus by increasing the amount of soluble salt, one decreases the resistivity. A wide variety of salts typically exists in the soil solution  Common cations or  base-forming elements include potassium, sodium, magnesium, and calcium. Common anions or acid-forming elements are carbonates, sulfates, chlorides, and oxides. Each of these ions has its own reactivity with metallic surfaces.  Calcium and magnesium are  generally not very soluble, whereas chloride is extremely soluble and thus completely removed by an aqueous extract. Two soils having the same minimum resistivity value may react differently depending upon the specific ions available in each soil. For example, chloride and sulfate are the most deleterious  salts, whereas carbonate forms a precipitate in basic  environments, which can create a protective layer over the metal surface and reduce the corrosion. Thus, if all conditions are given equal importance, a high resistivity soil of low alkalinity is more corrosive than hard scaling soil of much lower resistivity. The amount of total salts, and thus minimum resistivity, may not always be indicative of the corrosivity of the solution. Chloride and sulfate, the most detrimental of soluble salts, decrease resistivity, promote flow of corrosion currents, and obstruct the formation of protective layers. The effects of chlorides and sulfates on resistivity can be seen in Figure 1.2.2 for both theoretical and controlled laboratory tests. From these data, it can be concluded that at equal concentrations, chlorides are more corrosive than sulfates.  10  Also, at a resistivity  above 3000 ohm/cm, the chloride and sulfate levels will lie below 100 ppm and 200 ppm, respectively (Elias, 1990) . 11  Figure 1.2.2 Effect of chloride and sulfate concentrations on soil resistivity (Elias, 1990). HfcSJSI'IVfTY Ui,cm)  100000  I  •  t  |  I  ,  I  i  : I  Ji  -i_Li  I i i l  i I ii  i  I  i i  •  Loire Sand  A  Fontain. Sand  x  Sea Water Fill i  1—r i i  T  i  i  I I I  TT1  r  i  I  I  ••• Theoretical  iI  10000  I  4-4-+-  i I Ii i  i i 11 r r r  i  i  i  I  I  m  \ 1000  i  •  TT  i  ±3t  I  I  rrr  11  CI'  100 10  100 1000 SALT CONTENT (ppm)  11  10000  1.2.5 Moisture content As water is added to dry soil, the soil's resistivity drops quickly, and reaches a minimum level at 100 percent saturation. Also, as one increases the water content of the soil, the access to free oxygen decreases, which is needed at the cathode for reduction. Since the presence of oxygen is also necessary for soil corrosion to occur, there is an optimum water content, which is not necessarily at the lowest resistivity level, associated with the maximum corrosion rate.  Figure 1.2.3 shows that with an increase in water  content of the soil, the corrosion rate will at first increase rapidly due to a larger wetted area and improved conductivity, until it reaches an optimum value. Furthermore, a rapid decrease in the corrosion rate due to inaccessibility of oxygen is introduced when the water content has exceeded its optimum value. Studies have shown that for granular materials, the saturation needed for maximum corrosion rates occurs at a saturation of 60 to 85%. For granular materials, this saturation range also roughly corresponds to the range of moisture content required in the field to achieve the necessary compaction. Figure 1.2.3 Resistivity and corrosion rate as functions of water content in soil (Wranglen, 1985).  ~  UJ  \p  1-  CORROS  RESISTIV  11  /K  of>t  w  sat  w  WATER C O N T E N T C/o)  Soil moisture content primarily affects the activity of any chloride ions present. Thus, by increasing soil moisture, the chlorides become activated and corrosion is consequently accelerated. Tests have shown that when moisture contents were below  12  17.5%, the chloride ion concentration did not have a significant effect on the corrosion of zinc. With water contents exceeding 17.5%, chloride had a significant bearing on the corrosion rate (Corrpro Co. Inc., 1991) . 12  Water-carrying, multi-plate culverts in B.C., which have continuous flow, will most likely, due to seepage of stream water through seams and at the culvert's inlet, experience 100% saturation on the soil side of the inverts. According to Figure 1.2.3, reduction in corrosion rates should be expected within this area. However, corrosion tests conducted at 100% saturation levels most likely used water which contained less oxygen than that of highly aerated B.C. streams. Thus, Figure 1.2.3 will most likely be misleading with regard to corrosion rates experienced at locations of 100% saturation in B . C . streams. With the higher oxygen availability offered through highly aerated stream water seeping behind the structure through seams at the plate boundaries, rather high corrosion rates can be expected in invert soil-side locations where full saturation of the contacting soil exists. Moisture content is determined by soil structure, permeability and porosity of the soil. Sand and gravel are coarse-textured, rapidly-draining soils of high permeability, and constitute the most suitable and least corrosive backfills for steel-soil structures.  Clays  and silty soils are characterized by fine texture and poor drainage and are therefore inappropriate for use as backfill.  Soils can be divided into four groups for visual  classification concerning potential corrosiveness due to drainage conditions and aeration as shown in Table 1.2.2.  13  Table 1.2.2 Corrosiveness of different types of soil (Roads and Transportation Association of Canada, 1987). ' Corrosiveness Lightly corrosive Moderately corrosive Badly corrosive Unusually corrosive  Type of soil 1. Sands or sandy loams 2. Light textured silt loams 3. Porous loams or clay loams thoroughly oxidized 1. Sandy loams 2. Silt loams 3. Clay loams 1. Clay loams 2. Clays  Drainage Good  Aeration Good  Color Uniform color  Water table Very low  Fair  Fair  Slight mottling  Low  Poor  Poor  1. Muck 2. Peat 3. Tidal marsh 4. Clays & organic soils 5. Abode clays  Very poor  Very poor  Heavy texture- 0.5-1.0m Moderate below mottling surface Bluish - gray At surface; mottling or extreme impermeability  1.2.6 Aeration Aeration is a major factor governing corrosion in soils, as it affects the access of oxygen and moisture to the metal.  A porous soil has an abundant supply of oxygen  associated with it, initializing corrosion by combining oxygen with metal ions to form oxides, hydroxides, or salts of metals. If these corrosion products are soluble or removed from the anodic areas, corrosion continues; if, however, the products accumulate freely, they may act as a protective layer to reduce corrosion over time. Actually, in coarsetextured soils, such as sands and gravels, where there is free circulation of air, corrosion approaches the atmospheric type. On the other hand, in poorly aerated soils, the initial corrosion rates decrease very slowly with time due to the tendency of the corrosion products, which are in an deoxidized state, to diffuse outward into the soil, thereby not functioning as protective layers to the corroding metal. Oxygen is usually transported within a soil by dissolution in water or by diffusion. The penetration rate of oxygen within a soil will vary according to the physical properties  14  of the soil. Depth from the surface affects the amount of oxygen, dissolved or free, that reaches the metal surface. As the distance from the surface is increased, the rate at which oxygen is transported is reduced.  Differential oxygen transfer due to such factors as  differential compaction or dissimilar soil conditions may create areas of low oxygen concentration which become anodic and corrode preferentially (differential aeration cell). Thus, important qualities to obtain a low corrosive backfill are an even soil compaction and a homogeneous soil. Aeration alone should not be a guideline for corrosion, since even in a well aerated soil the presence of soluble salts can prevent the precipitation of protective layers of corrosion products and therefore not limit the disintegration of the metal (Escalante, 1989) . 13  1.2.7 Redox potential Redox potential of an area of soil has also been considered as a measure of corrosivity. High values of redox potential are obtained in well-aerated soils and low values in soils lacking oxygen. A low redox potential is associated with a high corrosion rate, since an anodic area is expected at such places. The redox potential on its own cannot indicate corrosivity. Rather, it is the variation of the redox potential along a certain distance which gives an indication of where differential aeration cells may occur, ultimately leading to a potential difference.  However, the measurement of the redox  potential is difficult and often of poor reproducibility, as the redox system is easily influenced by the introduction of atmospheric oxygen during testing (Escalante, 1989) . 14  1.2.8 Soil-related bacterial corrosion The contribution of bacteria to corrosion was recognized in the 1930's and since then much research has been done to identify the problem. Anaerobic sulfate-reducing bacteria are most commonly active in poorly drained, organic soil of near-neutral pH (between 5.5 and 8) and are associated with a low value of redox potential. Thus field  15  testing for sulfate-reducing bacteria in soil is a measurement of the redox potential of the soil. These bacteria have a physiological make-up which gives them the capability of utilizing hydrogen for growth purposes with sulfate or other reduced sulfur compounds as their terminal electron acceptor. Their metabolic product is sulfide. Another field test for the bacteria's presence is therefore a test for the presence of sulfide in the soil. The rate of corrosion which can arise from sulfate-reducing bacterial activity can be extremely high.  The aspects of the mechanism related to sulfate reducing bacteria  corrosion of ferrous metals are still quite unclear. Thorough studies are still needed for the understanding of the complexity of the deterioration factors involved in this type of bio corrosion (Popescu, Beschea, 1990) . 15  Where anaerobic sulfate-reducing bacteria are  active in the soil, the potential for corrosion can be reduced by removing the organic soil and using a more granular backfill such as sand (Patenaude, 1984) . 16  1.2.9 Climatic conditions Total annual rainfall is the most fundamental climatic feature controlling soil chemistry by determining the rate of leaching loss of soluble salts. Variations in the total annual rainfall are thus the primary cause of the extreme variations in soil chemistry which are found in nature. In arid regions where rainfall is inadequate to leach away even highly soluble salts like soluble chloride and sulfate, soils can be rendered very corrosive. In California, it was observed that soil resistivity values were related to their hydrologic location. In areas of high rainfall, the soil resistivity was high; conversely it was low in areas of meager rainfall (Figure 1.2.4) (Beaton, 1962) . 17  This susceptibility of high soluble salt content of soils in dry regions leads to a concern about the corrosive characteristics of soils already being utilized in the interior of British Columbia as backfill for existing soil steel structures.  The Ministry of  Transportation and Highways' 1993 galvanized structures study reported soil resistivity  16  Figure 1.2.4 Soil resistivity vs. average annual rainfall (Beaton, Stratfull, 1962). IOJOOO  5,000 su 2 X  SOIL R E S I S T I V I T Y -  o  s — £ Z  2  / 100  )  10  20  30  40  AVERAGE ANNUAL RAINFALL - INCHES  50  values as low as 2225 ohm/cm at structures along the Coquihalla Highway, by Kamloops, 7which is lower than the minimum resistivity limit of 3000 ohm/cm required for select backfill in the U . S .  18  As well, obtaining backfills for future structures in these areas of  lesser rainfall can come to involve a trial and error process if only a limited amount of backfill is readily available to meet the required electrochemical limits specified. The temperature of the soil, which is dependent on the ambient air temperature, is also a contributing factor in the corrosion process, and some interesting effects has been observed. The resistivity of a soil is inversely proportional to temperature, and an increase in temperature would therefore tend to increase the corrosion rate. However, an increase in temperature would reduce the solubility of oxygen, and reduce the rate of reduction  17  reaction at the cathode. These observations offset one another, resulting in a small net effect of temperature variation on corrosion (Escalante, 1989) . 19  1.3 Water chemistry and waterflowcharacteristics affecting corrosion 1.3.1 General Water, due to its ability to act as an electrolyte, plays an important role in the corrosion process. Water chemistry is the most important factor which governs the rate at which metal will degrade when exposed to water, the essential elements of which are soluble salts (usually denoted by resistivity), pH, hardness and alkalinity.  The water  chemistry of streams is influenced by the geological characteristics of watersheds, atmospheric dry depositions, the amount and quality of precipitation, and other pollution sources due to land use in the vicinity of the stream, e.g. fertilizer from agricultural land. Water velocity, which causes abrasion and aeration, also contributes to degradation of metals immersed in water. 1.3.2 Corrosive salts Corrosive salts such as chlorides and sulfates have a very detrimental effect on metal when dissolved in water. Their presence lowers the resistivity of the electrolyte, thereby promoting current flow and increasing corrosion rates. In addition, chloride has a tendency to dissolve otherwise adherent protective corrosion products present on the parent metal. Natural highly saline waters occur mainly in arid regions where little salt leaching loss occurs. Highly saline sea water or estuary water is a problem wherever it is found, and galvanized structures will corrode quite rapidly in these hostile environments. Galvanized steel structures are not advised installed wherever contact with saline water is likely, unless excessive sacrificial thicknesses are applied. 1.3.3 Protective salts Alkaline salts, mainly bicarbonates of calcium and magnesium, are common in varying concentrations in most natural waters. If the water contains high concentrations  18  of such compounds, it is called hard; otherwise it is soft.  These salts tend to form a  protective membrane composed largely of calcium carbonate, CaCC>3 that hinders the corrosion of reactive metals such as zinc and steel which would otherwise tend to corrode excessively. When present, the protective salts tend to modify metal corrosion products to encourage formation of more protective insoluble corrosion products.  The precipitated  salts combine with modified insoluble corrosion products to form the protective scale. Soft, high resistivity water containing very little of any type of dissolved salt, including hardness and alkalinity salts, tends to be corrosive because it possesses no scaling tendency. Under these conditions, a less effective unmodified corrosion product is the main type of protection available (Bednar, 1989) . 20  A galvanized-iron hot-water tank was placed in Chicago Great Lakes water (34 ppm Ca" ") where it lasted less than 20 years before failing by corrosion. On the other 1-1  hand, a similar tank was observed to only last for 1 to 2 years in Boston water (5 ppm Ca" *"). Protection to the tank in hard water is afforded by a film that protects the metal 1-  by retarding diffusion of dissolved oxygen to the metallic surface (Uhlig, 1963) . 21  Coastal B.C. stream waters are generally considered to be very soft, with alkalinity levels usually below 5 ppm Ca" " " (Sullivan, Samis, 1988) . Thus, the formation of any 1  4  22  protective insoluble corrosion-product scaling is not expected; rather, the formation of less adherent corrosion products that can easily be removed by seepage of water within and underneath multi-plate culverts results. The present study actually showed that zinc was lost very rapidly when exposed to continuous water flow. Soil-side invert zinc was lost in almost all water carrying structures, most of which showed otherwise favorable electrochemical properties for both water and soil.  19  1.3.4 Water pH Variable p H levels of water affect the corrosion rate of galvanized steel differently. As the zinc corrodes, it creates a protective layer, which acts as an oxygen diffusion barrier. This layer is created by the reaction of Z n ^  +  ions from the anodic zinc together  with OH" and C O 3 2 - ions present in the water, to give rise to compounds such as zinc carbonate[ZnC03], zinc hydroxide[Zn(OH)2], and zinc hydroxycarbonate[2ZnC03, 3Zn(OH)3, or Zn5(C03)3(OH)3].  At medium pH levels, between 6.5  and 8.5, the  hydroxide and carbonate ions in the water are present in high enough concentrations so that the zinc ions can react with them, giving rise to a protective deposit on the zinc surface. However, at lower p H levels, the concentrations of OH" and CO-p"~ ions in the water are not available in sufficient quantities for the zinc ions to react with and produce a protective scale. Consequently, there is less formation of a protective layer. Also at high pH levels, no formation of a deposit will occur. Here a reversible reaction will transpire in which zinc hydroxide reacts with the OH" ion in the water, almost completely converting the zinc hydroxide into Zn02^" ions.  In addition, zinc carbonate,  and zinc  hydroxycarbonate cannot form on the wall because the Zn^ " ion concentration in the 4  surface layer is negligible due to the reversible reaction (Legrand, Leroy, 1990) . 23  1.3.5 Water related bacterial corrosion Culvert investigations conducted in Wisconsin have found unusually high corrosion rates on the water-covered inverts of some galvanized steel pipes at sites of near neutral pH and high water resistivity value. Closer investigations indicated the cause of corrosion to be microbial in nature. A common characteristic of these high corrosion sites was the presence of nodular oxidation products on the water-covered inverts of the pipes. The metal surface under the nodules were observed to be pitted, and when the tubercles were removed, the underlying pitted surface appeared bright and shiny. Samples of rust lumps were removed and taken  20  to the laboratory for investigation, where they were identified microscopically as bacterial colonies of anaerobic sulfate-reducing bacteria. The colonization of surfaces results in the formation of a biofilm that adheres to the surface of the metal. Biofilms up to 100 micrometers in thickness are not unusual and in nearly all cases contain entrapped bacteria. Nodular corrosion of the metal surface then results from activity of sulfate-reducing bacteria, which are independent of the soil and which obtain electron acceptors in the form of sulfate from the water at the culvert site. The absence of oxygen in the nodules of rust is required for the bacteria to function. The mechanism of corrosion involving these bacteria is very complex and is still far from being completely understood (Howsam, 1990) . 24  Two types of localities can be noted where nodular bacterial corrosion has been most common. One is at sites of continuous flow which drain marshes or shallow lakes with large amounts of bottom vegetative growth. The other type has drainage consisting of intermittent surface runoff, where the source of sulfate is related to local geology or to industry. The appearance of nodular bacterial colonies have been observed at sites where sulfate in the water ranged from 1 to 30 ppm.  The Wisconsin study reported that  development of the nodules did not appear to correlate directly with the sulfur content of the water, but that other factors might be involved. It has also been observed that the presence of these bacteria had no corrosive effect on concrete and aluminum pipe (Patenaude, 1984) . 25  1.3.6 Velocity, abrasion, aeration, and temperature Velocity is a significant parameter in determining disintegration of culverts when water is sufficiently turbulent or contains enough suspended solids to promote the scouring, or abrasion, of metal surfaces. Factors involved are the frequency, duration, and amount of runoff events which carry significant amounts of abrasive materials.  21  In  addition, the characteristics and volume of the bed load are important elements to consider. In mountainous and hilly terrain, where the culvert is installed at steep grades, abrasion due to high velocity runoff in combination with bedload, is likely to control durability. Figure 1.3.1 displays the relative importance of abrasion at different stream energy levels. As the severity of the abrasive flow increases, an exponential growth in abrasion's importance on culvert degradation can be observed in the figure (Koepf, Ryan, 1986) . 26  Figure 1.3.1  Rate of wastage vs. severity of abrasive flow on metal culverts (Koepf, Ryan, 1986).  LU  z o z  NONE  LOW  MEDIUM  RAPID  Rate of wastage of culvert pipe  Water movement  also serves to help aerate water and ensure the high  concentrations of dissolved oxygen necessary to support corrosion. In river and stream water, dissolved oxygen is almost always present in sufficient quantity to cause corrosion. The higher the dissolved oxygen concentration in the water, the higher the cathodic potentials and, consequently, the corrosion rates.  22  The temperature of the water is another parameter which determines a metal's rate of corrosion. For carbon steel, each degree centigrade rise increases the rate of corrosion by 2.5%, provided the oxygen concentration does not change.  However, the reduced  oxygen concentration of higher temperature waters reduces the overall effect of temperature on corrosion. 1.4 Atmospheric corrosion Air itself is not generally corrosive to metals. For atmospheric corrosion to be of any importance at all, water must be present. It needs not occur as rain; merely humidity in the air is sufficient. Water is necessary as an electrolyte, and the more there is present in the air, the higher the rates of corrosion. Chloride salts in the air, as found in the immediate vicinity of the coast, produce high rates of corrosion attack, in particular on iron. Table 1.4.1 shows the dependence of the rate of corrosion of iron and zinc when exposed to sea-salt present in the atmosphere. As can be observed from Table 1.4.1, corrosion decreases with increasing distance from the coastal line, due to the diminishing salt content in the air (The Institution of Metallurgists, 1965) . 27  Table 1.4.1 Effect of sea-salt on atmospheric corrosion (The Institution of Metallurgists, 1965). (one year exposure) Distance from surf (yards) 50 200 400 1,300  Corrosion rate in Lim/year ingot iron 965 381 56 41  zinc 38 15 3 .05  Salt content of Air *  11.1 3.1 0.8  * Collected on a wet cloth; mg NaCl/day/100 cm  23  —  2  The most dangerous type of atmospheric corrosion, however, is man-made. Figure 1.4.1 shows the detrimental effects different sulfate pollution particles have on a polished steel specimen. As the figure shows, the rusting is low until a threshold relative humidity of about 70% is encountered (Wranglen, 1985) . 28  In general, atmospheres vary considerably with respect to moisture, temperature, and pollutants. Metals which resist a certain atmosphere successfully may perform poorly in another.  Galvanized steel, as an example, resists corrosion to a great extent in rural  atmospheres, but is less resistant in an industrial atmosphere (Figure 1.4.2). As a guideline, except in the most corrosive atmospheres, corrosion rates of metals are much lower when exposed to the atmosphere than when exposed to natural waters or soils. Thus, the parts of the multi-plate culverts exposed to air, such as inside above the high water mark, are not expected to control durability of the structure as whole.  Figure 1.4.1 Rusting of steel with increasing relative humidity due to various impurities (Wranglen, 1985). TEST TIME IN DAYS 20 r / SOOT PARTICLES+/  E cn UJ  0.01% S 0  120  / RETICLES OF /AMMONIUM SUL/f E+0.01%SO -  /  2  H A T  2  UJ OC  o  i  /  /  80!  <  o7  1  1  /  If  40  ^ / N O PARTICLES, /L\_. .__0.017. S 0 ONLY 2  -C  tx.  \  PARTICLESOF AMMONIUM SULPHATE, NO S 0 2  PURE AIR-^ A 0  3050 60  80  99  SUCCESSIVELY INCREASING RELATIVE HUMIDITY (UP TO 99%)  24  99  Figure 1.4.2  (American Galvanizers Association, 1990).  LIFE OF PROTECTION VS. THICKNESS OF ZINC AND TYPE OF ATMOSPHERE * Service Life is defined as the time to 5% rusting of the steel surface.  11  21  32  43  54  65  75  86  97  108  118  i  i  i  129  Thickness of Zinc in Micrometers so  r  i  i  i  i  i  i  i  i  Oz. of Zinc/Sq. Ft. of Surface .25  .50  .75  1.00  1.25  0.4  0.8  1.3  1.7  2.1  1.50  1.75  2.00  2.25  2.50  2.75  3.00  3.4  3.8  4.2  4.7  5.1  Thickness of Zinc in Mils 2.6  25  3.0  1.5 Zinc as a protective coating 1.5.1 General The service life of steel structures can be greatly increased by the use of metallic coatings as corrosion protection. A coating which has proven itself to be one of the most beneficial and economic alternatives, especially for underground service, is zinc.  The  corrosion resistance of zinc coating varies greatly with the environment in which it serves. A comparison between the corrosion rates in different environments for unprotected steel and zinc-coated steel is shown in Table 1.5.1. Here it is shown that zinc, in almost every environment encountered, has a slower rate of corrosion than steel (Wranglen, 1985) . In 29  addition, Table 1.4.1 illustrates the superior performance of zinc compared to iron when exposed to sea salt in the atmosphere. Table 1.5.1 Comparison of corrosion rates of uncoated steel and zinc-coated steel in various moist environments (Wranglen, 1985). Air Industrial  Corrosion rates in urn/year Water  Urban  Rural  River  Sea water  Soil Hard tap  Highly  water  corrosive  water  Unprotected steel * Zinc-coated steel  corrosive  slightly corrosive  100  50  10  50  100  10  100  30  5  10  5  2  30  20  15  15  10  3  * Pits may be 5 times as deep as the average thickness reduction  1.5.2 Hot dip galvanizing process Many different processes are used to apply the zinc coating to the steel. However, the most commonly utilized process for large steel sheets is hot-dip galvanizing. In this application the zinc is heated to its molten state and the steel is then immersed in the zinc bath, which usually holds a temperature between 430 and 470 degrees Celsius. During the immersion, three different zinc-iron alloys are formed when zinc and iron react metallurgically through a process called diffusion. The alloy compositions contain 6.25%,  26  11%, and 22% iron respectively. These alloys ensure that the galvanized layer adheres well to the steel surface. Figure 1.5.1 shows the three layers formed, where the layer richest in iron is closest to the steel substrate and that lowest in iron is adjacent to the pure zinc outer layer. The zinc-iron alloy tends to be more brittle than the pure zinc, which can lead to delamination from the substrate during bending of the galvanized plates if excessively thick amounts of alloy are produced (Slunder, Boyd, 1983) . 30  Figure 1.5.1 Hot dip galvanized coating on steel (Uhlig, 1963).  50-100 nm  1.5.3 Coating thickness The galvanized coating produced is usually specified in weight per unit area per plate, i.e. the coating on both sides of the substrate is taken into account.  A coating  weight of 2 oz./sq. ft., including both sides, is equivalent a coating thickness of 43 u,m on each surface.  Because of the many variables and changing conditions that are  characteristic of hot dip zinc coating, the mass of coating is not always evenly divided between the two surfaces of a zinc coated plate; nor is the zinc coated evenly distributed on one side of a plate. Thus, when a plate is dipped for a minimum of 2.0 oz./sq. ft. for example, a good chance exists that more than the required minimum is applied. This was observed to be the case in all metal core sampled structures in this survey (Carter, 1977) . 31  27  1.5.4 Protection offered by zinc One can distinguish between two separate stages in the protection given by zinc on steel. In the first stage, the zinc completely covers the steel, isolating the steel from the electrolytes in the environment. Here the corrosion of the zinc is due to the appearance of micro couples on the zinc surface. These areas of different electrochemical potentials are caused by zinc micro crystals at the surface of the galvanized layer with different crystal orientation and chemical composition. The micro crystals are all different; one may be anodic and dissolve into the electrolyte, while the underlying micro crystal may be cathodic. Hence the anodic and cathodic areas are constantly changing, giving rise to a fairly uniform corrosion. Zinc is oxidized at the anodic areas, where the zinc ions react with the anions present in the electrolyte. On the cathodic areas, reduction of oxygen takes place. At this first stage of protection it is of great importance that the zinc corrodes slowly and lasts as long as possible. The formation of protective deposits on the coating surface is very helpful in prolonging the life of the zinc coating. Such corrosion products slow corrosion down considerably, but do not stop it. The second stage in the life of the zinc as a protective metal begins when the steel substrate is exposed. This may be caused by damage to the coating, or because of gradual corrosion of the coating itself.  From this time onward, the protection is offered by  sacrificial action of the zinc, also called cathodic protection by a sacrificial anode. Due to a now higher potential difference between the steel cathode and the zinc anode than between the micro couples on the zinc, the zinc coating is now consumed at a more rapid rate, leaving the steel unharmed.  Even in the case of gross discontinuities in the zinc  coating, this type of protection persists.  It has been demonstrated that steel remains  protected in an industrial environment even when strips of 8-10 mm are left exposed (Basalo, 1992) . 32  Figure 1.5.2 illustrates the situation that prevails when steel is left  28  unprotected at a discontinuity in the zinc coating.  Here it can be seen that the  discontinuity is filled with corrosion products originating from the zinc itself. The zinc corrodes in the presence of a aqueous solution where zinc ions react with hydroxide ions producing Zn(OH)2-  It should be noted that in a natural environments, basic zinc  carbonates and sulfates rather than Zn(OH)2 are created.  Also, in atmospheric  environments periodic drying of the surface causes the formation of a non-conducting plug of corrosion product, which lowers the galvanic interaction due to high electrical resistance between the steel and the surface of the zinc. Figure 1.5.2 Corrosion plug resulting in corrosion resistance (Uhlig, 1963). corrosion plug  /  base metal (steel)  1.5.5 Service life of zinc coating The relatively constant rate at which zinc is consumed makes its useful life as a coating proportional to the thickness of the coating. Initially, zinc corrodes at a higher rate, due to formation of protective corrosion products.  However, after a year or two  when the protection of the corrosion products has come into effect, the much lower corrosion rate can be approximated as virtually constant as long as the environmental conditions remain the same. Thus the usual criterion for determining the expected service life of zinc coating, when the rate of deterioration of a particular environment is known, is coating thickness: the thicker the coating, the longer the service life. It is thus important that the zinc corrosion products remain on the structure in order to prolong the life of the galvanized coating. Introduction of chlorides contained in  29  road salt will remove the corrosion product through initiation of initial higher zinc corrosion rates.  The structure should avoid such deleterious agents as road salt by  diverting the drainage water flow from the salt covered road away from the crown of the culverts. This can be implemented by use of longitudinal drains which would lead the high saline water away from the structure. 1.5.6 Wet storage stain Wet storage stain is the voluminous white or gray deposit formed by accelerated corrosion of freshly galvanized articles which has been stored or transported under damp condition with limited access to air. Freshly galvanized zinc, due its reactive nature, easily reacts with the surrounding air to form a film of thin, hard zinc oxide, which is the first step in the development of a protective layer associated with zinc coatings. This surface then reacts with rainfall or dew under normal atmospheric exposure conditions to form a porous, gelatinous zinc hydroxide corrosion product. When drying, this product reacts with carbon dioxide in the atmosphere and is converted into a thin, compact and tightly adherent layer of corrosion product which consists mainly of basic zinc carbonate. However, if the continuously wet zinc product is not able to dry for a long time with the additional lack of air present, the modification of the gelatinous zinc hydroxide into a more tightly zinc carbonate will not occur. The relatively soluble non-protective zinc hydroxide will constantly be forming and corrosion will proceed as long as the original conditions prevail. The extent of the damage produced by wet storage stains depends on the duration of exposure to the above mentioned environment. During sustained storage in wet and poorly ventilated conditions, the protective value of the coating may be impaired and even completely removed. However, due to the voluminous nature of the wet storage stains, small superficial attacks may appear more serious than they actually are. Wet storage  30  stains can be avoided by proper storage of articles which will prevent moisture retention and allow for the circulation of air (AGA, 1992) . 33  1.6 Corrosion of steel Carbon steel is generally thought of as a metal with low corrosion resistance. Its low chemical resistance is indicated by its position closer to the electropositive(more corrosive) end of the galvanic series (Table 1.1.1). Furthermore, carbon steel's corrosion product is porous, non adherent, and a relatively good conductor, thus offering very small barrier protection in reducing the flow of corrosion current.  Steel finds extensive  application primarily because of its low cost, good mechanical properties, and ease of fabrication. Dry steel surfaces will not corrode, but as soon as any rust appears, corrosion will proceed if the relative humidity exceeds 80%, because rust absorbs moisture from the air. Steel is susceptible to pitting corrosion, which is a very localized form of corrosion, when immersed in an electrolyte such as soil or water. Pitting of steel is usually initialized due to physical conditions of the steel itself such as surface inhomogenities or by local variations in the electrolyte such as differential aeration cells, where the oxygen starved area acts as the corroding anode. Pitting proceeds as the differential aeration cell is sustained due to corrosion products developing in the pit, thus blocking oxygen access. Although unprotected steel is susceptible to rather high corrosion rates, it is widely used as the most economical material of construction even under ambient aggressive conditions where corrosion protection methods such as protective coatings offer reasonable life expectancies.  31  1.7 Design Service life considerations 1.7.1 General Durability studies of culverts have been conducted to a great extent over the last forty years by many local government agencies. Many of these studies have usually turned out to be inconclusive or controversial due to the large number of parameters influencing a structure's service life. Furthermore, trying to correlate information from various studies into a comprehensive database has proven to be an impossible task since durability studies conducted have not used a common rating system, but rather subjective interpretations by inspectors.  Also, different environments and climates make it difficult to transfer one  region's results to another. Thus, different ways of computing design service life currently exist. Due to this lack of common performance data, suppliers and manufacturers of galvanized steel products and B.C.'s Ministry of Transportation and Ffighways have not yet agreed to which approach to apply in establishing design service lives for soil steel structures. Until recently, the method used extensively in the United States has consisted of various modifications of the California Method of determining service life. However, a new design philosophy has recently emerged, which is a variation of that currently in use in Britain, Germany, and France.  The American Association of State Highway and  Transportation Officials (AASHTO) has also recently incorporated this method into their designs. 1.7.2 California Method of determining service life The California Method is a survey technique to estimate corrosion potential of proposed culvert sites.  The method estimates the time to first perforation of metal  culverts and is based on a survey of over 12000 culverts conducted in California prior to 1960. It was developed using data from a wide range of site conditions. The California data included non-corrosive, non-abrasive sites, corrosive sites, abrasive sites, and  32  abrasive-corrosive sites. The method correlates p H and resistivity of either soilside or the waterside to estimate number of years to first perforation. Some average level of abrasion is therefore implicitly included. This estimation technique has in some instances been found to be overly conservative, and under other circumstances, overly liberal. In areas where soluble salts which will decrease the resistivity appear as protective scaling salts, e.g. calcium, the chart tends to underestimate the life of a structure. On the other hand, in high resistivity areas, where no scaling salts are available, corrosion can proceed quickly, and the California method will overestimate life under these circumstances. For example, in Florida, the average service life determined by inspection was estimated to be 70 years or more, whereas perforation time estimation from the California Chart was less than 40 years (Bednar, 1989) . 34  In the design of culverts, the designer must first select a culvert fulfilling the structural requirements.  High safety factors are applied to account for uncertainties in  stresses induced during installation and early years of service when the soil cover is adjusting to the weight of the embankment and other loads. During these early years, little metal is lost, thus these required safety margins are essentially upheld over this period. However, when the design life of the structure is approached, the soil surrounding the culvert has fully developed and attained static equilibrium. Now most of the load is carried by the soil, and the culvert primarily provides containment, suggesting that a smaller safety factor is required at the end of a culvert's life. Thus, the difference between the initial thickness required for full structural support and the reduced final thickness needed after static equilibrium has occurred would be available for corrosion.  The  California Method uses this philosophy to account for allowable metal loss due to corrosion. One method of determining service life which has commonly been used by suppliers is a slight modification of the California Method depicted in American Iron and Steel Institute's "Handbook of Steel Drainage & Highway Construction Products". This 33  variation of the California Method acknowledges the fact that first perforation does not mark the end of a pipe's service life. The National Bureau of Standards soil tests indicate that overall metal loss at time of first perforations is only about 13% in soils.  The  modification predicts that service life ends at about twice this metal loss. This seems like a more valid assumption, since a minor perforation due to pitting does not constitute a loss of serviceability in culverts. Actually, there are reports of entire inverts being lost while the culverts remain standing, resisting the imposed loads (Worley, 1971) . 35  Figure 1.7.1 shows a modified design chart to determine the average service life of a 1.32 millimeter thick galvanized pipe based on the California Method. Due to an estimated linear relationship between thickness of pipe and service life, gage factors have been devised as multipliers for obtaining desired design life for thicker pipes (American Iron and Steel Institute, 1983) . 36  Figure 1.7.1 Chart for estimating average life of plain galvanized culverts (American Galvanizers Association, 1983). Figure 5-4 CHART FOR ESTIMATING AVERAGE LIFE OF PLAIN GALVANIZED CULVERTS  100.000  Minimum Resistivity (R) ohm cm  34  1.7.3 Available corrosion rate data The National Bureau of Standards (NBS) of the U.S. started burying metal pipes and sheet steel for corrosion testing purposes as early as 1910. Due to this extensive field testing and the comprehensive database existing, one general conclusion has been that corrosion is greatest in the first few years of burial, and then levels off to a significantly lower steady rate. N B S suggested an exponential equation to predict the amount of general corrosion at some time (t) after burial: x=K * t  n  where x is the metal loss after a time "t". "K" and "n" are soil and site dependent factors (Romanoff, 1957) . "K" is associated with short term corrosion rate trends, whereas n, a 37  number less than unity, decides the long term characteristic of the corrosion rate. This generalized corrosion rate relationship has been found to be a reasonable prediction model to determine the range of corrosion rate for single phase materials. Problems arise in the determination of the " K " and "n" constants for specific environments and also in the integration of the transition between the galvanized phase and the subsequent bare steel phase. Great Britain are currently using the above-mentioned model to calculate the steel loss after galvanizing is consumed for the life of their buried structures. For the wide range of soils included in the N B S study, not necessarily reflective of select backfill used in today's construction, ranges of "K" and n factors were produced for different steels. For low alloy and carbon steel, a "K" factor ranging from 150 to 180 u.m was obtained, whereas "n" values lay in the 0.5 to 0.6 range. Galvanized steel was found to have a "K" value varying from 5 to 70 u,m, but an "n" value was not determined. France has done a 10 year study of buried box samples and electrochemical cells in the same fills as used in soil-steel construction. This research resulted in an "n" value of approximately 0.6 when zinc is present, and then varying from 0.65 to 1 after depletion of the zinc. A "K" value for galvanized steel (only the zinc phase) was measured to be in the 35  range of 3 to 50, with the higher values representing a backfill of lower resistivity and higher chloride and sulfate concentration. For resistivities greater than 5,000 ohm/cm, the range of "K" reduces to 8 to 45 with a mean value of 25. Darbin has extended the "n" value of 0.6 to 0.65 for galvanized steel to also account for loss in the carbon steel consumption phase. Such an extrapolation, according to F H W A seems unconservative in light of higher carbon steel corrosion rates, and rather an "n" value of 0.8 for pure carbon steel is believed to be warranted (Elias, 1990) . 38  Researchers at the University of Stuttgart undertook an analysis of the N B S data obtained from sites characterized as well-draining with a predominately granular soil, deemed to be representative of select backfill.  The soil's electrochemical limit were as  follows(Elias, 1990): pH range Resistivity Chlorides Sulfates  4.5-9.5 > 1,000 ohm/cm < 50 ppm < 200 ppm  However, most of the backfills were associated with soils classified as moderately to mildly corrosive (Table 1.2.1). The data indicate a rapid loss during the first years for both galvanized and bare steel specimens, then attenuating to a reduced rate. A model was proposed assuming constant corrosion rates for the galvanizing and the bare steel after all zinc is lost. Rates of corrosion were developed for non-saturated soils that conform to the above electrochemical limit, and they are: For Zinc: 6 um/year 2 u.m/year  for the 2 first years until zinc depletion  For Carbon Steel: 45 um/year 9 rum/year  for the 2 first years thereafter  36  For conditions in which the soil is saturated or exceeding the chloride and sulfate concentration of 50 ppm and 200 ppm, respectively, but still having resistivities greater than 1,000 ohm/cm, somewhat greater losses were projected. The maximum loss rates are as follows: For Zinc: 17 u,m/year 2 u.m/year  for the 3 first years until zinc depletion  For Carbon Steel: 80 u,m/year 12 u.m/year  for the 2 first years thereafter  Reinforced Earth Company Ltd. reports electrochemical criteria recommended for Reinforced Earth backfills as follows: pH range Resistivity Chlorides Sulfates  5 > < <  - 10 1,000 ohm/cm 200 ppm 1000 ppm  Reinforced Earth Company Ltd. further reports metal loss rates applied in design life calculations of galvanized steel tensile reinforcing strips utilized in reinforced earth walls. For non-saturated backfill (Reinforced Earth Company Ltd., 1988) : 39  For Zinc: 6 u,m/year 2 u.m/year  for the 2 first years until zinc depletion  For Steel (resistivity > 10,000 ohm/cm): 6 (im/year  after complete zinc removal  For Steel (resistivity 1,000 - 10,000 ohm/cm): 9 u,rn/year  after complete zinc removal  For saturated backfill: For Zinc: 10 u,m/year  for the 3 first years  37  2 um/year  until zinc depletion  For Steel (resistivity > 10,000 ohm/cm): 6 urn/year  after complete zinc removal  For Steel (resistivity 1,000 - 10,000 ohm/cm): 9 um/year  after complete zinc removal  Corrosion of metallic strips buried in chloride rich backfills (areas of highway deicing salt use) can, according to Reinforced Earth Company Ltd., be conservatively estimated using the rates listed below: For Zinc: 17 um/year 2 |j,m/year  for the 3 first years until zinc depletion  For Steel: 12 um/year  after complete zinc removal  Other research data with special emphasis on corrosion prediction in marine environments indicate that the splash zone is usually the area hardest hit by corrosion. Furthermore, corrosion of metal in soils is approximately one-fifth of the corrosion in the splash zone. Thus, it is expected that water-carrying culverts should experience greater metal loss in the invert part of the structure exposed to water and wet soil, than the area above the high water mark not continuously affected by drainage water. Also, it should be noted that corrosion does not occur in a uniform manner, but sometimes takes the shape of pitting, thus loss of cross-sectional area will be greater where significant pitting is expected  Therefore, recent work done by the American  Federal Highway Administration (FHWA) stated that consideration must be given to the effects on tensile strength by the pitting mechanism when choosing uniform corrosion rates (Elias, 1990). Data have shown that a factor of approximately 2 exists between average tensile strength loss to average thickness loss.  Thus using the N B S model,  F H W A has proposed that the galvanized steel loss determination may be:  38  x = 25 * t x = 50 * t ° 0  65  65  (average) (maximum)  And for carbon steel, the equations would be: = 40 * 0.80 x = 80 * t x  t  0  80  (average) (maximum)  Furthermore, by using the constant rate model proposed by Stuttgart University, F H W A suggests maximum rates to be: For Zinc: 15 um/year 4 [am/year  for the 2 first years until zinc depletion  For Carbon Steel: 12 (xm/year These last rates are essentially equal to the corrosion rates employed by A A S H T O to calculate sacrificial thicknesses in their design for M S E structures (Elias, 1990) . 40  39  1.7.4 The new design life philosophy The present general design philosophy for reinforced soil structures such as multiplate galvanized steel culverts and other reinforced earth works can be summarized in 3 steps: 1. Place the reinforcement in a select granular soil backfill whose electrochemical properties are within certain limits generally associated with mildly corrosive regimes.  2.  Choose a desired design life, depending on the structure classification.  Generally in the range of 70 to 100 years.  3. Based on a set of corrosion rates associated with the mildly corrosive regime, a sacrificial thickness is added to the structural thickness required, which should be equal to the thickness assumed lost due to corrosive attacks at the termination of its required service life.  This design philosophy was adopted because it, to some extent, controls the environment in which the metallic reinforcement is placed, thus limiting corrosion to acceptable levels. Furthermore, the study of soil-corrosion is a very inexact science, with a high number of interrelated factors influencing initial and long term corrosion rates. N o well defined relationship between any soil parameters and corrosion has yet proven to exist. This design procedure considers the worst case scenario and should therefore, as long as the soil conform to the required electrochemical criteria, provide a reasonable safety margin in design life determination of soil-steel structures This design practice has been in operation in Europe for a number of years, and has recently (1991) been adopted by the U.S., where it now appears in AASHTO's "Standard Specifications for Highway Bridges" (AASHTO, 1992) . 41  40  The general design philosophy is agreed upon by countries such as Germany, France, United Kingdom, and United States.  However, within the 3 above-mentioned  steps, some variations are apparent. British design requirements The current method employed for design life requirements of buried structures in Great Britain (G.B.) has been in operation since 1982, and was refined in 1988 as Departmental Standard B D 12/88 (Brady, McMahon, 1994) . 42  It is based on a  classification system which rates the different environments in contact with the buried structure according to their corrosivity. Sacrificial thicknesses are then assigned according to the corrosivity rating of the site. Aggressive sites are consequently assigned higher sacrificial thickness than non-aggressive sites. Air-exposed accessible surfaces of underpasses and culverts are not given any sacrificial thickness. It is assumed that these surfaces will last their design life by periodic maintenance when and if necessary. For soil exposed surfaces, either the existing soil and/or a backfill must conform to certain physical and chemical requirements in order to determine their corrosion rating. If the structure is partially or wholly immersed in trenches, the corrosion properties of the existing soil within a horizontal distance Y=0.4+0.23*S meters of the structure has to be evaluated, where S is the structural span in meters. The assessment criteria for the soil is reproduced in Table 1.7.1, and the corrosion classification, based on total points obtained in the assessment tests, can be viewed in Table 1.7.2. For structures installed in embankments, or where the existing soil was removed for a minimum distance of (Y) from the structure, a different set of backfill requirements has to be fulfilled, which will immediately put the backfill in the nonaggressive category. The select backfill's physical and chemical requirements are depicted in Table 1.7.3.  41  Table 1.7.1 Corrosivity classification test for existing soil in G.B. (Brady, McMahon, 1994). Property Soil Type  Ground water level at buried position and drainage culverts  Measured Value  Points  Fraction passing 63 Lim sieve Plasticity Index (PI) of fraction passing 425 Lim sieve  < 10 % <2  +2  Fraction passing 63 Lim sieve Plasticity Index (PI) of fraction passing 425 uun sieve  < 10 % 2-6  0  Fraction passing 63 urn sieve Fraction passing 2 |im sieve PI of fraction passing 425 lim sieve  >10%but<75% < 10 % <=6  0  Any grading outside the above limits PI of fraction passing 425 Lim sieve  <=6  -1  PI of fraction passing 425 Lim sieve Any grading  > 6 but < 15  -1  PI of fraction passing 425 Lim sieve Any grading  >= 15  -2  Organic content or material containing cinder or coke  > 0.2 %  -4  Well drained area Poorly drained area and drainage culverts with periodic flow Ground water above invert level of structure and drainage culverts with continuous flow'  +1 -1 -4  >= 10,000 < 10,000 but >= 3,000 < 3,000 but >= 1,000 < 1,000 but >= 100 < 100  +2 +1 -1 -3 -4  pH of soil and ground water  6 <=pH<=9 5<=pH<=6 Less than 5 or more than 9  0 -2 -4  Soluble sulfates (ppm)  <=200 > 200 but <= 500 > 500 but <= 1000 > 1000  +1 0 -2 -4  Chloride ion (ppm)  <=50 > 50 but <= 250 > 250 but <= 500 >500  0 -1 -2 -4  No discoloration of lead acetate paper Slight darkening of lead acetate paper Moderate darkening of lead acetate paper Rapid blackening of lead acetate paper  0 -2 -3 -4  Resistivity (ohm - cm)  Sulfide and hydrogen sulfide  Table 1.7.2 Corrosivity Classification for existing soil in the U K (Brady, McMahon, 1994). Points total 0 or more -1 to -4 -5 or less  Corrosivity Classification Non-aggressive Aggressive Very aggressive  42  Table 1.7.3 G.B. specification requirements for backfills to corrugated steel buried structures (Brady, M c M a ton, 1994) Maximum fines content  10 (passing 63 |im sieve)  Maximum plasticity index of material passing 425 urn sieve 6  Minimum resistivity (ohm/cm)  Maximum Sulfate Content (Expressed asS0 )  Maximum Chloride ion Content  pH  0.25 gm/liter (250 ppm )  0.025 % (250 ppm)  >= 6 but <=9  3  2,000  All water- or effluent-carrying culverts require concrete paved inverts. For grades less than 2%, invert protection shall be 100 mm, whereas 125 mm is required for grades greater than 2%. Also, if bedloads of diameter greater than 100 mm are anticipated to be carried through the culvert, a minimum concrete thickness of 170 mm is required. In addition, assessment of any water or effluent running through the structure is also necessary for any seepage underneath the culvert. Table 1.7.4 shows the classification scheme for water or effluent. No sacrificial thickness is required on the water side due the concrete paved invert which should offer satisfactory protection to the underlying steel. Table 1.7.4 G.B. corrosivity classification of water or effluent (Brady, McMahon, 1994) Corrosivity Classification  —  J  —  v  j,  —  ,  —  Non-aggressive  6 <= pH <= 9  Properties of water and effluent Chloride ion (ppm) <=50  Aggressive  5 <= pH < 6  > 50 but <=250  > 200 but <= 500  Very aggressive  < 5 or > 9  >250  >500  PH  Soluble sulfates (ppm) <=200  -  • / •  The expected rates of deterioration for galvanizing and steel installed in aggressive and non-aggressive environments are given in Table 1.7.5. Great Britain design standard does not encompass structures installed in very aggressive environments or those carrying sea water. All permanent buried structures are specified a design life of 120 years (Brady, McMahon, 1994) . 43  43  Table 1.7.5 G.B .'s specified rates of deterioration (B rady, McMahon, 1994). Rate of corrosion of galvanizing (urn/year)  Position  Buried surfaces  Non-Aggressive  Aggressive  4  14  Calculation for thickness of sacrificial steel (M in \im) * Non-Aggressive M = 22.5 * t  067  Aggressive M = 40 * t °  80  * M is the thickness in urn of the sacrificial steel for the corroding face, and t is the number of years for which sacrificial metal is required American design requirements The U.S. has just recently adopted the new design life philosophy for buried structures.  The American Association of State Highway and Transportation Officials  (AASHTO) presented their design service life requirements for mechanically stable earth (MSE) walls in a 1991 interim report, and was further expanded in the 1992 fifteenth edition of "Standard Specifications for Highway Bridges". "Division I -Design" of the A A S H T O code states that buried steel reinforcing elements are to be designed with a sacrificial thickness assumed to be lost by uniform corrosion during service life of the structure. Sacrificial thicknesses shall be computed for each exposed surface according to the depletion rates as follows: Galvanization loss Carbon steel loss  = 15 |im/ year for the first 2 years = 4 u,m/ year for subsequent years =15 u,m/ year after zinc depletion  These rates are assumed to be maximum average loss rates. Thus, for tensile structural members, such as anchors, the reduced minimum thickness at end of the design life should remain proportional to tensile strength of the member. A minimum design life of 75 years should be assured for permanent structures such as highway rural retaining walls, and 100 years for critical structures, which include bridge abutments, urban retaining walls, and railway structures.  44  In addition, certain physical and chemical restrictions are imposed on the backfill to limit corrosion. According to "Division II - Construction" of the A A S H T O code, the backfill should be a sound, durable, granular material free from organic matter or other deleterious material (such as shale or other soft particles with poor durability). The backfill should meet the gradation requirements given in Table 1.7.5. (Test methods used for by A A S H T O are also included and are described in "Standard Specifications for Transportation materials and methods of sampling and testing", A A S H T O ,  1992.)  Furthermore the material must also conform to the quality and electrochemical requirements given in Table 1.7.6 and Table 1.7.7, respectively. The backfill must also be placed dry of optimum, in the saturation range of 60 to 80 %, which will ensure a conservative range of resistivities in the field compared to the laboratory test which are evaluated at 100% and are the indices used in the current specifications. Division II Construction further prohibits use of cinders, which are highly acidic and therefore corrosive. If the use of de-icing salt is a possibility in the wintertime, an impervious membrane should be installed between the pavement structure and the backfill, due to the deleterious effects produced by a high chloride concentration. This membrane should be sloped to drain beyond the reinforcing zone ( A A S H T O , 1991) , (AASHTO, 1992) . 44  45  The Federal Highway Administration (FHWA) has close to identical design life specifications for buried structures as AASHTO's for federal construction projects. However, its structural backfill electrochemical limits were found to be somewhat more liberal, in that they allow up to 200 ppm of chloride and 1000 ppm of sulfate, compared to AASHTO's limits of 100 ppm and 200 ppm, respectively (FHWA, 1992) . 46  45  Table 1.7.6 U.S.'s select granular backfill gradation requirements (FHWA, 1992). Percent by weight Sieve size (mm) passing designated sieve (AASHTO T27 and Til) 100  100  75  75-100  No. 200 (75 Lim)  0-15  Table 1.7.7 Quality requirements for select granular backfill in the U.S \ (FHWA, 1992). . Shear angle of internal Sodium sulfate Los Angeles abrasion Plasticity index friction soundness loss (AASHTO T96) (AASHTO T90) (AASHTO T236) (5 cycles) (AASHTO T104) >= 34 degrees  15 % max  50 % max  6 max  Table 1.7.8 Electrochemical requirements for select granular backfill in the U.S (FHWA, 1992). Minimum Resistivity pH Sulfate Content * Chloride Content * (AASHTO T288) (AASHTO T289) (AASHTO T290) (AASHTO T291) any method any method any method min 3,000 ohm/cm  5-10  200  100  * Test for chloride and sulfate not required if pH between 6-8 and minimum resistivity is > 5,000 ohm/cm.  46 Design life comparisons Table 1.7.9 shows a comparison of the different electrochemical limits imposed on structural backfill utilized in conjunction with buried galvanized steel structures. Table 1.7.9 Comparison of electrochemical limits for select backfills used in various countries (Elias, 1990). Property  U.S.(AASHTO)  France  G.B.  Germany  Min. resistivity (ohm/cm)  > 3,000  >1,000 dry * >3,000 wet **  >2,000  >3,000  pH  >5 but < 10  >5 but < 10  >6 but < 9  >5 but < 9  Chloride Content (ppm)  <100(AASHTO) <200(FHWA)  <200 dry <100 wet  <250  <50  Sulfate Content (ppm)  <200(AASHTO) <1000(FHWA)  <500 dry <1,000 wet  <250  <500  Sulfides (ppm)  —  <300 dry <100 wet  —  —  Organic Content (ppm)  —  >100  —  —  Biochemical Need ofoxygen  —  Minimal  —  —  — — Redox Potential + mV * Dry is upland structure ** Wet is structure repeatedly or permanently submerged  —  100-200  Table 1.7.10 illustrates a comparison of design lives applied to different classes of structures in U.S., France, Great Britain, and Germany. In Table 1.7.11 can be seen total sacrificial thicknesses on the buried side, presumed depleted by corrosion throughout the structure's service life span.  By  normalizing the sacrificial thickness with respect to a common design life, it is evident that the British rates of corrosion applied to aggressive sites are the most stringent of the compared corrosion rates, followed by U.S.'s rates of corrosion.  47  Table 1.7.10 Comparison of required design life for reinforced soil structures in various countries (Elias, 1990). Structure Classification  U.S. (AASHTO)  France  G.B.  Germany  Temporary Structure  —  30  —  —  Provisional Structure  —  5  —  —  Permanent Structure Abutment and Rail Supporting Structures  75  70  120  70  100  100  120  100  Table 1.7.11 Comparison of total required sacrificial thickness for galvanized steel reinforcement in select backfill to last the intended design life (Elias, 1990). Total sacrificial Thickness(buried side) (both galvanizing and steel included) (mm) Structure Classification  Temporary Structure Provisional Structure Permanent Structure  U.S. (AASHTO) *  France *  ...  0.25 dry 0.75 wet  —  —  0  —  —  —  0.95  0.5 dry  Abutment and Rail Supporting 1.35 0.75 dry Structures 1.0 wet Minimum galvanizing thickness is 86 u.m/side ** Minimum galvanizing thickness is 100 urn/side  48  G.B. **  1.85 - aggressive 0 .6 - nonaggressive 1.85 - -aggressive 0.6 - nonaggressive  Germany **  0.5 0.85  CHAPTER 2 MoTH'S STRUCTURES SURVEY 1994 2.1 Location selection Reviewing different reports prepared for various agencies in the United States led to the decision of limiting the sample location to Vancouver Island.  In a galvanized  culvert study conducted in Oregon, it was concluded that the only significant variable influencing the performance of galvanized steel culverts was their location west or east of the cascade mountain range, which is the divider of the State into wet and dry regions. It was found that in the drier Eastern Oregon galvanized steel culverts had performed satisfactorily, while in Western Oregon, where the climate is wet, galvanized steel deteriorated quite rapidly (Wolfe, Victor, Macnab, 1976) . 47  Vancouver Island, belonging to British Columbia's coastal climatic zone, receives substantially more precipitation than the interior, making it suitable as a "worst case scenario". Since a province-wide survey could not be carried out due to time constraints, this will provide us with an upper bound result. In addition, stream water alkalinity in streams of the surveyed area were reported to possess a low alkalinity, in the range of 030 ppm (Sullivan, Samis, 1985) . It is therefore expected that scaling tendencies in the 48  soft stream waters of the surveyed structures will be minor. Initially MoTH'S Bridge Information System (BIS) was used to select suitable culvert sites for inspection.  However, the BIS proved of limited usefulness for this  purpose in that it lacked such vital information as culvert type (wood, concrete, or multiplate) and installation year.  Hence, requests were distributed to Vancouver Island's 3  M O T H district offices as well as local Highway maintenance foremen for knowledge of multi-plate structures' locations and ages. Successful replies led to the sample size of 35 structures, with a fairly even distribution of locations and ages.  Table 2.1.1 displays  locations, and a frequency distribution of ages is presented in Figure 2.1.1.  49  Table 2.1.1 Location of structures examined through summer, 1994. NAME (ID #)  AGE  LOCATION  (WTS.)  H W 14 BILSTON CR.(MAIN RD.) BILSTON CR.(SIDE RD.) JORDAN RIVER + 4.5 KM PAT PHILLIPS (8064) CIRCO(8187) JORDAN RIVER + 9.6 KM JORDAN RIVER + 10 KM MILE 61 JORDAN RIVER + 12.8 KM LINES CR. (7084)  4 4 12 3 3 30 12 3 30 2  UNDER NEW 4 LANE HW STRETCH BY HUMPBACK RD; TOWARDS SOOKE NEXT TO BILSTON CR.(MAIN RD.) PIPE 4.5 KM FROM JORDAN RIVER TOWARDS PORT RENFREW 7.1 KM FROM JORDAN RIVER TOWARDS PORT RENFREW 7.4 KM FROM JORDAN RIVER TOWARDS PORT RENFREW 9.6 M FROM JORDAN RIVER TOWARDS PORT RENFREW 10.0 KM FROM JORDAN RIVER TOWARDS PORT RENFREW 12.3 KM FROM JORDAN RIVER TOWARDS PORT RENFREW 12.8 KM FROM JORDAN RIVER TOWARDS PORT RENFREW 16.0 KM FROM JORDAN RIVER TOWARDS PORT RENFREW  9  UNDER SEYMOUR AVE.  9  UNDER SEYMOUR AVE.  35 16  IN LADYSMITH; UNDER NEWLY CONSTRUCTED BRIDGE UNDER NANAIMO RIVER ROAD; 15 KM FROM HW 1  10 44 36 36 16 16 16 16 16 16  QUALICUM BAY; ON CORCAN RD. OFF DORMAN RD. OFF BAYLIS RD. DOWNTOWN BOWSER ON HW 19; NEAR GEORGIA PARK STORE SOUTH OF CAMPELL RIVER; UNDER HW 19; LKI: segm. 2340 Km 111.08 SOUTH OF CAMPELL RIVER; UNDER HW 19; LKI: segm. 2340 Km 112.85 APPROX. 38 KM NORTH OF SAYWARD RD. AND ISLAND HW. JUNCTION APPROX. 43 KM NORTH OF SAYWARD RD. AND ISLAND HW. JUNCTION APPROX 46 KM NORTH OF SAYWARD RD. AND ISLAND HW. JUNCTION NORTH OF TSITIKA RIVER; JUST SOUTH OF CROSSING LOGGING ROAD 300 M SOUTH OF RELOAD UNDERPASSES(2 BRIDGES CROSSING HW 19) 300 M NORTH OF TURNOFF TO WOSS HW YARD  20+  ON BOTTOM OF LAST HILL LEADING TO MT. WASHINGTON SKI AREA.  BULLERCR.(6775)  8  COWLEY RD. (6774) BEAVER CR. (6676) PRICE CR. (6679) SCHLEY (7894) HYDRO HILL WEST (7889) HIRSCH (7892)  20+ 20+ 8 27 2 2  3.5 KM WEST OF INTERSECTION OF HW 4 AND 4A ON HW 4 INTERSECTION OF KRISCOTT RD. AND CHATSWORTH RD. PORT ALBERNI; ON CHERRY CREEK RD. ACROSS FROM SCHOOL PORT ALBERNI; ON BEAVER CREEK RD. 150 M NORTH OF BAIN BRIDGE RD. PORT ALBERNI AREA; SPROUT LAKE; 1.1 KM EAST OF LAKESHORE DRIVE UNDER HW 4; 64 KM WEST OF PORT ALBERNI UNDER HW 4; 65 KM WEST OF PORT ALBERNI UNDER HW 4; 70 KM WEST OF PORT ALBERNI  VICTORIA  SEYMOUR AVE (WALK WAY) SEYMOUR AVE (BIKE PATH)  HW 1 HOLLAND CR. (8811) BOULDER CR. (8787)  H W 19 KINKADE (6766) THAMES RV. (6763) WILLOWS CR. SIMMS CR. ISLAND HW.+38 KM (6979) ISLAND HW.+43 KM (6981) ISLAND HW.+46 KM (6982) CROMAN (6984) WOODENGLE EAGLES NEST (6988)  COURTNEY PARADISE ARCH (7053)  HW 4  H W 28 LUPIN CR. (6784)  8  FALLS CR. #1 (6788)  25  FALLS CR. #2 (6787)  25  WESTMINSTER MINE RD.; BY LUPIN FALLS TRAIL; BETWEEN CAMPELL RIVER AND GOLD RIVER ON RD PAST GOLD RIVER TOWARDS PULP MILL; (EQUIPPED WITH SPRAY SCREEN ON SIDE OF ROAD) ON RD PAST GOLD RIVER; 300 M NORTH OF FALLS CR. #1  50  Figure 2.1.1  0  4 2  8 6  12 16 20 24 28 32 36 40 44 10 14 18 22 26 30 34 38 42 46  AGE (YEARS)  2.2 Field work 2.2.1 Core sample extraction Generally three metal cores were extracted from the structures, usually from the current wet/dry line, or the invert, and above the high water line at the 3 or 9 o'clock positions. The reason for the choice of these sampling locations is as follows:  when  sampling above the high water mark, the soil side is not affected by creek water seeping behind the culvert wall, avoiding raised moisture content and subsequently corrosion rates. Metal loss measured at these locations will therefore be representative of other buried structures not in constant contact with running water, e.g. bin walls and underpasses. Samples obtained from the wet/dry line or in the invert represents the worst conditions encountered in the case of water carrying multi-plate structures. Furthermore, the invert samples were sampled from the upstream faces of the corrugations, since these locations would likely sustain more metal loss than downstream faces because of the former's greater exposure to abrasives (Bellar, Ewing, 1984) . 49  51  Samples were generally obtained approximately 2 culvert diameters into the structure from the inlet or outlet side, depending on the most convenient accessibility. The 1.5 inch diameter cores were all drilled on the tangents (flat part) of the corrugations, and when zinc was present, the metal coupons were sampled from clean and uniform zinc surfaces. Drilling was performed using a forming tool guided by a pilot drill and driven by a fast speed electric drill, which in turn was powered by a gas generator. As water levels continued to decrease during the course of the field testing period due to dry weather, an extension shaft to the drill was made and used in drilling cores in the bottom of the invert of structures that were not lined with concrete.  Exact core  sample locations are depicted in Table 2.2.1 The sample holes were at first patch-repaired with galvanized steel patches. Their use was discontinued due to the time-consuming installment and poor durability. A patch that had been installed at Willows Creek 16 months ago displayed complete zinc loss and heavy rusting. Subsequently rubber plugs were used in the upper holes, and the invert and wet/dry line holes clamped with steel washers connected through a bolt to a steel bar.  52  TABLE 2.2.1 Visual inspection summary SAMPLE LOCATIONS  SITE  1&2  BILSTON  INVERT CROWN STREAM CONDITION CONDITION SEDIMENTS 3  (SIDERD.)  TOPOGRAPHY  LANDUSE  RUST  SPANGLE  BOULDERS  HILLY  EXPOSD. STEEL  WHITE RUST  COBBLES  URBAN  WTR. SPD. SPECIFIED DATE (m/s)(SLOPE) T H C K N . (MM) INSPECTED MILD  3.56  JUNE 27  MILD  3.56  JUNE 27  SEDIMENTS ALL CORES SAMPLED FROM SAMI• PLATE  ADDITIONAL INFO:  VEGETATION IN STREAM/ LOW W,VTER LEVEL 1&2  BILSTON  .  3  (MAIN RD.)  ADDITIONAL INFO: JORDAN R.  RUST  SPANGLE  BOULDERS  HILLY  EXPOSD. STEEL  WHITE RUST  COBBLES  URBAN  SPANGLE  BOULDERS  MOUNTAINOUS  WHITE RUST  COBBLES  RURAL  VEGETATION IN STREAM/ LOW WATER LEVEL 3 EXPOSD. STEEL  1&2  SEDIMENTS  + 4.5 KM ADDITIONAL INFO:  CONCRETE  SPANGLE  COBBLES  MOUNTAINOUS  (CRACKED)  WHITE RUST  GRAVEL  RURAL  JULY 14  STEEP  4.32  JUNE 28  STEEP  4.32  JUNE 28  MILD  4.32  JULY 15  MILD  4.27  JULY 14  INTERMEDIATE  4.32  JUNE 30  INTERMEDIATE  3.56  JULY 15  4.32  JUNE 30  3.32  JUNE 10  7.11  JUNE 10  CORE SAMPLE #1 AND # 3 SAMPLED FROM SAME PLATE  CIRCO  CONCRETE  SPANGLE  COBBLES  MOUNTAINOUS  (CRACKED)  WHITE RUST  GRAVEL  RURAL  SPANGLE  BOULDERS  MOUNTAINOUS  WHITE RUST  COBBLES  RURAL  NO SIGN OF WATER LEVEL EXEEDING CONCRETE  JORDAN R.  EXPOSD. STEEL  + 9.6 KM ADDITIONAL INFO:  4.32  INVERT BOLTS HARD HIT BY DEBRIS/ SOME BOLTS MISSING IN INVERT  PAT  ADDITIONAL INFO:  TOP:  BOTTOM: 7.11  PHILLIPS ADDITIONAL INFO:  INTERMEDIATE  GRAVEL  PITTING  BOLTS SLIGHTLY MARKED BY ABRASION  Y  NODULES OF RUST COVERING DEEP PITS ON DOWNSTREAM SIDE OF CORRUGATIONS IN INVERT  JORDAN R. + 10 KM ADDITIONAL INFO:  THIN ZINC  BOULDERS  MOUNTAINOUS  CARBONATE  COBBLES  RURAL  EXPOSD. STEEL  LAYER  PIPE(NOT MULTIPLATE) ; PIPE DIAMETER 2.5 METER MULTIPLATE EXTENSION CONNECTED AT OUTLET OF PIPE  MILE 61  CONCRETE  SPANGLE  BOULDERS  MOUNTAINOUS  (CRACKED)  WHITE RUST  COBBLES  RURAL  GRAVEL ADDITIONAL INFO:  ALL SAMPLES FROM SAME PLATE  t \  CONCRETE LINING PLACED CONVEX  JORDAN R. + 12.8 KM ADDITIONAL INFO:  LINES  EXPOSD.STEEL  SPANGLE  RUST(FLAKING)  WHITE RUST  BOULDERS  MOUNTAINOUS RURAL  NODULES OF RUST COVERING DEEP PITS ON DOWNSTREAM SIDE OF CORRUGATIONS IN INVERT  2  1  CONCRETE  SPANGLE  GRAVEL  WHITE RUST  MOUNTAINOUS RURAL  STEEP  FLAT  N/A  ADDITIONAL INFO:  SEYMOR  CONCRETE  SPANGLE  AVENUE  WHITE RUST  (WALKW.)  SOME RUST  ADDITIONAL INFO:  SEYMOR AVENUE (BIKEPATH) ADDITIONAL INFO:  cx 1.2&3  BIKEPATH  N/A  URBAN  SPANGLE  N/A  WHITE RUST  FLAT  N/A  URBAN  CORROSION PRESENT AT CONCRETE FOOTING/ STEEL PLATE INTERPHASE DUE TO RETENTION OF WATER  53  TABLE 2.2.1 Visual inspection summary (contd) SAMPLE LOCATIONS  SITE HOLLAND  INVERT CROWN STREAM CONDITION CONDITION SEDIMENTS  243 .  CR.  TOPOGRAPHY  LANDUSE  RUST  SPANGLE  COBBLES  HILLY  PERFORATIONS  WHITE RUST  GRAVEL  URBAN  W. SPEED (m/s)  SPECIFIED DATE THICKNESS INSPECTED  MILD  5.54  JUNE 28  INTERMEDIATE  TOP: 4.78  JUNE 16  SEDIMENTS SEA WATERINGULVERT AT HIGH TIDE  ADDITIONAL INFO:  BOULDER  is r"  REMOVED SUMN>IER 1994  \  EXPOSD.STEEL  SPANGLE  BOULDERS  HILLY  NOBOLTABRS.  WHITE RUST  COBBLES  RURAL  GRAVEL  BTM: 7.11  BAFFLES  ADDITIONAL INFO:  WATER DRIPPING OUT OF SAMPLE HOLE #2 AND #3  KINKADE  1&.2  .  SEDIMENTS  SPANGLE  COBBLES  HILLY  WHITE RUST  GRAVEL  RURAL  TOP: 5.54  JUNE 17  BTM: 4.32  SAMPLE #1 AND #2 TAKEN 5 CM ABOVE CURRENT WATER LINE  ADDITIONAL INFO:  WATER DRIPPING OUT OF SAMPLE HOLE #1 AND #2 THAMES  MILD  EXPOSD.STEEL  X ?  ADDITIONAL INFO:  RUST UNDER WET/DRY LINE  NO SPANGLE  COBBLES  HILLY  WHITE RUST  GRAVEL  RURAL/URBAN  MILD  5.54  JUNE 17  INTERMEDIATE  3.56  JULY 22  MILD  4.78  JULY 18  TOP: 4.32  JULY 18  SAND LOCATED 50 METERS FROM SEA  SAMPLE 1 A <1D 3 FROM SAME PLATE; SAMPLE 2 AND 4 FROM SAMPE PLATE PARADISE  1  2  ARCH  NATURAL  SPANGLE  STREAMBED  WHITE RUST  BOULDERS  MOUNTAINOUS RURAL  CONCRETE FOOTINGS EXPERIENCE UNDERMINING  ADDITIONAL INFO:  RETENTION OF WATER AT FOOTING/STEEL PLATE INTERFACE==> RUSTING WILLOWS  2  1  HEAVY RUSTING  NO SPANGLE  (FLAKING)  WHITE RUST  COBBLES  HILLY URBAN  SEA WATER IN CULVERT AT HIGH TIDE  ADDITIONAL INFO:  SAMPLES OBTAHMED FROM SAME PLATE SIMMS  i  CONCRETE  SPANGLE  GRAVEL  HILLY  (PREVIOUSLY  WHITE RUST  SAND  URBAN  MILD  PERFORATED) POSSIBLE TIDAL ACTION AT HIGH TIDE  ADDITIONAL INFO:  BULLER  BTM: 3.56  ....  14.2  3 EXPOSD.STEEL RUST  SPANGLE  GRAVEL  FLAT  WHITE RUST  SAND  RURAL  SPANGLE  COBBLES  FLAT  WHITE RUST  GRAVEL  URBAN  MILD  2.82  JUNE 20  MILD  4.78  JUNE 22  MILD  4.78  JUNE 22  INTERMEDIATE  3.56  JUNE 20  3.56  JUNE 21  SEDIMENTS ADDITIONAL INFO:  COWLEY  2  CONCRETE  ROAD  SAND ADDITIONAL INFO:  CONCRETE LINE DIN 1993  2  BEAVER CREEK  HEAVY RUST  SPANGLE  COBBLES  FLAT  EXPOSD.STEEL  WHITE RUST  GRAVEL  AGRICULTURAL  ADDITIONAL INFO:  PRICE  EXPOSD.STEEL  SPANGLE  COBBLES  MOUNTAINOUS  CREEK  SEDIMENTS  WHITE RUST  GRAVEL  RURAL  SAND PIPE HAS PERFORATED IN INVERT AT OUTLET END DUE TO WATER IN POOL AT OUTLET  ADDITIONAL INFO:  1&2  SCHLEY  C ADDITIONAL INFO:  CONCRETE  SPANGLE  GRAVEL  MOUNTAINOUS  WHITE RUST  SAND  RURAL  WATER DOES NOT APPEAR TO HAVE EXCEEDED CONCRETE LINING  54  STEEP  TABLE 2.2.1  SITE  Visual inspection summary (contd)  SAMPLE LOCATIONS  INVERT  CROWN  STREAM  CONDITION CONDITION S E D I M E N T S HYDRO  CONCRETE  SPANGLE  BOULDERS  HILL WEST  TOPOGRAPHY  W. S P E E D  LANDUSE  (m/s)  SPECIFIED  MOUNTAINOUS RURAL  INTERMEDIATE  CLEARCUT ADDITIONAL INFO:  TOP: 4.78  JUNE 21  BTM7.11  RETENTION OF WfATER AT CONCRETE/STEEL PLATE INTERFACE  HIRSCH  CONCRETE  SPANGLE  COBBLES GRAVEL  MOUNTAINOUS RURAL  INTERMEDIATE  TOP: 4.78  JUNE 21  BTM-7.11  CLEARCUT ADDITIONAL INFO:  DATE  T H I C K N E S S INSPECTED  BAFFLES  LUPIN  EXPOSD.STEEL  SPANGLE  BLOULDERS  MOUNTAINOUS  SEDIMENTS  WHITE RUST  GRAVEL  RURAL  MILD  4.32  JULY 19  INTERMEDIATE  3.56  JULY 19  STEEP  3.56  JULY 19  TOP: 2.82  JULY 20  SAND ADDITIONAL INFO:  NO WATER IN CULVERT DURING INSPECTION ZINC PRESENT ON WATERSIDE IN INVERT UNDER S  R SEDIMENTS  EXPOSD.STEEL  SPANGLE  COBBLES  MOUNTAINOUS  WHITE RUST  SAND  RURAL  FALLS CR. #1  ADDITIONAL INFO:  STEEPER THAN FALLS CR. #2 BACKFILL HAS SLID OUT AT INLET AND EXPOSED PARTS OF THE SOIL SIDE OF PIPE  FALLS CR.  EXPOSD.STEEL  #2  ADDITIONAL INF D:  SPANGLE  COBBLES  MOUNTAINOUS  WHITE RUST  BOULDERS  RURAL  STEEPER THAN FALLS CR. #2 BACKFILL HAS SLID OUT AT INLET AND EXPOSED PARTS OF THE SOIL SIDE OF PIPE  ISLAND HW  1 2  EXPOSD. STEEL MOUNTAINOUS  + 38 K M  SEDIMENTS  RURAL  BOULDERS  MOUNTAINOUS  COBBLES  . RURAL  INTERMEDIATE  BTM7.U ADDITIONAL INF y.  BOULDERS UP TO 1 METER DIAMETER PILED UP IN PIPE  BOLTS MARKED HEAVILY BY ABRASION  CORE SAMPLE #1 AND #2 SAMPLEDFROM THE SAME PLATE ISLAND HW  EXPOSD.STEEL  MOUNTAINOUS  BOULDERS  MOUNTAINOUS  + 43 K M  RUST  RURAL  COBBLES  RURAL  INTERMEDIATE  TOP: 3.56  JULY 20  BTM7.11 ADDITIONAL INFO:  BAFFLES NODULES OF RUST COVERING DEEP PITS ON DOWNSTREAM FACING SIDE OF CORRUGATIONS IN INVERT  ISLAND HW  EXPOSD.STEEL  + 46 K M  MOUNTAINOUS  COBBLES  RURAL  MOUNTAINOUS RURAL  STEEP  TOP: 3.56  JULY 20  BTM7.11 ADDITIONAL INFO:  CROMAN  ADDITIONAL INFO:  NO SOIL PRESENT DIRECTLY BEHIND THE SAMPLE HOLE  dY 1  2  EXPOSD.STEEL  MOUNTAINOUS  COBBLES  MOUNTAINOUS  SEDIMENTS  RURAL  GRAVEL  RURAL  MILD  TOP: 3.56  JULY 20  BTM: 2.82  BAFFLES BOTH SAMPLES OBTAINED FROM SAME PLATE  WOODENGLE  1 2  EXPOSD.STEEL  MOUNTAINOUS  BOULDERS  MOUNTAINOUS  SEDIMENTS  RURAL  COBBLES  RURAL  MILD  4.32  JULY 21  4.32  JULY 21  GRAVEL ADDITIONAL INFO:  NO SOIL PRESENT BEHIND SAMPLE #1 AND #2  CULVERT HAS SAGGED AT OUTLET  SAMPLE #1 AND #2 OBTAINED FROM SAME PLATE EAGLES  RUSTING  MOUNTAINOUS  LAKE  NEST  (FLAKING)  RURAL  UPSTREAM OF  RURAL  INLET SIDE  MARSH  ADDITIONAL INFO:  PITTING UNDER WATER BOTH SAMPLES OBTAINED FROM SAME PLATE  55  MOUNTAINOUS MILD  2.2.2 Coating thickness Zinc coating thickness was measured next to each core sample location where zinc was present. Twenty-five readings were taken at each location in a 5x5 grid within an area of 25x25 mm. Coating thickness was measured using a portable Elcometer 256 Microprocessor Coating Thickness Gauge, which has the ability of retaining the minimum, maximum-, mean-, and standard deviation values of each batch of 25 readings.  The  meter was periodically calibrated using calibration foils of known thickness. Care was taken not to include measurements caused by additional substances, such as dirt present on the coating surface; i f abnormally high standard deviations were recorded, the entire batch of readings was retaken.  A l l readings were taken on the tangents of the  corrugations since the curved surfaces proved to influence the accuracy of the readings. A quick test conducted  showed readings  increased by  measurements were taken on top of a corrugation.  10-15 micrometers when  This was probably due to the  calibration procedures used in calibrating the Elcometer Thickness Gauge, which was conducted on a flat piece of metal. 2.2.3 pH collection Water p H was determined using a standard field p H meter.  A plastic cup was  rinsed with the sample water and a sample extracted and tested on site. The p H meter was left in the cup until a stable reading was obtained. For determining soil pH, a sample of the backfill taken directly from the metal core hole was generally used.  However, due to the presence of large rocks located  directly behind the core hole at many locations, obtaining a representative backfill sample proved to be either difficult or impossible. In such cases a soil sample close to the structure was obtained. Soil p H was determined according to the California Test Method (American Iron and Steel Institute, 1983) . The method specifies 2 rounded teaspoons of soil mixed with 50  56  2 teaspoons of distilled water. Soil was then dispersed by stirring , and tested by a field pH meter. The p H meter was allowed to stabilize before a reading was recorded. 2.2.4 Water and soil sample extraction A water sample for lab p H and resistivity was collected in each stream, using a 2 liter bottle rinsed twice with sample water before filling. A soil sample was collected in standard soil sample plastic bags and sealed with tape.  The soil sample was approximately 2-3 kilograms in order to accommodate  laboratory soil box resistivity measurements which require 1.3 kg of dry sieved soil. The soil samples were extracted as close to the structure as possible. Once a soil sampling location had been determined, the topsoil was removed in order to exclude any organic material, which would not be characteristic of the true backfill, before the soil sample was acquired. In addition, it was desired that smaller soil samples of approximately 100 grams be extracted from the actual backfill behind the drilled metal core hole.  This proved a  difficult task, however, since much of the backfill contained material of large grade which often was found to block the access to the wanted soil.  Whenever backfill could be  obtained from the core sample holes, it was stored in small containers until tested in the laboratory. 2.2.5 Water speed The water speed was calculated by measuring the travel time of a cork between 2 points of a known distance. Three trials were timed and averaged to obtain a mean water speed. As the weather remained dry throughout the field investigation period, the water speed measurements were made impossible in many structures due to low water levels.  57  2.2.6 Visual inspection A visual inspection of each structure was also performed. Information recorded included: -Topography -Land use -Stream bed sediments present -Crown condition - white rust present -zinc condition -Invert condition -high water mark location -corrosion type -pitting -surface rust -exposed steel -abrasion -sediments present This information, along with other information pertinent to the study which was also recorded, can be viewed in Table 2.2.1. 2.2.7 Photo documentation A series of photographs showing the overall structure and its internal condition was obtained from each location. 2.3 Laboratory work Analysis of the soil and water samples was performed at the M o T H Geotechnical Laboratory in Victoria, B.C. 2.3.1 Soil moisture content The soil samples were transferred into pre-weighed pans upon arrival at the laboratory, and left to dry in the oven at ca. 200 degrees centigrade after weighing the soil filled pan. A drying period of 24 hours was deemed sufficient for all moisture to escape. The samples were left to cool, and then weighed again in order to calculate moisture content.  58  2.3.2 Soil resistivity The soil's resistivity was first measured according to the California Method. This method requires determination of resistivities at various levels of soil moisture, and is a very time-consuming procedure. In addition, the method exhibits inherent problems such as changing repacking density, uneven moisture distribution, equilibrium time, electrode contact variation by packing density and textural differences, which lead to a low degree of accuracy and reproducibility. The California Method of measuring soil resistivity was discontinued and a better method was sought to reduce the testing time and increase the accuracy of the results. A preferred method for laboratory analysis was found in "Method of Soil Analysis - Part 2" from the American Society of Agronomy (Page, 1982) , which uses an aqueous 51  extract on a 2:1 soil-water ratio. This procedure allowed for soil quantities of less than 0.5 kg in order to extract enough liquid for testing. After the extraction, the electrical conductivity of the solution was measured using a Y S I Model 32 Conductance Meter with a 3403 (k= 1.0/cm) Conductance Cell from Yellow Springs Instrument Co., Inc. The meter was allowed to stabilize before each reading was taken. reciprocal of conductivity, was then easily calculated.  Resistivity, being the  The conductivity meter was  calibrated against a 0.5 mol/1 NaCl solution with a conductivity of 42 milli-mhos. This relatively simple and rapid test procedure is very accurate and extremely reproducible.  Also, this procedure determines resistivity in the most adverse case  (saturated), a comparable resistivity independent of seasonal and other variations in soilmoisture content. 2.3.3 Water resistivity The water conductivity was measured using the same conductance meter and conductance cell that were used for the soil resistivity measurements. The cell was kept in  59  solution until the reading had stabilized. Conductivity data were then converted to resistivity. 2.3.4 p H Water and soil p H values were remeasured in a more controlled laboratory environment. The lab p H values were not in agreement with measurements taken in the field. Overall, the lab value for p H had dropped compared to p H values obtained in the field.  The p H changes over time can be contributed to C O dissolution and micro 2  biological action in soil and water. The extent of pH change for both water and soil was on average approximately 1. The laboratory p H data set were subsequently rejected. For future testing, due to possible changes of the overall p H after sampling, it is recommended to take p H measurements on site. 2.3.5 Coupon evaluation To simplify the evaluation of the condition of each of the 94 coupons extracted, each coupon's exterior and interior surfaces were photo-documented. Every coupon extracted that contained any trace of zinc was tested for coating thickness. Zinc tends to build up a tightly adherent layer of corrosion products consisting of basic zinc carbonate when present in the atmosphere. This layer is present on the air side of the coupons, and a less adherent modification on the soil side. These corrosion products, due to their composition, occupy larger volumes than the original zinc and thus had to be removed before the true thickness of the zinc could be measured. A sufficient method for removing zinc corrosion products, while not affecting the true zinc, was to immerse the coupon in a 10 % acetic acid solution for 10 minutes.  The coupon was  removed and the sides containing zinc bristle brushed using a hard non-metal brush. The coupon was re-immersed for another 10 minutes, brushed again, rinsed in water, and dried. Thickness measurements were obtained using the same Elcometer 256 Coating Thickness Gauge used in the field. To preserve consistency with field data, 25 readings  60  were obtained on each side of the sample coupon. Readings were taken midway between the pilot drill hole perimeter and the coupon perimeter, at each hour mark until 25 readings were taken. When working with acids, proper safety was assured by always working in a fume hood, and by wearing proper hand and facial protection. Coupons of bare steel with small amounts of surface rust were bristle brushed and rinsed in water.  Specimens containing heavy rust deposits were immersed in 50 %  hydrochloric acid for 30 minutes, then removed and bristle brushed. A sharp object was used to remove rust in pits. This dipping and cleaning action was continued until all corrosion products were removed. All the cores, including the ones containing zinc, were then tested for overall thickness. A digital caliper specially equipped with arms capable of measuring pit depth was used for thickness testing.  Measurements were obtained midway between the  perimeter of the pilot drill hole and the coupon perimeter at eight equally distributed locations around the core sample.  These eight measurements were then averaged to  obtain an overall mean coupon thickness. In addition, for each metal sample , the deepest pit was located and a minimum thickness recorded. All this information was compiled in this study's preliminary report.  61  CHAPTER 3 RESULTS AND ANALYTICAL PROCEDURES 3.1 General Of the 35 structures from which metal samples were taken, 32 experienced continuous flow of water throughout the year. One culvert, Lupin Creek, was dry at the time of inspection. However, it was apparent that large flows of water were periodically experienced during snow melt and storms due to the large boulders and smaller debris found within the culvert.  Two of the structures surveyed, located in Victoria, were  underpasses, namely the Seymour Avenue bike path and walkway. Concrete invert protection was found in all 7 culverts installed within the last three years.  In addition, 2 culverts which had shown signs of deterioration had recently  received an invert concrete protection layer. These were Simms Creek and Cowley Road culverts. Culvert grades, which were not measured quantitatively, varied from virtually flat (Kinkade Culvert) to very steep (Pat Phillips) (> 15 %), with most having mild and intermediate slopes (< 5 %). Abrasions, ranging in degree of severity, were noted in virtually all culverts, due to abrasive bedload present in all streams. Debris ranged in size from sand and gravel to boulders up to one meter in diameter. In the steeper culverts that were not concrete lined, invert bolts were badly deteriorated and sometimes missing. This was a clear indicator of large bedloads being carried through the structure at high flows. Soil sampled from behind the structures was typically found to be granular backfill material without any inclusions of organic material.  Some of the structures, and in  particular Seymour Avenue underpass and Thames River Pipe where found to have a very uniformly graded sand backfill.  62  The inside of the crown, only affected by air, showed no signs of deterioration, even in the oldest culverts. White storage stains, which were present in most installations, did not seem to have affected the zinc's ability to protect the underlying steel. Two general regimes with different deterioration rates were noted. One was the soil side of the steel not affected by creek water. Here zinc was still present in almost every structure. The second regime was the more attacked area of the culvert in contact with stream water. This constituted both the inside, water-exposed face, and the invert soil side in constant contact with backfill saturated by stream water. Overall, the structures generally appeared in sound condition, due to their relatively young age (2-44 years).  However, some structures were experiencing  somewhat rapid invert deterioration because of the environment in which they existed. 3.2 Corrosion parameters 3.2.1 pH A rather narrow neutral range of pH's was found for both soil and streams in contact with the surveyed structures. The average soil p H for all the sampled sites was 7.0, with minimum and maximum values of 6.0 and 7.7, respectively. Stream pH's were found to be overall slightly more acidic than the soil, with a mean of 6.3 and minimum and maximum values of 5.9 and 7.8, respectively. Frequency distributions of the in situ p H of soils and stream waters can be seen in Figure 3.2.1 a) and b), respectively. Figure 3.2.1 pH frequency distribution for soil and water samples obtained in survey. a) ; b) FREQUENCY DISTRIBUTION  F R E Q U E N C Y DISTRIBUTION  WATER pH  SOIL pH  ui  6  6.25 6.5 6.75  7  7.25 7.5 7.75  V)  4  O  2  8  1 llllll 6  6.25 6.5 6.75  7  7.25 7.5 7.75  WATER pH  SOIL pH  63  8  All p H values measured lay within the p H limits utilized in the new design standards to determine the service life for soil-steel structures discussed in Chapter 1. Since only very low values will show significant effects of metal corrosion, p H is not a likely contributing factor in the rapid degradation observed in some of the B.C. structures. 3.2.2 Soil moisture content During the site investigations it was noted that the soil samples obtained from next to the multi-plates contained increasingly less moisture as the summer progressed. Moisture content is a fluctuating variable dependent on incidence and intensity of precipitation, and hence difficult to use as a service life indicator. Also, the soil sampled adjacent to the structure, normally in the upper layer of the soil, was found to have a lower moisture content than backfill sampled from the coupon holes. Therefore, in order to get a soil sample with moisture content representative of the one directly behind the culvert (which controls in-situ resistivity and to a large extent the corrosion), the sample should be obtained from sample holes drilled in the structure. By measuring the moisture content of the soil samples obtained from the coupon holes, it was generally noted that the holes drilled directly above the current water level contained backfill of a higher moisture content than those obtained from the upper sample holes which were not affected by the stream water. One extreme example of this was noticed at the Falls Creek #1 structure. Midway into the culvert, water was flowing out from 2 holes located approximately 20 cm above the current wet/dry line.  Thus it is  apparent that ground water can have a tendency to find its way behind the culvert wall, raising the water table above the current inside water level.  This condition has likely  transpired due to highly compacted soil in the culvert invert that forces the water into an alternate route of less compaction further up the culvert wall. This moistening of the soil reduces soil resistivity to its lowest level and increases corrosion rates.  64  It was also noted that the Seymour Avenue backfill had one of the lowest moisture contents (3.7%) compared to the other backfills sampled from the coupon holes. This lack of presence of water in the backfill is likely due to the headwall concrete barrier above the underpass which limits the access of rain and drainage water behind the galvanized steel reinforcement.  Also, the uniformly graded, well draining sand found  behind every coupon hole plays a role in the low moisture content experienced.  The  Seymour Avenue underpass possesses the second lowest minimum soil resistivity (3100 ohm/cm, which was measured in soil obtained from the coupon hole) in this survey, and a higher corrosion rate would presumably have been encountered if water had been given free access to the backfill. With such low moisture contents present, the effective on-site soil resistivity can be orders of magnitude lower than the minimum resistivity measured in the laboratory, which is obtained at saturation level. 3.2.3 Resistivity Distributions of this survey's measured minimum resistivities for soil and stream water are depicted in Figure 3.2.2. Figure 3.2.2 DISTRIBUTION OF RESISTIVITIES OBTAINED IN B.C.  •  SOIL FES IS W H Y  O WATER RESISTIVITY WATER RESISTIVITY: # of measurements: 32 minimum value: 5200 maximum value: 71400 mean value: 25000 standard dev.: 14400  1000  10000  1  LOG (RESISTIVITY) (ohm-cm)  65  SOIL RESISTIVITY: # of measurements: 35 minimum value: 2800 maximum value: 62500 mean value: 15400 standard dev.: 16100  The soil resistivity measurements shown in Figure 3.2.2 are all measured in the laboratory at, or somewhat beyond, the saturation point. Therefore, the soil resistivities are minimum values, independent of seasonal water content fluctuations experienced in the in-situ backfill. One soil sample was obtained from each site, and its measured resistivity is represented by a point on the graph. According to the corrosion classification system based on resistivity depicted in Table 1.2.1, all soil and water resistivities obtained in this survey were found to be in the moderate to non-corrosive range. Values of resistivity of the soil and backfill surrounding the structures were generally found to be lower than stream resistivities. The lowest soil resistivity obtained from the sampled sites was 2800 ohm/cm at Falls Creek #1, whereas the highest was measured to 62500 ohm/cm obtained from Thames River Pipe, which also was the oldest pipe surveyed. The 25 year old Falls Creek #1 structure, located in the high rainfall area of Gold River, still displayed zinc present on the soil side though traces of rust were observed. The Thames River Pipe was backfilled with granular beach sand and had still an excellent protective layer of zinc present after 44 years of service. Minimum soil resistivities of such a high value as observed in this survey are in agreement with reported measurements obtained from other projects where the backfill contains very small amounts of soluble ions. A culvert survey conducted across the U.S. found minimum resistivities (measured at saturation) as high as 111,000 ohm/cm (Corrpro Co. Inc., 1991) . 52  The lowest stream water resistivity was measured to be 5200 ohm/cm in Simms Creek, and thus was classified as mildly corrosive, whereas Island HW. + 46 km possessed the highest measured minimum resistivity value, 71,400 ohm/cm. Simms Creek Culvert, was found in an earlier study to have excessive perforations throughout its invert. This could result from the vicinity to the ocean, where sea water is likely introduced at high tides. In the Island HW. + 46 km structure, complete loss of zinc was observed in the invert. However, no pitting, and few rust marks were detected throughout the culvert. 66  Zinc loss can most likely be attributed to abrasion in this culvert of grade greater than 10 %. Overall, minimum soil resistivities were found to be lower than the resistivities obtained from the streams.  This is made apparent in Figure 3.2.3, which shows a  distribution of soil/water resistivity of all structures surveyed according to corrosiveness. Nevertheless, inverts were always found to be attacked more rapidly by corrosion than the upper parts of the structure, only exposed to soil. This is likely due to the fact that the water resistivities measured in the laboratory are as experienced in situ, whereas the soil resistivities measured at saturation level, are not representative of the drier soils, of higher resistivity, encasing the structural steel in the ground. Figure 3.2.3 MINIMUM RESISISTrVTTY DISTRIBUTION ACCORDING TO CORROSIVENESS  IB Soil I! Water  very corr. moderately non-corr. corrosive mildly corr  Compared to other studies conducted in the U.S., the surveyed structures had very high minimum soil resistivity values in many backfills installed before the introduction of more stringent backfill requirements. This has to do with the climate of western British Columbia, which is one of high annual rainfall.  Over the last few thousands of years,  much of the soluble salts causing lower resistivities has leeched out due to prolific rainfall 67  and percolation of water through the soil lattice to great depths, leaving the soil highly resistant to ionic movements. Only one backfill resistivity did not conform to the U.S. standard with respect to the limit of 3,000 ohm/cm for minimum soil resistivity. Currently, no newer U.S. standard was formed that requires a limit on water resistivity. The British system classifies the conductivity of water only with respect to chloride and sulfate concentrations (Table 1.7.6), none of which was measured in this study. However, there is reason to believe that the classification scheme for soil is not adaptable to water in all instances. In the geographic area of this study, very few soluble ions in the stream water are present as protective salts, such as calcium and magnesium (Sullivan, Samis, 1985). When relatively low stream water resistivities are found, they are most likely the result of aggressive salts, such as chloride and sulfate. A galvanized steel study conducted by Corrpro Companies, Inc. all over the U.S in 1987, reported a water sample resistivity of 5200 ohm/cm with a chloride content of 109 ppm, which places it in the aggressive range within the British water classification system (Blonska, 1987) . 53  3.2.4 Microbiological corrosion No tests were conducted to prove any existence of bacterial corrosion. However, the Eagles Nest culvert, which also showed the second largest invert corrosion rate, possessed a surrounding environment which could indicate the presence of bacterial corrosion. The heavily corroded invert showed nodules of rust at the present water table (Figure 3.2.4), which is one indication of sulfate-reducing bacteria.  Also, the culvert  drained a shallow lake where a lot of organic growth was observed (Figure 3.2.5). Other culverts in the same area between Sayward and Port Hardy, mostly appeared to contain clear steel throughout their inverts, with much lower deterioration rates than those observed in Eagles Nest culvert.  68  Figure 3.2.4 Heavily corroded invert of Eagles Nest culvert displays nodules of rust.  69  3.3 Structural condition 3.3.1 Inside condition - above the high water mark The inside areas not affected by creek water showed no significant signs of degradation. N o structures showed signs of rust, and most displayed original spangle, representative of newly galvanized plates. However, virtually all sites displayed presence of wet storage stains, which seemed to have no effect on the protection offered by the zinc coating. The oldest structure surveyed, the Thames River Pipe, still exhibits spangle and the inside appears to be in sound condition after 44 years in service, with a measured zinc thickness of 96 u,m. This culvert is also located approximately 50 meters from the ocean. Based on Thames River Pipe's excellent condition despite its proximity to the ocean, and coupled with the pristine condition of the air exposed inside of all the other surveyed structures, atmospheric corrosion does not appear to govern the durability of any of the structures inspected.  The excellent behavior of the zinc coating can most likely be  attributed to the lack of moisture accessibility to these particular well-shielded areas. Corrosion products removed from the inside of the coupons obtained above the high water mark showed an average thickness of only 3 u,m. This demonstrates that the measured zinc coating thicknesses are close to the original values applied. Figure 3.3.1 shows a frequency distribution of the measured zinc coating thicknesses measured after corrosion product removal from coupons obtained above the high water mark.  The  average thickness measured was 70 u,m, with minimum and maximum values of 49 u,m and 103 u.m, respectively. All thicknesses measured were therefore greater than the 43 u. m of galvanized coating, or 1 oz/ft. per side of galvanizing, currently specified by M o T H as the minimum requirement for original plate coating thickness. The resolution of the measurements taken with the Elcometer Thickness Gauge was to the closest 1 |im.  70  Figure 3.3.1  FREQUENCY DISTRIBUTION OF INSIDE ZINC COATING THICKNESS  40 50 60 70 80 90 100 45 55 65 75 85 95 ZINC COATING THICKNESS (micro-meter)  3.3.2 Soil side condition - above the high water mark Soil side inspection of coupons not affected by the creek water revealed substantial amounts of zinc coating still remaining on new as well as old structures.  Only 4 of 30  structures sampled in areas not affected by creek water, showed complete zinc loss. One of these structures was 12 years old and the others 20 years or older. The 44 year old Thames River pipe showed presence of soil side zinc, most likely due to the high resistivity sand used as backfill for the structure. At the 25 year old Falls Creek #2 structure, part of the backfill at the entrance to the culvert was removed due to erosion. After removal of more backfill covering parts of the pipe, an area affected by soil corrosion was visible which displayed rust marks with most of the zinc coating still intact, protecting the steel.  71  One example indicating the exemplary condition of the soil side being not in contact with stream water was the Holland Creek culvert. This 35 year old multi-plate pipe had recently been removed and a new bridge erected. The removed pipe section was still on site, and a few general observations about the soil side crown were noted during this field visit. The crown soil side showed only small signs of degradation. Traces of zinc were still present on the soil side, but surface rust was observed on most of the crown soil side, with no pitting noted. Three square holes had been cut in the crown to assist in the removal of the pipe. The plate thickness at the locations of the holes was measured to be within 5.46 to 5.52 mm with the digital calipers, and thus no significant metal loss was observed when compared to the original thickness of 5.45 to 5.70 mm obtained at the culvert's end plates. Taking coating thickness measurements of the coupons before and after removal of corrosion products resulted in an overall mean corrosion product loss of 25 urn from the soil side. This gives rise to the belief that zinc corrosion products adhere quite well to the zinc surface and thus offer excellent barrier corrosion protection and extend the life of the underlying zinc at locations not affected by continuous flows of creek water. The initially higher rates often experienced with corrosion of zinc are due to the initial degradation of the unprotected pure zinc before the corrosion product protection come into effect.  3.3.3 Invert inside (water side) condition - Unprotected inverts Varying degrees of invert deterioration were observed.  Some culverts showed  heavy rusting and deep pits throughout, whereas others displayed only light surface rust and mostly clear exposed steel. None of the coupon-sampled culverts, except Holland Creek, showed signs of invert perforations.  A smaller water-carrying installation in  Courtney was observed to be completely invert perforated. However, this structure was concrete paved only days after inspection by the local highway maintenance crew.  72  The survey indicated that the galvanizing offered very little protection for the underlying steel in the inverts. Complete zinc loss, only a few years after installation, was apparent at virtually all sites surveyed. The two 4 year old structures in Bilston Creek displayed complete zinc loss and pitting of the steel when inspected. Only one site, an 8 year old Lupin Creek culvert of intermittent flow, showed presence of zinc on the water side in the invert. However, the zinc appearing here was only present under sediments deposited in the culvert, presumably during the first couple of years of service. Premature zinc loss is most likely due to abrasive bedload being brought through the structure during storm and snow melt runoffs. The high water mark is defined as the zinc/steel interface (the extent of the zinc coating along the inside of the pipe) (Figure 3.3.2), which is a measure of bedload transportation and frequency and amount of snow melt and winter flow levels. The high water mark is not the extent of max flood level, but rather represents the regular winter and spring flow level of a pipe. At this particular location, the greenish zinc/steel alloy produced during galvanization can be observed.  The high water mark was found  anywhere from 30 cm above the invert as seen in Figure 3.3.2, to the extreme of approximately 1.5 meters above the invert in Boulder Creek structure. In the majority of the structures, the high water mark was observed about 30 to 50 cm above the invert.  73  Figure 3.3.2 High water mark indicated by steel/zinc transition in Crowman Pipe.  Almost all water-carrying structures showed signs of abrasion due to transport of bed sediments through the invert. This was noticed by areas of exposed steel on the inletfacing sides of the corrugations  Heavy exposure to abrasion was generally noted by  observing the invert bolts, which were often found to be quite deformed, and in extreme cases missing (Figure 3.3.3). In certain structures, the corrugations in the plates were deformed because of large boulders of diameter up to 1 meter (Figure 3.3.4) hitting the invert plates while being carried through the culvert. This can be noticed in Figure 3.3.3, which is taken from Kameko Culvert along Island Highway, a culvert of intermittent flow, and was therefore not sampled during this survey. Figure 3.3.2 also shows the presence of baffles, which are installed to allow for easier passage of fish through the structure. Sediment retention behind the baffle blocks was a good indicator of the kind of bedload that would be transported through these culverts every year.  74  Figure 3.3.3 Extreme bolt damage noticed in water carrying structures exposed to large boulders.  75  Many culverts showed in varying degrees the pitting of the structural steel in the invert. Pitting was generally noticed under nodules of rust, which mostly appeared on the outlet facing end of the corrugations in structures susceptible to abrasion (Figure 3.3.5). Pitting, however, is not the best measure of durability for steel drainage structures. On the other hand, in pipes flowing under pressure, such as water mains, pitting rate would determine the life of the pipe. Average metal loss is a better measure of remaining life of Figure 3.3.5  Pitting corrosion on down-stream side of corrugations.  drainage structures experiencing only gravity flow.  In drainage structures, only  approximately 13 percent of the overall thickness is lost at time of perforation according to studies conducted by U.S.'s NBS (Bednar, 1989) . They further concluded that at 26 54  percent metal loss, or twice the time to perforation, the end of a drainage structure's service life would be reached  This should be somewhat questionable, since from the time  of perforation, erosion and undermining of the underlying backfill through the perforations  76  could cause the structure to become unstable. One smaller installation running under HW. 19 in Royston, south of Courtney, showed excessive perforations in the invert. Undermining of the culvert, with excessive loss of backfill halfway through its length, was apparent during inspection.  Stream water was leaking out through perforations and  eroding and undermining the underlying backfill (Figure 3.3.6).  Figure 3.3.6 Heavy perforation in Roy Creek culvert.  77  The structures constructed close enough to the shoreline to be affected by sea water at high tides, displayed heavy rusting due to accelerated corrosion from chloride ions in the intrusive water. Two culverts affected by sea water at high tides, Holland Creek (5.54 mm invert plate thickness) and Simms Creek (3.56 mm invert plate thickness), which were 35 and 36 years old, respectively, were both perforated at numerous locations throughout the invert. Another structure which showed heavy corrosive attacks due to contact with sea water was 36 year old Willows Creek culvert located close to Simms Creek culvert. This culvert, constructed with 4.78 mm thick plates, was not yet observed to be perforated, but the overall invert thickness was reduced measurably. The smaller, Roy Creek culvert in Royston and Beaver Creek culvert in Pt. Alberni displayed the highest amount of invert loss of all water carrying culverts. Both structures were also observed to carry water from nearby farmlands, which most likely explains their poor invert conditions. Liquid manure has the ability to reduce water resistivity to as low as 200 ohm/cm if concentrated enough. In addition, fertilizers can contain ammonium sulfate, which are very aggressive towards steel. However, ammonium sulfate are not typically used as fertilizer in high rainfall areas. 3.3.4 Invert inside(water side) condition - Concrete protected inverts Lining the invert with concrete reduces the corrosive attacks from the stream water and shields the steel from wear due to abrasive bed sediment transported through the water-carrying structure. The barrier's function is to shield the steel from abrasive bedload and limit the stream water contact to only periods of very high flow. It is thus important that the stream water does not overflow the concrete lining at regular flows, and that the lining is designed in a concave fashion in order to divert abrasive bedload into the middle of the pipe and minimize sediment contact with the steel at high flows. Invert protection by lining the invert with concrete has only recently been introduced in newly erected drainage structures in B.C. Nine of the multi plate culverts  78  sampled in this survey were concrete lined, and 7 of them were installed within the last 3 years. Cowley Road and Simms Creek culverts were older structures which were recently concrete lined due to excessive corrosive attacks occurring in the invert. It was clear from the observed concrete linings already in place, particularly along HW. 14, that no standards currently exist regarding method of construction of the protective concrete lining. At the Mile 61 culvert, the concrete was laid down in a convex fashion, tunneling both sediment and stream water over to the steel concrete interface, causing excessive abrasion (Figure 3.3.7).  Also, some thinner concrete inverts were  noticed to be cracked at many locations throughout the structure. In other culverts, the extensions of the concrete up the sides of the structures were clearly insufficient to offer the proper protection required. Figure 3.3.8 shows the McKay culvert one year after installation.  The zinc is starting to disappear rapidly, with only the zinc/steel alloy  remaining at the concrete/steel interface.  Figure 3.3.9 displays the relative limited  amounts of concrete lining placed in the structure. The newly installed 2 year old culverts along H W 4 possessed concrete protective linings which seem to perform relatively well, with higher quality of workmanship involved in installing the concrete. Figure 3.3.10 to .3.3.12 show Schley, Hydro Hill West and Hirsch pipes, respectively. The abrupt transition at the concrete/steel interface, which can be noticed in Schley and Hirsch pipes, however, is a questionable design since debris and consequently moisture have a tendency to accumulate at these locations (Figure 3.3.12 and Figure 3.3.13).  Puddles of water were also observed at these locations, which  resulted in the attack of the galvanizing layer. Extending the concrete to produce a more gradual transition between the steel and the concrete should mitigate this problem. The concrete invert protection was observed to offer exemplary abrasion resistance from the bedload in almost all structures.  However, one extreme case was  encountered where the concrete had deteriorated within two years after installment (Figure 3.3.14). This steep culvert of grade greater than 10 %, which is located 34 K m 79  from Sayward towards Port Hardy on Highway 19, carries vast quantities of abrasive material, up to 1 meter in diameter, during times of heavy rainfall. Maintenance personnel did inform the author that the deterioration of the concrete invert lining had transpired within two years after its installation. One solution to reduce the cost of maintaining this particular culvert would be to weld railroad rails to the invert to allow the boulders to skid through the pipe leaving the pipe itself unharmed.  This was reported to be a viable  solution in a steep culvert in West Virginia which experiences the passage of large boulders at high flow (Bealey, 1984) . 55  Figure 3.3.7 Abrasion caused by improperly laid concrete invert lining in Mile 61 culvert.  80  Figure 3.3.8 McKay pipe showed zinc loss at the concrete/steel interface after one year in service.  81  Figure 3.3.9 McKay culvert shows a limited amount of concrete lining in the invert.  82  83  84  Figure 3.3.13 Accumulation of debris along the concrete/steel interface in Schley pipe.  85  Figure 3.3.14 Extreme abrasion of concrete invert lining due to transportation of vast quantities of large boulders.  86  3.3.5 Invert soil side condition Concrete lined inverts, arches with no metal invert, and other conditions such as high water levels limited invert coupon sample extraction to only 19 water carrying culverts. Of the 19 invert sampled structures, 16 displayed complete zinc loss on the soil side of the invert. Bilston Creek (Main Rd.), only 4 years old, showed a complete absence of any galvanizing on the invert soil side. This shows that galvanizing has very little protective effect when stream water seeps underneath the culvert and constantly keeps the soil wet. At the 3 sites displaying zinc, a lack of backfill was generally noted directly behind the metal core sample, thus at low flows air-exposed conditions by the metal resulted. This possibly explains the presence of still existing zinc at locations exposed for 16 years. There is reason to believe that the low scaling effect produced by possible lack of alkalinity of the surveyed streams causes the rapid deterioration of invert soil side zinc. Attack on the soil side steel was mostly found to be fairly uniform. Only in 4 out of 19 instances were the soil sides observed to be more pitted than the water sides. Due to the very young age of the concrete lined structures, in addition to coupon sampling restrictions under the concrete, it was hard to conclude the effectiveness of the concrete on soil side corrosion around the invert. Four of the 2-3 year old concrete protected culverts were coupon sampled at the concrete/steel interface.  A l l samples  contained coatings of 68 um, or more, of zinc. Thus, there is reason to believe that the concrete lining might offer some protection to the soil side by limiting water access to the soil through seams in the structure, thus keeping the soil resistivity high by leaving the soil in a drier state. However, more sampling in the nature is needed to prove this assumption. 3.4 Corrosion rates 3.4.1 Establishing corrosion rates above high water mark Because of the many variables and changing conditions that are characteristic of the continuous hot dip galvanizing process, the original zinc coating thickness is not 87  always evenly divided between the two surfaces of a zinc coated plate. Also, from plate to plate, the zinc thickness will change according to conditions of the hot-dip galvanizing process.  Parameters  such as zinc bath temperature,  time of immersion, grain  characteristics of the steel, and drainage conditions all make the galvanizing a highly variable process with respect to the finished zinc coating thickness. Canadian manufacturers of galvanized products specify a minimum zinc coating mass that each finished product shall conform to. N o upper limit for coating mass is specified.  Armco, which has generally been the manufacturer for most galvanized  structural steel used in drainage structures in B.C, specifies a minimum of 610 g/m (2.0 2  oz./sq.ft.) per plate. This amounts to 43 um of zinc coating per side of each plate. As can be seen from Figure 3.3.1, the coating thicknesses of the plates surveyed were, in all instances encountered, found to be greater than the specified minimum thickness of 43 u. m. This can result in some difficulties when trying to establish zinc corrosion rates based on observed galvanizing thicknesses from sampled metal coupons. To obtain a preliminary basis of the general soil side zinc loss conditions, rates were calculated based on difference of coating thickness measured on the inside and soil side for each coupon sampled. It should be realized that, due to the original coating variances discussed above, there is great deal of uncertainty involved with the accuracy of the rates produced. The original zinc coating thickness was taken to be the inside mean thickness measurement obtained after cleaning the coupon.  Loss rates were then  calculated by subtracting the mean soil side thickness from the inside thickness and lastly dividing by the age of the structure, assuming originally equal quantities of zinc on both sides. Six structures were measured to have larger amounts of zinc on the soil side than the inside after removal of corrosion products; these were therefore excluded from the corrosion rates computations. Figure 3.4.1 illustrates the distribution of zinc loss rates observed for the structures that displayed greater zinc loss on the soil side than the inside. For comparison are AASHTO's 88  presumed corrosion rates for the same structures, based on their zinc loss rates presented in Section of chapter 1. The graph clearly demonstrate that AASHTO's supplied rates are somewhat on the conservative side compared to those experienced in this survey of galvanized structures sampled on Vancouver Island. In conformity with AASHTO's higher initial rate of 15 urn/year for the first two years, the sampled structures clearly showed similar greater initial rates. As time progresses, the observed corrosion rates level off to a reduced value, also in agreement with AASHTO's predictions. However, this surveys preliminary rates were substantially lower than AASHTO's predictions which is clearly demonstrated in the graph. Figure 3.4.1 Preliminary rates  DISTRIBUTION OF ZINC LOSS RATE A B O V E HIGH WATER MARK  ZINC LOSS RATE (micro-meter/year)  An actual indication of the amounts of soil side zinc found on the structures at different age levels can be seen in Figure 3.4.2. Forty-four coupons taken from above the  89  high water mark are the basis for the graph, which shows distributions of average soil side zinc thickness from different age groups. The figure displays a clear reduction in zinc coating thickness as the structural age increases. However, in the structures which are 20 years or older, over half the coupons still possess protective layers of zinc.  As a  comparison, the measured inside zinc thicknesses for the 44 coupons are also included in the graph.  These inside values are believed to be close to the original galvanizing  thickness, because of the pristine appearance of the insides of the structures. Figure 3.4.2 DISTRIBUTION O F A V E R A G E SOIL SIDE ZINC THICKNESS A B O V E THE HIGH WATER MARK  • 0-9 Y E A R S O 10-19 Y E A R S A 20-44 Y E A R S INSIDE  0  20  40  60  80  100  120  140  ZINC THICKNESS (mcro-meter)  A method producing a better representation of mean zinc loss rates included splitting the coupons into groups of different ages, thus obtaining an average rate for each age group. Coupons were separated into somewhat overlapping age groups as seen in Table 3.4.1. A mean average zinc coating thickness for all coupons in each group was calculated for both the soil side and the inside of each coupon. Zinc loss was calculated in  90  two different ways. One method was to subtract the mean average obtained from the soil side from that of the inside of the coupons in one age group. A corrosion rate was then calculated by dividing the zinc loss by a weighted average of the coupons' ages in each group. Another approach was to assume the original zinc thickness to be the average of all the inside thicknesses of the 44 coupons used in the analysis. This average inside thickness was 70 um. The small variations in the two ways of accounting for original thickness resulted in very similar rates in both instances.  Figure 3.4.3 depicts the  calculated average corrosion rates graphed against the weighted average ages for each age group. One outlier, a coupon which had a soil side thickness measured to 123 um, 42 um higher than the second highest measured thickness, was left out of the calculations. Also shown in Table 3.4.1, at least eight coupons were the basis for calculations of loss rates for each age group, which was deemed sufficient to establish mean loss rates. Table 3.4.1 Zinc loss rates by division of sample coupons into age groups. age group (years) # of coupons used inside  2-4  2-9  8-9  8-12  10-20  16-20  25-36  25-44  13  21  8  11  13  10  8  9  avg. thickness  72  71  69  67  69  71  62  67  (Mm) stand, devtn.  13  13  14  16  15  13  9  15  59  56  50  48  31  27  22  24  stand, devtn (Mm)  12  13  12  20  27  24  22  21  weighted avg. age(years)  2.7  5  8.8  9.5  16.5  18  30.5  32  avg. loss(|im)  13  15  19  19  38  44  40  4  4.8  3  2.2  2  2.3  2.4  1.3  1.3  avg. loss(jun) (70 nms.side)  11  14  20  22  39  43  48  46  loss rate (um/year)  4.1  2.8  2.3  2.3  2.4  2.4  1.6  1.4  (Mm) soil side  Loss 1  avg. thickness (Mm)  3  (ins.-soilside) loss rate (Mm/year) Loss 2  91  FIGURE 3.4.3 SOIL SIDE - ZINC L O S S R A T E S ABOVE HIGH WATER MARK  •••one original thickness • calc. orig. thckn.f or each age group  5  10  15  20  25  30  STRUCTURAL AGE (YEARS)  With respect to steel loss rates above the high water mark, the young age of the structures studied, manufacturer's tolerances on original steel plate thickness, and the excellent protection offered by the zinc, rendered it impossible to establish any corrosion rates for steel in areas not affected by creek water. The few coupons displaying complete zinc loss were all observed to have no measurable loss of steel, and mainly uniform surface rusting with slight pitting as the corrosion mechanism observed. 3.4.2 Establishing invert corrosion rates Loss rates are based on measurements taken from coupons obtained in 19 water carrying culverts where no concrete invert protection existed. Invert zinc loss rates were impossible to estimate since galvanizing was non-existent on virtually all invert coupons. However, from the observed structures, zinc is believed to be lost within the first year on the water carrying side of the structure. Soil side invert zinc was also noticed to deplete rapidly, due to the likely soft waters of the surveyed streams which hinder the formation of adherent zinc corrosion products which would generally stifle corrosion.  92  The steel loss rates referred to in this section are all based on losses from both the soil and the water side of the structure. It was initially believed that the soil side corrosion observed above the high water mark could be subtracted from the overall invert corrosion observed, thus quantifying the relative soil and water side corrosion observed. However, the smaller extent of corrosion in the drier part of the structure rendered this approach unfeasible. Assigning relative amounts of corrosion from the soil and water side is thus impossible. Invert loss rate estimations based on the coupons obtained will always contain a certain degree of uncertainty due to plate thickness variations caused by manufacturers' tolerances. Also, over the past 20 years, thickness specifications used by Armco have changed twice.  Steel plate specifications usually categorize each plate by a nominal  thickness. However, this nominal thickness has a lower bound, but no upper thickness limit is specified. Based on measurements taken from the 40 coupons containing zinc on both sides and measurements obtained from the structures' end plates, structural thicknesses were observed to vary a great deal from plate to plate within each structure. Table 3.4.2 shows the specified thicknesses presently employed by Armco, together with their previous plate thicknesses available.  Also displayed are the thickness intervals  observed for each of the gages encountered in the surveyed structures. Mean steel loss rates and pitting rates for the inverts were calculated based on the mean and minimum overall thicknesses, respectively, measured from the coupons sampled in the invert or directly above the current water level. Due to the observed variations of the original steel plate thickness, upper and lower bound rates were calculated based on the lowest and highest original thickness observed within each gage (from Table 3.4.2). Mean rates were then calculated by taking the average of the upper and lower bound rates. Steel loss invert rates of more confidence were obtained in instances in which two coupons were obtained from the same plate. Original metal thicknesses were then taken to be equal to the coupon thickness of the sample obtained above the high water mark, 93  which possessed coating on both sides. More confident rates were thus calculated from five structures. Figure 3.4.4 shows distribution of the upper and lower bounds for mean steel loss rates in 19 structures. Figure 3.4.5 illustrates the mean of the lower and upper bounds calculated from the non-confident mean rates together with more confident rates. It is clear from these graphs that a wide variety of steel corrosion rates was encountered at the different structures. Table 3.4.2 Armco's specified total thicknesses for structural plate in mm. gage  before mid 70's minimum maximum >1993 * to 1992 mid 70's measured measured (CSA) 12 2.77 2.66 2.74 3.12 10 3.51 3.42 3.38 3.57 S 4.27 4.18 3.91 4.54 3.00 7 4.78 4.67 4.78 5.10 4.00 5.54 5 5.45 5.32 5.70 5.00 — — 3 6.32 6.23 6.00 1 7.11 7.01 6.91 7.22 7.00 * Thickness is the base metal thickness excluding zinc coating  F I G U R E 3.4.4 IMEAN INVERT STEEL LOSS RATE DISTRIBUTION (Loss from both soil and water side) 100 S L O W E R BOUND -©.UPPER BOUND  i  :  ^SK'^^I  E  : :  : t  t :  : :  : i  : :  20 —i—l_i—i 0  20  40  .i—i  _i—i.Jr~T-Tli  60  80  100  120  140  MEAN INVERT LOSS RATE (micro-meter/yr)  94  F I G U R E 3.4.5 MEAN INVERT STEEL LOSS RATE DISTRIBUTION (both w ater and soil side of invert included) 100 • NON-CONFIDENCE • CONFIDENCE  ••I  0  20  40  60  80  100  *  120  140  MEAN INVERT STEEL LOSS RATE(nicro-meter/yr)  Pitting rates of the invert steel were generally found to be 4 to 5 times greater than mean rates. Distribution of lower and upper bound pitting rates are depicted in Figure 3.4.6. As pitting rates increased, there was also a tendency for the mean corrosion rates to increase as well, which is manifested in Figure 3.4.7. Estimation of time to perforation for the invert sampled culverts was calculated based on a further constant deterioration of the steel at their respective pitting rates. Figure 3.4.8 shows remaining years, as well at total years since installation, expected until perforations will likely be observed in the culverts' inverts. F I G U R E 3.4.6 INVERT PITTING RATE DISTRIBUTION (Loss from both soil and w ater side) 100  0  40 80 120 160 200 240 280 INVERT PITTING RATE (rricro-meter/yr)  95  • F I G U R E 3.4.7 PITTING R A T E S WITH ACCOMPANYING MEAN RATES  O  STRUCTURES RA NKED BY PITTING CORROS ION  F I G U R E 3.4.8 I ESTIMATED A V E R A G E Y E A R S TO PERFORATION FOR SURVEYED STRUCTURES io 250  4  4  8 12 12 16 16 16 16 16 16 20 25 25 27 30 30 36 44 STRUCTURAL A G E (years)  96  3.5 Observed corrosion rates vs. American and British design loss rates A comparison between calculated soil side zinc loss rates from this study and the specified design zinc rates from Great Britain (U.K.) and the U.S. are presented in Figure 3.5.1. Rates calculated from structures in B.C. are based on grouping coupons of similar age obtained from above the high water mark, where a non-saturated soil state is generally experienced. The B.C. rates were on average observed to be lower than any of the design rates employed in Great Britain and U.S. FIGURE 3.5.1  to  COMPARISON OF DESIGN ZINC LOSS RATES WITH OBSERVED RATES IN B.C.  CD U.K.-aggressive CD U.K. - no n-aggressive  O O  ;§  U.S.-(AASHTO)  LU  measured B .C. study  if)  co  O _j o  N  1  0 10  20  30  40  STRUCTURES AGE (years)  Design steel loss rates from the U.S. and Great Britain, graphed in Figure 3.5.2, are compared to the observed mean rates found in 19 of this study's continuous water carrying culvert inverts. Great Britain's data depict equivalent annual rates at each age level.  The observed metal loss rates displayed in Figure 3.5.2. are only half of those  97  actually observed on the sampled coupons. This presentation stems from the fact that corrosion of the steel was virtually always observed to take place on both water and soil side, with the relative amounts not known. However, stream water, which was found to have some correlation to the observed steel loss, was always existent on the soil side of the structure as well, due to seepage through seams in the structure. Abrasion loss did not appear to have a large enough impact on the overall metal loss to substantiate the assignment of more loss to the water side to account for abrasion. The figure clearly illustrates the varying extent of metal loss observed in the B.C. structures.  Some of the structures experienced average metal loss rates far above the  design rates in operation in U.S. and Great Britain. FIGURE 3.5.2  COMPARISON OF STEEL LOSS RATES (calculated in B.C. vs U.K./ U.S. codes)  2  0  20  40  structural age (years)  98  60  3.6 Time to perforation: California Method vs. Extrapolation in B.C. Comparisons of estimation of time to first perforation in this study's water carrying structures with the perforation time estimated by the California method based on the minimum resistivity and p H parameters obtained at the structures' sites are shown in Figure 3.6.1. The figure includes data points from 20 surveyed structures where time to first perforation are, due to the relative young age of the structures, extrapolated from the pitting rates calculated for each structure.  The expected time to perforation based on  observed rates were then plotted against the expected time to first perforation based on the California Chart, which considers the observed environmental conditions at the structure's site for calculating time to first perforation.  If the California Method had  predicted the actual time to perforation with some accuracy, the data points would lie closer to the line of equality (line with slope of 1) in the figure. However, it can be seen that most data points are located under the line of equality, thus the California Method actually over-predicts the time to first perforation and the modified California Chart would thus be unconservative in predicting average service life. The lack of correlation in B.C. with the California predictions can probably be attributed to the high rainfall area over which this study was conducted, compared to the overall drier state of California. The California data were obtained from many culverts with intermittent flow, whereas the culverts studied in B.C. had virtually all continuous flow of various abrasive nature.  99  Figure 3.6.1 Comparison of time to first perforation. YEARS TO FIRST PERFORATION (CALIFORNIA CHART V.S. B.C.)  0  50  100  150  200  years to first perforation (California ch.)  3.7 Corrosion rates/corrosion parameter correlation With respect to the mean zinc loss rates, no correlation between parameters observed at the sites and corrosion can be made due to the grouping of different structures in producing mean zinc corrosion rates. However, the preliminary galvanizing loss rates were graphed against soil resistivity, but no apparent correlation was detected (Figure 3.7.1). Lack of correlation can be attributed to the low reliability of these rates because of the zinc coating differences that might occur on both sides of newly galvanized plates. The zinc corrosion observed is also susceptible to local corrosion cells created by soil differences and differential aeration cells due to variable compaction which was not quantified in this study.  Other factors such as chloride and sulfate content of the soil  might have been better indicators of corrosion susceptibility, since resistivity values are also affected by other soluble salts which are not particularly deleterious toward zinc.  100  Water and soil p H values were all observed to be in the non-aggressive range for both steel and zinc, and an effort was therefore not made to correlate these to any of the calculated loss rates. FIGURE 3.7.1 [SOIL RESISTIVITY V.S. PRELIMNARY ZINC LOSS RATES ABOVE HIGH WATER MARK i i Ml! —  :  i i i  t  i t :  i i i !i i  . -4-444. 1 1! i  i i  '^~TTT  1 11 1  4+1-14)  J sir * 1000  •!  i  i i i  i  i t :  M i  ...  • *  10000  100000  LOG (SOIL RESISTIVITY ohm-cm)  It is difficult to attribute the corrosion of the steel observed on the invert to any one particular factor. This is so because both soil side and water side losses are inherent to the steel loss rates observed. By graphing the loss with respect to soil resistivity, the corrosion taking place at the water side, which is not soil related, will be included. Figure 3.7.2, which graphs the distribution of mean invert steel loss rates at different soil resistivities, shows no relationship of increase in corrosion as the soil resistivity decreases. FIGURE 3.7.2 I MEAN INVERT LOSS RATE DISTRIBUTION |j A T DIFFERENT SOIL RESISTIVITIES (ohm-cm)  100 80  a *  • 0-5000 O 5000-10000  1  410000-2000C 20000+  40  •  20 n 0  • .  1  1  1  T  20  40  60  80  , • 100  120  MEAN INVERT LOSS RATE (micro-meter/yr)  101  Correlating abrasion to loss rates was tried by graphing steel loss rates at different culvert slopes  A l l culverts surveyed were observed to carry abrasive material, and  therefore the slope of the culvert was believed and observed to be a good indicator of abrasion sensitivity. However, Figure 3.7.3 shows no apparent relationship between slope and observed loss rates. Abrasion was observed on intermediate and steep culverts in the form of heavily degraded invert bolts; however, abrasion losses on the plates are most likely overridden by other deleterious factors attacking the steel. FIGURE 3 . 7 . 3 MEAN INVERT L O S S DISTRIBUTION A T DIFFERENT C U L V E R T S L O P E S 100  • MILD S L O P E  o z o  [  & INTERMEDIATE  LU UJ O  1 STEEP  Lu  i  0  20  40  60  80  100  •  120  MEAN INVERT L O S S (micro-meter/yr)  The corrosion tests conducted by N B S showed a decrease of metal loss over time. This was not experienced in the surveyed structures, where corrosion was noticed to be more affected by site specific conditions. Figure 3.7.4 illustrates the lack of correlation between the observed mean loss rate at the structures and their ages. Stream water is both in contact with the water side and soil side (seepage through seems) of the culverts, and therefore should be somewhat suitable as a indicator of the corrosion observed on the invert coupons. Relatively low water resistivity was observed to be an indicator of metal degradation according to Figure 3.7.5, which presents the distribution of mean invert loss rates at different intervals of water resistivity. To support  102  these observations, it should also be mentioned that two other perforated culverts not invert sampled were observed to have the lowest resistivity of stream water measured. Roy Creek culvert in Royston by Courtney had a completely perforated invert, and the resistivity of the stream water was measured to 2500 ohm/cm. At Holland Creek, the resistivity of the stream was measured to be 5300 ohm/cm. FIGURE 3.7.4  M E A N INVERT L O S S R A T E V S STRUCTURAL A G E O  g  CD  -4—»  CO l _  co co  o > C  8 12 16 16 16 25 27 30 44 12 16 16 16 20 25 30 36  c CD  STRUCTURAL A G E (years)  FIGURE 3.7.5 MEAN INVERT L O S S R A T E DISTRIBUTION A T DIFFERENT W A T E R RESISTIVITIES (ohm-cm) 1 00 O z Q LU LU  80  LU  40  O X  60  A  nO-10000  O  010000-20000  A A  A20000-30000 A  40000+  S o  20 0  30000-40000  • i  i  A i  O  •  0 20 40 60 80 100 120 MEAN INVERT L O S S R A T E (micro-meter/yr)  103  Contributing the observed rates solely to the low water resistivities encountered might be jumping to conclusions, because resistivities above 10,000 ohm/cm are generally considered to be non-corrosive This was clearly not the case in many of the surveyed structures, which showed excessive corrosion for resistivities considered non-corrosive. One could argue, that the relative low resistivity value are most likely caused by presence of aggressive ions than scaling ions, which stems from the very low alkalinity existent in Western British Columbia streams. One could also speculate that other factors, such as microbiological corrosion, could be the cause of high metal degradation at sites of stream resistivity greater than 10,000 ohm/cm. Linear regression analysis was performed to correlate the observed corrosion rates with their possible corrosion contributing factors. The only parameter found to have some correlation to mean steel corrosion rate was stream water resistivity. In this case, a sample correlation coefficient (r) of 0.64 was calculated. The lack of correlation obtain from soil resistivities likely stems from the soil's sampling locations not being directly in contact with the corroding metal.  104  3.8  Conclusions Results of this galvanized steel structures study of 35 structures located on  Vancouver Island indicate the following: 1. Varying degrees of metal loss for the structures surveyed were noticed. It can be differentiated between two very different regimes with respect to observed metal loss, one possessing much lower loss rates than the other. The part of the structure showing relatively low rates of corrosion is the structure's soil side where backfill areas are not affected by continuous soil saturation due to running water exposure.  This area  constitutes the part of the structure above the high water mark. Much higher corrosion loss is experienced in the invert areas of water carrying culverts saturated by stream water, be it the soil side or water side. 2. Zinc loss rates were established in areas above the high water mark not affected by creek water.  Average rates were found to be higher the first few years and then  leveling off to lower values. A n initial zinc loss rate of approximately 5 (xm/year was deduced for the first 2 to 3 years, with subsequent rates below 3 |j.m/year.  The  galvanizing showed overall good resistance to the soil environment and was even found to be present in a 44 year old structure. Formation and adherence of zinc corrosion products were found to offer protection to the underlying galvanizing. 3. Invert zinc was observed to be lost within the first year from the water side by abrasion in virtually all structures. The invert soil side also experienced rapid zinc loss, with one structure even indicated a complete loss of zinc within 4 years.  The lack of  scaling salts such as calcium and magnesium in the possible soft waters of the surveyed streams interrupts the formation of adhering protective layer. The less adherent corrosion product produced is easily removed by the seepage of drainage water behind the structure. No zinc loss rates were established on the invert soil side, due to lack of samples displaying the presence of zinc.  105  4. Invert steel loss rates were calculated and found to vary a great deal from structure to structure.  The lack of zinc on both soil and water side of sampled invert  coupons prevented a determination of relative corrosion rates from these two areas. Therefore the corrosion rates calculated include losses from both soil and water side of the invert. A wide range of steel loss rates was observed, with most mean rates observed to be below 40 um/year. The minimum and maximum mean values were calculated to be 6 | i m/year and 118 um/year, respectively. 5. Pitting of the water side inverts was observed in almost all culverts. However, the extent of the pitting appeared to increase as the water resistivity decreased. Pitting was mostly observed behind nodules of rust found on the sides of corrugations pointing downward in the culvert. This is likely due abrasives removing corrosion products trying to adhere to the leading edges of the corrugations, subsequently interrupting formation of the oxygen concentration cell producing the pits. Pitting rates were on average found to be 4 to 5 times higher than the mean steel loss rate. 6. The invert water side indicated abrasion loss, by wear on the leading edges, in virtually all culverts.  The relative extent of the abrasion was not determined.  correlation was found between culvert slope and invert metal loss.  No  However, as the  culvert slope increased, invert bolts showed a marked increase in deterioration, and were in extreme cases missing. 7. Soil and stream water pH was found within a range between 5.9 and 7.8, which is, from a corrosion viewpoint, considered favorable to both zinc and steel. Thus p H is not a corrosion-inducing variable in the surveyed structures. 8. Minimum soil resistivities for the surveyed structures were found to be all within the moderately to non-corrosive range, with the lowest value measured to be 2800 ohm/cm. N o correlation was found between observed corrosion loss and minimum soil resistivity values.  106  9. Water resistivities were all found to lay within the mildly to non-corrosive range.  The lowest stream resistivity for the sampled structures was 5200 ohm/cm.  Despite the indication of low corrosion susceptibility due to the stream water, a clear correlation was established between low water resistivities and high steel loss rates in the invert areas. Another galvanized study had observed chloride levels exceeding the U.S. threshold of 100 ppm in water with resistivity value of 5200 ohm/cm, a resistivity value which is normally considered mildly corrosive..  It is thus possible that chloride  concentration, rather than resistivity value, is a better indicator of stream waters corrosion susceptibility. 10. For some of the structures showing excessive loss of steel and where the measured  corrosion-causing  parameters  indicated  only  a  moderately  corrosive  environment, it is possible that some other factor, such as microbiological corrosion, is the cause of the rapid deterioration occurring. 11.  Concrete invert protection had only been in place for three years in the  structures surveyed, thus it is difficult to conclude anything about its long term effectiveness against corrosion and abrasion.  However, it was apparent that when  improperly laid it can speed up the degradation process of the steel structure. Accelerated degradation of the inside steel was observed in the surveyed structures by creation of a hostile environment due to retention of organic material and humidity at the steel/concrete interface, or by diversion of abrasives towards the steel/concrete interface in areas where the concrete lining was not placed in a convex shape. In structures where the lining did not suffer cracking and was extended above the regular water flow level, a reduction of stream water access and consequent corrosion, of the soil side steel is likely. 12. In water carrying culverts which were not concrete lined, durability was observed to be governed by invert metal loss due to a combination of the presence of abrasive bed 107  sediments and corrosion attacks from the stream water and saturated invert soil. It is apparent from this study that a number of the non-protected structures will probably have to be replaced before they reach the end of their 75 to 100 year design life, unless protective measures, such as the installation of concrete invert lining, are taken. -  108  CHAPTER 4 RECOMMENDED PRACTICE 4.1 General Due to the current lack of a specific design life consideration practice coupled with the sometimes aggressive nature of structures' sites, many of the newly designed buried steel structures will have to be replaced before the end of an economically viable design life. It is therefore in MoTH'S best interest to adopt specifications covering design life considerations for galvanized steel buried structures. The present recommended design procedure is consistent with the philosophy presently in effect worldwide. It considers mandated electrochemical and physical limits on a select backfill; furthermore, an addition of sacrificial thickness to the structural reinforcement thickness, presumed to be consumed within the designated lifetime of service, is required. 4.2 Backfill requirements Backfill should be a sound, durable, granular material free from organic matter or other deleterious materials (such as shale or other soft particles with poor durability) It is recommended that the AASHTO's electrochemical and quality requirements for backfill be adopted, as illustrated in Table 4.2.1 and Table 4.2.2, respectively.  Furthermore, to  achieve a well-draining granular backfill, AASHTO's gradation requirements as seen in Table 4.2.3 should be employed. Test methods to check for conformity are included in the tables, with descriptions given in "Standard Specifications for Transportation materials and methods of sampling and testing", A A S H T O , 1992. The reason for using coarse draining backfill is that the natural resistivity is lower when the soil is drained. In addition, uniform oxygen penetration, in damp and saturated conditions, lowers the possibility of oxygen concentration cells developing. Furthermore,  109  good aeration of soil promotes the formation of a protective layer on a zinc surface. The above should lead to overall lower corrosion than experienced in soils of finer grade. For drainage conduits, the select backfill should at least extend Y meters away from the structure, with Y = 0.4 + 0.23 * S where S is the span of the structure. For bin walls, S is the height of the structure. Table 4.2.1 Recommended electrochemica requirements for select granular backfill. Minimum Resistivity (AASHTO T288) any method  pH (AASHTO T289) any method  Sulfate Content * (AASHTO T290)  Chloride Content * (AASHTO T291) any method  min 3,000 ohm/cm  5 -10  200  100  * Test for chloride and sulfate not required if pH between 6-8 and minimum resistivity is > 5,000 ohm/cm.  Table 4.2.2 Recommended quality requirements for select granular backfill. Shear angle of internal friction (AASHTO T236)  Sodium sulfate soundness loss (5 cycles) (AASHTO T104)  Los Angeles abrasion (AASHTO T96)  Plasticity index (AASHTO T90)  >= 34 degrees  15 % max  50 % max  6 max  Table 4.2.3 Recommended select granular backfill gradation requirements. Sieve size (mm)  Percent by weight passing designated sieve (AASHTO T27 and Til)  100  100  75  75-100  No. 200 (75 >im)  0-15  By limiting the soil resistivity to 3000 ohm/cm, moderate corrosion of the galvanized steel is anticipated, although the economic benefits, such as availability of a  110  wider range of backfills outweigh this negative aspect.  All measured soil resistivities,  except for one, were higher than this minimum soil resistivity limit of 3000 ohm/cm, indicating a possible abundant availability of fill conforming to this standard in western B.C. Fill availability in the low rainfall parts of the province is expected to be somewhat more limited due to the lack of leeching out of soluble salts in soil deposits causing overall a generally lower soil resistivity. Testing of the backfill before installation behind the structure should be done by taking multiple samples, due to significant variability in backfill sources which can occur. Samples should be laboratory-tested to assess the mean conditions of the select backfill. 4.3 Corrosivity classification Due to the limited number of structures observed and corrosion data gathered in this survey, only a very general corrosivity classification of the structures can be produced. This study clearly indicated two distinct corrosivity regimes in water carrying culverts, one being the non-saturated areas not affected by stream water, which display rather low corrosion rates. This area was represented by the area above the high water mark where the inside zinc appeared free of corrosion attack. The other more deteriorated areas were the stream water saturated soil and the water sides of the invert. It is thus logical at this point in time, due to the lack of more performance data, to assign two different sets of rates to the corrosion-affected areas: a lower set of corrosion rates indicative of the relatively small metal loss observed in the non-saturated zones, and a moderately higher set of loss rates for the invert soil and water side. The extent of the saturated zone in a water carrying culvert depends on the overall flow characteristics of the stream. The high water mark was in round culverts observed to lie about 30 to 45 degrees from the invert of the structure. In elliptic culverts, the height  ill  of the high water mark was observed to be no higher than 30 to 40 cm above the invert This should give preliminary guidelines to define the extent of the saturated zone. Inside surfaces of underpasses and culverts above the high water line were observed to have virtually no corrosion. It is therefore recommended that these areas not be given a sacrificial thickness; it is expected that they will last their design life with periodic maintenance if necessary. However, underpasses exposed to traffic and the use of deicing salt, should be given a sacrificial thickness for the bottom plates of the structure. This thickness, due to splashing of very corrosive water from traffic, should be equal to that specified for the saturated zone. 4.4 Concrete invert protection In line with the British standard and current practice in B.C., concrete paved inverts should be specified for all water carrying structures. The protection offered from the concrete lining, if properly placed, is primarily in the form of elimination of both abrasion and corrosion of the water side metal. In addition, the continuous flow of water through a structure with correctly placed concrete may reduce any anticipated corrosion of the buried surfaces. However, the extent of this soil side protection is as yet somewhat questionable due to the lack of performance data. Concrete linings should be constructed in a concave fashion to funnel debris efficiently through the channel, thereby avoiding contact with the steel. The concrete protection should extend up the walls of the culvert to a level where overflow of the stream water at the concrete/steel interface at regular flows is avoided. If the concrete is extended 30 to 45 degrees, depending upon the water level, from the invert in a round structure, then proper protection would be assured in the water-affected zone of these kinds of structures.  Furthermore, the concrete cut-off at the concrete/steel interface  should be a gradual one (Figure 4.4.1), in order not to allow for debris and water accumulation and retention, both of which are catalysts for corrosion.  112  113  With respect to thickness of the concrete protective layer, only Great Britain was observed to have standard specifications. Due to the short service time since installation of the B.C. concrete lined culverts and therefore lack of performance evaluation, the effectiveness of the British thicknesses in the heavy sediment-laden B . C . streams is somewhat questionable. As a preliminary approach, the British concrete lining thicknesses should be sufficient. These are described using the following parameters: for grades less than 2%, invert protection shall be 100 mm, whereas 125 mm is required for grades greater than 2%. Also, if it is anticipated that bedloads of diameter greater than 100 mm be carried through the culvert, a minimum concrete thickness of 170 mm is required. Instead of the grates constructed outside of steep culverts, expected to experience flows containing large boulders, railway rails welded to the plates and cast into the concrete throughout the length of the structure could be a viable alternative for abrasion control. Only about a one meter wide strip of the very bottom of the invert was observed to be completely deteriorated in the most abrasion susceptible structure (Figure 3.3.14). Thus three or four railway rails should offer sufficient protection in the very bottom of the invert. This method was reported to work well in West Virginia culverts and is likely a viable solution in addition to the massive grates constructed to date (Figure 4.4.2), which prevent culvert contact with trees and large boulders carried by the streams. Concrete lining would also likely reduce maintenance costs by eliminating sediment buildup inside the culverts, which is often experienced in bare steel inverts (Figure 3.3.4). The concrete and steel rails would offer a smoother ride for the larger boulders, thereby speeding up the transported material and hopefully eliminating deposition of sediments within the steeper grade structures. Concrete invert lining is also an excellent tool to retrofit existing water carrying structures which demonstrate excess invert deterioration.  With the higher loss rates  observed in the inverts of culverts in this study, previously constructed B.C. culverts, 114  which sometimes were constructed with thinner gages in the invert due to the more lenient structural requirements of the crown of a culvert, will have their service lives extended considerably by the application of concrete to the invert of the structure.  It is  recommended that existing water carrying structures displaying excessive invert corrosion be concrete invert lined to prolong their service lives. Structuresfromthis survey that will not likely fulfill their service life without installation of concrete protection due to rapid invert degradation are the multi-plate culverts which include Beaver Creek, Eagles Nest, Bilston Creek, Jordan + 4.5 km, Buller Creek, Willows Creek, and Boulder Creek. Figure 4.4.2 Grate installed to stop larger bouldersfromcontact with the heavily abraded Island HW. 34 km culvert.  If high sulfate concentrations are expected in the stream water, testing should be performed to determine if sulfate-resistant Portland cement might be required. If drained stream water contains sulfate levels above 200 ppm, moderate sulfate-resistant cement Type 20 should be applied. If levels above 1000 ppm are experienced, it is questionable if  115  galvanized steel should be used at all. However, concrete placed in this environment should be of Type 50, Sulfate resistant Portland Cement. 4.5 Sacrificial thickness requirements For non-saturated areas not affected by stream water, maximum mass presumed lost per side due to corrosion at the end of the designated design life may be calculated assuming a uniform loss model with the following rates. Galvanization loss Carbon steel loss  = 6 urn/ year for the first 2 years = 3 um/ year for subsequent years =15 urn/ year after zinc depletion  Galvanizing loss rates are based on loss rates obtained from the Stuttgart study. Data are based on sites characterized by minimum resistivity greater than 1000 ohm/cm. In addition, compared to the observed corrosion rates in this study these specified zinc rates appear to be somewhat conservative. Furthermore, the specified zinc thickness on galvanized products is 43 um; it was found however to be in almost all instances at least 1.5 times as thick. Sacrificial thickness calculations when designing should, however, use the 43 um of zinc coating as their basis. This assumes complete zinc loss within 12 years, which is reasonable, but conservative, compared to the average observed zinc loss in this survey. Due to lack of performance data from this study to determine carbon steel loss rates, AASHTO's carbon steel rate should be used for calculating sacrificial thickness and this rate should render a conservative result.  This is believed to be so, because the  Stuttgart study reported carbon steel loss rates of 45 um/yr (for the two first years) and 9 um/yr (for subsequent years) associated with the zinc rates discussed above. After 90 years of exposure, using the aforementioned rates, an average carbon steel rate of 10 (j, m/yr. would be calculated, thus rendering AASHTO's rate of 15 um/yr as conservative. For saturated soil areas and water side inverts corrosion rate per side is proposed as follows:  116  Galvanization loss Carbon steel loss  =15 (xm/ year until zinc depletion = 20 um/ year after zinc depletion  The protective value of zinc in the invert soil and water side was observed to be as minimal, which stems from lack of corrosion product formation due to the likely soft waters of the surveyed streams. Thus for the present standard of employing 43 \xm of zinc coating, protection will be offered for only 3 years, in accordance with observed behavior. For other areas of the province with harder scaling stream waters, this high zinc rate will provide a conservative result. Many of the steel loss rates observed in B.C. were found to lie above the 20 u, m/year margin. However, with the lining of concrete to all water carrying inverts, as well as the provision of this sacrificial thickness to both water and soil side, it is believed that this will result in an overall conservative approach to invert sacrificial thickness. With concrete properly placed, a virtual elimination of invert metal loss from the water side is expected. However, cracking of the concrete cover, which enables the access of water to the metal surface can reduce the protective value of the lining. In addition, overflowing of the concrete will also cause corrosion attack of the water side metal surface. With respect to future corrosion loss of the invert soil side, reductions are expected upon comparison to this survey's measured rates. In the unprotected culverts, stream water has free access to saturate the invert soil side by seepage through the plate seams. Concrete invert lining is believed to be able to mitigate this water access which has proven to be a catalyst for corrosion, and should lead to a reduction in soil side corrosion of the steel plates. Abrasion metal losses which are inherent to the observed rates measured to date, will also cease when concrete lining is installed. Figure 4.5.1 illustrates the proposed metal loss rate for the invert of water carrying structures, compared to the observed rates and specified rates for U.S. and G.B.  The  British rates which are based on a decaying model, are here normalized to uniform  117  corrosion rates over a life span of 100 years. The British rates are 5 u.m/year and 16 u, m/year for non-aggressive and aggressive sites, respectively. It can be seen from the graph that this proposed invert steel loss rate will cover most of the observed rates in this study. If in extreme cases the proposed rates are not acceptable, the realization that the invert carries less structural load than the upper part of the culvert, coupled by the fact that the structure will most likely still be standing after complete invert loss, will assure a safety factor greater than one. In addition, the concrete itself will also carry much of the imposed invert loads if the structural invert metal has completely degraded. Some structures did show losses somewhat below the specified rate. However, there are also uncertainties to future land use of the drainage areas of today's nonaggressive stream waters. Areas which today are occupied by forests might in 50 years have been turned in to agricultural land or residentially developed property. This will of course increase the aggressive nature of the waters flowing through a water carrying structure. F I G U R E 4.5.1  COMPARISON OF MEAN STEEL LOSS RATES  (U.K. rates normalized to 100 yrs) Ei  Calculated in B . C . (total rate/2)  o i_ o  E  Pro posed B . C .  CD  "4-»  CO  U.K. aggressive  L-  10  U.K.  co  o "cU  <D  non-aggressive  A.A.S.H.T.0  ^—»  w  CD O)  co  a> >  20 40 structural age (years)  118  60  4.6 Design life Choosing the length of the design life of the structures is required to determine the total sacrificial thickness to be added to the structure. This report has not explored the qualitative or quantitative details involved in choosing a design life which minimizes the overall cost of the structure. A n optimal design life will probably lie somewhere between the 75 years as specified by A A S H T O and 120 years which is currently the number in use in the G.B. 4.7 Deicing salt A potential crown corrosion problem exists on culverts placed under roadways where large amounts of deicing salt is used to keep roadways open during the winter months. Chloride ion concentrations beyond the threshold level of the specifications of 100 ppm are very aggressive towards galvanized steel; this is due to the fact that drainage water, now high in chloride content, seeps into the soil and lowers the soil's resistivity. Chloride also has the ability to break down otherwise protective corrosion product deposits. In the areas of this study, deicing salt was not regularly used and did therefore not affect the performance of the structures . However, in colder areas of the province, it is recommended that longitudinal drains be constructed on the roadway directly above galvanized steel culverts to drain any seepage water from the road beyond the reinforcement zone of the structure. 4.8 Further testing If further studies of galvanized steel structures are to be conducted, some recommendations regarding methods and test equipment to simplify the collection of data and to produce more accurate results can be offered. When obtaining invert samples, the variability of the plate thicknesses from plate to plate due to manufacturers tolerances created uncertainties as to the correct original thickness. This can be circumnavigated by 119  obtaining, whenever possible, two samples from one plate, in which one is affected by the stream water and the other is not. The upper sample will most likely contain zinc on both sides and represent the original thickness of that particular plate. Coupon samples gathered above the high water mark and in other non-saturated areas, such as underpasses, should be taken at, or above, the 9 and 3 o'clock mark to receive a worst case scenario for this non-saturated zone. Below this area, if not affected by stream water, the soil side has not likely experienced the worst case conditions due to shielding from the rain offered by the structure itself. When obtaining a soil sample, more effort should be put into the acquisition of a soil sample directly behind the actual drilled coupon hole. This sample will of course give an indicative representation of the actual environment of the exposed sample. If small rocks are blocking the smaller grade backfill, a mandrel and hammer can be used to split the rock to achieve access to the soil. Non-destructive in situ wall thickness measurements by use of an ultrasonic wall thickness gauge could also be a viable alternative to drilling coupon samples. This type of equipment sends an ultrasonic pulse through the plate and is reflected back when it reaches a dissimilar surface. Thickness is measured by converting the time lapse between transmitting the pulse and receiving its echo. It is important that the probe be in direct contact with the material to be measured, so any rusted surfaces would have to be cleaned. Thickness measurements could then be taken at 3 to 5 meter intervals throughout the structure, thereby resulting in far greater numbers of samples than the number procured in the present study. A field engineering chemistry kit should be purchased and used to indicate stream water chloride and sulfate levels on site.  120  4.9 Future developments Future testing of stream waters involving the determination of sulfates, chlorides, and perhaps scaling tendencies due to alkalinity, might result in the sub-grouping of water aggressivity with appropriate deterioration rates for non-aggressive, aggressive, and even very aggressive waters. This is presently done in Great Britain, and their classification system for water aggressivity can be observed in Table 1.7.4. If this study is undertaken, this survey's water samples, which are stored at MoTH'S Geotechnical Laboratory in Victoria, could also be incorporated into the possible future development of a water aggressivity classification system. Such a system will result in reduced costs and proper allocation of materials in order for structures to achieve durability and to last their intended design lives. 4.10 Other design features Two of the concrete-footing multi-plate arch structures in this survey showed excessive localized corrosion attack at the concrete/steel interface due to a design feature which generated an aggressive environment at this particular location. The large Seymour Avenue underpass had a dip in the concrete along the concrete/steel interface which allowed organic material and drainage water to collect. Figure 4.10.1 depicts the current situation encountered at this location. The zinc at this location was already removed and rusting of the steel was evident (Figure 4.10.2). A similar problem with early rusting was apparent in the Paradise Arch where retention of debris and water was found (Figure 4.10.3. and Figure 4.10.4).  121  It is recommended that this design feature be discontinued, and in the cases of the already existing structures, removal of the corrosive environment and subsequent filling of the gaps with concrete to eliminate the possibility future buildup of debris which causes the corrosive environments in these locations.  Figure 4.10.1 Dip in the concrete at the concrete/steel interface results in retention of organic material.  122  123  124  4.11 Measurement techniques Suggestions are given for measurement techniques for key index parameters for backfill and stream water which govern corrosivity. Many methods are currently in use to measure soil resistivity directly or electrolytic conductivity.  A S T M 557-78 employs four electrodes to measure the  resistivity. Another resistivity test is the California method, which is outlined in chapter 2. The method preferred for laboratory analysis of soil resistivity is the one used in this survey and determines the electrolytic conductivity of an aqueous extract of soil water. All of these methods measure the minimum resistivity of the soil. Resistivity of a water sample is most easily determined by the use of a conductivity meter and then converted to resistivity values as described in chapter 2. Chloride ion concentrations measurements can done by a Mohr argentometric method of soil extracts, namely Method 65-3.5 outlined in Black (Black, 1965) . 56  A recently adopted standard, which measures anions including chloride and sulfate ion concentrations simultaneously by ion chromatography, is A S T M D4327-88. This is a very accurate and reproducible method. Its basic operation consists of ion separation by an exchange resin succeeded by quantification with a conductivity detector.  Most errors  usually associated with other classical sulfate detection methods are alleviated by this procedure. This method measures concentration extracts and can therefore also be used to measure stream water chloride and sulfate concentrations. A number of field kits which are specifically designed for such measurements are in existence but their accuracy is lower. The p H of soil samples is most easily determined by a glass electrode-calomel reference electrode p H meter on a 1:1 weight ratio of soil to water. The same electrode can be used to measure water sample p H (Elias, 1990) . 57  125  The presence of microbiological bacteria such as the sulfate-reducing bacteria could be tested for in stream waters where traces of sulfate are detected. Nodules of rust on the water side of the invert can be sampled, brought to the laboratory and detected by placing the nodules in an environment conducive to their growth. One suggested chemical test is the immersion of the nodules in a nutrient broth called Baars solution. The sample should be sealed in an airtight bottle and incubated at 30 degrees Celsius. A positive result is indicated by the solution turning black, due to formation of iron sulfide when anaerobic sulfate-reducing bacteria are present in the nodules. Other culture media which are proposed for detection of these bacteria include Postgate and API media (Brady, McMahon, 1994) . 58  126  BIBLIOGRAPHY Smith, W. Harry. Corrosion Management in Water Supply Systems. Van Nostrand Reinhold, New York,  1  1989, pp. 1-7. ^Uhlig, Herbert H. Corrosion and Corrosion Control. John Wiley & Sons Inc., New York, London, 1963, pp. 6-13. ^Illustrated Case Histories of Marine Corrosion. European Federation of Corrosion Publications, 1990, pp. 1-3. ^Turner, Jane M. "Controlling Galvanic Corrosion in Soils with Cathodic Protection", Galvanic Corrosion. 1988, pp. 178-180. ^Elias, Victor Durability/Corrosion of Soil Reinforced Structures. Federal Highway Administration, 1990, p. 33. ^Palmer, J. D. "Environmental Characteristics Controlling the Soil Corrosion of Ferrous Piping", Effects of Soil Characteristics on Corrosion. STP 1013, ASTM, 1989, pp 7-11 Meacham, D. G., J. O. Hurd, and W. W. Shisler. Culvert durability study. Ohio dept. of Transportation,  7  1982, p. 32. ^Worley, Herbert E. Corrosion of Corrugated Metal Pipe. State Highway Commission of Kansas, 1971, p. 5. Galvanizing for Corrosion Protection. A specifies Guide. American Galvanizers Association, 1990  9  10  Elias, Victor. Durability/Corrosion of Soil Reinforced Structures. Federal Highway Administration,  1990, pp. 15. ^Elias, Victor. Durability/Corrosion of Soil Reinforced Structures. Federal Highway Administration, 1990, pp. 38-39. 12 Soil Side Durability of Corrugated Steel Pipe. Corrpro Companies, Inc., March 1991, pp. 2-3. Escalante, Edward. "Concepts of Underground Corrosion", Effects of Soil Characteristics on Corrosion.  13  STP 1013, ASTM, 1989, pp. 85-86. Escalante, Edward. "Concepts of Underground Corrosion", Effects of Soil Characteristics on Corrosion.  14  STP 1013, ASTM, 1989, pp. 86-87. Popescu, A. and T. Beschea. "Major biodeterioration Aspects of Buildings in Romania", Microbiology  15  in Civil Engineering. Federation of European Microbiological Societies Symposium No. 59, 1990, p. 100. ^Patenaude, Robert. "Bacterial Corrosion of Steel Culvert Pipe in Wisconsin", Transportation Research Record 1001 1984, pp. 66-69. Beaton, J. L. and R. F. Stratfull. "Field Test for Estimating Service Life of Corrugated Metal Pipe Culverts". Highway Research Board Proc. Vol. 41, 1962, pp. 260-261. 18 17  Harrison, John and Salem Bahamdun. Galvanized Structures Study. Ministry of Transportation and Highways, British Columbia, 1993, p. 8. 127  /  l%scalante, Edward. "Concepts of Underground Corrosion", Effects of Soil Characteristics on Corrosion. STP 1013, ASTM, 1989, p. 86. 90 Bednar, Lawrence, "Durability of Plain Galvanized Steel Drainage Pipe in South America: Criteria for Selection:", Transportation Research Record 1231.1989, pp. 80-83. oi Uhlig, Herbert H., Corrosion and Corrosion Control. John Wiley & Sons Inc., New York, London, 1963, pp. 98-103. ^Sullivan, M. A. and S. C. Samis. Assessment of Acidification Potential of Selected Lower Mainland and Vancouver Island. British Columbia Streams. Department of Fisheries and Oceans, 1988, pp. 52-53. Legrand L. and P. Leroy, Prevention of corrosion and scaling in water supply systems. Ellis Horwood, New York, 1990, pp. 193-199. 24 Howsam, P. Microbiology in Civil Engineering. E. & F.N. Spon, London, 1990, p. 28. 25 Patenaude, R. "Bacterial Corrosion of Steel Culvert Pipe in Wisconsin" Transportation Research Record 1001 1984, pp. 66-69. Koepf, A. H. and P. H. Ryan "Abrasion Resistance of Aluminum Culvert Based on Long-Term Field Performance" Transportation Research Record 1087 1986, pp. 15-16. 27 The Institution of Metallurgists, Corrosion and protection of metals. New York American Elsevier Publishing Company Inc., London, 1965, pp. 86-88. 28 Wranglen, Gosta. An Introduction to Corrosion and Protection of Metals. Chapman and Hall, London, 1985, pp. 120-123. 29 Wranglen, Gosta. An Introduction to Corrosion and Protection of Metals. Chapman and Hall, London, 1985, pp. 48-49. Slunder, C. J. and W. K. Boyd. Zinc: Its Corrosion Resistance. International Lead Zinc Research Organization, Inc., Second Edition, 1983, pp. 61-65. 31 30  Carter, V. E. Metallic Coatings for Corrosion Control. Newnes-Butterworths, London, Boston, 1977, pp. 67-72. "^asalo, C. Water and Gas Mains Corrosion. Degradation and Protection. Ellis Horwood, New York, 1992, pp. 125-127. Wet Storage Stain. American Galvanizers Association, 1992 34 33  Bednar, Lawrence "Plain Galvanized Steel Drainage Pipe Durability Estimation with a Modified California Chart", Transportation Research Record 1231 1989, pp. 70-78. Worley, Herbert E. .Corrosion of Corrugated Metal Pipe. State Highway Commission of Kansas, 1971, p.  35  A 15. 36 American Iron and Steel Institute Handbook of Steel Drainage & Highway Construction Products. W. P. Reyman Associates, Inc., New York, 1983, pp. 231-239. Romanoff, M. Underground Corrosion. NBS Circular 579 - U.S. Det. of Commerce., 1957 128  37  38  Elias, Victor. Durability/Corrosion of Soil Reinforced Structures. Federal Highway Administration,  1990, pp. 18 Reinforced Earth Company Ltd. "Reinforced Earth: 24 Years of Experience and Lessons for the Future",  39  Vancouver, B.C., 1988, pp. 6.8-6.11 40 Elias, Victor. Durability/Corrosion of Soil Reinforced Structures. Federal Highway Administration, 1990, pp. 17-32. 4 1  Standard Specifications for Highway Bridges. American Association of State Highways and  Transportation Officials, Fifteenth Edition, 1992 Brady, K. C. and W. McMahon. The durability of corrugated steel buried structures. Transportation  42  Research Laboratory, 1994, p. 5. ^^Brady, K. C. and W. McMahon. The durability of corrugated steel buried structures. Transportation Research Laboratory, 1994, pp. 5-9. 44 Standard Specifications for Highway Bridges. American Association of State Highways and Transportation Officials, Interim, 1991 45 Standard Specifications for Highway Bridges. American Association of State Highways and Transportation Officials, Fifteenth Edition, 1992 46 Standard specifications for construction of roads and bridges on federal highway projects. U.S. department of transportation, Federal Highway Administration, 1992, pp. 585-588. Wolfe, Victor D. and Stephen H. Macnab. Corrugated Metal Pipe Comparison Study. Oregon State  47  Highway Division, Salem, Oregon, 1976 Sullivan M. A. and S. C. Samis. "Survey of selected British Columbia and Yukon Salmon streams for  48  sensitivity to acidification from precipitation", Department of Fisheries and Oceans, No. 1388,1985, pp. 44-47. 49  Bellar Peter J. and James P. Ewing. "Metal Loss Rates of Uncoated Steel and Aluminum Culverts in New York", Transportation Research Record 1001. TRB, National Research Council, Washington D.C., 1984 pp. 75-77 American Iron and Steel Institute. Handbook of Steel Drainage & Highway Construction Products.  50  Canadian Edition, W. P. Reyman Associates, Inc., New York, 1983, pp. 196. Page, A. Methods of Soil Analysis Part 2. Chemical and Microbiological Properties. 2nd Edition  51  Agronomy #9, American Society of Agronomy Inc., Soil Science Society of America, Inc., Madison, Wisconsin, USA, 1982 52  Soil Side Durability of Corrugated Steel Pipe. Corrpro Companies, Inc., March 1991, pp. 13-21.  Blonska, Frank. Condition and Corrosion Survey on Corrugated Steel Storm Sewer and Culvert Pipe.  53  Corrpro Companies, Inc., 1987, p. 57.  129  Bednar, Lawrence. "Plain Galvanized Steel Drainage Pipe Durability Estimation with a Modified  54  California Chart", Transportation Research Record 1231. 1989, p. 78. Bealey, Mike. "Precast Concrete Pipe Durability: State of the Art", Transportation Research Record  55  1001. 1984, p 89. Black, C.A. Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties. Agronomy # 9,  56  American Society of Agronomy, 1965 ^Elias, Victor. Durability/Corrosion of Soil Reinforced Structures. Federal Highway Administration, 1990, pp 8-15 5&Brady, K. C. and W. McMahon. The durability of corrugated steel buried structures. Transport Research Laboratory, 1994  130  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items