Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Using refractive index to monitor oil quality in high voltage transformers Kisch, Ryan John 2008

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

Item Metadata

Download

Media
24-ubc_2008_fall_kisch_ryan_john.pdf [ 2.48MB ]
Metadata
JSON: 24-1.0067031.json
JSON-LD: 24-1.0067031-ld.json
RDF/XML (Pretty): 24-1.0067031-rdf.xml
RDF/JSON: 24-1.0067031-rdf.json
Turtle: 24-1.0067031-turtle.txt
N-Triples: 24-1.0067031-rdf-ntriples.txt
Original Record: 24-1.0067031-source.json
Full Text
24-1.0067031-fulltext.txt
Citation
24-1.0067031.ris

Full Text

Using Refractive Index to Monitor Oil Quality in High Voltage Transformers  by  Ryan John Kisch  B.Sc.E.E., Saginaw Valley State University, 2004  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR DEGREE OF  MASTER OF APPLIED SCIENCE in The Faculty of Graduate Studies (Electrical and Computer Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  June 2008  © Ryan John Kisch, 2008  Abstract Insuring reliable operation of high voltage electrical equipment, such as transformers and cables, is of great importance to the power industry. This is done by monitoring the equipment. A large portion of this monitoring includes analyzing the quality of the insulating oils and observing various compounds formed in the oils during aging.  Most often, transformer monitoring  includes routine oil sampling and analysis, which has proven to be very effective at diagnosing faults and determining the insulation condition. Many techniques have been demonstrated for the purpose of online monitoring, and various commercial products are available. However, utility companies are still looking for more cost effective methods to monitor their equipment between sampling intervals. The work presented here was performed in order to investigate the use of refractive index for monitoring insulating oils.  The refractive indices of various oil  samples obtained from the field were measured and differences were observed. Accelerated aging experiments were conducted in a laboratory and increases in the refractive indices of these artificially aged oils were observed. Experiments were conducted to determine what by-products would contribute to this increased refractive index by investigating the effects of individual groups on the refractive index change.  These groups included aromatic compounds, polar  compounds, furans, acid, and fault gases. We observe that the formation of furans, acids, and fault gases cannot be detected using refractive index for the concentrations typically found in the field. We conclude that changes in the refractive index of an oil can be used as an indicator of the oil’s aging and its break down and the formation of aromatic and polar compounds.  11  Table of Contents  Abstract Table of Contents List of Tables List of Figures List of Symbols and Abbreviations Acknowledgements Introduction and Motivation 1 1.1 Overview  ii iii vi viii xi xiv 1 2  .  1.2  2  Review of Applied Equipment Monitoring Techniques  4  1.2.1  Dissolved Gas Analysis  5  1.2.2  Furans  9  1.2.3  Moisture  10  1.2.4  Oxygen  10  1.2.5  Interfacial Tension (IFT)  11  1.2.6  Neutralization Number/Acid Number  12  1.2.7  KV BreakdownlDielectric Breakdown  12  1.2.8  Color  13  1.2.9  Polar Compounds  13  1.2.10  OnlineMonitoring  14  1.3  Review of Research into Insulation Diagnostics  16  1.4  Our Investigation  20  Measuring Refractive Index 2.1 Introduction to Sensors  23 23  2.2  Why Use Refractive Index’  23  2.3  Introduction to Sensors  28  2.4  The D-Fiber Sensor  28 111  2.4.1  D-fiber Sensor Fabrication  36  2.4.2  Placing the D-Fiber Sensor into the Measurement System  37  2.4.3  Sensor Calibration  39  2.4.4  D-fiber Sensor Resolution  43  2.5 3  49  Experiments 3.1 Introduction to Chapter  53 53  3.2  54  Samples Obtained From the Field  3.2.1  Dissolved Gas In Oil Samples From the Field  54  3.2.2  Other Measured Properties of Oil Samples Obtained From the Field  59  3.3  Effects of Accelerated Aging on Refractive Index of Oils  64  3.4  Polar Compounds in Oil  76  3.4.1  Introduction to Section  76  3.4.2  Methanol Extraction  76  3.4.3  Oil Samples  77  3.4.4  Refractive Index Measurements  79  3.4.5  Polar Compound Extraction From Naturally Aged Oils  84  3.4.6  Discussion  87  3.5  4  FISO Refractive Index Sensor System  Effects of other Contaminants in Oil  89  3.5.1  Oil Samples Spiked with Furans  89  3.5.2  Acid Artificially Introduced into Oil Samples  94  3.5.3  Gas Artificially Introduced into Oil Samples  97  Summary, Conclusion, and Suggestion for Future Work 4.1 Summary  102 102  4.2  104  Conclusion iv  4.3  .106  Suggestions for Future Work  109  References  V  List of Tables Table 1-1: Common commercially available transformer oils with type and refractive index listed  4  Table 2-1: Results of non-relative measurements conducted to find resolution for constant system operation of two and a half hours  46  Table 2-2: Temperature results of relative duration period measurements to show average temperature variation over relative measurement period  47  Table 2-3: Transmission ratio results of relative duration period measurements to show average transmission variation over relative measurement period  47  Table 2-4: Refractive index results of relative duration period measurement conducted to find resolution of system using relative measurement  48  Table 2-5: Results of refractive index resolution test using two oils with very close refractive index values  49  Table 3-1: Refractive index measurement and DGA results of cable oil samples taken from the field  56  Table 3-2: Refractive index measurement and DGA results of transformer oil samples taken from the field  57  Table 3-3: Refractive index measurement and DGA results of load tap changer samples taken from the field Table 3-4:  58 Measured refractive indices of oil samples obtained from the field with some  physical and chemical property values shown  60  Table 3-5: Measured refractive index versus time for accelerated aging samples at 120°C  65  vi  Table 3-6: Measured refractive index versus time for accelerated aging samples with varying contents at 150°C  68  Table 3-7: Measured refractive index versus time for accelerated aging samples with varying contaminants at 150°C  71  Table 3-8: Aging conditions for oils used in polar compound measurements  78  Table 3-9: Measured properties of aged oils  79  Table 3-10: Refractive index measurements of oil and methanol samples and concentration of polar compounds measured by HPLC  80  Table 3-11: Refractive index measurements of naturally aged oil and methanol samples and area of polar compounds measured by HPLC  84  Table 3-12: Measured refractive indices of oil samples varying in 2-furaidhyde concentration. 90 Table 3-13: Measured concentrations of furans in l2mL Luminol samples spiked with 3 drops of furan mixture  91  Table 3-14: Measured refractive index change due to acid added to Luminol TRi oil samples at varying concentrations  95  Table 3-15: Measured refractive index change due to ethane injection into Luminol TRi oil samples at varying concentration levels  98  Table 3-16: Measured refractive index change due to acetylene injection into Luminol TRi oil samples at varying concentration levels  98  vii  List of Figures Figure 1-1: Generation of combustible gases in transformer oils due to temperãturë and faults (not to scale). This figure is similar to the gas generation chart found in [1] Figure 2-1:  6  Normalized plot of the real and the imaginary value of refractive index as a  function of frequency. A similar figure is found in [45]  25  Figure 2-2: Real value of refractive index versus wavelength illustrating change in refractive index values with different resonant frequencies  27  Figure 2-3: (a) Magnified cross section of a typical step-index circular single mode fiber. (b) Magnified cross section of the core showing the refractive index profile and the optical field distributions. Decaying optical fields in the cladding are called evanescent fields. A similar figure found in [44]. (Figure not to scale)  30  Figure 2-4: D-fiber cross section (not to scale), showing the core dimensions, cladding thickness “d” between the core and outer cladding flat side, and the protective jacketing surrounding the cladding  31  Figure 2-5: (a) Section of D-fiber: For a section of D-fiber, with length “L”, the distance “d” between the core and planner side of the cladding is reduced by Ad giving a new distance dr. (b) and (c) show the respective refractive indices and optical field distributions in the “cut-out section” shown below (a) [note, co-ordinate system rotation]. (b) shows a section not etched, with d between core/cladding interface and field confined to the fiber. (c) shows a section after etching, with reduced distance dr and field extending into the external medium  33  Figure 2-6: Calibration curve measured by sweeping the refractive index of the three thermo optic oils by temperature control, and recording the power transmission. Region I, II, III, and the lossless region are shown  34  Figure 2-7: Diagram of experimental set-up showing D-fiber sensor and FISO sensor viii  38  Figure 2-8: Measured power transmission of D-.fiber sensor at various optical wavelengths.  ...  42  Figure 2-9: Calibration curve shown for operating wavelengths of 1550nm and 1500mm The operating point is moved by increasing the temperature. When the temperature control has been exhausted the wavelength can be shifted to move the operating point further  44  Figure 2-10: Diagram of FISO system setup  51  Figure 3-1: The refractive index of transformer oil samples minus the refractive index of load tap changer oil samples obtained from same equipment from the field  62  Figure 3-2: Plot of measured oil refractive index versus aging time when exposed to a temperature of 120°C  66  Figure 3-3: Plot of measured oil refractive index versus time when exposed to a temperature of 150°C with different contents present  69  Figure 3-4: Plot of measured oil refractive index versus time when exposed to a temperature of 150°C with different contaminants present Figure 3-5:  72  Examples of different types of hydrocarbon compounds.  parraffinic compound (hexane).  (a) example of a  (b) example of a naphthenic compound (cyclohexane).  example of a aromatic compound (benzene)  (c) 74  Figure 3-6: Methanol extract refractive index versus the area of polar compounds measured by HPLC in nitrogen blanketed oil samples  82  Figure 3-7: Methanol extract refractive index versus the area polar compounds measured by HPLC in free breathing oil samples  83  Figure 3-8: Methanol extract refractive index versus the area of polar compounds measured by HPLC in naturally aged oil samples  85  Figure 3-9: Change in refractive index of naturally aged oils after methanol extraction versus the area of polar compounds measured by HPLC ix  86  Figure 3-10:  (a)  Chemical structure of benzene.  Chemical structure of 2-furaldehyde  (b)  Chemical structure of Furan  (c) 93  Figure 3-11: Change of refractive index of Luminol oil samples versus approximate acid number  96  Figure 3-12: Change of refractive index of Luminol oil samples plotted versus approximate ethane gas concentrations injected  99  Figure 3-13: Change of refractive index of Luminol oil samples plotted versus approximate acetylene gas concentrations injected  100  x  List of Symbols and Abbreviations IEEE  Institute of Electrical and Electronic Engineers  PCB’s  Polychiorinated biphenyls  IFT  Interfacial tension  ASTM  American Society for Testing and Materials  DGA  Dissolved gas Analysis  02  Oxygen  2 N  Nitrogen  2 H  Hydrogen  4 CH  Methane  CO  Carbon monoxide  2 CO  Carbon dioxide  6 H 2 C  Ethane  4 H 2 C  Ethylene  H 2 C  Acetylene  NPLC  High pressure liquid chromatography  KV  Kilovolt  ppm  Parts per million  UV  Ultraviolet  n  Refractive index  n’  Real part of refractive index Imaginary part of refractive index  N  Number of atoms per unit volume xi  Vacuum permittivity e  Electron charge  in  Mass of electron  co  Frequency Resonant frequency  7  Damping coefficient Wavelength  FOT  Fiber Optic Temperature sensor  FRI  Fiber Optic Refractive Index sensor  d  Cladding thickness of D-fiber  0 n  Refractive index of fiber core Refractive index of fiber cladding  HF  Hydrofluoric acid  next  Refractive index of external medium  fleff  Mode effective refractive index  Tr  Power transmission ratio Propagation constant  fir  Real part of propagation constant  3 fl  Imaginary part of propagation constant  L  Length of etched section of D-fiber  Pt  Power into leaky section of D-fiber  0 P  Power out of leaky section of D-fiber  DI  De-ionized  Pmeas  Power measured xii  Fmax  Maximum power  25  Refractive index at 25°C  T  Temperature Refractive index sensor resolution Change in refractive index  Ulvil  Universal Multicharinel Instrument  df  Distance between reflecting surfaces  F  Finesse  R  Reflectance  V35  Voltesso 35 oil  LTC  Load tap changer  TX  Transformer tank  RI  Refractive index  ppb  Parts per billion  xiii  Acknowledgements First I would like to thank my family and close friends for always being there for me, and showing constant love and support throughout my education.  The encouragement from my  mother, father, brother Shawn, and relatives helped keep me going when faced with challenges during my studies. I would like to thank my supervisor, Dr. N. A. F. Jaeger, for his expert guidance, continual patience, and support. He has taught me about more than just engineering during my studies at UBC. I would like to thank my colleagues in the optics lab for the company, support, and ideas they shared. Special thanks to Sameer Chandani for the time he set aside to discuss problems and provide assistance while he was conducting his own studies. I would like to thank Powertech Labs and its employees for their collaboration and the resources that were provided for this investigation. Special thanks to Salim Hassanali for his sharing of knowledge, helpful suggestions, and technical support. Thanks to Stevo Kovacevic and Edward Hall for their technical support as well. I would like to thank FISO Technologies Inc. for their collaboration in this work by providing their equipment to us. Thank you to Francois Bouchard for his support and interest in this investigation. Finally, I would like to thank Becky for her love and support.  xiv  Chapter 1 1 Introduction and Motivation The purpose of this chapter is to provide an introduction to the topic of high power electrical equipment monitoring and set the stage for the research conducted here. It begins with a brief overview describing the types of equipment that are of interest and their liquid and paper insulating systems.  The most common way of monitoring these types of equipment is by  monitoring the condition of the insulation. Hence, we review some of the methods currently used by the utility companies to monitor the condition of these equipment insulation systems. Some key indicators that are commonly used to assess the operating condition of such equipment, as well as the condition of the insulation itself, are presented. These key indicators include, but are not limited to, various aging by-products such as dissolved gases, acids, furans, and water, and an oil’s dielectric strength and color.  Hence, several of the tests used for  equipment monitoring are briefly described. In addition to the tests performed in the laboratory, many companies have developed systems for online monitoring in order to aid in diagnosing the operating condition of equipment. Research groups are exploring new techniques to improve on current monitoring methods. This research includes improvements to current in-lab methods, as well as in-situ techniques including online monitoring. techniques was conducted.  A literature review of some of the investigated monitoring This resulted in our decision to investigate how the aging of  transformers would affect the refractive indices of their oils, This investigation is the topic of this thesis.  1  1.1 Overview The electrical power industry uses a complex system to generate, transmit, and distribute electricity that is used by commercial, industrial, and residential consumers. This industry has grown rapidly since the early society.  th 20  century, and has now become an integral part of our modem  Lack of power has major social and economical impacts, as was observed in the  Northeast Blackout of 2003 (on August States and Canada.  th), 14  which affected many eastern cities in the United  Although such serious power outages are not a common occurrence,  electrical utility companies do often experience outages to a smaller degree which not only inconvenience both the companies and the consumers, but also generate a loss of revenue. For this reason extensive research has been conducted in order to develop techniques to indicate potential faults and future failures so that preventative action can be taken [1] [2] [3] [4] [5] [6]. Currently there are many new research areas being explored to make the electrical system even more efficient and to minimize equipment failure [7] [8] [9] [10] [111. In some cases, failure of a single piece of equipment used by an electric utility company may be the cause of a power outage. High voltage equipment is generally very expensive, so maintenance and care is taken not only to prevent failures from occurring, but to prolong the life of these expensive assets. This equipment includes, but is not limited to, oil filled transformers and oil filled high voltage power cables. Transformers that are used in an electrical transmission system to step up and step down voltage levels, in order to minimize power loss on transmission lines, are called power transformers. Transformers that are used at various points in the system to measure the voltage and current at different locations, are known as instrument transformers. An insulation system using both liquid and paper is commonly used for both types of transformers as well as for underground high voltage power cables.  There are many  maintenance activities performed in order to extend the life of equipment such as inspecting the 2  physical condition of a transformer’s bushings, tanks, and gaskets, but most experts would agree that the most important maintenance procedures involve checking the condition of the equipments’ insulation.  According to many standards organizations such as the Institute of  Electrical and Electronic Engineers (IEEE), the average life of a power transformer is 20 to 25 years, and the lifetime is usually related to the condition of the transformer’s insulation [12]. Paper insulation, comprised of cellulose such as Kraft-paper, has been used historically to insulate transformer conductors and can be used to insulate high power cables as well [3]. Other papers that can be used include Nomex Aramid paper and Polyester Composite based papers [13]. Good dielectric properties, high thermal rating, and low moisture absorption are all key characteristics of a good insulating paper. Over the lifetime of a transformer, the condition of the paper will degrade due to exposure to high temperatures, oxygen, moisture, and numerous other contaminants found in the insulation system. In many cases, the paper will work in parallel with the oil to provide insulation, in which case the condition of both the oil and the paper affects the equipment lifetime. Oil is used in electrical equipment not only due to its ability to provide good electrical insulation, but also because it is very stable at high temperatures. Initially mineral oils were used due to their availability, as they were fabricated by refining hydrocarbons collected in the distillation process of petroleum [7]. Mineral oils consist of basic hydrocarbon liquids such as paraffin, naphthene, aromatic hydrocarbons, and olefin [5]. Mineral oils are still most commonly used today in high voltage equipment, although companies are trying to find other liquids that may be better.  Synthetic oils based on polychlorinated  biphenyls (PCB’s) were introduced due to their low flammability, but in the 1970’s their use declined as the toxic effects on the environment became a concern, and restrictions regarding their use were put in place. In searching for substitutes that were nontoxic and noncombustible, ester liquids, silicone fluids, and vegetable oils were proposed, although they were more costly 3  and less readily available [7] Some commercially available transformer oils that are commonly used are shown in Table 1-1 with the type and refractive index listed. Table 1-1: Common commercially available transformer oils with type and refractive index listed. Insulating Oil GE SF97-50 Dow Corning 561 Rhodorsil 604 V 50 Clearco STO-50 Envirotemp FR3 Midel 7131 Biotemp ECO Fluid Shell Diala AX Volteso-35 Lurninol Tn  Type Silicone Silicone Silicone Silicone Ester Ester Vegetable Oil Mineral Mineral Mineral Mineral  Refractive Index @ 20*C 1.4000 1.4040 1.402 @ 25°C 1.4000 1 .47 50 1.4555 1.4708 1 .4600 1.4815 1.4743* 1.4552*  1.2 Review of Applied Equipment Monitoring Techniques In the next section we will discuss applied equipment monitoring techniques which are those insulating monitoring techniques most commonly used by utility companies today. Oils can be tested in many different ways, and data can be collected over time, in order to spot trends and to provide insight into factors that can reduce equipment lifetime. Initially changes in the insulation due to the influence of service conditions will occur at the molecular level, which will eventually lead to chemical reactions resulting in the formation of new chemical compounds [6]. For this reason, a variety of tests are performed to indicate the insulation condition. Oil may be tested for its gas content, dielectric breakdown strength, acidity, water content, oxidation inhibitor, ash content, viscosity, metal content, and interfacial tension (IFT) [4] [12] [14] [15] [16]. Test standards have been set by organizations such as the American Society for Testing  *  Measured using FISO FRI sensor at (discussed later in this thesis).  4  and Materials (ASTM) [16] and, in many cases, tests are performed accordingly.  In what  follows in this section, a variety of these tests currently being used will be discussed 1.2.1  Dissolved Gas Analysis  The most common test performed when analyzing the oil insulation of equipment is dissolved gas analysis (DGA) [l][2][12][14][15][17][18j. Over the life cycle of the equipment, many gases will be dissolved in the insulating oils for various reasons, and detection of these gases can be indicators of a piece of equipment’s condition. Absorption from the atmosphere, the breakdown of hydrocarbon chains present in the oil, and the breakdown of cellulose in the insulating paper can all contribute to the addition of gases such as oxygen (02), nitrogen (N ), 2 hydrogen 2 (H ) , methane 4 (CH ) , carbon monoxide (CO), carbon dioxide 2 (C0 ) (C ) 6 H , ethane 2 , ethylene 2 (C ) 4 H , acetylene 2 (C ) H , and other hydrocarbons to the oil [7]. Since various gases will be generated under various conditions, the presence and quantity of a particular gas can be a significant indicator of a particular problem. By measuring the changes in relative levels of these gases, faults such as arcing, corona, overheating of the oil, and overheating of the cellulose  can be revealed. Temperature is one factor that can cause the generation of fault gases and, in general, can greatly reduce the lifetime of a transformer. An increase in operating temperature of 10°C above the equipment rating may reduce a transformers operating life by about one half [12]. Higher temperatures can be created due to failures of cooling systems, equipment overloading, and the presence of arcing or electrical discharge. Various fault gases will be generated at different temperatures and at different rates as shown in Figure 1-1. Electric utility companies will often take oil samples from their oil filled equipment and send them to a laboratory such as Powertech Labs, in Surrey. British Columbia, Canada, where the oil can be analyzed for fault gases using  5  Hydrogen H2  Methane  Ethylene  1•  C2H4  -  Nomial Op erafion  C2H2  Acetylene C2H2  Hot Spots  I  100 200 0 Partial Discharge Not Temperature Dependent  300  Arc  400 600 500 Temperature °C  700  800  900 1000  Figure 1-1: Generation of combustible gases in transformer oils due to temperature and faults (not to scale). This figure is similar to the gas generation chart found in [1].  6  the DGA method.  This is most often done by separating the gases from the oil using gas  chromatography so that the quantity of each gas can be detected [6] [9]. Gases such as hydrogen and various hydrocarbons, in particular, methane, ethane, ethylene, and acetylene are formed due to the breakdown of hydrocarbon chains present in the oil. This breakdown is part of the oil degradation process. As shown in Figure 1-1, partial discharge will lead to the formation of many of these gases, known as combustible gases, although the concentration of H 2 will be much more prevalent as compared to others. At about 150°C, hydrogen and methane will begin to be generated which can be attributed to hot spots or corona discharge [1]. Further overheating of the oil to 250-300°C, can lead to the production of more methane and ethane, which may occur in the presence of hot spots as well.  As the  temperature rises to yet higher levels, and eventually reaches about 350°C, ethylene will be produced while the production of other gases will not be as pronounced. Acetylene is produced at extremely high temperatures starting around 500°C and becomes more commonly found in oil experiencing arcing conditions, which typically generates temperatures in excess of 700°C. Large amounts of hydrogen are also generated under arcing conditions.  By analyzing the  concentration of each gas, decisions can be made regarding the state of equipment, and it can be determined whether or not the equipment is operating safely or if it should be taken offline and repaired or replaced. Faults may be revealed when the concentrations of gases are known, and the type of servicing needed for the transformer may be established using diagnostic procedures [1] [2]. The IEEE has developed a guide to classify the state of transformers based on gas levels which rates them from condition 1, which indicates that the transformer is operating safely, to condition 4, which indicates that the gas concentrations exceed safe limits and the equipment should be taken offline [19]. Another popular method used to analyze the type of fault occurring in the 7  equipment is the Rogers Ratio Method where ratios between acetylene and ethylene, methane and hydrogen, and ethylene and ethane indicate the type of fault that is likely occurring [1]. The amount of dissolved gas found in an oil sample can be used as a guideline for fault analysis, but even more important is the rate of increase of these gases. The total concentration of gases found in an oil includes the collection of gases over the entire lifetime of the equipment which may be due to small amounts being generated over a long period of time, or large amounts being generated over a short period of time. Therefore, it is also important to know the history of the equipment as well, so that current levels can be compared to previous levels to see if there is a sudden increase. A general rule that may be used is, if the gas concentration increases by 10% of the maximum allowable concentration in a month, then there is a problem [12]. Other than hydrogen and the hydrocarbons thus far discussed, other gases such as carbon monoxide, carbon dioxide, oxygen, and nitrogen may also be detected when performing DGA. Carbon dioxide, oxygen, and nitrogen may be present in the oil, as they can be absorbed from the surrounding air, depending on the construction of the transformer. In most cases, insulation exposure to air is avoided by sealing and pressurizing the equipment. This not only aids in keeping the oil isolated, but also helps stabilize the system to account for large pressure changes that can occur due to temperature fluctuations and/or arcing [15]. There are many different transformer designs used, including the older free breathing style where air exposure is more common, those using bleeder valves that allow air to leave and enter the enclosure to compensate for pressure changes, free breathing conservator types (where oil contained in a separate conservator will be in contact with air), conservator system types having a bladder to reduce air exposure further, and systems filled with an inert gas such as nitrogen which minimizes the amount of oxygen and moisture that comes in contact with the oil [12]. In any case there will most often be these atmospheric gases present in a sample as well as moisture. Oxygen and 8  moisture can greatly reduce the lifetime of a transformer, and detecting certain levels can indicate a leak in the system. Carbon gases such as carbon monoxide and carbon dioxide can be generated due to the breakdown of cellulose contained in the paper insulation, which can give an indication of the insulating paper condition. When temperatures reach about 100°C, the cellulose insulation will begin to decompose, so detecting these gases can give some indication of overheating [12]. Although detecting these gases can give some indication of the breakdown of paper insulation, detecting the amount of furans is most often used to measure a paper insulation system’s strength. 1.2.2  Furans  It is not practical to directly measure the insulating paper’s tensile strength, or degree of polymerization, as this would require the removal of a strip of paper from the winding when the transfonner is not in service [3]. Other means are, therefore, necessary in order to assess the condition of the paper.  Oil can be tested for the concentration of furan compounds which  provide an indication as to the degree to which the paper has been degraded [7] [20].  In  particular, the concentration of 2-furaldehyde has been directly related to the degree of polymerization of the paper [21].  Furan testing can be performed using the same samples that  are extracted for DGA according to ASTM Method D 5837-99 [22].  Companies, such as  Powertech Labs, will perform screen testing to measure the amount of 2-furaldehyde contained in a sample and, if high levels are found, further testing for more furans using high pressure liquid chromatography (HPLC) are conducted [23].  Furans are formed in the presence of high  temperatures, oxidative compounds, acids, and moisture, and can be used to estimate the residual lifetime of the paper insulation [17].  9  1.2.3  Moisture  Water may be present in an insulation system. It may be in the form of tiny droplets suspended in the oil, it may be dissolved in the oil, or it may be in a free state usually at the bottom of the taik holding the oil [14]. The dielectric strength of oil is weakened when moisture is added to it. The combination of moisture and oxygen will degrade paper insulation at an accelerated rate and can lead to the formation of acids and sludge. Moisture can be present in oil filled electrical equipment if it is absorbed by the paper insulation during manufacture, or can enter the system through a leak in the form of water or humidity. The amount of water that can be dissolved in oil is temperature dependent, and higher temperatures allow for larger concentrations [15]. Also, moisture will constantly redistribute itself between the paper and the oil, depending on the temperature. For example, at 20°C the ratio of water in the paper as compared to in the oil may be as high as 3000:1 whereas at 60°C the ratio may only be 300:1 [12]. It is, therefore, important to record the temperature of the system when extracting an oil sample in order to estimate the water content in the paper insulation. In most cases, levels of moisture ase analyzed in oil samples when performing DGA and the concentrations are measured in parts per million (ppm). The concentration of moisture can be compared to the percent saturation at the measured extraction temperature to determine whether the level is acceptable and if it is increasing.  The moisture content in oil samples can be measured  according to ASTM Method D 1533-00 [24]. 1.2.4  Oxygen  Another important factor that can aid in analyzing insulation quality is the concentration of oxygen in an oil sample.  Levels of oxygen exceeding 2000 ppm may deteriorate paper  insulation at an accelerated rate and a concentration of 10000 ppm indicates that the oil should be de-gassed [17]. The presence of oxygen can lead to chemical reactions that form acids and 10  poiar compounds which can eventually lead to sludge. Sludge will coat the windings and may cause and/or contribute to heat transfer problems. In sealed transformers oxygen can enter the system through leaks or be produced in the presence of water. Open breathing type transformers will obviously have a higher concentration of oxygen in the oil and will, therefore, experience a more rapid rate of oxidation and moisture related effects. Higher temperatures also contribute to faster oxidative breakdown, which emphasizes the need to keep operating temperatures at suitable levels and to avoid overloading [14]. A transformer operating with normal levels of oxygen present in the insulation system may have a lifetime up to ten times longer than one operating with higher levels [17].  In order to reduce the levels of oxygen, and its effects,  antioxidants and oxygen inhibitors may be used. An antioxidant such as 2,6-di-tert-butyl-pcresol can be added to a transformer oil to prevent oxidation; decreases in the antioxidant concentration can be used to characterize the condition of the liquid insulation [6].  The  oxidation process can also be slowed down naturally as some oils contain chemical compounds that act as natural inhibitors. Laboratories that perform DGA will often perform tests to measure the amount of oxygen inhibitor present and, if levels become too low, the antioxidant may be replaced when the oil is treated. 1.2.5  Interfacial Tension (IFT) Interfacial tension is a measure of the boundary strength at an oil/water interface. An oil  sample should “float” when added to water, creating a distinct boundary between the two. The interfacial tension between oil and water is weakened in the presence of polar compounds and other contaminants formed by oxidation [14]. The force, in dynes per centimeter (ASTM D 971), needed to pull a small wire through the oil/water interface is a measure of IFT [25]. One dyne is equal to iO newtons. New oil should measure about 40-50 dynes per centimeter [17]. Over the lifetime of a transformer the IFT will generally decrease exponentially. An increase in 11  the acidity is usually observed as well, although it usually lags the decreasing IFT. When both are measured, low IFT and high acidity can provide an excellent indication of poor oil quality. Sludging will often begin at IFT levels around 22 dynes per centimeter and at high levels of acidity [14]. 1.2.6  Neutralization Number/Acid Number  Acids tend to accelerate the breakdown of paper insulation and act as catalysts to the degradation of transformer oils [13]. Acids are formed during the aging of insulation through oxidation and will attack metals, will form sludge at a neutralization number of about 0.4, and will attack the cellulose in the paper, greatly decreasing the equipment lifetime [12].  The  neutralization (or acid) number is found by measuring the amount of potassium hydroxide (KOH) required to neutralize the acids in 1 g of oil. The neutralization number test can be performed according to ASTM D-974 [26].  This test is not a direct indication of the oil’s  dielectric strength, but does indicate the presence of contaminants and lowered oil quality. 1.2.7  KY Breakdown/Dielectric Breakdown  The KV (Kilovolt) Breakdown or Dielectric Breakdown test is a measure of the oil’s dielectric strength or ability to withstand electric stress [15]. Laboratories test for dielectric strength by applying large voltages to oil samples and recording the levels at which the oils break down. Factors that can affect the measured dielectric breakdown voltage of insulating oils include water content, oxidation products, size and number of particles in the oil, and, if saturation levels are exceeded, concentration of dissolved gases [27]. The acceptable minimum breakdown voltage, according to the IEEE [28], is 30 kV for transformers operating at 230 kV and above, 28 kV for those rated between 69 kV and 230 kV, and 23 kV for those rated at 69 kV or less, using a 1 mm gap between electrodes as outlined in ASTM D 18 16-04 [27]. Other tests  12  must be done in addition to measuring the dielectric strength of oil since, for example, the paper insulation may be severely degraded long before the oil insulation begins to break down. 1.2.8  Color  New or unused oils will most often be clear before being placed into a transformer tank and stressed. Some older types of oil will have a light yellowish appearance when unused. As oils are aged, and contaminants are formed, they will begin to change color, becoming darker. An oil’s color is often measured by comparing it to a color wheel and assigning a number, wherein a higher number indicates a darker color [15]. The test procedure for measuring color is described in ASTM D 1500 [29]. 1.2.9  Polar Compoundst  As a transformer ages, and the oil and paper degrade, various contaminants are produced and can be found in the oil. Polar compounds such as aldehydes, ketones, and alcohols, may be formed during the aging cycle as by-products of chemical reactions [30].  These chemical  reactions may occur due to oxidation, hydrolysis, and/or polymerization. If polar compounds are present in an oil the insulation quality of the oil will be weakened, and detecting them generally shows that the oil has been degraded. Labs often monitor the level of polar compounds in order to help detennine the quality of the oil.  The concentration of polar compounds can be measured using the same HPLC  equipment used to measure the concentration of furans. At Powertech Labs an HP Series lIt  —  Part of this subsection is pending publication. Kisch, R.J., Hassanali, S., Kovacevic, S., and Jaeger, N.A.F. (2007) The effects of polar compounds on refractive index change in transformer oils, Proceedings of High Voltage and Electrical Insulation Conference ALTAE 2007. Equipment by Hewlett Packard test and measurement division now known as Agilent Technologies, Santa Clara, CA, USA  13  liquid chromatogram is used. This piece of equipment can measure the concentration of various chemical compounds by mixing a solvent (or mobile phase) with the sample (or analyte) and passing them through a column containing solid material (or stationary phase).  When the  mixture is passed through the stationary phase, individual compounds originally contained in the sample will be eluted at specific retention times. The retention time is the length of time which elapses between in injection of the solution, containing the compound, into the stationary phase and the detection of the compound after it has passed through the stationary phase.  A  chrornatograph will be produced, containing a series of peaks plotted as functions of time. The type of compound can be identified and the concentration can be calculated using the retention time and area under the peak, respectively. As will be discussed later in the thesis, this method was used in one of our experiments. The relative concentration of polar compounds in an oil sample was found by adding the total area under the peaks of its chromatograph. This was repeated for many samples and comparisons were made. 1.2.10 Online Monitoring Online monitoring is a very efficient way to detect faults in real time and increases the capabilities of an electric utility company to prevent failures. The purpose of online monitoring is not to eliminate current techniques such as oil sampling and laboratory analysis, but can be used to assist in decision making regarding the health status of the insulation and the operation of the equipment. It aids in monitoring the equipment between sampling intervals and can reduce the sampling frequency needed. The value of online monitoring has been strongly expressed by professionals working in the field and extensive research has been, and is currently being, performed to this end. A few companies have developed online oil monitoring equipment, including systems measuring levels of dissolved gases, moisture, and dielectric strength. Combustible gas monitors 14  have been available commercially for quite some time and are widely used [311.  The GE  Hydran was one of the first commercially available gas monitors and it functions using a membrane that allows hydrogen and other combustible gasses to permeate through it. After being passed into a cell, the gases are detected by measuring the electric current that is generated as they are “burned”. This product can be used to measure the change in concentration of all gases collectively; however, it cannot be used to measure the change of each gas individually. For companies that prefer the detection of hydrogen, the Morgan Schaeffer Calisto** may be used. This device uses a polymer barrier to separate the gases from the oil and a capillary tube which extracts hydrogen only. A thermal conductivity detector is used to measure the amount of hydrogen extracted from the oil. This device can also measure moisture using a solid state detector [32]. Morgan Schaeffer, along with many other companies, also offer portable dissolved gas analyzers, however, a technician is still required to set up the measurement apparatus at the equipment site. These portable units often work using gas chromatography, which allows for the detection of the eight key fault gases. A few companies, such as Serverontt, have taken the same type of technology used in the portable analyzers and designed stationary online monitors. Although instruments such as the Serveron Online Transformer Monitor have the accuracy to meet a laboratory grade DGA, few utility companies can afford to outfit many transformers with this equipment due to its high cost [311.  § Equipment by General Electric Energy, Atlanta, GA, USA  **Equipment by Morgan Schaeffer, LaSalle, QC, Canada Equipment by Serveron, Hilisboro, OR, USA  15  1.3 Review of Research into Insulation Diagnostics In this section we will discuss some of the techniques that have been investigated for the purpose of insulation diagnostics and fault detection but that have not commonly been applied in the field.  Numerous groups have investigated various high power equipment monitoring  techniques. Gas and chemical sensors have received significant attention for this purpose. These sensors have become important to several industries.  They are often used for chemical  processing, for medical applications, and for molecular biotechnology. Since the types of gases and chemicals present in the insulating oils can provide helpful information about the equipment, gas and chemical sensing has become widely researched by groups in the power industry. There are a variety of techniques that can be used for gas and chemical sensing, all which have advantages and disadvantages, depending upon the application.  For example, electrochemical  or solid state detectors may be very useful for chemical sensing but, when used in a substation to monitor transformer oils, the interference caused by high electromagnetic fields can affect the reliability of their readings. Optical means are useful to the power industry because the required electronics can be located remotely.  Optical signals are typically transmitted using optical  fibers, which are immune to electromagnetic interference. One optical technique that has been widely used to determine the presence of chemical species and gases is optical spectroscopy [33]. The application of optical spectroscopy to assess the condition of transformer insulation has been extensively investigated and is currently the subject of ongoing research [9] [34] [35] [36] [37]. In absorption spectroscopy changes in specific atoms’ and molecules’ energy levels, due to the absorption of light, are used to identify their presence in a sample. By directing light through samples and analyzing the transmission or absorption at particular wavelengths, one can determine whether a particular gas or chemical compound is present in the sample. 16  Many groups have designed systems that showed sensitivity to gases in air, but some of these sensors have not been used in transformer oil yet. In a study conducted in 1992 [38], aDfiber sensor was used to detect methane in air by measuring the absorption, of the evanescent field, at 1660mn, with a sensitivity in the 1000 ppm range. More recently, in 2004, a fiber optic system was constructed which could detect the presence of multiple gases by placing silica tube between hollow sections of fiber, and measuring the molecular absorption [34]. Acetylene and carbon monoxide lines were observed from 1520-1540 and 1560-1570 nm, respectively.  Both  of these sensors, however, were used for proof-of-principle, and were not used to detect gases present in insulating oils.  It is unknown how detectable the gases would have been in  transformer oils, or if the sensors would function as efficiently. Studies have been performed to develop sensors for the detection of gases and chemicals, based on absorption, which could operate in transformer oils. Some of these methods, however, require the aid of another technology that either separates the gases or chemicals from the oils or otherwise aids in their detection, before absorption is measured. In 1998, absorption of light at 530 nm was used to measure the concentration of furans, as low as 0.lppm, in oil samples [39]. Here, a novel material, which was invented by the authors, was formed using the sol-gel process. The material was placed between plates which were immersed in oil samples. When the plates were immersed, furans were absorbed by the material and the concentrations of furans present were determined by the amount of absorption of light transmitted through the material. This technology was developed to form the basis of a portable instrument which could be used in the field, but which required an operator.  The authors outlined the possibility of developing a  continuous monitoring system, which could be permanently installed on a transformer, but to the best of our knowledge, no such system has been built.  17  In 2002 an optical fiber sensor was developed which used absorption as a means of detecting the presence of methane in transformer oil. A polyflon membrane was first used to separate the methane from the transformer oil before it was detected as a gas. This methane sensor was used for proof-of-principle, and demonstrated detection at 28500 ppm.  Other  techniques used for separating gases from oils have found their way into commercial products, as was discussed in the previous section. Nevertheless, the most powerful method for separating gases and measuring the concentration with the highest sensitivity is gas chromatography. In the last decade, some groups began to characterize transformer oils by their absorption profiles, however, they did not only look at gases and furans in oils. A few groups have linked the formation of aromatic compounds to oil degradation, and have related the absorption peaks at specific wavelengths to their presence [9][35][36].  In the years between 2000 and 2004,  publications were released which stated that the formation of aromatic compounds in the transformer oils contributed to changes of the absorption profiles in the UV region (200-3 90 nm) and that levels of these compounds increased as oil was degraded[9] [35]. Experiments were conducted using oils taken from failing transformers, as well as using oils in which various transformer faults such as arcing, overheating, and increased oxidation where simulated. The authors claimed that measuring absorption at 390 nm could help differentiate between a transformer failing from either thermal or arcing faults. Around the same time, in 2001, the authors of [36] claimed that the grade of a transformer oil could be determined by the amounts of aromatic compounds as well.  In this study, however, the authors looked at the spectral  characteristics between 4000 and 1710 cm’, or 2500 and 5882 nm. Various transformer oils were used that differed in technical grade, service life, and content of antioxidant additive and dissolved water.  The authors concluded that the degree of service deterioration could be  18  determined by measuring the optical density of the oil at 1710 cm , or 5848 nm, and that small1 size IR spectral equipment could be used for this purpose. Methods other than using absorption have been researched as well.  One of these  methods is optical hydrogen detection using either a palladium film or using a palladium-silver alloy film. The interaction between hydrogen and palladium results in the formation of a metal hydrogen alloy, or hydride. When the sensing area of these types of detector are exposed to hydrogen, the electrical and optical properties of the metal change as a result of the shift in electronic structure [40] [41]. A few groups have created sensors by measuring the optical power reflected from a surface with this type of coating.  In 2002 a sensor was tested using  transformer oil and detected hydrogen concentrations from 200-1500 ppm [42].  Higher  sensitivity was demonstrated previously in 1996, however, by a group which coated the end of a fiber with a palladium-silver alloy and detected hydrogen in transformer oils to concentrations as low as 50 ppm [40]. Although this type of system could be constructed at a relatively low cost, the sensitivity does not match that of gas chromatography. The authors also revealed that high levels of other gases can contribute to faulty readings. In 2005, a group measured the complex permittivity of oils in the frequency range of 20Hz to 1MHz where ionic and molecular polarization processes are expected to dominate [7]. It was observed that the addition of aging by-products, such as water and polar compounds, could be detected in transformer oils. The change in the real part of the permittivity due to temperature change increased with increasing moisture present in the oil. Additionally, the real part of the permittivity increased with moisture content, and the imaginary part increased with “polarizable inclusions”. A prior study was performed in 1998, by J. Unsworth and N. Hauser, showing similar results.  Changes in oils’ permittivities were measured and related to the  formation of polar compounds [43]. In this study, however, changes in oil refractive indices 19  were also measured. The authors measured increasing permittivities of transformer oils as they were degraded, which they attributed to the addition of polar compounds. The changes in the oils’ refractive indices were very small, however, and they did not recommend using these changes as an indicative parameter of polar compounds. At this time the group used a refractive index sensor with a resolution of 0.0002, whereas today we have sensors with resolutions at least one order of magnitude higher [44]. Tn 2004, T. Aka-Ngnui et al. claimed that changes in refractive indices of transformer oils occur due to the generation of oil degradation products [11]. Voltages large enough to “break” an oil were applied across electrodes which were immersed in the oil. The cladding of an optical fiber was removed from a 2 cm section and this sensing section was immersed in the oil as well. Changes in refractive index were detected by a loss in the optical power transmitted through the fiber, however, actual refractive index measurement values were not obtained. The change in refractive index was attributed to the formation of degradation products, however, further investigation was needed in order to determine what products were formed and the concentration of each.  1.4 Our Investigation The applied equipment monitoring techniques being used for laboratory diagnosis are very effective for determining the health status of equipment when samples are provided. However, there is still a need for systems which can “flag” the electrical utility companies if the operation of the equipment or quality of the oil degrades greatly between sampling intervals. Since the importance of devising new techniques for online monitoring has been thoroughly expressed in literature, as well as by professionals working in the field, we decided that exploring online monitoring further would be of value to the power industry, as well as to the scientific community. After conducting a thorough literature review it was clear that the various 20  methods studied for the purpose of online monitoring each had their advantages and disadvantages. One single method did not seem to have emerged as a clear favorite over others for determining the quality of the oil and the operating condition of the equipment. Using refractive index as an indication of oil quality and equipment operation had not been extensively researched. The studies performed by J. Unsworth and N. Hauser and by T. Aka-Ngnui et a!. indicated that measuring refractive indices of transformer oils might be useful with higher resolution sensors.  Since measuring refractive index is relatively inexpensive,  compared to other diagnostic techniques, can easily be incorporated into a system and implemented for online monitoring in the presence of high voltages, and can provide high resolution using a method devised by a previous member of our group, we decided to explore its use further for determining oil quality and operation of the equipment. Since this topic has not been thoroughly explored by other groups, we did not know how the changing properties of oils would affect the oils’ refractive indices and, if measuring the refractive indices of oils would be a useful “flag”. In what follows, we have studied the effects of transformer oil aging, and of some of the by-products formed during aging, on the refractive index changes of the oils.  Several  experiments have been conducted in order to observe the degree to which the refractive indices of transformer oils change during the aging process. Initial experiments included measuring the refractive indices of oils obtained from the field and trying to detect trends. Oil samples were prepared in the laboratory as well, through accelerated aging, in order to study the effects of aging on refractive index in a controlled manner. Using our high resolution sensor, changes in the oils’ refractive indices due to the addition of various aging by-products such as acids, furans, polar compounds, and gases, that where not previously detected, are made observable when introduced in sufficient quantities.  These changes, due to the inclusion of individual by 21  products, are compared to the overall changes measured in refractive indices of oils when they are aged at an accelerated rate. This work constitutes an initial investigation conducted with the intention of contributing to the development of online monitoring systems for oil filled high voltage equipment.  22  Chapter 2 2 Measuring Refractive Index 2.1  Introduction to Sensors The purpose of this chapter is to discuss our methods for measuring refractive index. The  chapter begins by briefly discussing why we thought measuring refractive index of insulating oils may be promising for the purpose of transformer condition monitoring. The sensors that were used for our measurements are introduced, one of which was fabricated in our lab (the D fiber sensor) and one of which was obtained commercially (the FISO sensor). The fabrication, system setup, calibration, and testing procedures used to make a reliable D-fiber sensor are discussed in detail, since until now it has only been used for demonstrating proof-of-principle. Since the FISO sensor is already a commercially developed product, the theory of operation is briefly discussed.  2.2  Why Use Refractive Index? The interaction of an electromagnetic wave’s electric field with atoms and molecules  present in a medium will affect the propagation of the wave and, therefore, the dielectric constant (and refractive index) will be dependent upon the manner in which atoms and molecules are assembled [45j. The refractive index can be represented as a complex value: n=n’—in”  The reader is directed to [45] if they are not familiar with issues related to the refractive index of a material.  23  (2-1)  where n’is the real part of the refractive index and n” is the imaginary part. The real part of the refractive index accounts for the effect a medium will have on the velocity of the electromagnetic wave traveling through it and the imaginary part gives the absorption and, therefore, is sometimes referred to as the extinction coefficient [45]. For the simple case of an isotropic medium occupied by N atoms per unit volume, the complex refractive index is given as follows [45]: jj-  2-  2  —  2  2  “.0 —  2  22 [(cv —cv 22 2m& 0 ) +y cv ]  2  22 [(w —cv 22 2m& 0 ) +y 0) J  —  where s is the vacuum permittivity, e is the charge of an electron, m is the mass of an electron, 0 is the resonant frequency of the electron cv is the frequency of the electromagnetic wave, cv  motion, and ‘I’ is the damping coefficient. Looking at this equation we see that the real term becomes 1 and the complex term is maximized when cv is equal to cot,. The imaginary term is , and can be often neglected. 0 significant only when cv is very close to co  Equation (2-2) illustrates that the refractive index is dependent on the frequency of the electromagnetic wave. This phenomenon is referred to as chromatic dispersion. Figure 2-1 shows a plot of the real and imaginary parts of the refractive index as functions of frequency for a material that may be represented by (2-2). One can see that the absorptive part of the refractive index is a maximum when cv  , and approaches 0 as one moves away from this point. For the 0 co  plot of the real part in the region where o < w 0 the refractive index increases with the frequency, until a point very near ca. where the slope becomes negative. The negative slope only exists where absorption is significant, and as the frequency increases beyond this region the slope becomes positive again.  g Note: a similar figure is presented by A. Yariv and P. Yeh in [45].  24  I  —  I?’’  n’-l  0-  I -‘+  -3  I  -2  I  -1  I 0  I  I  1  2  I  I>  3  4  (w-w y Figure 2-1: Normalized plot of the real and the imaginary value of refractive index as a function of frequency. A similar figure is found in [45].  25  Equation (2-2) only represents the refractive index that would be calculated due to the collection of single atoms having electrons with only one  Wc,  value. This equation is, therefore,  only used for demonstration purposes. The refractive index of a dielectric material such as a mineral oil, would obviously be affected by a multitude of electrons, atoms, and molecules, and would constitute a summation of the contributions of each.  Using this simple illustration,  however, it is easy to visualize how the refractive index would change due to the formation of new compounds and the breakdown of others. Figure 2-2 shows a plot similar to that shown in Figure 2-1, although only the real part of the refractive index is shown and the frequency has been converted to wavelength on the x axis. In Figure 2-2, line A represents the real part of the refractive index of a medium with “type 1” atoms present, having only one resonant wavelength Aj. If the refractive index is measured at Am the value of n’ will be n ’. If the type 1 atoms are 1 removed from the medium, and replaced with “type 2” atoms, a new resonant wavelength may exist at , 02 and the real part of the refractive index will now be n A ’ shown on line B. The 2 shifting resonant frequency will obviously result in the shifting of the imaginary refractive index profile as well. Similar behavior would occur in the break down of compounds and formation of new compounds, in that the real part of the refractive index and the absorption profile would both change. In the previous chapter, we discussed various techniques used as indicators of oil degradation, and many of them involved measuring the changing absorption profiles. For example, in [9] increased absorption was observed between 200 and 390 nm due to the formation of aromatic compounds. If the absorption profile of the medium changed, the real part of the refractive index must have changed as well. Since the composition of the oil changes during aging, we expect the refractive index to change as well. We have, therefore, studied the changes that occur in the refractive indices of the oils, by using our sensors to measure the change in the real part. 26  A  B  712  Figure 2-2: Real value of refractive index versus wavelength illustrating change in refractive index values with different resonant frequencies.  27  2.3  Introduction to Sensors In order to measure refractive index, two sensors at very different stages in their  development were used. One of the sensors was fabricated, here at UBC, by etching D-shaped optical fiber to produce a sensing region. When the sensor was immersed in a liquid or gas medium, the power transmission measured through the fiber could be related to the medium’s refractive index value.  This method was developed by Sameer Chandani, until recently a  member of our lab, who has demonstrated proof-of-principle prototypes [44].  To our  knowledge, other than in our lab, this sensor has not been used in any other experimental setting and has not yet been made into a commercially available product. This sensor was chosen because it had a higher resolution than many of the commercially available sensors.  One  limitation of our D-fiber sensor is its comparatively narrow operating range. The other sensor used was a commercially available measurement system on loan to our lab by FISO Technologies Inc. The system included both a Fiber Optic Temperature sensor (FOT) and a Fiber Optic Refractive Index sensor (FRI). FISO Technologies is a company (located in Quebec City, Quebec, Canada) that offers a variety of fiber optic sensors such as pressure, strain, refractive index, and temperature, all of which function using the Fabry-Perot cavity and Fizeau interferometer principles. Using the FISO system complemented the use of the D-fiber sensor, as it has a wider operating range.  Nevertheless, the FISO sensor did not have as high a  resolution when compared with that obtainable in the range where the D-fiber sensor was most sensitive, so would only be used for less sensitive measurements.  2.4  The D-Fiber Sensor In circular core, step index, single mode fibers a “guided mode” will exist. This is a  mode in which the optical field is confined to the core and the field is in the shape of a Bessel 28  function J 0 in the core and decaying in the cladding in the shape of a modified Bessel function of the second kind K 0 [44]. Figure 2-3(a) shows a magnified cross section of a typical circular core, step-index, single mode optical fiber. The core is surrounded by a cladding with a large thickness, and the cladding is surrounded by an external medium, which is typically a protective jacket.  Figure 2-3(b) shows the magnified core surrounded by the cladding, with the radial  refractive index and optical field distributions in the core and cladding regions. The field in the cladding is evanescent and decays as one moves away from the core into the cladding of the fibers, as shown in Figure 2-3(b). Depending on the thickness of the cladding, the evanescent field could extend into the outer medium surrounding the fiber. However, standard single mode fibers have large cladding thicknesses and, therefore, the interaction of the evanescent field with the outer medium is virtually non-existent. For our sensors, we wished to access the evanescent field and, therefore, we used a single mode D-shaped fiber. A D-shaped fiber is a specialty fiber that has an outer cladding with a regular cylindrical shape on one side and a planar side extending the length of the fiber as shown in Figure 2-4. The cladding is surrounded by a protective jacket. The thickness of the cladding on the planar side of the fiber, or distance, d, between the core and the protective jacket, is only 13 jim. The core of the fiber has an elliptical shape. The refractive index of the elliptical core, , 0 n  is greater than the refractive index of the cladding, n. As is the case for a typical standard  single mode fiber, light launched into the D-fiber will normally be guided with minimal loss. In our sensors, accessing the evanescent field will allow a decrease in the transmission, depending  on the refractive index of the surrounding medium. Hence, in order to access the evanescent field, we must reduce d by etching the fiber. For our sensors, the evanescent field is accessed by removing the fiber’s protective jacketing over a small section, and etching the fiber. 29  This etching process, done by exposing  flci  nfr) flco  t -  lid  o Ez(r) N N  o  N  a  (b)  (a)  Figure 2-3: (a) Magnified cross section of a typical step-index circular single mode fiber. (b) Magnified cross section of the core showing the refractive index profile and the optical field distributions. Decaying optical fields in the cladding are called evanescent fields. A similar figure found in [441. (Figure not to scale).  30  Figure 2-4: D-fiber cross section (not to scale), showing the core dimensions, cladding thickness “d” between the core and outer cladding flat side, and the protective jacketing surrounding the cladding.  31  the cladding to hydrofluoric acid (HF), reduces d. By reducing d, the evanescent field can be made to interact with a medium external to the fiber. Figure 2-5(a) illustrates this, showing a section of D-fiber that has been etched over a length, L. Figure 2-5(b) and Figure 2-5(c) show a “cut-out” section of the D-fiber (note the cut out has been rotated) shown in Figure 2-5(a). When d is reduced to d?. the evanescent field, which is normally confined to the cladding, extends into the external medium. If the refractive index of the external medium, ex becomes greater than the mode effective index, flef the mode of the overall waveguide structure becomes a “leaky mode”.  Leaky modes are a subset of radiation modes, which are characterized by  oscillatory fields in the cladding that are not highly iossy [44]. We exploit this “leaky behavior” to form the basis of our sensor, and relate the power transmission ratio, Tr, to the refractive index of the measurand. Since the propagation constant, J3i, for a leaky mode is complex (J3 1  =  fir  +  /3,,  where fir is the real part and / j is the imaginary part), Tr will depend on the length of the “leaky” 3 section, L, and on the imaginary part of the propagation constant/3 [44]: =  =  i 2 e  (2-3)  where P, is the power into the leaky section, and P 0 is the power out of the leaky section. A typical transmission profile, as a function of next, of this type of sensor is shown in Figure 2-6. If the external medium is our measurand and n is lower than n the sensor will operate in its lossless region. There is a narrow range of refractive index values to the right of the lossless region that we refer to as Region I. This is the region in which the steepest decrease in Tr occurs as a function of next.  We have defined a second region, Region II, in which a  minimum point in the transmission occurs. Region II starts to the right of Region I where the slope is not as steep, and ends at Region III. Region III is the region with the largest range of next, where the slope is positive for further changes  32  Of next,  but is not as steep as that of Region I.  (a) Etched Section  4d 1  d y x p  —  Cladding  I  I  External Medium  I’d  External  Medium  I’exl  Planar Side Cladding Boundary  Iy  Reduced Planar Side Cladding Boundary  L  L  I’)  n(y)  ilco lid  flco I’d  I’ext  -  0  a  I  o  a+d  Ez(y)  i  a±dr  tz  -d,. 4 a  E4’y)  N.  0  ;  a+d  0  -.7  (c)  (b)  Figure 2-5: (a) Section of D-fiber: For a section of D-fiber, with length “L”, the distance “ci” between the core and planner side of the cladding is reduced by Ad giving a new distance dr. (b) and (c) show the respective refractive indices and optical field distributions in the “cut-out section” shown below (a) [note, co-ordinate system rotation]. (b) shows a section not etched, with d between core/cladding interface and field confmed to the fiber. (c) shows a section after etching, with reduced distance dr and field extending into the external medium. 33  I  0.8 0  ?  0.4  0  0.2  0.  I I I  I I I  I  I  I I I I  I I I I  I I I I I  I I I I I  I I I I I I I  I I I I I I I  I I I  I I I  I  I  Lossless  ...“  —  ——  0i13 0112 0111  III  II WI I  I  1.4400  I  I  1.4500  I  1.4600  I  I  I  1.4700  1.4800  1.4900  1.5000  Refractive Index Figure 2-6: Calibration curve measured by sweeping the refractive index of the three thermo optic oils by temperature control, and recording the power transmission. Region I, II, Ill, and the lossless region are shown.  34  Decreasing d results in increasing the amount of power that will be lost in the leaky section of the fiber. This will give a lower minimum point, in Region II, and a larger change in Tr for a smaller change in  In [44] it was shown that in Region I minimizing d would  maximize the resolution, but in Region III the maximum resolution occurred at d  4.0 p.m. The  sensor’s refractive index resolution, M, improves with increasing transmission ratio slope, and can be calculated by [45]: (2-4) ônj  where tTr is the resolution to which the transmission ratio can be measured. Referring to Figure 2-6, the sensor will operate with the highest resolution in Region I, and, therefore, we perform our measurements in this region. It is apparent from Figure 2-6 that the sensor is not very useful in Region II, and that the resolution is much lower in Region III. The next three sub-sections will outline the various steps used when making a D-fiber sensor. In sub-section 2.4.1 we will discuss the fabrication process, which includes: (a) exposing and cleaning a section of D-Fiber (b) etching a section of D-fiber in order to decrease d to a few p.m In sub-section 2.4.2 we will discuss how to place the etched D-fiber into the measurement system. This process will include: (c) arranging the optical equipment (d) cleaving the D-fiber ends and inserting it into the system (e) making adjustments to maximize the power and recording the maximum power transmission Finally, in sub-section 2.4.3 we discuss the calibration process which involves:  35  (f) filling the trench with the calibration oils and measuring the power transmission as the temperature is swept (g) generating calibration curves so that the power transmission can be related to the refractive index of the measurand. (f) selecting a sensor with high resolution Section 2.4.4 will discuss the approach used to measure the resolution of our D-fiber sensor. A relative measurement method was used in order to minimize the effects of system drift. By comparing our samples to control samples we could increase the resolution. 2.4.1  D-fiber Sensor Fabrication The D-shaped fiber used to make the sensor was KVH Industries E-Core single-mode  polarization maintaining optical fiber 205 170-1550S. In Figure 2-4 we show the fiber having an elliptically shaped core with dimensions of approximately 4 jim and 2  jim.  The minimum  cladding thickness, d, between the fiber core and the outer cladding flat is approximately 13 jim. The indices of refraction of the core and cladding are 1.4756 and 1.4410 respectively. The length of fiber used for a sensor was 32 cm, however, the sensing region was only 1 cm long. The fabrication process begins by removing approximately 1 cm of the protective jacketing, to expose the cladding. The cladding is cleaned by immersing it in acetone for 20 mm. The exposed cladding section becomes the sensing region after etching. The fibers are etched by immersing the exposed cladding in a 10% hydrofluoric acid (HF) solution. The entire surface of the exposed cladding is etched during this process, however, the effect of decreasing the distance between the core and cylindrical cladding surface is negligible, since this distance is much larger than the minimum distance between the core and flat surface.  The protective  jacketing keeps the cladding isolated from the HF, therefore, only the cladding area which has been directly exposed will be etched. The etch time that produced sensors with high resolution 36  was between 180 and 215 minutes, depending on factors such as the temperature of the room and non uniformity of the fiber. Many fibers were, therefore, etched for different duration periods, and the one having the highest resolution was selected.  When a fiber was removed from the  acid it was immediately immersed in de-ionized (DI) water (for at least 15 minutes) in order to stop the reaction. Once the sensing region had been etched, the fiber could be tested. 2.4.2  Placing the D-Fiber Sensor into the Measurement System  The sensor is placed into the measurement system so it can be tested and calibrated. The ends of the sensors are cleaved in order to produce an optically flat fiber end with minimal loss. The sensor is placed into the measurement system shown in Figure 2-7, which can be automated using LabView.  An HP 81682A*** tunable laser, housed in an HP 8164A lightwave  measurement system, was set to 1500 nrn and the output was connected to an HP 11 896A polarization controller. The optical fiber used at the output of the polarization controller is a circular core single mode fiber, which must be coupled to the elliptical core D-fiber. The two types of fiber are coupled using a mechanical splice into which cleaved ends of each fiber are inserted. The other cleaved end of the D-fiber sensor is fixed so that the light will be emitted onto an HP 81521B optical detector. The detector was connected to an HP 81533B optical detector head interface, which was housed in the same HP 8164A lightwave measurement system as the tunable laser. When coupling the sensor to the circular fiber, minimum power loss is desired and achieved by turning the laser power on and carefully adjusting the position of the cleaved ends until maximum power is detected. This could be a delicate procedure since the mechanical splice was made to couple light between two circular fibers and, typically, the best connection resulted in a 3 dB loss in power. Once coupling is maximized, the sensor region is placed in a trench that can be filled with a liquid, such as acetone, to clean the sensing area.  Equipment with HP abbreviation from Agilent Technologies, Inc., Santa Clara, CA, USA.  37  Communication Cable Standard Fiber D-Fiber Copper Wire  Thermo-Electric Cooler  Figure 2-7: Diagram of experimental set-up showing D-fiber sensor and FISO sensor.  38  After cleaning the sensing area the maximum power transfer, Pm,,.,, can be measured by surrounding the fiber with a medium having a refractive index value in the sensor’s lossless region (discussed below). Before Fmax is recorded, the polarization is altered until the maximum optical power is read at the detector.  By selecting this polarization state before each  measurement period, we keep the sensor properly calibrated. The temperature of the liquid in the trench is controlled using an ILX Lightwavettt modular laser diode controller, with a thenno-electric cooler.  Precise temperature control is necessary during both the sensor  calibration and the experimental measurement process, as the refractive indices of many materials are very sensitive to temperature. 2.4.3  Sensor Calibration  The power transmission ratio is calculated using the maximum power transfer reading Fmax, and the power measured with a sample present in the trench Ppieas as follows: T  —  -  meas max  (2-5)  In order to relate the power transmission ratio to the refractive index, a calibration curve must be generated. The refractive index of the medium surrounding the sensor could be changed by known amounts by using thermo-optic oils as the medium and controlling their temperatures. Three oils were ordered from Cargille Labs, which were prepared so as to have specific refractive indices at specific temperatures.  The refractive indices at 25°C as functions of  wavelength could be calculated for each oil using the provided Cauchy equations as follows [46]:  tt  Equipment from ILX Lightwave, Bozeman, MT. USA.  ::: Oils purchased from Cargille Labs, Cedar Grove, NJ, USA 39  (A) = 1.490962+ 1 n  (A) 2 n  n(A)  =  =  599807.3 2 A  +  2.131038x10’ 2 4 A  (2-6)  +  1.556322 x 1012 4 A  (2-7)  398447.1 3.980245x 1011 2 A + 4 A  (2-8)  1.4690106 +  1.449033+  510457 2 A  where A is the wavelength in Angstroms and n (A), n 1 (A), and n 2 (A) represent the refractive indices 3 of oils 1, 2, and 3, respectively. As will be discussed below, for a few key oils used in our experiments, our sensor operated most effectively at l500nm so, in what follows, this wavelength will be used when specifying several equations and values measured using the D fiber sensor. At 25°C and l500nm, the 3 oils had refractive indices, , 25 of 1.49367, 1.47131, n and 1.45081, respectively. The refractive index as a function of temperature can also be calculated using the provided temperature coefficients for each oil, d(nD) and n 25 of each oil by: dt ,  n(T)  =  25 n  + d(nD)  (T  —  25° C)  (2-9)  therefore, (T)=1.49367—(3.9lxlOj(T—25°C) 1 n  (240)  (T) 2 n  =  )(T—25°C) 4 1.47131—(3.86x10  (2-11)  (T) 3 n  =  1.45081 —(3.83 x 10 )(T 25°C) 4  (2-12)  —  where T is the temperature in degrees Celsius, and 1 n ( T), 2 n ( I), and n (r) are the refractive indices 3 of oils 1, 2, and 3 as functions of T, respectively. Using these oils we could calibrate the sensor over a wide range of refractive index values. A sensor is calibrated by filling the trench in the experimental apparatus with one of the thermo-optic oils and immersing the sensor region in the oil. As the temperature is swept the 40  power out of the D-fiber sensor is measured. This is repeated for all three thermo-optic oils. A calibration curve is generated by plotting T as a function of the thermo optic oils’ refractive indices, as shown in Figure 2-6. After the calibration curve is generated, the refractive index of a sample, or measurand, can be found by measuring T,. and using the curve to find the corresponding refractive index value of the measurand. When using the D-fiber sensor, it is best operated in the high resolution Region I. One way to move the operating point of the sensor into Region I is by changing the temperature of the measurand. This method is only useful in a relative measurement, however, where knowing the refractive index, of a sample, at a specific temperature is not crucial. In some cases it may be more important to measure the change in a sample’s refractive index compared to a control value at the same temperature, which is what we would like to do. This method has a limitation, however, as the amount that the refractive index can be shifted is dependent upon the measurand’s temperature coefficient, which may not be large enough to move the operating point into Region I. As well, if the temperature is increased far beyond the ambient temperature, the resolution may be lowered since the temperature stability may decrease (depending on the system). Our system could maintain a suitable stability to about 35.7°C, but not far above this value. If we reach the sensor’s limit for moving the operating point by temperature, further adjustment can be had by shifting the wavelength. Figure 2-8 shows the power transmission of our sensor for several tested wavelengths. The figure was produced using the calibration oils as discussed previously and using different operating wavelengths. Also, the refractive indices of the oils at the respective wavelengths, using the Cauchy equations and the thermo-optic coefficients, were calculated to produce the figure. As can be seen in Figure 2-8, the resolution of the sensor changes slightly for different operating wavelengths. For the wavelength range that was used, lowering the wavelength 41  1 0.9 0.8 0,7 0  0.6 0.5 0.4 0  0.3 0.2 0.1 1.448 Index of Refraction  Figure 2-8: Measured power transmission of D-fiber sensor at various optical wavelengths.  42  decreased the resolution, however, it was still quite high. The nominal operating wavelength of our D-fiber, given by E-Core, is 1550 nm. Some of the transformer oils obtained from the field, such as Luminol oil, had refractive index values that were around the edge of Region II and Region III when using 1550 mm For these oils, the operating point is shifted from the edge of Region II and Region III to the edge of Region I and Region II by setting the temperature to 35.7°C (see Figure 2-9). The resolution is still not maximized, however, so the wavelength is shifted to 1500 nm. This moves the operating point between Tr = 0.4 and Tr = 0.7, in Region I, where the steepest slope occurs, as shown in Figure 2-9. Using 1500 nrn provided us with the high resolution needed for our measurements. We, therefore, calibrated the sensor using 1500 nrn, and all measurements using the D-fiber sensor were performed using this wavelength. 2.4.4  D-fiber Sensor Resolution We selected our sensor by comparing the slopes of all the sensors’ calibration curves and  selecting the one with the steepest slope in Region I. As previously mentioned, the resolution of our sensor can be calculated using (2-4). The sensor’s resolution was measured using the settings as discussed in the previous subsection, i.e. the temperature was set to 35.7°C and the wavelength was set to 1 500nm.  The resolution was tested for different operation times. It was  found that the best resolution could be achieved using a relative measurement method. The “relative method” was used to help minimize the effects of system drift. System drift can occur due to various shifting parameters such as slight temperature changes, laser noise, or polarization drift. Relative measurements were performed by measuring the refractive index of a sample oil and of a control oil, and calculating the difference. This was done three times and then the average of the three measured differences was used as the final measured refractive index change of the sample, or the n.  43  l—,—__  Operatmg pornt at room temp and l5SOnna  \ \  ‘  0.8  -  \  \  ‘  j  0.6  Shift in operating point by temperature  S  -  e  Operating point at 35.7°C and 1550mu  I  Shift in operating point by wavelength  0  Operating point at 35.7°C and l500mn  0.4I  1500mm 0.2-  1 -  — — — 0— 1.4300  1.4400  1.4500  1 1.4600  -  1550mn  I  I  1.4700  1.4800  I 1.4900  1.5000  Refractive Index of External Medium n Figure 2-9: Calibration curve shown for operating wavelengths of 1550nm and l500nm. The operating point is moved by increasing the temperature. When the temperature control has been exhausted the wavelength can be shifted to move the operating point further.  44  The sensor’s resolution was first tested over a long period of time with minimal averaging. The minimum and maximum refractive index and temperature values recorded over the entire period were compared to see the system stability without using relative measurements. The length of time selected for this measurement was two and a half hours, which was needed to make several relative measurements.  We call this first test the “non-relative measurement  method”. The resolution was next tested with what we call the “relative duration period method” by using the same oil for the sample and for the control, and recording values over the duration of several relative method measurement periods (a relative method duration period is the amount of time it took to perform one measurement of a sample and its corresponding control). Finally, we test the resolution of the sensor using a sample oil and using a control oil that had very similar values of refractive index using the “relative method”. Table 2-1 shows the results obtained using the non-relative measurement method. The temperature of the oil and the power transmission through the sensor were measured over a twoand-a-half hour period. A power and a temperature value were recorded every second for thirty seconds, and the average of each was used to produce one data point. This process was repeated for two and a half hours. For every data point, Tr was calculated and the refractive index, n, of the measurand was found.  T is the temperature measured by the FISO FOT.  The column  labeled max corresponds to the maximum value recorded over the entire duration period, and rn/n is the minimum value.  The column labeled max-rn/n shows the fluctuation of each  parameter. The temperature fluctuates by approximately 0.10°C over the measurement period, which is just above the FOT resolution. Over the duration period the power transmission drifts and the recorded Tr fluctuation was 0.0383. This power fluctuation corresponded to a refractive index fluctuation of about 0.000 14, which would give us a resolution comparable to the FISO 45  system. We would like to perform more sensitive measurements than this and have set our resolution goal to be an order of magnitude greater. Table 2-1: Results of non-relative measurements conducted to find resolution for constant system operation of two and a half hours. Parameter T(°C) Tr n  max 34.55 0.6343 1.446624  mm 34.45 0.5960 1.446485  max-rn in 0.10 0.0383 1.39E-04  Using the non-relative measurement method, we observed that the power transmission changes slightly due to system drift over long periods of time. Hence, in order to minimize the effects of system drift, we would like to make relative measurements between sample oils and control oils over a shorter period of time. This relative method is tested by using the same oil for both the sample oil and the control oil and using the relative duration measurement method. We, therefore, observe the amount that the power fluctuates (A Tr) over the time duration period required to make one sample and one control measurement, due to changing system parameters. The change in refractive index due to system drift can then be found, and the sensor resolution can be calculated. Table 2-2, Table 2-3, and Table 2-4 show the results of conducting several relative duration period measurements, and will be referred to in this paragraph. The total time required to perform one relative measurement is 30 minutes, 15 minutes for the sample measurement and 15 minutes for the control measurement (keep in mind that the sample oil and the control oil were the same for this relative duration period measurement). A power and a temperature value were recorded every second for thirty seconds, and the average of each was used to produce one data point. Data points were recorded for 30 mins. For every data point, Tr was calculated and the refractive index, n, of the measurand was found. T is the temperature measured by the FISO FOT. One complete “Run” would consist of recording data points for 30 46  minutes, using the same oil, and breaking that data into two separate 15 minute data sessions. The two recordings could then be averaged separately and compared to find the resolution over the 30 minute “Run”. Referring to Table 2-2, Table 2-3, and Table 2-4, for each  ii,  i’, and T,  value, the av] and av2 values were the average of each 15 minute recording. The A value is the average fluctuation measured, which is the difference between the two averaged values. If we average all the measured parameter A values, we obtain the following; an average AT value that is less than the resolution of the temperature sensor, a reduced ATr value of 0.0034, and a correspondingly reduced An 3 of 1.2x10 . These values were deemed to be adequate for our 5 measurements as they met the desired order of magnitude improvement. Table 2-2: Temperature results of relative duration period measurements to show average temperature variation over relative measurement period. Run 1 2 3 4 5  Tav](”C) 34.48 34.50 34.52 34.51 34.50  Tav2(°C) 34.48 34.50 34.52 34.50 34.47  AT(°C) 0.00 0.00 0.00 0.01 0.03  Table 2-3: Transmission ratio results of relative duration period measurements to show average transmission variation over relative measurement period. Run 1 2 3 4 5  T,av] 0.5992 0.6082 0.6158 0.6229 0.6297  Trav2 0.6030 0.6115 0.6201 0.6256 0.6325  47  AT 0.0038 0,0033 0.0043 0.0027 0.0028  Table 2-4: Refractive index results of relative duration period measurement conducted to find resolution of system using relative measurement. Run 1 2 3 4 5  av1  av2  AJ2  1.446615 1.446584 1.446555 1.446528 1.446502  1.446603 1.446572 1.446539 1.446518 1.446491  1.2E-05 1.2E-05 1.6E-05 1.OE-05 1.1E-05  In order to test the resolution of the system further, two oils were used that had very close, but not the same, refractive indices. Here, one was the “sample” and one for the “control”, see Table 2-5. Again, for the measurement of a sample, a power and a temperature value were recorded every second for thirty seconds, and the average of each was used to produce one data point. Data points were recorded for three minutes. For every data point, Tr was calculated and the refractive index, n, of the measurand was found. All values of n could be averaged to give the “Average Sample n”. This process was repeated for the control to give the “Average Control n” value. The difference was found by subtracting the two which we call the “Run Measurement An”. Three consecutive run measurements were averaged to give a “Trail Measurement An  “.  As one can see in Table 2-5, there were slight variations between the Run Measurement An values, but the difference between them all fell within our measured refractive index resolution. When comparing the Trial Measurement An values they were very close to one another. Since there was greater consistency between the “Trail” values than the “Run” values, it was decided to use the Trial value when performing experiments using the D-fiber sensor.  48  Table 2-5: Results of refractive index resolution test using two oils with very close refractive index values  Trial# Trial #1  Trial #2  Trial #3  Run Trial Measurement Measurement IS.n An  Run#  Average Sample n  Average Control n  7 8  1.447170 1.447174  1.447157 1.447166  1.34E-05 7.67E06  9  1.447179  1.447167  1.18E-05  10 11  1.447176 1.447177  1.447168 1.447159  7.71E-06 1.79&05  12  1.447169  1.447162  7.13E-06  13  1.447176  1.447160  1.59E-05  14  1.447167  1.447159  7.89E06  15  1.447174  1.447165  9.OOE-06  1.10E-05  1.09E-05  1.09E-05  By performing this analysis, and repeating it several times, we decided to adopt this relative measurement method, since, for our experiments we are only concerned with relative changes in refractive index of oils and not the actual index value. Hence, for the remainder of this thesis, when we make high resolution measurements using the D-fiber sensor, we use the relative method We have achieved a maximum resolution of 1.1 x i0. 2.5  FISO Refractive Index Sensor System A commercially available sensor was lent to the lab by FISO Technologies, Inc. The  FISO system complemented the use of the D-fiber sensor, as it has a wider operating range. The FISO sensor, however, did not have as high a resolution as compared to that obtainable in the range where the D-fiber sensor was most sensitive.  This section will explain the operating  principles of the FISO system. As previously mentioned, a Fiber optic Refractive Index (FRI) sensor, and a Fiber Optic Temperature (FOT) sensor were provided to us by FISO Technologies. The FOT was used to determine the actual temperature of the calibration and sample oils during measurement periods. 49  FISO also supplied a “Universal Multichannel Instrument” (Ulvil), which had 8 sensor channels that could be used simultaneously. This measurement unit converted an optical signal, from the sensor, to a measured parameter. Figure 2-10 shows a schematic diagram of the FISO system. A broadband light source in the Ulvil produces an optical signal with wavelengths between 600 and l000nm. The light is transmitted, by multimode fiber, to the Fabry-Perot cavity located at the end of the sensor.  A Fabry-Perot cavity is made by separating two parallel partially-  reflecting surfaces by a distance dj; with a medium between the two surfaces having a refractive index n. A spectrally varying transmission or reflection function is produced due to interference between the multiply-reflected waves. If the reflected waves are in phase, they will interfere constructively causing a power transmission peak, whereas those that are not in phase will produce lower transmissions. The transmission function can be represented as follows [47]: f \ Tf , 2 df)=  (  1 r2,d  (2-13)  l+Fsin I 2 L2  where df is the distance between the reflecting surfaces,  is the wavelength of the light, n is the  refractive index of the material in the cavity, and F is the finesse which is equal to  4R [(1-R)  where R is the reflectance of the mirrors. This type of cavity can be useful for many different measurement applications. The transmission function will change with d, which can be used for the case of stress or pressure measurements. It will also change with the refractive index of the medium in the cavity, which can be used to measure the refractive index of a liquid or to measure any other parameter which can be related to a change in a material’s refractive index, such as temperature. The spectrally varying transmission signal created by the Faby-Perot cavity is sent back through the fiber to the Ulvil, where it is projected onto a Fizeau interferometer. This second 50  r.  UMI  Broadban 1 /Light Source’  ( FRI  ‘.  IK:  j’\ /  \  Fizeau  Interferoñi’V1  Photo-diode 1 Array —  Fiber Optic Cables Fffl -  FOT  OiIFilled Trench  Fabry-Perot Interferometer  I  C  Th  Temperature  -  Cop:::::  Thea mo-Electric Cooler  Figure 2-10: Diagram of FISO system setup.  51  mterferometer is similar to the Fabry-Perot interferometer. Two semi-reflecting surfaces are separated to form a wedge shaped cavity, instead of the rectangular shaped cavity. The varying distance between the reflecting surfaces is used to spatially separate the different wavelengths of light.  The spectrum is projected onto a photodiode array.  The refractive index which  corresponds to the peak wavelength reflected back from the Fabry-Perot is calculated by the Ulvil and displayed. The refractive index sensor had a very broad measurement range from 1.0000 to 1.7000, which would be useful for oils that did not have refractive index values within the narrow range of the D-fiber sensor.  The resolution of the FISO FRI was 1x10 , however, which is 4  approximately an order of magnitude less than that of our D-fiber sensor. The FOT could perform temperature measurements from -40 to 300°C, with a resolution of 0.05°C. We used the FOT simultaneously with the D-fiber sensor or the FRI in order to set and measure the temperature of the oil samples.  52  Chapter 3 3 Experiments 3.1  Introduction to Chapter In this chapter, the experiments that were performed will be discussed and the results will  be presented. Many sample oils that were tested using the methods described in Chapter 1 were provided to us by Powertech Labs. The refractive indices of many of these oils were measured, by us, to see if any changes could be observed due to varying amounts of contaminants. Although there were a large number of oils available, the samples oils did not have control oils to which they could be referenced. This posed a problem when trying to determine how much the refractive index actually changed. The work involved in detennining what particular type of oil the sample was would be very costly and time consuming, at Powertech’s expense, and was not practical for the sake of these experiments. In order to overcome the challenge of working with unknown oils, direct comparisons were only made between oils taken from a particular piece of equipment or between oils taken from pieces of equipment from the same station (e.g., a transformer station), since most often they would be the same type of oil. Equipment located in the same station would most likely have been installed at the same time, been constructed by the same manufacturer, and been filled with the same type of oil. Another challenge we faced when using these oils was that many variables change during the natural aging process, and relating the refractive index change to one contaminant was difficult.  For this reason “clean oils” were used to prepare samples with varying levels of  53  specific contaminants. The first clean oil used was Voltesso 35 (V35) which is a mineral oil that has been commonly used by transformer manufacturers and is found in many of the aging transformers in the field.  Although V35 was used quite often in many experiments, the  refractive index did not fall within the most sensitive region of the D-fiber sensor. In fact, this was the case for most oils obtained from the field. In order to measure the refractive indices of these oils, only the FISO sensor was used. For many of the experiments conducted, the FISO sensor provided sufficient resolution. Nevertheless, after performing many measurements using the samples obtained from the field, and the samples prepared using V3 5, it was found that higher resolution was necessary for some of the experiments. The D-fiber sensor would be used to conduct experiments measuring changes due to the addition of contaminants such as furans, acids, and gases. Having tested many oil samples for refractive index, a few samples had refractive index values which fell within the D-Fiber sensor’s high resolution range.  One of these oils was Luminal Tri****.  Fortunately, many newer transformer installations throughout western Canada have been filled with this oil type.  This mineral oil was also recently obtained by Powertech Labs in large  quantities. Hence, access to large amounts of this oil was relatively easy to obtain for our experiments.  3.2  Samples Obtained From the Field  3.2.1  Dissolved Gas In Oil Samples From the Field  The initial experiments that were conducted involved measuring the refractive index of oil samples obtained from the field which had been tested for fault gases using DGA (Dissolved  Oil from: Imperial Oil Limited, Calgary, Alberta, Canada Oil from: Petro-Canada Lubricants, Mississuaga, Ontario, Canada  54  Gas Analysis). Relating refractive index changes to gas content could be a very valuable tool for equipment diagnostics, and could serve as an important online monitoring tool. As discussed in Chapter 1, extensive research has been conducted in order to find ways to detect fault gases, and more efficient and cost effective methods are still needed. Samples from industry were obtained from three different types of equipment, including oil filled cables, transformers, and load tap changers. The concentration of gases found in each equipment type should vary, as each piece of equipment functions differently.  The oils  contained in them are subjected to different fault conditions, including overheating and high voltage discharges. Of the three types of equipment, the oil filled cables experience the least  amount of fault activity due to their simple construction and purpose. Oils found in transformers experience a higher degree of faults and harsher environmental stressors than oils found in cables, due to the higher complexity of the equipment and more involved function. Load tap changers experience the highest degree of faulting and siressors since they function as electrical switches, most often with moving parts including high voltage contacts. Oils contained in load  tap changer tanks will be exposed to the highest degree of arcing and gases may be generated any time a switch operates. When comparing the results of DGA, various concentrations of fault gases were found in each piece of equipment which seemed to reflect the fault behavior discussed above.  The  samples obtained from the cables had relatively low levels of all seven fault gases typically monitored, i.e., hydrogen, methane, acetylene, ethylene, ethane, carbon monoxide, and carbon dioxide, as well as the two atmospheric gases, oxygen and nitrogen, as shown in Table 3-1, especially those generated under arcing and extremely high temperatures. Although the samples contained relatively low concentrations of the gases, the levels did vary by a small amount. As shown in Table 3-1 the refractive indices did not vary between samples 1-2 and 1-3, and was 55  only different in sample 1-1, by a small amount equal to the resolution of the FISO refractive index sensor. It seemed reasonable to assume that changes in concentrations at these small levels do not have large effects on the refractive index, and that the changes are not detectable using the FISO sensor. Table 3-1: Refractive index measurement and DGA results of cable oil samples taken from the field. Sample ID Refractive 21.60°C  Index  @  Gas Content (ppm) Oxygen Nitrogen Carbon Dioxide Carbon Monoxide Hydrogen Methane Acetylene Ethylene Ethane Water Total Combustible Gases Gas Content(%v/v)  1-1  1-2  1-3  1.4763  1,4762  1.4762  2390 7640 32 0 60  1410 3840 25 0 33 2 1 1 1 5 38 0.53  3420 11700 27 0  5  1 1 2 6 69 1.01  57  3 1 1 1 6 63 1.52  The concentrations of gases found in the three transformer samples were much higher than those found in the cables, as shown in Table 3-2. In particular, the hydrogen levels were much higher, as well as carbon dioxide and carbon monoxide. The presence of hydrogen often occurs due to its formation through partial discharges, and the carbon dioxide and carbon monoxide are typically present due to overheated paper insulation [12]. The gas concentrations varied between samples as well, and the refractive index of sample 2-3 was 0.004 lower than the other two samples.  Since sample 2-1 and 2-2 had the same refractive index values, it was  assumed that any gas that differed in concentration by a relatively large amount between these 56  samples could not be a factor in the refractive index difference of sample 2-3. Also, by looking at the gas concentrations of the three samples together, gases measured in sample 2-3 that lay between the concentrations of the other two samples could also be eliminated. Based on these observations, it seemed that hydrogen, oxygen, nitrogen, methane, ethane, carbon dioxide, and acetylene were not contributing to the large refractive index change observed in sample 2-3. Table 3-2: Refractive index measurement and DGA results of transformer oil samples taken from the field. Sample ID Refractive Index 21.60°C  @  Gas_Content (ppm) Oxygen Nitrogen Carbon Dioxide Carbon Monoxide Hydrogen Methane Acetylene Ethylene Ethane Water Total Combustible Gases Gas Content(%v/v)  2-1  2-2  2-3  1.4717  1.4717  1.4713  6030 77900 5730 614 2520 28 0 67 15 15  3460 70700 5230 516 9990 39 0 54 24 16  3030 78700 6160 893 3720 74 0 19 17 12  3244 9.29  10623 9.00  4723 9.26  As shown in Table 3-3 the load tap changer oils had a much higher concentration of some gases and lower concentrations of others, when comparing them with the transformer oils. In particular the levels of acetylene, ethylene, and ethane were much higher, and levels of hydrogen, carbon monoxide, and carbon dioxide were much lower. This lower value is expected for carbon monoxide and carbon dioxide since no paper is present in the load tap changer tank, and the higher levels of acetylene, ethylene, and ethane would be produced during the arcing that occurs when the tap is changed. Sample 3-1 had a much lower refractive index than the other 57  two samples. It also had much higher levels of most gases except carbon monoxide which was at about the same level. The water content was also lower. Comparing samples 3-2 and 3-3, the concentrations found in 3-2 were higher for every gas, and much higher for hydrogen, methane, acetylene, ethylene, ethane, and water content. Since the measured refractive index was only slightly different between the two samples, it seemed very unlikely that any of these gases were a factor in lowering the refractive index to such a degree in sample 3-1.  Table 3-3: Refractive index measurement and DGA results of load tap changer samples taken from the field. Sample ID Refractive Index 21.60  @  Gas_Content (ppm) Oxygen Nitrogen Carbon Dioxide Carbon Monoxide Hydrogen Methane Acetylene Ethylene Ethane Water Total Combustible Gases Gas Content(% v/v)  3-1  3-2  3-3  1.4725  1.4752  1.4751  31200 67700 1270 23 725 369 6080 1600 252 13  31200 65100 676 29 352 248 2510 745 114 31  30800 65500 631 24 17 12 271 80 9 17  9049 10.92  3998 10.09  413 9.73  By conducting these experiments, we did observe varying refractive index values of the oils. It did not seem, however, that any of the gases present in the oils were large factors in the observed changes in the refractive indices.  More experiments were necessary in order to  determine if the addition of particular gases would result in small changes in the refractive indices of the oils. Since the samples obtained from Powertech Labs all had multiple gases in  58  the oils, a specific gas would have to be injected into new samples to isolate its effect on the refractive indices. Also, it should not be assumed that the presence of gas is the only factor that contributes to the changes in refractive indices of oils. As discussed in Chapter 1, many physical and chemical properties of oils change during the aging cycle.  Besides the changes in the  concentrations of gases, there are other changes in the oil properties such as varying acidity, concentration of poiar compounds, IFT (Interfacial Tension), or concentration of furans. Further investigation was required to determine if other changes contribute to changes in the refractive indices of the oils. In what follows we present the results of further experiments that were conducted for this purpose. 3.2.2  Other Measured Properties of Oil Samples Obtained From the Field  After performing the experiments described in the previous section, we observed that the refractive index of an oil sample would change over time, but low levels of fault gases did not appear to contribute significantly to the change. We, therefore, used oil samples that had tests other than DGA performed on them to determine if the addition of contaminants and changing properties of the oils affected their refractive indices. The refractive indices of a large number of oil samples having various physical and chemical properties are shown in Table 3-4. These properties include KV breakdown (Kilovolt breakdown), IFT, color, and acid number. Oil samples were provided in sets which included two samples taken from the same transformer, one from the load tap changer (LTC), and one from the transformer tank (TX). The combination of load tap changer and transformer oil samples could be used for comparison. Some sets of samples were extracted from equipment from the same station as well. It was assumed that equipment from the same station would be filled with the same type of oil and, so, could be used for comparison as well. As shown in the table, oils 59  Table 3-4: Measured refractive indices of oil samples obtained from the field with some physical and chemical property values shown. Sample ID Al-TX Al-LTC A2-TX A2-LTC B3-TX 133-LTC B4-LTC B4-TX CS-TX C5-LTC C6-TX C6-LTC D7-TX D7-LTC D8-TX D8-LTC E9-TX E9-LTC E10-TX E10-LTC Eli-TX E11-LTC F12-TX F12-LTC G13-TX G13-LTC H14-TX H14-LTC 115-TX 115-LTC 116-TX 116-LTC J17-TX J17-LTC J18-TX J18-LTC  Refractive Index (n) 1.4867 1.4859 1.4863 1.4852 1.4860 1.4851 1.4859 1.4849 1.4853 1.4845 1.4850 1.4808 1.4836 1.4803 1.4810 1.4811 1.4812 1.4807 1.4811 1.4802 1.4810 1.4805 1.4781 1.4796 1.4789 1.4790 1.4783 1.4792 1.4789 1.4789 1.4785 1.4786 1.4787 1.4764 1.4782 1.4771  n7x  -  KV Breakdown 16 22 25 26 31 21 44 28 18 22 17 24 23 20 33 16 25 20 38 19 21 19 20 15 24 17 25 19 24 20 26 26 23 18 30 23  LTC  0.0008 0.0011 0.0009 0.0010 0.0008 0.0042 0.0033 -0.0001 0.0005 0.0009 0.0005 -0.0015 -0.0001 -0.0009 0.0000 -0.0001 0.0023 0.0011  60  IFT 22.3 13.9 19.6 14.2 19.9 14.9 20.0 17.1 18.3 37.7 18.2 14.6 22.3 23.1 18.8 14.1 23.3 21.2 18.0 19.4 18.4 18.0 21.4 28.6 23.2 22.2 23.9 14.4 23.4 14.6 27.3 19.6 17.8 25.9 23.6 30.5  Color 3.5 2.5 2.5 2.5 2.0 2.0 1.5 2.0 2.5 1.0 3.0 2.5 1.5 2.0 4.5 3.0 2.0 2.5 3.0 3.0 4.5 3.0 1.5 1.5 2.0 1.5 1.0 1,5 1.0 2.5 0.5 1.0 1.5 1.0 1.0 1.0  Acidity 0.06 0.63 0.09 0.46 0.07 0.28 0.05 0.12 0.11 -  0.10 0.37 0.03 0.03 0.15 0.65 0.04 0.04 0.13 0.06 0.23 0.08 0.05 0.01 0.04 0.04 0.06 0.45 -  0.44 -  0.06 0.12 <0.01 <0.01 <0.01  extracted from the same station will be identified by the first character in the Sample ID, and oils extracted from the same transformer are identified by the second character. For example, the samples Al-TX, Al-LTC, A2-TX, and A2-LTC are all from the same station (station A). Samples Al-TX and Al-LTC are from transformer 1 and samples A2-TX and A2-LTC are from transformer 2. After making comparisons between oil samples and trying to relate a single property to a refractive index change, it did not seem likely that any one property could be directly related. For example, in some cases it seemed that a lower IFT would produce higher refractive indices for some comparable oils, but the reverse seemed to occur for others. The colunm in Table 3-4 labeled rx  —  flLTC  represents the refractive indices of the  transformer samples minus the refractive indices of the load tap changer samples, for the same piece of equipment, e.g., transformer Al. Figure 3-1 shows a graph of these values for each piece of equipment. By looking at the table and the figure, we see that for twelve of the eighteen samples the refractive indices of the transformer samples were higher than those of the load tap changer samples. Of those, nine of them, or 75%, had a difference between 0.0005 and 0.0011. The refractive indices of the load tap changer samples were higher than those of the transformer samples for only five of the eighteen cases, and of those three of them were lower on the order of the resolution of the sensor, i.e., -0.0001. As discussed in Chapter 1, [9] and [35] present studies that were performed to detennine if transformers could be characterized by the UV absorption of their insulating oils. The authors concluded that through aging, an increase in aromatic compounds is observed and that this increase in aromatic compounds will increase the absorption in the UV region, between 200 and 400 nm. As discussed in Chapter 2, when the absorption profile of a medium changes, a change will also be observed in the real part of the refractive index. If increased absorption is observed in the UV region, an increase in the real part of the refractive index should be observed as well 61  0.0050  0.0040  0.0030  0.0020  0.0010  •.  0.0000  -0.0010  -0.0020  Equipment ID  Figure 3-1: The refractive index of transformer oil samples minus the refractive index of load tap changer oil samples obtained from same equipment from the field.  62  for the wavelength region over which we performed our measurement (provided the increased absorption in the UV region is the dominant change). Hence, it is assumed that the addition of aromatic compounds to the aging transformer oils will also increase the refractive indices of the oils. In [9], samples taken from failing transformers were tested to determine if a relationship existed between the UV absorbance and type of fault. The oils tested had failed from either thermal or arcing faults.  Transformer oils which failed due to thermal faults had higher  absorbance of light, in the 360  —  400 nm region, than those which failed due to arcing. The  authors concluded that measuring the absorbance at 390 nm could be used to differentiate between a transformer failing from either of the faults. The higher absorption of the thermally failing oils in this wavelength region would correspond to a higher refractive index value measured using the FISO sensor. It is expected that the oils in transformer tanks would exhibit thermal faults more often and that the oils in load tap changers would exhibit arcing faults. The refractive index of the transformer oils measured using the FISO sensor should, therefore, be higher than the changes measured in the load tap changer oils. Generally speaking, the apparent trend which we observed was that, for the same equipment, the refractive indices of the oils in the transformer tanks were higher than those in the load tap changer tanks, with only 28% of the samples showing the opposite and only 11% showing a significant negative difference. There are various reasons why the refractive indices of some oils may not have followed this trend. Some oils could have been filtered or changed at some point in the equipments’ lifetime, which would obviously lead to changes in the refractive indices. No information regarding the fault activity of the equipment has been provided either, and there is a possibility that some transformers could have experienced a high level of arcing when compared with others. The effects of other contaminants on the refractive indices of the 63  oils are also unknown. Hence, no solid conclusions can be drawn from the results of these measurements other than that a trend has been observed, which seems to be consistent with the results of[9]. When using samples from the field it seemed that too many variables could affect our ability to make any solid conclusions regarding how contaminants affected the refractive indices of the oils. We have, therefore, continued our investigation using a more systematic approach. This was done by conducting measurements using clean oils that had been aged, degraded, or contaminated in a controlled fashion.  3.3  Effects of Accelerated Aging on Refractive Index of Oils Since there were no oil samples available that had been collected over time from the  same piece of equipment, it was decided that any experiments that were to be conducted measuring the effects of aging on refractive index would require artificial aging of the oils. Thermally accelerated aging experiments are often performed in order to predict the lifetime of insulation systems and to generate contaminants in samples for experimental investigation [48] [49]. Accelerated thermal aging involves exposing oil samples to high temperatures, in order to simulate the aging effects that naturally occur over the life of a transformer, in a much shorter time span. It has been found that increasing the temperature of oil 10°C above its normal operating value will decrease its lifetime by up to one half [12]. If the oil samples are exposed to extreme temperatures, the aging process can be accelerated to a point that new oil samples aged for a few weeks will possess the properties of oils that have been in use for over 30 years. Using this technique, however, does not expose the oil to many conditions that oils taken from the field would have experienced.  Therefore, many of the contaminants, fault gases, and chemical  compounds found in a typical sample may not be present. For example, oil that is aged in a lab would not be exposed to arcing, unless purposely introduced, and would not contain the by 64  products associated with it. Nevertheless, these aging experiments are still useful, as there are chemical and physical changes that the oil will undergo that would be common to both oils naturally aged and aged at an accelerated rate. Three accelerated aging experiments were conducted in order to study the changes in refractive indices as oils are aged. The first experiment was conducted using both new and used V3 5 oils. Approximately 100 g of each oil were placed in separate tin cans and were sealed. Pin holes in the tops of the cans would also allow oxygen to reach the samples, accelerating the aging even further.  The samples were placed in a laboratory grade oven, and exposed to a  temperature of approximately 120°C. Samples were extracted at intervals shown in Table 3-5 and the refractive indices were measured and the observed color was recorded. Table 3-5: Measured refractive index versus time for accelerated aging samples at 120°C.  Days Aged 0 (new) 15 45 90  New Oil 120 C Refractive Index 1.4743 1.4746 1 .47 52 1.4755  Color clear light yellow yellow/orange orange  Days Aged 0 (new) 15 45 90  Used Oil 120 C Refractive Index 1.4743 1 .4 749 1.4753 1.4755  Color clear yellow/orange orange orange  As shown in the Table 3-5 and in Figure 3-2 the refractive indices increased over time, and after 90 days changed by 0.0012. The color of the samples also changed over time as shown in Table 3-5 (note: the colors of the oils shown in Section 3.3 are subjective and reflect the opinion of the author). This can be explained by the electronic absorption edge of the oils being shifted into the visible wavelengths as was observed in [50]. 65  1.4756 1.4754 1.4752  ‘  I  1.4750 1.4748  = C 1.4746 1.4744 1.4742 0  20  40  60  80  100  Days Aged Figure 3-2: Plot of measured oil refractive index versus aging time when exposed to a temperature of 120°C.  66  In the next two aging experiments, the samples were exposed to temperatures of 150°C, but for shorter periods of time. In the first of these two experiments, that being the second aging experiment, new V35 was used for four cases. One tin was filled with 75g of oil only, a second tin with 75g of oil and a 5g copper coil (12 gauge wire), a third tin with 75g of oil and 5g of insulating papertttt, and a fourth tin was filled with 75g of oil, a 5g copper coil, and 5g of insulating paper. These combinations were chosen to investigate the change of refractive index in each case, since copper and paper are commonly found in high power equipment, and are known to affect the degradation of oil [3][12][21][50]. Since the samples in the second aging experiment were exposed to a higher temperature than the first aging experiment, samples had to be extracted at shorter intervals as shown in Table 3-6. At 150°C both the color and the measured refractive indices changed rapidly. After 7 days of aging the oils were a brownish color and their refractive indices were higher than that of the new V35 aged for 15 days at 120°C. Figure 3-3 shows that the oils’ refractive indices increased progressively again, and that there were small variations between the four cases mainly after the  th 7  day of aging, however, these variations were only plus or minus one resolution unit  of the FISO sensor. The sample of oil exposed to copper only measured the greatest refractive index change. It also seemed, by observing the color, that this sample degraded the fastest. Copper acts as a catalyst to oil aging, and oil mixed with copper and oxygen will oxidize faster, producing larger concentrations of carbonyl compounds [50]. The sample having oil, copper, and paper had a slightly lower refractive index value, at the end of the experiment, than the copper only. The addition of paper to the oil results in some of the oxidation products being absorbed by the paper, which helps counteract some of the negative effects produced by the copper catalyst [3] [51]. fttt  was Kraft upgraded paper from Algonquin Industries, Guilford, CT, USA.  67  Although the paper would absorb some of the oxidation products and some moisture contained in the oil, when the cellulose begins to break down it would also add contaminants such as furans, carbon monoxide, and carbon dioxide. This could explain why the refractive index of the sample with paper only was slightly higher than the sample with just oil. Although these slight variations in refractive index values were measured after aging, the differences between the four samples were not appreciable comparing them with the resolution of the sensor. More important was the observation that the refractive index increased in each case with aging. Table 3-6: Measured refractive index versus time for accelerated aging samples with varying contents at 150°C. Days Aged 0 (new) 1 4 7  Oil Refractive index 1.4743 1.4743 1.4745 1.4747  Days Aged 0(new) 1 4 7  Oil and Copper Refractive index 1.4743 1 .4 743 1 .4 746 1.4749  Color clear yellow/orange dark orange dark brown  Days Aged 0(new) 1 4 7  Oil_and Paper Refractive index 1.4743 1 .4 743 1.4745 1.4748  Color clear yellow orange brown  Days Aged 0(new) 1 4 7  Color clear yellow orange brown  Oil, Copper, and Paper Refractive index Color 1.4743 clear 1.4743 yellow 1.4745 dark orange 1.4748 dark brown  68  1.4750 1.4749  1.4748 x  1.4747 >  1.4746 II  1.4745 0  1.4744 1.4743 1.4742 0  1  2  3  4  5  6  7  8  Days Aged  Figure 3-3: Plot of measured oil refractive index versus time when exposed to a temperature of 150°C with different contents present.  69  In the third aging experiment, different contaminants were added to the samples to see if they affected the aging process and refractive index values. Five tins were filled with 75g of new V35, and a 5g coil of copper (12 gauge wire). No contaminants were added to the first tin, and 0.1 2g of oxygen inhibitor was added to the second to see if aging effects could be reduced. The third tin had 0.34g of water added to it, the fourth 0.lg of acetic acid (99.7% purity), and the fifth had 0.34g of water and 0.lg of acid. In the third experiment, the color and refractive index changed faster than those of the first aging experiment, as was observed in the second experiment, due to the higher temperature, especially in the samples containing acid as shown in Table 3-7 and Figure 3-4. The refractive index measurement of the oil and the oil with inhibitor samples were almost exactly the same throughout. Since V35 has an oxygen inhibitor concentration of 0.08% to begin with, adding more had no effect. The water did not affect the refractive index either, although the color seemed to vary in comparison to the plain oil sample.  This was unexpected, however, as  moisture should increase the effects of oxidation. It is assumed that the high temperature dried out the oil before the water could have any significant affect. The oils containing acid and both acid and water, were observed to have an increased change of refractive index and color. The acid was found to initially lower the indices for both cases by a very small amount, but after only a day they increased by 0.0002. After 8 days the refractive indices of the samples containing acid were 0.0004 greater than the oil only sample. The acid initially lowered the refractive indices of the oils but, during aging, this decrease was more than compensated for by the increase caused by the additional aging by-products. Since acid is a catalyst to aging, it is assumed that oils with acid present would age faster than oils  Acetic acid was from Fischer Scientific, Ottawa, ON.  70  without it, which in these experiments translated to a darker oil color and higher measured refractive index.  Table 3-7: Measured refractive index versus time for accelerated aging samples with varying contaminants at 150°C. Days Aged 0 1 4 8 Days Aged 0 1 4 8 Days Aged 0 1 4 8 Days Aged 0 1 4 8 Days Aged 0 1 4 8  Oil Refractive Index Color 1.4743 clear 1.4743 dark yellow/orange 1.4749 orange/brown 1.4753 brown Oil and Inhibitor Refractive Index Color 1.4743 clear 1 .4744 light orange 1.4749 orange/brown 1.4753 brown Oil and Water Refractive Index Color 1.4743 clear 1.4744 light yellow 1.4750 orange/brown 1.4754 brown Oil and Acid Refractive Index Color 1.4742 clear 1 .4 744 light orange 1.4752 brown/orange 1.4757 brown Oil, Acid, and Water Refractive Index Color 1.4742 clear 1 .4 744 darker yellow 1.4752 brown/orange 1.4757 brown  71  1.4758 1.4756 1.4754 1.4752 1.4750 —.  -  CleanV-35  1.4748 ‘7  -  1.4746  -.  -  Inhibitor  A  Water  X  Acid  /  1.4744 1.4742  —)E  —  Water and Acid  1.4740 0  1  2  3  4 5 Days Aged  6  7  8  9  Figure 3-4: Plot of measured oil refractive index versus time when exposed to a temperature of 150°C with different contaminants present.  72  By conducting these aging experiments, it was observed that the refractive index does, in fact, change to a relatively large degree during the aging process. As oils were aged and by products were formed, the refractive indices of the oils increased.  This result supports the  assumption that the addition of aromatic compounds will contribute to the increase in the refractive index, as discussed in the previous section. Paraffms, also known as Alkanes, are saturated hydrocarbons which are composed of hydrogen and carbon atoms linked by single bonds (see Figure 3-5(a)). Naphthenes, also called Cycloalkanes, are paraffms containing one or more carbon rings (see Figure 3-5(b)). The hydrogen and carbon atoms present in naphthenes are linked by single bonds as well. Aromatic hydrocarbons, also known as an Arenes, contain one or more aromatic ring. An aromatic ring consists of six carbon atoms that form a conjugated system of alternating single and double covalent bonds between the carbon atoms, which are also linked to hydrogen atoms by a single bond (see Figure 3-5(c)). During thennal decomposition of mineral oils, paraffinic compounds dehydrogenate forming naphthenic compounds which further dehydrogenate to form conjugated C=C double bonds and aromatics [35].  The bonding  electrons found in the it-orbitals of conjugated systems can be excited to higher energy levels, and these energy transitions are frequently observed in the near UV region (1 90-400nm)  [331.  The larger the conjugated system becomes (the more alternating double and single bonds found in a compound), the lower the energy required for a transition, which corresponds to light at longer wavelengths being absorbed. The electrons that form the a bonds, that paraffinic and naphthenic compounds are generally made up of, require a higher energy and, therefore, shorter wavelength, usually below 150 nm, in order for a transition to occur [33].  This, in part, could  explain our observations of an increased refractive index and change in color of the oils during aging. The decomposition of paraffins and naphthenes and formation of aromatic compounds in the oil produces absorption in the near UV region. The formation of more aromatic compounds 73  H  H  H  H  H  I  I  I  I  I  H—C I  H  —  H  C—C—C—C—C—H I I H H H H H (a) H  H H  \,i  \ C  H /  H—C  C—H  H—C  C—H  / C’  /\ HH (b)  H  I-I  H -C  H  H  H C  C  H H (c)  Figure 3-5: Examples of different types of hydrocarbon compounds. (a) example of a parraffinic compound (hexane). (b) example of a naphthenic compound (cyclohexane). (c) example of a aromatic compound (benzene).  74  over time leads to the polymerization of these conjugated systems and, therefore, shifts the electron absorption edge further into the UV and eventually visible region. The shifting of the absorptive behavior of the oil would obviously affect the refractive index, as discussed in Chapter 2, Although the addition of aromatic compounds is one factor that may be linked to the increase in refractive index of the oils, there are other compounds, or factors, that could also contribute. For samples used in any particular experiment, it was observed that the refractive index increased with a darkening of the oil color. This does not mean, however, that color change can be directly related to the refractive index. For example, when comparing the refractive indices of the oil only sample shown in Table 3-7 at day eight, which had a brownish color, with the samples shown in Table 3-5 at 90 days, which had an orange appearance to them, the brown sample had a lower refractive index than the orange samples. However, oils from the same experiment with a darker color had a higher refractive index that oils with a lighter color. This leads us to believe that the darkening of an oil in a piece of equipment could result in an increased refractive index of that oil but this does not necessarily mean that this darker oil would have a higher refractive index than a lighter oil extracted from a different piece of equipment. It also seems that catalysts added to the oils, such as copper and acid, increased the degradation of the oils which, in turn, increased the refractive index over time. These catalysts may accelerate the formation of aromatics and other compounds, which not only indicate a decrease in the oil quality, but increase the refractive index as well. Hence, we conclude that a change in the refractive index can be linked to the aging of oils.  Nevertheless, further  experiments are needed to determine how individual contaminants such as polar compounds, ftirans, acid, or dissolved gases affect the refractive indices of the oils.  75  3.4  Polar Compounds in Oil  3.4.1  Introduction to Section  The following experiments were conducted in order to determine the extent to which the concentration of polar compounds affected the refractive index of the oils.  Much of the  following section was taken from [52], which was presented at a conference by the author of this thesis. A method is described which could be used to indicate relative concentrations of polar compounds using refractive index as an indicator. The results are presented using this method. Samples with varying levels of polar compounds were obtained using oils taken from transformers used in accelerated aging experiments at Powertech Labs.  The actual  concentrations of polar compounds found in the samples were measured using HPLC (high pressure liquid chromatography) as discussed in Chapter 1, and the refractive index changes due to poiar compounds were measured using the FISO system. 3.4.2  Methanol Extraction In order to analyze oil samples, the polar compounds were removed from each sample  using a liquid-liquid extraction technique. This method is often used to remove furans from oil samples prior to using HPLC [22][53j. When a polar solvent such as methanol, which we used for extraction, is mixed with a sample of oil the polar compounds will be partitioned into the methanol due to their affinity. For our experiments, the refractive indices of the oil samples were first measured using the FISO FRI (Fiber optic Refractive Index) sensor. One gram of high purity grade 99.9%  Parts of this section are pending publication. Kisch, RJ., Hassanali, S., Kovacevic, S. and Jaeger, N.A.F. (2007) The effects of polar compounds on refractive index change in transformer oils, Proceedings of High Voltage and Electrical Insulation Conference ALTAE 2007.  76  methanol  was then added to ten grams of each oil sample and the solutions were mixed  vigorously using a vortex mixer. A centrifuge was used to separate the two phases and the methanol phase, containing all of the polar compounds, was extracted from each sample. The refractive index of each oil sample was measured again after the polar compounds were removed, and the methanol extract samples were analyzed using NPLC. The refractive indices of each methanol extract sample containing the polar compounds were also measured. 3.4.3  Oil Samples Oil samples with varying levels of polar compounds were made available through  accelerated aging and filtering experiments conducted at Powertech Labs. V35 was used for an accelerated aging experiment in surrogate transformers designed to mimic free breathing transformers and nitrogen-blanketed transformers. A set of four samples were extracted from four free breathing type transformers and a set of four were extracted from four nitrogen blanketed type transformers. Each set of four samples included two that had been filtered through different online purification units developed by Powertech Labs, which could be compared to two control samples which had no purification units connected to them. One of the purification units removed only moisture and particulate matter from the oil (dehydration unit), while the second unit removed all paper and oil degradation products such as carbonyl and acid compounds, polar compounds, oxygenated compounds, furanic compounds, dissolved metals, as well as moisture and particulates (decontamination unit).  The decontamination unit would  restore the aged oil quality to that of new oil and maintain it at a “near new” level as long as it was connected to the transformer.  The oil and paper insulation life would, therefore, be  extended and, subsequently, would extend the life of the transformer. The oils were thermally aged by heating the surrogate transformer windings with a high  Methanol was from Analabs, Inc., Crab Orchard, WV, USA.  77  amperage, low voltage DC current. The windings were heated in 40 hour cycles consisting of a 10 hour ramp from 30°C to 120°C, held at 120°C for 20 hours and, thereafter, cooled to 30°C. Table 3-8 lists each sample number and its associated aging conditions. Aging effects differed between each sample as shown in Table 3-9.  The samples  extracted from nitrogen blanketed transformers (samples 1-4) did not show as severe signs of aging as compared to those exposed to oxygen, as the effects due to oxidation were reduced. For each of the two sets of oils, samples that were extracted from transformers that were filtered for moisture (samples 2 and 6) indicated more aging as compared to unfiltered samples, as effects due to moisture were reduced. Those samples extracted from transformers connected to the decontamination unit showed the least aging, as effects due to many contaminants were reduced. Table 3-9 illustrates this by showing some measured physical, chemical, and electrical properties of our samples. Table 3-8: Aging conditions for oils used in polar compound measurements. Sample Number  Blanketing Treatment Type  Cycles  1  Nitrogen  Decont. Filter  237  2  Nitrogen  Moisture  237  3  Nitrogen  None  237  4  Nitrogen  237  5  Oxygen  None Decont. Filter  6  Oxygen  Moisture  237  7  Oxygen  None  205  8  Oxygen  None  205  78  237  Table 3-9: Measured properties of aged oils. Sample Number  3.4.4  Power Neutralization Interfacial Color Factor Number Tension  1  0.039  0.003  43.5  0.8  2  0.031  0.003  42.5  0.5  3  1.037  0.004  33.5  1  4  1.21  0.004  32.5  1  5  0.06  0.003  43.2  0.5  6  8.268  0.092  21.2  3.5  7  11.85  0.152  15.7  4  8  30.79  0.432  14.6  5,5  Refractive Index Measurements  By analyzing each methanol extract sample using HPLC, it was observed that samples obtained from oils showing more aging did, indeed, contain higher concentrations of polar compounds.  The nitrogen blanketed samples contained much lower concentrations of polar  compounds than those extracted from the free breathing units.  Samples extracted from the  dehydration units contained lower concentrations as compared with unfiltered samples, and samples from the decontamination units contained the lowest. This is shown in Table 3-10, in the “Polar Compounds” column, which corresponds to the sum of the areas under predetermined peaks in the chromatograph (a larger area indicates a higher concentration of polar compounds). The refractive indices of the oil samples were measured at 24.30°C and of the methanol extract samples at 21.70°C. The measured refractive index of new V35 mineral oil was 1.4746. Other than the samples extracted from the decontamination unit, the refractive indices of all oil 79  samples increased during aging.  The refractive indices of the oil samples taken from  transformers with the decontamination units were very close to that of new V35 mineral oil and in the case of the decontamination filtered nitrogen blanketed oil sample it was lower than that of new V3 5. The authors believe that the decontamination filtered nitrogen blanketed sample’s lower refractive index is due to the filtering, which extracts some compounds originally present innewV35oil.  Table 3-10: Refractive index measurements of oil and methanol samples and concentration of polar compounds measured by HPLC. Oil RI Area of Change Methanol After Polar inRi RI Extraction Compounds  Sample Number  Oil RI  1  1.4743  1.4742  0.0001  1.3310  842  2  1.4761  1.4759  0.0002  1.3311  1422  3  1.4760  1.4758  0.0002  1.3314  3303  4  1.4760  1.4758  0.0002  1.3315  3572  5  1.4746  1.4743  0.0003  1.3352  4205  6  1.4761  1.4757  0.0004  1.3410  44080  7  1.4762  1,4758  0.0004  1.3413  47460  8  1.4765  1.4759  0.0006  1.3494  94720  After the polar compounds were removed, the refractive indices of the oil samples dropped by a small amount in each case. It seemed that the changes in refractive indices of the nitrogen blanketed samples were very small, and barely measurable. Although the changes in the refractive indices of the free breathing oil samples were somewhat larger, they were still close to the resolution of the FISO system. Nonetheless, the apparent trend was that an oil 80  sample that showed a larger amount of polar compounds would experience a larger change in refractive index when the polar compounds were removed. The refractive index of methanol extract that had been obtained from a new sample of V35 oil was 1.3307.  The refractive indices of the methanol extracted from all of the aged  samples were higher than 1.3307, and showed an increase with the amount of polar compounds. The differences between the methanol extract refractive indices from the free breathing samples were quite large in comparison to those from the nitrogen blanketed ones. Figure 3-6 and Figure 3-7 show that the refractive indices of the methanol extract samples, for each set, increased relatively linearly with the amount of polar compounds.  81  1.3316 41.3315 1.3314  EEE I  F 2 1  1.3310  • Measured  1.3309 1.3308  F 0  II  500  III  II  1000  II  I  I  II  ii  iii  I  1500 2000 2500 3000 Area of Polar Compounds  I  ill  3500  4000  Figure 3-6: Methanol extract refractive index versus the area of polar compounds measured by HPLC in nitrogen blanketed oil samples.  82  1.3520  8  1.3500 1.3480 1.3460 1.3440 1.3420 1.3400 1.3380 1.3360  . Measured Linear Fit  1.3340 1.3320  I  0  I  I  20000  I  I  I  I  40000 60000 80000 Area of Polar Compounds  I  100000  Figure 3-7: Methanol extract refractive index versus the area polar compounds measured by HPLC in free breathing oil samples.  83  3.4.5  Polar Compound Extraction From Naturally Aged Oils  Since there were a few samples from Section 3.2.2 that had been measured for polar compounds using HPLC, it was decided to see if similar results could be obtained using this extraction technique on oils obtained from the field. The total area produced by adding the area under the peaks on the chromatograph was available for four oil samples from the same station. These samples were Al-TX, Al-LTC, A2-TX, A2-LTC found in Table 3-4.  The same  procedure described above was used to measure the refractive indices of the oil samples before and after methanol extraction, and of the methanol extract itself. Table 3-11 shows the results, and Figure 3-8 and Figure 3-9 show the methanol refractive index and the change in refractive index of the oils after methanol extraction. Table 3-11: Refractive index measurements of naturally aged oil and methanol samples and area of polar compounds measured by HPLC. Oil RI Change Methanol After inRI RI Extraction  Area of Polar Compounds  Sample ID  Oil RI  Al-TX  1.4867  1.4863  0.0004  1.3394  51200  A2-TX  1.4863  1.4858  0.0005  1.3403  60561  A2-LTC  1.4852  1.4844  0.0008  1.3493  87317  Al-LTC  1.4859  1.4850  0.0009  1.3527  94860  84  1.3550 1.3530 1.3510 1.3490 1.3470 1.3450 1.3430 1.3410 1.3390 1.3370 1.3350 45000  55000  75000  65000  85000  95000  105000  Area of Polar Compounds Figure 3-8: Methanol extract refractive index versus the area of polar compounds measured by HPLC in naturally aged oil samples.  85  0.001 0.0009  0.0007 0.0006 0.0005 0.0004 0.0003 45000  55000  65000  75000  85000  95000  105000  Area of Polar Compounds Figure 3-9: Change in refractive index of naturally aged oils after methanol extraction versus the area of polar compounds measured by HPLC.  86  Using this methanol extraction technique provided similar results when naturally aged oils were used, as the refractive index of the methanol extract increased with increasing polar compound concentration. The change in refractive index after extraction was also increased when the concentration of poiar compounds was higher. Figure 3-9 shows that the change in refractive index was quite linear with the area of polar compounds measured. Large changes were measured for a few of the samples that were extremely degraded, but again, it did not seem that the polar compounds would be the only factor changing the refractive indices of the oils. 3.4.6  Discussion Small decreases in the refractive indices of the oils were observed after the methanol  extraction was performed. We can assume that the decrease is caused due to the removal of the polar compounds from the oil, since it was observed that when larger amounts of polar compounds were removed from the oils larger decreases in the refractive indices were observed. In particular, this is clearly demonstrated by referring to the more degraded oil samples such as accelerated aging samples 6, 7, and 8, and all naturally aged samples. Hence, we conclude that the refractive index of oil samples is increased slightly due to the formation of polar compounds. By concentrating the polar compounds using methanol extraction, larger changes in the refractive indices of the methanol extracts could be measured as compared to those of the oils. These larger changes that are measured in the methanol solution could be due to the addition of a solute (the polar compounds) having higher refractive indices than the solvent (the methanol). Also, the methanol molecules could react with some of the extracted polar compounds, to form new compounds having higher refractive indices. Regardless, the change of the refractive index induced in the methanol was a useful indication of polar compounds. During the aging process, different types of polar compounds are formed, some of which will not contribute to changes in the refractive indices of the oils or the methanol extracts. For 87  example, the refractive index (at optical frequencies) is affected by the electronic polarizability of a medium, but not by the molecular dipolar orientation polarizability [43]. Some oil samples could contain higher concentrations of the types of polar compounds which do not contribute to a refractive index change and would not be detected. Also, when we used HPLC we looked at the total and not individual polar compounds. This could lead to some measurements not fitting linearly into the results, since one oil sample could contain more of a compound that affects the refractive index change to a larger degree than other samples containing similar concentrations of those compounds which do not. Samples extracted from nitrogen blanketed transformers and samples extracted from free breathing transformers should be treated individually.  Different aging effects may produce  different ratios between the types of poiar compounds formed. Free breathing transformers will experience aging effects primarily due to oxidation processes, whereas aging in nitrogen blanketed transformers may be dominated by the formation of polar compounds due to other aging effects. Hence, the results of tests performed using samples extracted from different types of transformers should be treated separately. Nevertheless, since the presence of polar compounds does indicate a weakening of the insulation quality of the oil, detecting them is a valuable tool in analyzing the insulation quality. The measurement of the refractive index in methanol extract can be used to indicate an increase in certain types of polar compounds, so could, therefore, be used to indicate a decrease in insulation quality in cases where expensive HPLC equipment is not immediately available. If this method was to be put into practice, further experimentation would be necessary in order to develop relationships between the concentration of polar compounds and change in refractive index of methanol for different types of oils and different types of transformer aging. It would be helpful as well, to repeat this experiment using refractive index sensors with higher 88  resolution. We were not able to do this with our D-fiber sensor since the surrogate transformers used at Powertech Labs were not filled with oils having refractive index values in its high sensitivity region. Our observation was that the polar compounds in the oils do not seem to change the refractive indices to a tremendous degree, but still do contribute to the increases that are observed during aging as discussed in Section 3.3. In order to use refractive index to detect polar compounds only, the polar compounds must be extracted from the oil samples first, for example, using the methanol extraction technique. It is possible that an integrated “lab-on-a chip”, which detects poiar compounds in the methanol extract using refractive index, could be used as a cost efficient way for online monitoring of oil quality. This would require additional development of this technique and design of the system.  3.5  Effects of other Contaminants in Oil In the previous section we saw that the addition of polar compounds to oil samples  slightly increased the samples’ refractive indices.  Tests showing the effects of other  contaminants such as furans, acid, and dissolved gases are discussed in this section. Contaminants were artificially introduced into oil samples to observe their individual effects in a controlled manner. 3.5.1  Oil Samples Spiked with Furans Experiments were conducted in order to determine if the addition of furans affects the  refractive index of oil samples. Levels of furans in oil samples are measured at Powertech Labs as an indicator of paper degradation in oiL’paper insulated equipment. Oil samples have been prepared at Powertech Labs using V35 oil, with varying levels of 2-furaldehyde, which have been used as controls in system calibration for HPLC. The oil samples used for this experiment varied only by the concentration of furans present in them.  89  As shown in Table 3-12, the concentrations of 2-furaldehyde did not produce changes that were greater than the 0.000 1 resolution of the FISO sensor. The highest concentration of 2furaldehyde present in any of the control oils was 1 O0ppm, which far exceeds the typical values found in samples obtained from the field. Table 3-12: Measured refractive indices of oil samples varying in 2-furaidhyde concentration. Refractive Index 1.4746 1.4746 1.4746  Furan Concentration(ppm) 0.5 1 100  Typical concentrations of furans in oils taken from the field are generally measured in the hundreds of parts per billion (ppb). Table 3-12 indicates that such small concentrations would not produce appreciable changes in the refractive indices of the oils. This study was performed, however, to validate this assumption by finding the minimum detectable concentration that can be detected with our sensor. The D-fiber sensor’s higher resolution would allow smaller changes in the refractive indices of the sample oils to be detected when Luminol Tn oil was used. A mixture of furans was obtained from Powertech Labs containing four compounds including 2-furaldehyde (2-Fur), 5-acetylfuran (Acetyl Furan), 5-methyl-2-furaldehyde (5 Methyl Fur), and phenol.  l2mL  samples of Luminol Tn oil were spiked with drops of the furan mixture, shaken vigorously, and the refractive indices were measured. A sample with 3 drops of furans added to the oil was measured for furan concentration at Powertech Labs using HPLC, and the results are shown in Table 3-13. This measurement was used to correlate the number of drops added to a specific volume of oil to its concentration.  It was observed that very high levels the four compounds  were found in the measured sample. 90  Table 3-13: Measured concentrations of furans in 1 2mL Luminol samples spiked with 3 drops of furan mixture. Compound Name 2-Fur Acetyl Furan 5 Methyl Fur Phenol  Ret Time 6.3  Concentration (ppb) 24500  8.55  6364  10.07 11.89  36548 10585  Samples containing 1 and 2 drops of furans did not show any change in their refractive indices. Samples with 3 drops of furans changed to a slight degree, although the measurement became very unstable, and the refractive index drifted very quickly during a measurement period. This was only the case for the samples spiked with furans though, and not the control samples, so it was assumed that the varying concentration of furans throughout the sample was causing the unstable readings. The number of drops was doubled to 6 in each sample, and a small refractive index change of approximately 4.3 x 1 0 could be measured repeatedly. Although the four compounds shown in Table 3-13 were present in the oil, we will limit our discussion to the varying level of 2-furaldehyde for demonstration purposes. As previously discussed, the concentration of 2-furaldehyde has been directly related to the degree of polymerization of the paper [211. If we look at the concentration of 2-furaldehyde, we see an extremely high level, which would not normally be found in samples obtained from the field. A transformer having oil with a concentration of 2487 ppb has an estimated remaining lifetime percentage of 7% [17], and our concentration was about 10 times higher than this. The measured concentration was 24500 ppb for 3 drops of furans. Our refractive index change of 4.3 x i0 was measured using oil with 6 drops present. The concentration of 2-furaldehyde would be approximately 49000 ppb.  91  If we assumed that the refractive index change of the oil was due to the presence of 2furaldehyde only, and we extrapolate our data, we observe the need of a resolution improvement by a factor of at least 25. There are, however, very high levels of the three other compounds as well, which exceed the normal levels observed in the field.  The addition of these other three  compounds could be contributing to the small increase in refractive index as well.  By  performing this test, we have observed that higher resolution sensors are necessary in order to detect these furans by refractive index change. The concentration of these 4 compounds would not contribute to any large changes in refractive index as paper insulation degrades and it is very doubtful that any furans would either in the typically low concentrations. Also, it is likely that in an oil sample, taken from the field, the changes due to the furans would be masked by changes due to the addition of other aging by-products. The chemical compound Furan, has a structure similar to the aromatic hydrocarbon benzene, although an oxygen is present in the place of two of the six carbon atoms found in the aromatic ring (see Figure 3-10(a) and (b)).  Other chemical compounds having this same  aromatic ring structure with an oxygen present but having other compounds bonded to a carbon in the place of a hydrogen, such as 2-furaldehyde, are often referred to as “furans” (see Figure 3-10(c))  .  During the aging process of paper insulation, long polymers consisting of cellulose  molecules begin to break down into monomer units, which continue to break down into glucose molecules, which eventually break down to form furans [54]. Furans are known to absorb light in the UV region and are often detected using 220 and 280 nm wavelengths [53].  Since  dissolved furans in the oil would lead to higher absorption at these wavelengths, one would expect the refractive index to increase as well. As we have observed, we expect the increase to be very small since the concentrations are in the parts per billion. Hence, when comparing the refractive index change which would be observed due to the addition of furans to that which 92  H H  H H—C  H H (a)  H  H  H  H  c—c  H  H zH  ..,  •• (b)  :o,.  (c)  Figure 3-10: (a) Chemical structure of benzene. (b) Chemical structure of Furan. (c) Chemical structure of 2-furaldehyde.  93  would be observed due to the addition of other aging by-products, such as aromatic hydrocarbons, one expects that the furans would be overshadowed. Nonetheless, the extremely small increase that the furans contribute will add to the contributions of other aging by-products. 3.5.2  Acid Artificially Introduced into Oil Samples By adding acid to oil samples, it is possible to measure the effect that a specific acid has  on the oils’ refractive indices. Here, the type of acid used was acetic acidttttt (99.7% purity) which is commonly found in transformer oils. Drops of acid were added to 23mL of Luminol TRi oil and the samples were agitated vigorously to distribute the acid throughout the oil. The  concentration of acid in one of the samples was measured, which could be correlated to the number of drops added to a specific volume of oil. The refractive index was measured using the D-fiber sensor and the refractive index changes are shown in Table 3-14. A change in refractive index of -2.2 x i0’ was measured for samples with an approximate acid number of 0.72. Samples with an approximate acid number of 0.48 had a measured refractive index change of -1.54 x 1 0. The lowest change in refractive index due to the addition of acid, shown in the table, is -6.1 x i0 . The acid number was 5 measured for a sample showing this refractive index change and was 0.24 mgKOH/gOIL. This is a very high reading for acid, and would only be measured if the oil was extracted from a transformer that was at the end of its life. An acceptable acid number for an in-service oil is less than 0.05 and if it reaches 0.2 the oil should be reclaimed [12]. Figure 3-11 shows the change in refractive index plotted versus the approximate acid number.  Acetic acid was from Fisher Scientific, Ottawa, ON.  94  Table 3-14: Measured refractive index change due to acid added to Luminol TRi oil samples at varying concentrations. Approximate Acid Number (mgKOHIgOIL) 0.72 0.48 0.24 0  An -2.24E-04 -1.54E-04 -6.1E-05 0  Performing this acid experiment showed that transformers at their end of life would show a slight decrease in refractive index due to acid number, that is barely detectable by our sensor. For changes in acid only, using refractive index to detect increasing levels may be possible with better resolution. Although performing this experiment showed that the addition of acetic acid to transformer oil would slightly lower the refractive index, the degree to which it is lowered is very small in comparison with the expected increase in refractive index due to other compounds formed. We have previously observed that during aging the net refractive index of oils increases due to the formation of other aging by-products. Although some of the aging by-products may include acid which would slightly lower the net refractive index, the lowering would be overshadowed by the larger increase due to other compounds, such as aromatic hydrocarbons, being added to the system. Acid acts as a catalyst to oil aging as well, and the presence of acid in an oil would lead to an accelerated rate of formation of other by-products that tend to increase the refractive index.  95  0  0.1  Acid Number (mgKOHJgOIL) 0.2 0.3 0.4 0.5 0.6  0.7  0.OOE+00 -5.OOE-05 -1.OOE-04 .  -1.50E-04 0  -2.OOE-04 -2.50E-04 -3 .OOE-04  Figure 3-11: Change of refractive index of Luminol oil samples versus approximate acid number.  96  0.8  3.5.3  Gas Artificially Introduced into Oil Samples  By artificially adding a particular gas to an oil sample, it was possible to determine if that single gas had any affect on the sample’s refractive index. Gas insertion was provided using canisters containing compressed gases. It was decided that the use of the D-fiber sensor and Luminol TRi oil would be necessary for this experiment, since we assumed that very small changes in refractive index would occur due to the presence of a gas. Samples were prepared using the gases ethane and acetylene, since they were available at Powertech Labs, and had high Ostwald coefficients for Luminol Tn oil (i.e., they would stay in the oil for long periods of time) [55]. Experts working at Powertech Labs also suggested using these gases. Ethane and acetylene may be generated in larger quantities in load tap changers. Using these gases eliminated the use of a testing apparatus which reduces air exposure to the oil, since they have slower diffusion rates from the oils compared to the other fault gases. Oil samples with approximate concentrations of gases were prepared using air tight syringes. Standard bottles that are used by Powertech Labs to store samples obtained from the field for DGA analysis where used to store our samples. These bottles were filled completely with 28. 5mL of oil which minimizes the headspace where the gas could slowly diffuse over time. In order to prepare the samples with approximate concentrations of gas, the volume of gas that would be injected into the oil was first calculated. For example, a concentration of 100 000 ppm of ethane injected into a sample used the following procedure: lppm of a gas corresponds to 1111. per L of oil, so we needed to inject 2.85mL of gas into our oil samples. A tube was connected from the gas canister to the air tight syringe, and the valve on the canister was slowly opened to fill the syringe. When the syringe was filled, a stop at the end was closed to keep the gas inside the syringe, and the gas flow from the canister was stopped. The volume of the gas in the syringe was reduced to approximately 2.85mL by slowly opening and closing the stop until 97  the desired level was reached. Once the desired amount of gas was contained in the syringe, the oil from the bottle was sucked into it. A small space of gas existed above the oil in the syringe, and over time the space became smaller as the gas was being forced into the oil by applying pressure, and shaking the syringe. Eventually there was no space above the oil, as the gas was totally dissolved in the oil, producing an oil sample with the desired concentration of gas. This process was repeated to produce several samples with desired concentrations of ethane and acetylene which were tested in order to see if the refractive indices changed due to the gases. Table 3-15 and Table 3-16, and Figure 3-12 and Figure 3-13 show the results of these experiments.  Table 3-15: Measured refractive index change due to ethane injection into Luminol TRi oil samples at varying concentration levels. Ethane Concentration (ppm) 200000 100000 50000 0  An  -5.3E-05 -2.9E-05 -1.3E-05 0  Table 3-16: Measured refractive index change due to acetylene injection into Luminol TRi oil samples at varying concentration levels. Acetylene Concentration (ppm) 200000 100000 0  An -3.6E-05 -1.6E-05 0  98  Ethane Concentration (ppm) 0  50000  100000  150000  200000  250000  0. OE+00 -1.OE-05 -2.OE-05 .  -3.OE-05 -4.OE-05 -5.OE-05 -6.OE-05 -7.OE-05 -8.OE-05  Figure 3-12: Change of refractive index of Luminol oil samples plotted versus approximate ethane gas conceniTations injected.  99  Acetylene Concentration (ppm) 0  50000  100000  150000  200000  250000  0.OE+00 -5.OE-06 -1.OE-05 —  -1.5E-05 -2.OE-05 -2.5E-05 -3.OE-05 -3.5E-05 -4.OE-05 -4.5E-05 -5 .OE-05  Figure 3-13: Change of refractive index of Luminol oil samples plotted versus approximate acetylene gas concentrations injected.  100  The refractive index decreased by a very small amount for samples injected with either of the two gases. Ethane affected the refractive index a bit more than acetylene, although the changes were still very small. At a concentration of approximately 50000ppm of ethane and 1 00000ppm of acetylene, the refractive index changes were just barely measurable with our sensor. It is common to obtain oil samples from load tap changers that have gas concentrations in the thousands of ppm or possibly even in the tens of thousands of ppm, but not in the hundreds of thousands of ppm. Although we could measure small decreases in the refractive indices when gases were present in the oils, the concentrations that were needed to observe a small change were much higher than would be found in equipment in the field. Looking at the samples measured in Section 3.2.1, it does not seem likely that the relatively small concentrations of gases found in these samples would have contributed to the change in refractive index. Using refractive index change to detect gases does not seem to be a promising method that could be used for online monitoring. It is assumed that the small changes in refractive index that could be measured due to the addition of gas to the oil would be overshadowed by the larger increases in refractive index that occur naturally during aging.  101  Chapter 4 4 Summary, Conclusion, and Suggestions for Future Work 4.1  Summary In summary, we have performed experiments to determine if refractive index can be used  to monitor the quality of high voltage equipments’ insulating oils and subsequently the condition of the equipment. Based on a substantial literature review, we started the investigation with the belief that various contaminants and compounds would affect the refractive indices of oils to a certain extent. In fact, we did observe such effects and, the extents to which various aging by products affected the refractive indices of the oils were recorded. Two refractive index sensors were used in this investigation, one of which had only been used previously for demonstration of proof-of-principle (the D-fiber sensor) and another which was commercially available (the FISO sensor).  A maximum resolution of 1.1 x i0 was  achieved with our D-fiber sensor, which was needed for some of the more sensitive measurements. The FISO sensor was used when the measurements did not require the higher resolution of the D-fiber sensor and when the samples under test had refractive index values that were not in the D-fiber sensor’s most sensitive region. Many oil samples obtained from the field were tested using the FISO sensor. The first sets of oils were extracted from cables, transformer tanks, and load tap changer tanks, which provided us with oils having varying levels of gases in them.  The refractive indices were  observed to have changed in some of the samples, but through the process of elimination, it was 102  concluded that the gases could not have been a factor in the change. The second sets of oils from the field had varying physical and chemical properties that were measured. Here, a set of oils included a sample extracted from the transformer tank that could be compared to another sample extracted from the load tap changer tank. When comparing the sets of oils, it was apparent that, in most cases, the transformer tank sample had a higher refractive index than the load tap changer sample. This was consistent with a previous study performed by another group that measured UV absorption of oils [9]. Accelerated aging experiments were performed using insulating oils and increases in their refractive indices were measured as the oils aged and their colors changed and became darker. These increases in refractive indices were atthbuted to the decomposition of paraffms and naphthenes and the formation of aromatics and other aging by-products. The measured increases in refractive indices were quite large for these experiments. Experiments were conducted using oil samples varying in the concentration of polar compounds. We found that the addition of poiar compounds increases the refractive indices of the oil samples.  However, these increases were not large.  Using a methanol extraction  technique, the refractive index change of the methanol could be used as an indication of the concentration of polar compounds. These refractive index changes in the methanol were found to be much larger than those observed directly in the oil. In order to measure the extent to which the addition of other aging-by products changed the refractive index of an oil, furans, acetic acid, acetylene, and ethane where added to oils in controlled manners. It was observed that these contaminants did not affect the refractive indices to a large degree.  103  4.2  Conclusion During the aging process of oil found in high voltage equipment, various by-products are  formed which typically include (but are not limited to) fault gases, acids, furans, polar compounds, and aromatics. We have observed that the collection of all the compounds that were formed during the accelerated aging experiments increased the refractive indices of the oils to a relatively large degree.  We believe that the increased refractive indices that were observed  would be common to most oils naturally aged in the field. Hence, experiments were conducted to investigate what by-products would contribute to this increased refractive index by separating their effects. The addition of a few types of furans, the addition of acetic acid, the injection of acetylene, and the injection of ethane all changed the refractive indices of the oils to very small degrees. We expect similar results using other furans, acids, and fault gases. The addition of furans led to very small increases in the refractive indices that were only measurable when the concentration of furans greatly exceeded those typically found in oils taken from the field. It was observed that the addition of acetic acid led to a decrease in the refractive index of an oil, however, the concentration necessary to measure a small change matched that found in a sample extracted from a transformer at the end of its life. Decreases in the refractive indices were also measured when acetylene was injected and ethane was injected into the oils. The concentrations of acetylene and ethane needed to measure a refractive index change on the order of our D-fiber sensor’s resolution were over ten times higher than those normally found in oils taken from severely faulting transformers found in the field. The changes in refractive indices, due to the addition of the furans, acids, and gases which were tested, would obviously be masked by the larger changes due to the formation of other aging by-products.  104  Small portions of the increases that were measured in the refractive indices of the oils can be attributed to the formation of polar compounds. Larger portions of the increases in refractive indices can be linked to the formation of aromatic compounds and other aging by-products. The formation of by-products such as aromatic compounds and polar compounds occur as oils degrade and one would expect a more rapid rate of the formation of these compounds in oils being degraded at faster rates. Therefore, the refractive index may be used as a measure of the “break down” of the oil through aging. Hence, an oil with a refractive index that is increasing faster than is normally expected may be experiencing a fault or other undesirable condition which is increasing the rate of the “break down”. We conclude that the refractive index could be used as a “flag” indicating an increased aging rate. One can monitor the level of polar compounds formed in an oil, however, it was observed that directly measuring the refractive index of the oil samples was not useful for this purpose. Since the presence of other compounds affects the refractive index change to a larger degree, other techniques must be used. We believe that using the methanol extraction technique and measuring the refractive index change of the methanol is a useful indicator of polar compounds. This technique could be used in a fashion similar to the one that was presented here but such an approach would require an operator. The technique would become of much more value if an integrated “lab-on-a-chip” type system were designed. Oil could be monitored online for polar compounds if a system were designed that would automatically perform all the necessary steps involved in extracting polar compounds from the oil by mixing a small amount of oil with a small amount of methanol and measuring the refractive index change of the methanol. One can envision that such a system could be designed using MEMS technology (Micro-Electronic-Mechanical Systems technology).  105  The detection of furans by directly measuring the refractive index of oil samples does not seem feasible, as the small changes measured due to their formation would be overwhelmed by the changes caused by other aging by-products. In order to make the detection of furans by refractive index possible, one would first have to separate them from the oil first. Furans can be separated using a solvent such as methanol. However, the furans must then be separated from the poiar compounds that would be extracted by the methanol at the same time.  As was  previously discussed, this is normally done using HPLC. If a method similar to that of [39] were used, where a soT-gel process was used to create a material that absorbed furans in the oil and a corresponding change in the absorption of light at 530 nm was observed, the refractive index might be useful in the direct detection of furans in oils.  It is assumed that the changing  absorption profile of the sol-gel material would result in a corresponding change in the refractive index and, depending on the sensitivity of the material, may prove useful. Separation techniques would be required for the detection of acetic acid or fault gases. There are various techniques which can be used for the separation of gases from oil, such as using the membrane technology of the GE Hydran, or the polymer barriers used in the Morgan Schaeffer Calisto. The detection of gases using refractive index could provide a cost effective solution if made possible, however, it is very doubtful that the sensitivity could rival that of DGA.  4.3  Suggestions for Future Work We have observed the increase in refractive index of V35 oil due to aging and expect  similar results when using other oils. Nonetheless, one should show this for other oils. Our investigation would have been more consistent if Luminol Tn was used throughout the entire duration of our study, however, many experiments were already completed using V35 oil before it was found that the Luminol Tn’ s refractive index value fell within our D-fiber sensor’s high 106  resolution region. Hence, it might be useful to repeat the aging experiments using Luminol Tn. When repeating the aging experiments it would be of value to use a higher resolution sensor, such as the D-fiber sensor.  This higher resolution would allow for smaller changes to be  measured over shorter periods of time.  If possible, the properties of each oil sample (KV  breakdown, acidity, IFT, color) should be measured in addition to the measurement of the refractive index. Measuring the properties of the oils would provide more information regarding the state of oils and their associated refractive indices. It would also be useful if a study could be initiated in collaboration with a utility company, in which the refractive indices of transformers found in a sub station would be monitored over longer periods of time. The study may include online data collection which could commence in the field by comparing the operation and refractive indices measured of similar equipment. Since, to our knowledge, refractive index sensors have not been used in the field for the purpose of insulating oil monitoring, there is only limited data revealing the refractive index changes that would be expected.  The refractive indices of oils taken from  transformers with known histories might also be measured. These measurements could provide some initial data that would give an estimation of approximate changes that could be expected. As more data is recorded and examined trends may become more apparent.  This type of  approach might also be carried out in a lab using controlled conditions as well. In order to monitor the level of polar compounds, the integrated “lab-on-a-chip” approach should be explored. Automated “methanol extraction” of the polar compounds may be performed using a MEMS type device. Sensors which provide high resolution at the refractive index values of typical solvents such as methanol are, therefore, necessary. A member of our lab has begun to investigate the shifting of the high resolution range of our D-fiber sensor by depositing sol-gels onto the surface of the flat side of the cladding. We expect that by varying 107  the sol-gel material, a process can be employed to create sensors with high resolution regions tuned to a specific refractive index value. Others have afready begun to design integrated optic versions of the high resolution sensor as well.  These sensors with tuned regions of high  resolution may also prove to be useful for various applications other than the integrated “lab-on a-chip” or the monitoring of insulation oils. Finally, it may be worthwhile to explore the possibility of using a similar material and so! gel process to that used in [39] for measuring refractive index change due to the presence of furans. If a similar coating were deposited on the surface of our sensor, it might be possible to create a high resolution furan sensor that could be used directly in the insulating oils.  108  References  [1] R. R. Rogers, “IEEE and JEC Codes to Interpret Incipient Faults in Transformers, Using Gas in Oil Analysis,” IEEE Transaction on Electrical Insulation, vol. El- 13, no. 5, October 1978. [2] C. E. Lin, J. M. Ling, and C. L. Huang, “An Expert System for Transformer Fault Diagnosis Using Dissolved Gas Analysis,” IEEE Transactions on Power Delivery, vol. 8, no. 1, pp. 231238, 1993. [3] M. Mi, C. Eley, A.M. Emsley, R. Heywood, and X. Xaio, “Measuring and Understanding the Ageing of Kraft Insulating Paper in Power Transformers,” IEEE Electrical Insulation Magazine, vol. 12, no. 3, pp. 28-34, May/June 1996. [4] M. A. A. Wahab, M. M. Hamada, A. G. Zeitoun, and G. Ismail, “Novel Modeling for the Prediction of Aged Transformer Oil Characteristics,” Electric Power Systems Research, vol. 51, pp. 61—70, 1999 [5] I. Fofana, V. Wasserberg, H. Borsi, and E. Gockenbach, “Retrofilling Conditions of High Voltage Transformers,” IEEE Electrical Insulation Magazine, vol. 17, no. 2, pp. 17—30, 2001.  [6] V. G. Arakelian, “Effective Diagnostics for Oil-filled Electrical Equipment,” IEEE Electrical Insulation Magazine, vol. 18, no. 6, pp. 26—38, 2002. [7] C. T. Dervos, C. D. Paraskevas, P. Skafidas, and P. Vassiliou, “Dielectric Characterization of Power Transformer Oils as a Diagnostic Life Prediction Method,” IEEE Electrical Insulation Magazine, vol. 21,no. l,pp. 11-19, 2005 [8] C. T. Dervos, C. D. Paraskevas, P. D. Skafidas, N. Stefanou, “Dielectric Spectroscopy And Gas Chromatography Methods Applied on High-Voltage Transformer Oils,” IEEE International Conference ofDielectric Liquids, pp. 233-236, 2005. [9] R. Wood, R. Shoureshil, M. Godoy Simoes, and X. Wang. “Optical Sensor for Transformer Monitoring,” Symposium on Diagnostics for Electric Machines, Power Electronics, and Drives, pp. 142-145, 24-26 August 2003. [10] J. Bodzenta, B. Burak, Z. Gacek, W. P. Jakubik, S. Kochowski, M. Urbanczyk, “Thin palladium film as a sensor of hydrogen gas dissolved in transformer oil,” Sensors and Actuators, vol. B 87, pp. 82—87, 2002. [11] T. Aka-Ngnui, M. Benounis, N. Jaifrezic-Renaulti, 1. Stevenson, A. Beroual, “In Situ Monitoring of the Degradation of Insulating Oil Under AC Voltage,” IEEE Annual Report Conference on Electrical Insulation and Dielectric Phenomena, pp. 253-256, 2005.  109  [12] Hydroelectric Research and Technical Services Group, FIST 3-30, Facilities Instructions, Standards, and Techniques: TransformerMaintenance, United States Department of the Interior, Bureau of Reclamation, D-8450, Denver, Colorado, October 2000. [13] T. Golner, “Development of New Insulation System for Use in Oil-Filled Transformers,” Electronics Insulation Conference and Electrical Manufacturing and Coil Winding Conference, pp. 529-532, 4-7 October 1993. [14] Western Area Power Administration, Power Systems Maintenance Manual, Chapterl0: Maintenance of Liquid Insulation, United States of America Department of Energy, December 1990. [15] Department of the Army TM 5686, Power Transformer Maintenance and Acceptance Testing, Washington D.C., 16 November 1998.  [16] American Society for Testing and Materials International, Annual Book ofASTM Standards 2006, Section 10, Electrical Insulation and Electronics, vol. 10.03: Electrical Insulating Liquids and Gases; Electrical Protective Equipment, 2006. [17] Hydroelectric Research and Technical Services Group, FiST 3-31, Facilities Instructions, Standards, and Techniques: Transformer Diagnostics, United States Department of the Interior, Bureau of Reclamation, D-8450, Denver, Colorado, October 2000. [18] M. Duval, “Dissolved Gas Analysis: It Can Save Your Transformer,” IEEE Electrical Insulation Magazine, vol. 5, no. 6, pp. 22-27, November/December 1989. [19] IEEE Standard C57.104-1991, Guide for the Interpretation of Gases Generated in Oil Immersed Transformers, Institute of Electrical and Electronic Engineers, 1 May 1991.  [20] A. M. Emsley, G. C. Stevens, “Review of Chemical Indicators of Degradation of Cellulosic Electrical Paper Insulation in Oil-Filled Transformers,” lEE Conference on Science, Measurement and Technology, vol. 141, no. 5, pp. 324-334, September 1994. [21] D. H. Shroff and A. W. Stannett, “A Review of Paper Aging in Power Transformers”, TEE Proceedings, vol. 132, Part C, pp. 3 12-319, 1985. [22] ASTM Designation: D 5837-99, “Standard Test Method for Furanic Compounds in Electrical Insulating Liquids by High Performance Liquid Chromatography (HPLC),” ASTM International. [231 J. Unsworth and F. Mitchel, “Degradation of Electrical Insulating Paper Monitored with High Performance Liquid Chromatography,” IEEE Transactions on Electrical insulation, vol. 25 no. 4, pp. 737-746, August 1990. [24] ASTM Designation: D1533-00, “Standard Test Method for Water in Insulating Liquids by Coulometric Karl Fischer Titration,” ASTM International. 110  [25] ASTM Designation: D97 1 -99a, “Standard Test Method for Interfacial Tension of Oil Against Water by the Ring Method,” ASTM International.  [26] ASTM Designation: D 974, “Test Method for Acid and Base Number by Color-Indicator Titration,” ASTM International. [27] ASTM Designation: D 1816-04, “Standard Test Method for Dielectric Breakdown Voltage of Insulating Oils of Petroleum Origin Using VDE Electrodes”, ASTM International.  [28] IEEE Standard C57.106-2002, Guide for Acceptance and Maintenance of Insulating Oil in Equipment, Institute of Electrical and Electronic Engineers, (Revision of IEEE Std C57.1 061991), 2002. [29] ASTM Designation: D 1500, “Standard Test Method for ASTM Color of Petroleum Products”, ASTM International. [30] B. Pahiavanpour, and I. A. Roberts, “Transformer Oil Condition Monitoring,” lEE Colloquium on TransformerLfe Management, pp.6.1-6.6,22 Oct 1998  [31] T. Cargo!, “An Overview of Online Oil Monitoring Technologies,” Proceedings to Weidmann-ACTI Technical Conference, 30 November 2005. [32] Calistro, Morgan Schaffer Systems Inc., 2003. [33] J. W. Cooper, Spectroscopic Techniques for Organic Chemists, John Wiley and Sons Inc., New York, 1980.  [34] W. Peng, G. R. Pickrell, F. Shen, A. Wang, “Experimental Investigation of Optical Waveguide-Based Multigas Sensing,” IEEE Photonics Technology Letters, vol. 16, no. 10, pp. 23 17-2319, 2004 [35] J. A. Palmer, X. Wang, R. A. Shoureshi, A. Mander, D. Torgerson, and C. Rich, “Effect of Aging on the Spectral Response of Transformer Oil,” IEEE International Symposium on Electrical Insulation, pp.460-464, April 2-5, 2000. [36] M.P. Zakharich, 1.1. Zaitsev, V.P. Komar, F.N. Nikonovich, M.P. Ryzhkov, and I. V. Skomyakov, “Analysis of Transformer Oil Using JR Analyzers”, Journal of Applied Spectroscopy, vol. 68, no. 1, 2001. [37] Y. Wang, Z. Meng, and H. Sui, “Spectrum Absorption Detecting Fault Gas Dissolved in Transfonner Oil”, SPIE Advanced Sensor Systems and Applications, vol. 4920, 2002.  [38] B. Culshaw, F. Muhammad, G. Stewart, S. Murray, D. Pinchbeck, J. Norris, S. Cassidy, M. Wilkinson, D. Williams, I. Crisp, R. Van Ewyk, A. McGhee, “Evanescent Wave Methane Detection Using Optical Fibres,” Electronic Letters, vol. 28, no. 24, pp. 2232-2234, 1992. 111  [39] R. Blue, D. G. Uttamchandani, and 0. Farish, “A Novel Optical Sensor for the Measurement of Furfuraldehyde in Transformer Oil,” IEEE Transactions on Instrumentation and Measurement, vol. 47, no. 4, August 1998. [40] M. A. Butler, R. Sanchez, and G. R. Dulleck, “Fiber Optic Hydrogen Sensor,” Sandia Report SAND-96-1 133, Sandia National Labs, Albuquerque, NM, USA, 1 May 1996. [41] J. Villatoro, A. Diez, J.L. and Cruz, M. V. Andres, “In-Line Highly Sensitive Hydrogen Sensor Based on Palladium-Coated Single-Mode Tapered Fibers,” IEEE Sensors Journal, vol. 3, no. 4, pp. 533, 2003. [42] J. Bodzenta, B. Burak, Z. Gacek, W.P. Jakubik, S. Kochowski, and M. Urbanczyk, “Thin Palladium Film as a Sensor of Hydrogen Gas Dissolved in Transformer Oil,” Sensors and Actuators B. Chemical, vol. 87, pp. 82-87, 2002. [43] J. Unsworth and N. Hauser, “Relating Low Frequency Dielectric Properties and Refractive Index of Mineral Oils,” IEEE Conference on Properties and Applications of Dielectric Materials, pp 82-85. [44] S. M. Chandani, “Fiber Optic Sensors Based on D-Shaped Elliptical Core Fibres,” PHD Thesis, University of British Columbia, 2007. [45] A. Yariv and P. Yeh, Photonics: Optical Electronics in Modern Communications, edition, Oxford University Press, New York, 2007, pp. 211 219.  th 6  —  [46] Cargille Labs Thermo-Optic Oils Data Sheet, Immersion Liquid Code 1160, 5040, and 50350, Cargille Laboratories, Inc., Cedar Grove, NJ, USA, June 2006. [47] C. Belleville and G. Duplain, “Fabry-Perot Optical Sensing Device for Measuring a Physical Parameter,” U.S. Patent 5 202 939, Apr. 13, 1993. [48] T.K. Saha, M. Darveniza, D.J.T. Hill, and T.T. Le, “Electrical and Chemical Diagnostics of Transformers Insulation Part B: Accelerated Aged Insulation Samples,” IEEE Transactions on Power Delivery, vol. 12, no. 4, October 1997 [49] R. Blue, D. Uttamchandam, 0. Farish, “Infrared Detection of Transformer Insulation Degradation Due to Accelerated Thermal Aging,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 5 no. 2, April 1998 [50] P.M. Mitchinson, I. L. Hosier, P. L. Lewin, A. S. Vaughan, G. C. Chen, and P. Jarman, “An Experiment to Evaluate the Benefits of Processing Aged Transformer Oil,” IEEE Symposium on Electrical Insulation, pp. 89-92, 2006. [51] T. B. Whitfield, J. E. Castle, C. Saracco, M. Au, “Transport of Copper Over Paper in HighVoltage Electrical Insulation”, Surface and Interface Analysis, vol. 34, no. 1, pp. 176-179, 2002.  112  [52] R. J. Kisch, S. Hassanali, S. Kovacevic, and N. A. F. Jaeger, “The Effects of Polar Compounds on Refractive Index Change in Transformer Insulating Oils,” ALTAE High Voltage and Electrical Insulation Conference, 29 October 2007. [53] D.J.T. Hill, T.T. Le, M. Darveniza and T. Saha, “A Study of the Degradation of Cellulosic Insulation Materials in a Power Transformer Part III: Degradation Products of Cellulose Insulation Paper,” Polymer Degradation and Stability, vol. 51, pp. 211-218, 1996. [54] C. H. Zhang, and J. M. K. Macalpine, “Furfural Concentration in Transformer Oil as an Indicator of Paper Ageing, Part 1: A Review,” IEEE PES Power Systems Conference and Exposition, pp.1088-1091, Oct. 29 2006, [55] J. Jalbert, R. Gilbert, P. Tetreault, and M. A. El Khakani, “Matrix Effects Affecting the Indirect Calibration of the Static Headspace-Gas Chrornatographic Method Used for Dissolved Gas Analysis in Dielectric Liquids,” Analytic Chemistiy, vol. 75, no. 19, 1 October 2003.  113  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0067031/manifest

Comment

Related Items