James B. Moon
Many forms of instrumentation are utilized in the operation and maintenance of the modern utility. Information is knowledge and knowledge enables us to predict and control the processes and events that shape customer satisfaction and operating costs. Instrumentation continues to become less expensive and more capable. The pace of development since the introduction of the microprocessor has steadily increased and the Internet promises to bring new and innovative ways in which to rapidly deliver data and information. At the same time, the burden of interpretation in such a wealth of data provides new opportunities for error unless expertise and automation are fully exploited.
Information about the status of power transformers rated at 25 MVA and above has historically been limited to a top oil temperature alarm, the results of a yearly dissolved gas analysis (DGA) and a few other bits of information for larger transformers. Today’s technology provides for the immediate measurement of dissolved gases in transformer oil and other critical parameters on a near real-time basis. But herein lies the difficulty. The interpretation of the data provided by the instrumentation requires a fundamental knowledge of the way in which the measurement is made.
Each type of transformer defect or fault produces a different combination of gases. These gases can be measured in their dissolved state in the transformer oil or in the headspace at the top of the transformer. The gases resulting from a transient fault condition change with time, both in concentration as well as in the ratio of some of the more volatile gases. Almost all faults produce some level of hydrogen, which is why hydrogen monitoring is often used as a basic alert signal that is the basis for a more thorough DGA. The nature of that signal depends on the technology used to measure the dissolved gas level.
Three types of measurement technology have been demonstrated in field operations. Semiconductor sensor technology utilizes a silicon chip that provides an electrical signal when exposed specifically to hydrogen. While the response time is relatively fast, several minutes may be required in order to achieve the full response proportional to the hydrogen concentration. The specificity of the response to hydrogen and hydrogen alone is quite good.
Fuel cell technology has also been used to monitor hydrogen levels in transformer oil. An electro-chemical oxidation of hydrogen at a detection electrode produces an electric current proportional to the quantity of hydrogen. For small molecules like hydrogen, the response sensitivity is said to be 100 percent of concentration. Other molecules such as acetylene, ethylene and carbon monoxide can also undergo oxidation at the electrode resulting in additional electrical output. These additional electrical outputs are part of a composite signal and not distinguishable by individual gas.
Gas chromatography is the third technology that has been used in field measurements. Gas samples are either taken from the headspace of the transformer or are extracted from the oil and then passed through long, thin columns. Thermal conductivity measurements made on the gases provide a signal, which is translated into the concentration of the gas in the original sample.
Relating this data to the transformer fault is the key to making good maintenance and operational decisions. Some indications are benign, or at worst, relate to long-term reliability or the prediction of the end of useful life of the transformer. Other measured values, however, are indicators of serious faults that may result in sudden catastrophic failure.
Table 1 shows that some faults generate low levels of gases over a long period of time while other faults generate considerable concentrations of gas very quickly.
Consider two transformers of the same type and rating. The condition of the two transformers begins in the same, normal operating region, exhibiting low levels of gases that are common to most transformer operation. Background hydrogen is present at levels of 40 parts per million (PPM) and carbon monoxide at a level of 150 PPM. The readings from each of the three types of instrumentation are shown in Table 2. While the readings for hydrogen are very similar, the gas chromatograph provides the complete picture, much as would be obtained from DGA.
Now consider two different events in each of the two transformers. In transformer A, a partial discharge event occurs which generates mostly hydrogen. The hydrogen level rises to 600 PPM and the methane level rises to 80 PPM while the carbon monoxide level remains unchanged. Table 2 shows the information provided by each of the three types of instrumentation is quite different. The fuel cell and semiconductor technologies indicate that something has changed, but the specifics are inconclusive. The chromatography results indicate that partial discharge is a probable event.
Transformer B experiences an arcing event. The hydrogen level rises to 800 PPM and the level of acetylene rises to 200 PPM. In Table 2, the hydrogen level reported by the semiconductor sensor shows the level of hydrogen that is actually present in the transformer oil. The fuel cell instrument reports a combined response that is the sum of the instrument’s sensitivities to each of the gases present. The gas chromatograph reports each of the gases on an individual basis, providing a complete analysis of the status of the transformer.
In comparing the two events in the two transformers the semiconductor sensor provided information that was appropriate for the hydrogen level. Although the technology doesn’t provide any insight beyond the hydrogen level, the rise in hydrogen is a clear signal that there has been a change in the status of the transformer that demands an immediate DGA. The fuel cell technology clearly indicates a rise in dissolved gas levels that is demanding of DGA, but due to the combined response indicates a level for transformer B not significantly different than for transformer A. While the combined levels of the gases to which the fuel cell technology responds is 1,260 PPM, only 844 PPM is reported due to the different sensitivities of the instrument to hydrogen (100%), carbon monoxide (18%), acetylene (8%) and ethylene (1.5%). Only gas chromatography provides an analysis that would dictate an immediate change be made in the operation of the transformer, or even that it should be taken out of service due to the increase in acetylene.
Interpretation of instrumentation readings requires a thorough understanding of measurement capabilities of the instrumentation being used. It is important to carefully consider the objectives that are to be realized in instrumenting high value assets to insure that the proper instrumentation is specified at the time of initial procurement.
Moon is CEO of Micromonitors Inc., Bend, Ore. More information is available at 800-880-2552 or www.micromonitors.com.