Home » Destruction of Dibenzyl Disulfide in Transformer Oil
There is not one single corrosive sulfur compound that is responsible for all corrosive sulfur issues that are present in all mineral oil filled electrical apparatus. Depending on the oil, there can be tens to hundreds of different sulfur compounds present in the oil. Of these, only a small fraction are corrosive or are compounds that can degrade from stable species into ones that are reactive. This is usually based on time and temperature. Only a very few corrosive sulfur compounds have been identified of which dibenzyl disulfide (DBDS) is one. This paper only concentrates on DBDS. The reason is that it has been found in many oils that have resulted in recent failures (2000-2007) of transformers or reactors due to corrosive sulfur attack and the formation of copper sulfide.
There are five main classes of sulfur compounds found in crude oil but not all types are considered to be corrosive or reactive (see Table 1)[1]. Elemental sulfur and sulfur compounds in concentrations up to 20% [2] are present in the crude oil used to make transformer oil.
Sulfur is commonly found in crude oil, as it is a common element in the earth’s crust. As shown in Table 1, elemental sulfur and the sulfur-containing mercaptans are very reactive followed by sulfides. Reactive sulfur is mainly in the form of organic sulfur compounds like R-SH, where the sulfur is attached at the end of an organic molecule. When the molecule is more complex, for instance when the sulfur is surrounded or contained within the molecule then the sulfur compounds are more stable and less reactive, like in R-S•S-R. Just a few years ago, disulfides (of which DBDS belongs) were once thought to be very stable but it has been found that the disulfide linkage can be susceptible to cleavage resulting in the production of mercaptans. Thiophenes are the most stable of all these sulfur compounds. Research at Doble has shown that even thiophenes will break down given enough time and thermal stress. Whether they form corrosive sulfur compounds or not is unknown. It is known that a large percentage of the breakdown products will reform into smaller thiophene compounds. Crude selection and the refining process are the two main factors that dictate the presence of any of the five sulfur groups in a finished transformer oil.
Presently, the refining techniques are such that detectable amounts of elemental sulfur and mercaptans are very rare in a newly finished transformer oil. Other sulfur containing products, especially thiophenes, are considered advantageous as they may provide some degree of oxidation stability, although there is some debate over that claim.
DBDS in itself might or might not be corrosive. Some researchers [3] suggest that a DBDS-copper complex is formed in which the copper is removed from the conductor surface and goes through a series of reactions in which copper sulfide is then formed on the copper surface. The information presented in this paper suggests that DBDS degrades through cleaving of the disulfide linkage as the temperature increases in the oil resulting in the formation of mercaptans. These DBDS breakdown products are very corrosive. Experiments at 110°C involving DBDS showed corrosion of the copper surface occurs in a relatively short period. Degradation of DBDS at temperatures lower than 110°C can also occur and in experiments performed over the past several years corrosive sulfur attack on copper in oils with DBDS occur at temperatures as low as 80°C in just over 60 days. Subsequent testing of that oil shows a reduction of DBDS during that time. Other researchers using a combination of copper and paper have detected the development of copper sulfide at 80C in the Kraft paper insulation and on the copper surface [4]. It might be that both processes, as well as others, occur simultaneously.
A review of the literature indicates that DBDS is usually found in concentrations ranging from 100 to 1000 mg/kg (ppm) in certain lubricating oils. DBDS is added to lubricating oils to protect against wear, reduce friction and increase oxidation stability. Through a method developed at Doble to detect DBDS, concentrations of 100 to 180 ppm have been found in new transformer oils but only in a small percentage of products tested. Whether DBDS is formed as a result of the refining process, added or a combination of both is unknown. The chemical structure of DBDS is found in Figure 1.
The focus on DBDS is not without reasons. They are:
Initially, the first experiment was performed to make sure that DBDS did form corrosive sulfur compounds in the oil that attacked the copper and caused the production of a sulfide deposit on the copper surface. In these experiments DBDS was added to a white mineral oil in varying concentrations of 250, 125, 50 and 5 ppm (mg/kg). A white mineral oil has a very low total sulfur content measured at 2-3 ppm maximum. White oil that was used in the experiments was tested without the addition of DBDS and did not exhibit any signs of copper corrosion (ASTM D 1275B) or copper sulfide formation on the paper insulation by the Doble Covered Conductor Deposition (CCD) test. Please refer to the Addendum for an explanation of the Doble CCD test.
The white oil with 250 ppm of DBDS was tested using ASTM Test Method D 1275B (aging an abraded copper strip for 48 hours at 150°C). It was found to fail this test within 40 hours. In addition, the concentration of DBDS was determined before and after testing and found to have been reduced by more than 50% with a final concentration of 116 ppm.
Additional experiments were conducted to help in determining at what concentrations DBDS could still cause corrosion of the copper surface. Concentrations of DBDS in white oil were tested at 125, 50 and 5 ppm. The results of the testing are shown in Figures 2 (Picture of Corroded Strips) and 3.
As shown in Figure 3, the difference in failure times between concentrations of 250, 125 and 50 was not that significant (40-48 hours). Even a concentration of 5 ppm corroded the copper strip over an extended period of time. This indicates that even small concentrations of DBDS may cause corrosive sulfur issues in electrical apparatus.
As a result of this experiment, a theory was put forth that DBDS were cleaved to the corresponding mercaptans (thiols) and other single-ring products. Further, four possible compounds could be formed of which benzyl mercaptan was the most likely. The chemical reaction is presented in Figure 4.
Benzyl mercaptan is very volatile and would not ordinarily be present in a newly refined transformer oil, as it would be easily removed. However, once produced, there is no escape in a well-sealed transformer. It is very oil soluble and is very reactive to copper and silver surfaces.
The same is not true in open conservator transformers where oxygen is present in higher concentrations. It was concluded that degradation of organic sulfur compounds involves an oxidative attack localized at the sulfur atom [8]. As a result, benzyl mercaptan molecules are oxidized and some DBDS is actually reformed. Some of the benzyl mercaptan is likely lost through the free breathing nature of the conservator. Copper sulfide is formed but at least half or less than what would be formed in a sealed transformer. This chemical reaction is shown in Figure 5. It is most likely that some water is also formed, but the amount would be so minute in comparison to the water content already existing in the transformer, it would be indistinguishable.
Research was initiated to determine a method to remove DBDS and the degradation byproducts of DBDS from the oil as it was clear that these sulfur compounds cause copper and silver corrosion and the formation of sulfide films.
Numerous methods of removal or destruction were attempted with varying levels of success. However, one of the methods attempted yielded exceptional results with the destruction of DBDS in white oil. In order to conduct the destruction process, the information in a patent by Louis Pytlewski et. al of the Franklin Institute in Philadelphia, PA approved in 1983 [9] was consulted. Although several methods are provided in the patent to produce reagents for PCB dechlorination, one seemed best suited as the reagent was very easy to produce, did not use sodium metal and provided much safer operating conditions. The reagent was prepared by adding 10 grams of solid sodium hydroxide pellets to an open beaker with 50 mL of 400 molecular weight polyethylene glycol and mixing until totally dissolved. The ingredients were mixed at about 120°C until it turned a dark brown color that indicated that the decomposition reagent was formed.
About 500 mLs of sample oil from a large ABB GSU transformer manufactured in 2001, rated at 280 MVA, 200 kV, was added to the beaker at 120°C and mixed with the reagent. The sample oil used was known to fail the corrosive sulfur tests. By adding the oil this effectively lowered the temperature from 120°C to about 80°C in an air environment. Samples were then taken at 1, 5, 10, 30 and 60 minutes to test the DBDS concentration. The temperature of the mixture stayed between 80 and 100°C. The results are listed in Table 2 and shown in Figure 7.
As shown in Table 2 and Figure 7, the experiment was very successful in destroying the DBDS compound in the oil. It only took a little over 10 minutes of contact time to reduce the DBDS concentration to below 1 ppm. Testing via ASTM D 1275B and the Doble CCD with dielectric strength testing (CCD+DT) of the paper was performed to make sure that the destruction of DBDS by this method yielded a non-corrosive transformer oil. The results of the ASTM D 1275B test are shown in Table 3.
Results of the Doble CCD+DT tests are provided in Table 4. Usually the CCD test is carried out for 4 days but to make absolutely sure that the destruction of DBDS and benzyl mercaptan was successful, a 6-day test was performed.
Only enough oil was removed to perform a single vial test as opposed to both “Air Ingress” and “Sealed” so as to save enough oil for the D 1275B test as already reported in Table 3. Although the visual inspection of the CCD samples showed that they had a deposit on the paper, they were not metallic in nature. Additionally, the dielectric tests show that the deposits on the paper did not adversely affect the insulation qualities of the paper as new oil impregnated 3-mil thick paper ordinarily has a dielectric strength of approximately 1700 to 2000 volts/mil.
Two additional processes were conducted after the original soldium alkalai treatment. Thery were:
A 55-gallon drum of oil from a sealed GSU transformer with a known corrosive sulfur issue involving DBDS was shipped to Power Substation Services located in West Virginia. There a to pilot run was performed using a commercial sodum alkali processor. These systems have been used in the electrical utility industry for 20 years or more for either energized or de-energized transformer oil processing to dechlorinate polychlorinated biphenyls (PCBs). Besides the sodium reagant reaction chamber, these processors can also perform Fuller’s earth (clay), dehydrating/degassing treatment and reconditioning through filtration. Because most of these oil processing trailers require in excess of 300 gallons of oil just to prime the system, Power Substation Services (PSS) had to modify the original procedure to accommodate the small volume of oil. The conditions of the processing are provided in Table 7.
PSS arranged for two sister GE transformers to be processed using the sodium treatment processor technique. One transformer had no detectable level of DBDS and the other had only 7 ppm of DBDS present even though each transformer had around 900 to 1000 mg/kg of total sulfur. Even though the oil in these transformes had a high concentration of sulfur, it was not corrosive (refer to Table 11). It was only after DBDS was added that the oil became corrosive. Doble provided oil concentrates of DBDS to PSS to add to the transformer in order to increase the concentration to over 200 mg/kg in each unit.
Both transformers were single phase, substation type transformers that were retired from service. The primary voltage was 7.2 kV and the secondary voltage was 480 Volts. Each transformer was 667 kVA and contained approximately 350 gallons each. Figure 11 is a photograph of the two sister transformers.
Since these transformers were retired, there was no natural oil convection. In an effort to make the DBDS concentration homogeneous throughout, the oil in each transformer was circulated for approximately 30 minutes. The pump flow rate was approximately 10 gallons per minute, so about 300 gallons was circulated through the pump to aid in mixing the DBDS concentrated solution added to each transformer. The circulation was not performed using the oil-processing trailer in an effort to minimize the dilution of the DBDS and to ensure the additive remained in the transformer. For each transformer, an oil sample was taken before the DBDS was added, and after the oil circulation process was completed.
For simplicity of discussion, the process described, as well as the results, pertain only to one transformer. The exact same process was performed on both transformers, and samples were drawn at the same oil volume increments when the processing was conducted on each unit. The results, which are discussed later, were essentially the same for both transformers.
The initial PCB concentration of transformer ID 1207313 was 33 ppm. It was slightly low on oil so before the DBDS was added and mixed, mineral oil was added that met the requirements of ASTM D 3487. As a result of the mixing, a sample drawn from the transformer (bottom valve) just prior to oil processing showed an initial result of 54 ppm DBDS. This was considerably lower than what was expected, which indicates that the DBDS was not as thoroughly mixed throughout the oil as was desired. Part of the reduction in the expected DBDS concentration could also be due to some absorption of the DBDS from the oil by the paper insulation. In laboratory experiments some uptake of the DBDS into the paper has been observed. Because processing of the oil occurred so quickly after the addition of DBDS, it is unlikely that much was absorbed by the paper.
Table 10 lists the initial oil qualities of transformer 1207313 before it was processed through the rig.
After the hoses were connected to the transformer, the oil was pumped from the bottom valve into the processing trailer where it was heated. Next, the reagent was injected into the oil to react with the DBDS. The oil passed through a degassing column and then to a stage to quench any remaining reagent. Afterwards, the oil passed through a centrifuge to remove the reagent byproduct. Then, the oil passed through the fuller’s earth tanks, the vacuum dehydration column, and the final filter before being returned to the top of the transformer.
The first sample taken was an outgoing sample from the oil processing trailer at 250 gallons. This is the approximate midpoint of the first pass. A pass is defined as the transformer nameplate gallons. The second sample drawn was the mid point of the second pass. Another sample was taken near the midpoint of the third pass and also half way through the fifth pass. For the purpose of this project, additional passes were performed and a final sample was taken at the completion of the eighth pass. Figure 13 is a photograph taken of all of the oil samples drawn for transformer 1207313 for testing.
With the samples that were drawn during the processing of the oil from the two transformers, the following battery of tests were performed:
It appears that the original mixing of the DBDS into the total volume of the transformer oil should have proceeded for a longer period of time. Nevertheless, Figure 14 clearly shows that there was no DBDS present before it was added. Once it was added it was detected. Although the DBDS measured in the initial sample was only 54 ppm, enough DBDS was added to each transformer to elevate the initial ppm concentration to over 200 ppm DBDS. For the sister transformer, an initial contamination level of 81 ppm DBDS was measured. Despite this uneven mixing, once the processing of the transformers began, the DBDS was destroyed quickly and within 4 passes the concentration was below 1 mg/kg. At the same time, the PCBs were also destroyed in the process.
The results in Table 11 and Figure 15 provide the results of CCD+DT taken during the course of processing the oil. None of the copper rods used in the CCD tests ever developed a copper sulfide film that would indicate that the oil was corrosive. However, the oil was contaminated enough with DBDS to produce metallic deposits on the paper in the CCD test which are considered to indicate that copper sulfide films were present in the paper (Refer to the addendum).
Dielectric strength tests were performed on the Kraft paper from the CCD test to ascertain the effects of the deposit on the electric strength of the paper. In brand new 3-mil thick Kraft paper that has just been vacuum dried and oil impregnated values of 1700 to 2000 volts/mil are common. After performing this testing on several hundred samples as part of the CCD+DT test, Doble recommends that a 50% or greater reduction in dielectric strength is cause for concern.
Both the laboratory process and the PSS process (as used in the field) were completely successful in destroying DBDS (dibenzyl disulfide) from the oil. This process does not work on all types of sulfur compounds that are corrosive as experiments have been conducted in this respect. However, it may be just as succesful on other sulfur containing compounds such as certain types of mercaptans, sulfides and disulfides (but possibly not all) and has not been tested out in this regard. It is most likely that sulfur in the form of sodium sulfide or related compound is formed by the process and removed. Further research needs to be conducted in this respect.
Like other mitigation techniques, the removal of DBDS does not effect corrosion that has already taken place. However, this process will destroy DBDS and benzyl mercaptan that is remaining in the oil. For now, it is recommended that besides the process to remove the DBDS, clay treatment followed by the addition of synthetic oxidation inhibitor to a concentration between 0.22 and 0.30% (by weight) be incorporated into the process. In removing the DBDS, the oil might be less oxidatively stable and thus the reason for the addition of an oxidation inhibitor. The reason for this is that most oils with DBDS are not inhibited and therefore require some sort of stabilization for oxidation after the process of DBDS destruction and thus the recommendation to add DBPC (BHT) or DBP. For sealed transformers this is less of a concern.
The major advantage of the alkali process is that it should be able to treat most the oil in the transformer using just a few passes. This in contrast to a drain and flush process that will leave a good 10% of the total volume of the old oil behind. In addition, for larger transformers, an oil processing trailer would still have to be mobilized to vacuum-fill the transformer for drain and flush processes. The other advantage is, in most cases, for voltage classes 230kV and below, this DBDS removal process can be performed on an energized transformer so the utility is not required to take an outage.
Lance Lewand is the Laboratory Manager for the Doble Materials Laboratory and is also the Product Manager for the Doble DOMINO®, a moisture-in-oil sensor.
Scott Reed is President and CEO of MVA. Mr. Reed is an Electrical Engineer and is an executive officer of the IEEE Transformers Committee. At the time of this publication, Scott was working for another company.
TRANSFORM. INFORM. ADJUST. DIRECT. PROGRESS.