By Mark N. DeDecker, P.Eng.
Editor's Note: Having already run a paper from Eric Smith of PCB Disposal in our January issue entitled, "Reclassification vs. Replacement", we have chosen, in the interest of fairness, to run this paper from Mark DeDecker of ENSR Inc. These papers provide an excellent assessment of two dispoal options currently available on the market.
Background
Since the production of PCB transformers was outlawed in many countries in 1978, several options for the remediation of the existing PCB transformers have been proposed. Some have failed. Some have been a complete success. Still others have been in between.
ENSR's System 50 reclassification technology is one of the successes. This article will detail:
- the historical success of reclassification technology,
- present the advantages and precautions needed when using Per-chloroethylene,
- present mathematical evidence of the effectiveness of the System 50 process,
- backup the mathematical models with actual long-term performance of the System 50 technology and
- briefly touch on the life expectancy of a typical Askarel Transformer.
History of Retrofilling Technologies
Shortly after the US TSCA law was enacted, several companies devised methods to reclassify Askarel transformers. The reclassification methods were expected to be less expensive than replacement while at the same time complying with regulations regarding a ³Non-PCB² transformer.
One of the first technologies involved retrofilling Askarel transformers with silicone oil. This technology was invented by Dow Corning. After flushing the drained transformer with either silicone or a compatible solvent, the transformer was filled with silicone.
As most readers know , this practice would yield a concentration of PCBs in the silicone oil of approximately 2-5 per cent (20,000 to 50,000 ppm) in a typical transformer after a few weeks. After the retrofill, the PCBs are filtered out with Activated Carbon. Some initial problems were experienced with the technology, particularly if proper precautions were not taken to prevent water in the activated carbon from entering the transformer. After this problem was solved, the technology was effective at reaching a concentration of less than 50 ppm in the transformer after processing for several months and using multiple Activated Carbon canisters. However, that concentration rarely remained below 50 ppm PCBs. This is due to a number of properties that have been recognized since these initial retrofills:
- Silicone oil has limited solubility of PCBs
- Silicone oil has a relatively high viscosity
- Silicone oil has a very large molecular weight
- Transformer have two to 10 per cent (compared to the weight of the fluid) cellulosic materials such as wood and paper which are not well penetrated by silicone oil.
Together these factors give silicone oil a poor PCB extraction rate from porous cellulosic transformer components. The result yields an initial reclassification of transformers to less than 50 ppm PCBs. However, over extended time (frequently less than six months), PCBs would continue to leach out of the internals, increasing the PCB concentration to a value above 50 ppm, and in some cases to a value above 500 ppm (after several years). This leach back problem could continue for many years perhaps even decades.
Other fluids such as Rtemp or even mineral oil have limitations similar to Silicone oil although mineral oil is a somewhat better "solvent."
In 1985 Union Carbide began marketing a process in which chlorinated benzenes were used to extract PCBs from the transformer components for some period of time, before the final fill of silicone oil was introduced in the transformer. The process involved four or more retrofills to reclassify the transformer. In theory; if each retrofill was allowed to reach equilibrium, and the leach back from each retrofill was 5 per cent, any transformer would be less than 10 ppm PCB after the fourth retrofill forever. The only difficulty in this process was determining the proper timing for retrofills. With some transformers, six months will yield a permanent reclassification at less than 50 ppm PCBs. However, some transformers may require in excess of 18 months between retrofills.
Due to the nature of silicone oil for the final two retrofills of the transformer, an error in timing between retrofills could extend the time to permanent reclassification of the transformer by months or even years. This occurs when large amounts of wood and paper are present in the transformer and when the Silicone oil is introduced early. Thus, the procuedure failed for the same reason that direct retrofilling with Silicone oil failed: because of the poor solvency of Silicone oil for PCBs.
In 1986 ENSR began marketing a reclassification technology utilizing a Perchloroethylene (Perc) based dielectric fluid. Initially, it was thought that transformers could be reclassified after 6 months of extraction time in Perc. This led to a process which utilized a distillation system to continuously remove the PCBs from the Perc fluid starting immediately after the initial retrofill. After several years of experience with this technology, the fundamentals of transformer PCB extraction with Perc is now well understood. This has lead to the understanding that:
- All transformers will behave differently with respect to time to achieve permanent reclassification to less than 50 ppm PCBs.
- Correlations are generally found between the reclassification times of transformers with the same manufacturer and similar serial number series.
- Reclassification time is affected by the operating characteristics of the transformer, however, the range of operating temperature of most Askarel transformers is very narrow.
Attributes of Perchloroethylene Dielectric Fluids
At this point I would like to discuss the reasons that ENSR chose to use a Perchloroethylene-based dielectric fluid (Perc).
Perc is a widely studied fluid which has been used for more than 5 decades in the dry cleaning industry. During this time the precautions used in handling Perc have gone from essentially none to a fairly strict set of operating requirements for the dry cleaning operators. The current work place limit for 8 hour exposure to Perc is 25 ppm. Many people can detect the odor of Perc at about 5 ppm, which explains the familiar smell at the dry cleaners.
Table 1 contains additional detail of the thresholds of odor for various levels of Perc and the associated physiological effects.
The chart above shows that if a person can smell the Perc, they should be wearing some form of personal protective equipment such as a breathing mask with activated carbon filters, or a supplied air breathing apparatus (Supplied air is recommended for extensive spills of Perc, normal handling requires only an activated carbon mask). There is no reason for a person to remain in contact with Perc vapors long enough to cause deleterious.
Perc is an excellent fluid for the extraction of PCBs from transformers. Its low viscosity and molecular weight combine to give Perc a distinct advantage in the speed with which it can extract PCBs from transformer material.
A paper authored by GE testifies to this fact. Their conclusions found that Perchloroethylene is definitely superior to chlorinated benzenes and silicone in extracting PCBs. Their calculations also agreed with ENSR's experience in reclassification, determining that many transformers would not reclassify with only 6 months of processing. However, the effectiveness of Perchloroethylene was good enough for GE to launch their own reclassification process called Filtran Plus, in which Perchloroethylene was used for the initial retrofill of transformer.
Mathematical Modeling of the Extraction of PCBs from Transformer Materials
The permanency of reclassification of Askarel transformers has been a subject of discussion ever since the U.S. Environmental Protection Agency initiated the PCB rules into TSCA in May of 1979. Several reclassification technologies have evolved since that time. Mathematical modeling has evolved to explain the successes and failures of these technologies.
This section of the paper compares a mathematical model with actual field experience in reclassification technologies. To date, ENSR Operations has serviced over 8,000 Askarel transformers with over 7,000 of these reclassified to date.
The results of the study confirm the viability of applying this mathematical model to PCB reclassification technologies. The model data leads to the conclusions that:
- it is inevitable that any transformer can be permanently reclassified with the appropriate technology.
- reclassification time can be predicted based on limited transformer data.
- total residual PCBs within a transformer can be calculated from the model equations.
- all transformer materials will be < 50 ppm when permanent reclassification is achieved.
The model breaks down the diffusion of PCBs from the transformer into individual diffusion rates from various transformer internals; for example from wood, coil and cellulose components. Each of these components have different diffusion r>