Learnings

Outcomes

Three papers have been produced as part of the project, these are:

• Draft Measurement of PCBs in Transformer Oil - NPL (National Physical Laboratory)

The objective of the literature review was to understand the techniques available, both commercially and non-commercially, for measuring PCBs in the oil of pole-mounted transformers by either safely accessing the oil itself from a live transformer, or by measuring PCBs in the headspace of the transformer from the breather tubes. From extensive internet, literature and patent searches, it was found that a high number of the techniques that may give rapid results and that may be suitable for this application are in the developmental phase and not yet at a stage where they could be deployed for the work under consideration.

• Literature Review and Headspace Analysis report - NPL (National Physical Laboratory)

This feasibility study has shown that PCBs are not present at detectable levels in the headspace of transformers unless the oil in the transformer is heated. Even if the transformer is approximately at a temperature between 20°C and 60°C, PCBs may only be present in the low ppt levels, with the transformer containing at least 50 ppm of PCBs overall.

• Report on spectroscopic detection - RAL Space

The study focuses on optical spectroscopic detection techniques, with emphasis on the mid-infrared (mid-IR) wavelengths. 

Lessons Learnt

Multiple solutions have been discovered to identify PCB molecules within mineral oil, this was completed as part of a literature review carried out by National Physical Laboratory (NPL) and RAL Space, the solutions discovered by the review required physical samples of oil, but as a result the tests are able to determine the quantification of PCBs, this is important as there is an allowed threshold of 50ppm so a test that quantified results is far better than a positive/negative one.

Ultraviolet (UV) and infrared (IR) absorption spectroscopy are the two methods that are able to accurately quantify PCB molecules in a gaseous state, this is key when analysing the vapour within the headspace of an asset. There are known methods of detection for physical oil samples such as the tests that are currently being used such as ‘Chlor n Ol’ tests, the issue with these is they do not quantify PCBs and can provide false positive results.

• Key conclusions on spectroscopic data

Availability of quantitative spectroscopic data for PCBs appears extremely limited, such that a necessary first step in the development of any spectroscopic analysis device would be a laboratory measurement campaign to obtain quantitative spectra of ideally all 209 PCB congeners, or at least a significant subset of these known to be spectrally distinguishable. 

The UV (Ultra-Violet) absorption method has poor congener selectivity and so is not considered further. 

• Key conclusions on cross section data

The availability of quantitative PCB spectroscopic data is extremely limited; in order to enable PCB detection performance assessment, empirical methods have been established to derive semi-quantitative model spectra for a representative range of 12 PCBs in both vapour and liquid phase from the very limited set of quantitative spectroscopic data found in the literature. 

The absorption coefficients of model spectra have been generated, which provide the data required to estimate the minimum detectable quantity of PCB in a given sensing scenario. 

The positions of congener spectral features have been established, which, together with a knowledge of the spectral characteristics of the mineral oil or its vapour, informs the choice of spectral window for a sensing device. 

• Key conclusions on PCBs volatility

The low volatility of PCBs in general limits the concentration of PCB to be expected in a transformer headspace. At room temperature, the headspace above an oil sample containing 50 ppm of the most (monochloro-) and the least (decachloro-) volatile PCBs would show concentrations of ~ 1 ppb and ~ 0.001 ppt respectively. 

The gas phase PCB concentration in the transformer headspace increases approximately 10-fold for a temperature increase of 20 °C. Local heating of the transformer is a way to increase vapour phase sensitivity of a detection technique. 

For a detection technique to have equal sensitivity to all PCB congeners for a given concentration in the liquid, a dynamic range of 6 orders of magnitude would be required. 

The conversion factor relating a given concentration of PCB in the vapour to that of the same PCB in the liquid phase varies from ~105 to ~1011 for the most and least volatile congeners respectively. In consequence, even very low concentration measured in the gas phase implies a concentration between a million and a trillion times larger in the liquid phase. 

The ability accurately to relate a PCB concentration measured in the gas phase to that in the liquid phase would require knowledge of the temperature of the transformer oil bath to within a few degrees C. 

The ability accurately to relate a PCB concentration measured in the gas phase to that in the liquid phase would in addition require knowledge of the enthalpy of vaporisation of each PCB congener, information which is not generally available. Experimental determination of vaporisation enthalpies would be a non-trivial task, especially for the higher homologues. 

• General Key Conclusions

The regulatory requirement of < 50 ppm of PCB applies to liquid phase samples, irrespective of the actual composition of PCB congeners. Using vapour phase measurements to infer PCB concentrations in the liquid requires an accurate way to relate vapour phase to liquid phase concentrations, which requires an accurate knowledge of temperature of the mixture and the enthalpies of vaporisation of PCB congeners. 

The variability in composition of PCB trade formulations and the possibility that the oil in a transformer may have been supplied from a variety of uncontrolled sources means that any detection technique must allow for the presence of the full range of all 209 congeners if the total 50 ppm in liquid phase analysis is required. 

The requirements for vapour phase detection and quantification are extremely demanding, especially for the heavier, more chlorinated congeners. The outcome of this study suggests the practical solution should focus on liquid analysis, which has implications for the practical implementation of the solution and particularly on the method of obtaining samples. 

Vapour phase solutions are realistic only when applied to volatile monochlorinated PCB congeners. This would allow the development of a screening solution but not a full quantification solution. 

Using liquid phase analysis, mid-infrared spectroscopic analysis will fulfil the requirements, based on the partial knowledge we have and the assumptions we made at the time this project.