Anne McIntosh & David Hughes, EA Technology, UK
There is increasing pressure on all owners and operators of electrical systems to develop asset management strategies that are effective and cost efficient. Ensuring a corporately acceptable failure rate, prolonging the life of the assets and planning for appropriate replacement are of paramount importance. This is particularly true for transformers where reliable service is crucial, unexpected failures are extremely costly both in financial and operational terms and delivery for new transformer can be up to several years and represents a huge capital expenditure.
The condition based decision making (CBRM) process assists in implementing an asset management strategy that meets the corporate objectives. The process has been successfully applied to all major distribution and transmission asset groups including switchgear, cables, over head lines and transformers, but this article concentrates on the application to primary and grid oil filled transformers.
The CBRM Process
CBRM is a process developed by EA Technology in conjunction with several major electricity companies to assist with the tasks of defining, justifying and subsequently targeting spending to achieve defined levels of performance. The CBRM model provides a structured framework, which enables companies to fully utilise asset condition information, knowledge, and practical experience of their assets to understand the condition of their asset base.
The process also allows the future condition and performance to be estimated and the potential risk posed by failure of each asset to be quantified and compared. This information is then used to identify the probability of failure for individual assets and the resulting risk to the asset owners. The effect of different intervention strategies such as maintenance, refurbishment and replacement can be examined to identify the most cost-effective and efficient strategy to meet the corporate objectives.
The initial applications of CBRM used health indices and probability of failure (POF) calculations to define investment requirements to achieve a specified level of failure rate and is the topic of this paper. More recently risk models combining consequence of failure (COF) with POF have been populated to provide a more complete means of optimising investment but will not be discussed further in this article.
The four steps to success
To understand the mechanics of CBRM it is useful to define the process by a number of sequential steps.
Firstly, define asset condition by deriving ‘heath indices’ for individual assets and build health index profiles for asset groups. Secondly, link current condition to performance by calibrating the health index against relative probability of failure (POF). The health index profile is matched with current failure rate to determine health index/POF relationship.
Thirdly estimate future condition and performance by using knowledge of degradation processes to ‘age’ health indices, ageing rates dependent on initial health index and operating conditions. Future failure rates are calculated from aged health index profiles and previously defined health index/POF relationship.
Lastly evaluate potential interventions in terms of POF and failure rates by factoring in the effect of potential replacement, refurbishment or changes to maintenance regimes, modify future health index profiles and recalculate future failure rates.
Derivation of a health index
The first output and the foundation for the rest of the CBRM process is the definition of asset condition in the form of a health index (HI) derived for individual assets. These are built into a health index profile, a distribution of health indices, for a population.
The intention is to combine relevant information in order to provide a means of ranking equipment by proximity to end of life. The final number for each piece of equipment is normalised onto a scale of zero to ten; zero representing the best condition and ten the worst condition.
The detailed formulation of a health index for each population of equipment is specific to that population, based on the available condition information and the background history of the units in the population. However, the necessary information required to apply the process to transformers can be described in general terms.
A detailed understanding of degradation and failure processes, this enables the specific condition criteria that affect the performance of the transformer to be identified and collected. It is also important to understand the criticality of those processes on the current performance and also the end of life of an asset.
EA Technology’s expertise has been built up by completing failure investigations and CBRM projects with our clients, as well as from the weight of documented evidence of degradation and failure process for transformers that already exists.
In the case of transformers it is widely acknowledged that the paper condition is the primary factor in determining the technical end of life. Condition based information is required to identify the current performance of the asset. For transformers a large emphasis is placed on oil results, including oil quality, dissolved gas analysis (DGA) and furfuraldehyde content.
Measuring internal condition of transformers
Internal insulation degradation of transformers results from oxidation of the oil and paper components. The rate of degradation is very dependent upon the operating condition, in particular temperature and therefore load.
The rate of the oxidation processes increases exponentially with temperature and therefore a transformer that is heavily loaded for long periods of time will have a shorter life than a transformer that is subject to moderate loads.
Occasional overload situations in which the temperature of the transformer may be raised above the normal maximum temperature cause particularly rapid degradation and therefore significant shorten the transformer life.
The effects of internal oxidation can be sensitively and accurately monitored by oil tests. Oil test results provide information both on the degradation of the oil and the paper insulation.
Measurement of moisture, acidity and breakdown strength of the oil directly indicate the condition of the oil, and also give an indication of the overall internal condition.
Moisture, acidity and solid contamination are products of the oxidation of the oil and the paper. Furthermore, moisture and acidity accelerate the ageing of both the oil and the paper. Maintaining an acceptable oil quality will assist in reducing the rate of paper degradation and can be used as a means of prolonging the life of the transformer.
Dissolved gas analysis provides indication of abnormal electrical or thermal activity within a transformer. The energy available from overheating or electrical discharge breaks the oil down into the hydrocarbon gases, which can be detected by analysis.
The level and ratios of the different gases are a well-established means detecting and identifying a developing internal fault. By combining the information available from these different analyses, a very good understanding of the internal condition of the transformer can be obtained.
Other than internal insulation condition, possible end of life conditions can occur as a result of external degradation (corrosion of tanks, pipework, cooling systems) or degradation and failure of ancillary components (tap changers, bushings, cable boxes etc). External corrosion can be prevented by appropriate maintenance and is readily assessed by inspection. Relevant electrical tests can also be incorporated into the HI.
Operational information is included in the HI algorithm, for example the level of loading (e.g. past, present, expected future and overload occurrences), transformer rating and operational environment (e.g. the ambient temperature, humidity, any aggressive atmosphere).
Unit details including manufacturer and model, previous maintenance history and previous test details are recommended. Any defect or failure history of the specific population is also included in the derivation of the health index.
Linking Health Index to Probability of Failure
Standard curves are fitted to specific health index profiles in order to calculate the probability of failure (POF) for each population. A common framework relating health indices to POF is used allowing comparison of individual transformers in a population as well as comparison to other populations.
By design the relationship between the health index and the probability of failure is exponential. Individual transformers with a health index of <4.0 are in ‘good’ condition with a very low probability of failure, that would not be expected to deteriorate significantly in the short or medium term.
Values in the 4-7 range indicate ‘moderate condition’ with a low current probability of failure, but at risk of significant deterioration in the medium term.
Values >7 indicate ‘poor’ condition with a significantly increased probability of failure that will continue increase relatively fast in the short term.Table 1 shows a selection of results from a population of transformers. The process was initially applied in November 2004 and repeated in June 2007. The health index and POF are colour coded to enable easy identification of the asset condition, green indicates good condition, amber moderate condition and red poor condition (end of technical life).
The HI and POF calculations enable an immediate appreciation of the condition of the individual transformers. Tx 1 and 2 are at the end of life as the paper insulation is severely degraded and the POF is high, it is recommended that these transformers are given priority for replacement.
The results from Tx 3 demonstrate the degradation of the transformer over the test period, resulting in a slight increase in HI and POF over time.
Transformer Tx 4 shows an improvement in the HI and POF due to remedial treatment of the oil to remove particulate and improve the oil quality. As previously discussed this has the affect not only improving the condition of the transformer but also prolonging the life of the transformer by reducing the rate of degradation of the paper insulation.
Predicting Future Performance
The next stage of the CBRM process is to determine the future performance of the assets based on their current performance and rate of degradation.
Figure 1 shows the health index profile obtained for a population of grid and primary transformers. The population shows that the current profile is relatively good with the majority of transformers classified as being in good condition (HI of 0-3), with a low number as effectively being at end of life.
Figure 1 Current health index profile for a transformer population
The ageing algorithms can then be applied and the HI and POF recalculated for 5 years and/or 10 years. Figure 2 shows the calculated HI profile for the same population 10 years in the future.
Figure 2: Year 10 health index profile for a transformer population
In this case, different ageing algorithms are applied to transformers based on loading and environment, so those that are highly loaded will age quicker than low loaded transformers and those close to the coast will age faster than inland transformers.
The graph clearly demonstrates the change in the profile with more assets in a worse condition. In this case if there is no intervention the failure rate of the population will triple from 0.37 per cent to 1.29 per cent over a ten year period.
It is also possible to evaluate the different interventions available such as replacement, refurbishment, enhanced maintenance and monitor the effect on the condition and POF of the individual transformers and the entire group of transformers. The intention of these strategies is generally to maintain a specific, corporately agreed failure rate for the group. The asset management strategy can be simplistic or more complicated.
Using the previous transformer asset group the Year 10 profile indicates a threefold increase in rate of failure. Factoring in interventions reveals that to maintain the current (acceptable) failure rate would require 92 (out of 850) transformers to be replaced over the 10 years.
By considering the relevance and effect of refurbishment/remedial measures on an individual transformer basis and factoring these in, the same overall result can be obtained by a significantly cheaper option involving replacement of 45 transformers and refurbishment/remedial work on 70.
A feature of the health index, POF and ageing methodologies used in CBRM is that at all stages they are related back to physical condition and degradation and failure processes and assessed in light of the practical experience of the assets. Hence the process can be improved as further knowledge and experience is gained.
Application of the health index methodology for transformers offers a powerful tool to provide an efficient and effective asset management strategy. The process provides an effective means of linking the extensive engineering knowledge and condition information to corporate decision making, including maintenance and replacement.
Implementation has demonstrated it can deliver significant benefits with regard to effective asset management of transformers and other asset groups.