No Silver Bullets for Power Station Maintenance
Although today we have comprehensive methods for technical testing and sophisticated models for calculating and forecasting component behaviour, many of the defects identified in thermal power stations and process plants in recent years have been the result of inadequate servicing and maintenance.
The problems in thermal power stations can be attributed to the fact that the ideal maintenance strategy is found only rarely, because stakeholders frequently focus on short-term cost reductions based on methodological approaches. When cost pressures increase, plant managers’ willingness to invest often decreases.
Operating costs may drop when modernisation measures and the acquisition of new equipment are postponed or fail to materialise at all; the operating risks, however, will increase. Stakeholders seeking to counteract this trend must choose the maintenance strategy best fitting the case at hand.
Failure recovery, preventative maintenance, condition-based maintenance, predictive maintenance, risk-based maintenance, reliability-centred maintenance or total productive maintenance – the strategies to choose from are virtually boundless.
However, ”maintenance” can never be equated with ”remediation” and ”repair”. And as the operating performance of every plant is unique, none of the possible approaches can be applied ”as is”, but must be adjusted to the case at hand.
A glance at the applicable standards and regulatory acts also fails to provide a solution. They are applied during manufacturing, before the plant is placed into service and in periodic inspections. But how can we succeed in achieving the right cost-benefit ratio when it comes to maintenance during operation?
Filtering out the Relevant Influencing Factors
Deterministic approaches in maintenance focus mainly on causal ”if-then relations” based on detailed knowledge of the relevant components. While these approaches have proved their worth from the perspective of safety technology, they have repeatedly resulted in excessive maintenance measures because of the safety factors that must be observed in design.
Probabilistic approaches by contrast add probabilities to these ”if-then relations”. Risks are quantified by their frequency of occurrence and the severity of their consequences, thus enabling experts to prioritise actions and reduce costs. The downside includes uncertainties concerning the quality of the input data, and the high costs and efforts involved in complex events and fault-tree analyses.
A maintenance strategy tailored to the needs of a plant requires systematic knowledge of operations, plant condition and possible defects. Interconnected process equipment, machinery and electrical, instrumentation and control systems in particular are becoming the focus of particular interest.
Maintenance professionals should perform overall analysis to determine the key influencing variables and damage mechanisms. To do so, they combine deterministic and probabilistic tools, non-destructive testing, fracture mechanics and the finite element method. Some of the cracks and defects identified may not need to be repaired provided reliable assessment of their further propagation and behaviour can be made.
Leadership skills are one of the key success factors in ensuring maintenance is used preventatively to maintain and optimise function. As this type of maintenance requires both ”creativity” and a solution-orientated approach, the engineers in charge must be able to motivate staff while at the same time assume responsibility for risks. They need to establish an organisation that is based on division of labour and characterised by outstanding clarity, consistency, employee appreciation and trust.
A growing number of companies want to use big data analytics in their predictive maintenance and are also investing in the resources needed for this. Of the companies already using this technology, no less than 95 percent say that they have already achieved concrete results. This is the conclusion of a follow-up study conducted by PwC and Mainnovation in recent months, among 268 companies in the Netherlands, Germany and Belgium.