25. Meeting in Essen

Reliability Swaps - Tony Hudson

Summary
A reliability swap is an alternative way of managing short term plant reliability issues. The basic commodity swap mechanism was described and related to swapping unreliable generation for guaranteed replacement on failure at a fixed price. This protects the plant operator from the volatility of the energy market on failure and enables him to fully commit to supply contracts, the reliability risk company is able to sell on the unreliable generation to cover costs of replacement generation.

Access considerations for offshore wind turbine maintenance - Paul Thame

Summary
The siting of wind turbines offshore introduces the risk of delayed access times due to bad weather. Normal access is achieved via a service boat from which the maintenance staff reach a permanent ladder on the outside of the tower, then climbing up to the access deck. This is only practical when wave height is below about 1.5 m.

A simplistic production loss model of the access delay looks at the number of machine failures expected in a year, the chance that the waves exceed 1.5 m at any given time and the average duration of a high sea. However, this does not give a realistic estimate of the delay loss because:
High wind, high seas and failures are more likely in the winter. Thus, more failures will happen during the peak generating season when the machine is most difficult to reach.
To repair the machine, you actually need an adequate weather window once the seas have dropped in order to complete the work.

Developments to help reduce delay losses include reducing the failure rate of the machines (capital costs favour the largest machines which are the least proven and thus most prone to failures), a winch and cradle for transfer of men and equipment and a built-in crane in the nacelle to reduce the requirement for floating cranes. More expensive access options include a helicopter pad or a jack-up service vessel if the sea bed permits.

MAINTENANCE INDICATORS - Antoine Despujols
EDF/R&D Division, Industrial Risk Management Department
antoine.despujols@edf.fr

The definition of indicators is not separable from an improvement process. « What could be measured could be done » and a way to get better plant performances is to improve maintenance. More precisely, maintenance indicators meet different objectives which are :
to identify technical or organisational maintenance problems,
to report and justify value of maintenance to upper management
to motivate and to give goals to craftmen’s teams
to check progress in meeting goals.
Depending on these objectives, indicators may be defined at different levels (plant performances, maintenance actions, component reliability and maintainability). Their measures must be compared to references coming from previous evaluations or from other plants. Benchmarking is a good way to get references in order to diagnose maintenance activities but it should be used carefully. Many plant characteristics must be taken into account to allow comparisons such as nominal power, net capacity factor, fuel, manpower costs, etc.. Using different sources of information we found a list of about 170 maintenance indicators. The formula, usefulness, level, of each have been set and a selection of about 15 has been made for benchmarking. This limited list proposes synthetic indicators to measure impact of maintenance on availability, costs and individual safety, as well as to evaluate maintenance strategy and organisation ; maintenance effectiveness ; maintenance resources and activity. In order to rank plants according to maintenance practices we are working to define a global maintenance indicator which could take into account unavailability and direct costs due to equipment failures and preventive maintenance tasks.

SUMMARY SPREADSHEET MODELS FOR LOSS OF LOAD SIMULATIONS – Henk Wels, NRG
The making of a reliability block diagram to calculate reliability & availability of ONE power plant is straightforward. It includes failure rates and repair times per component and the results are expected systems states given by frequency, average duration, fraction of time. By adding these system states, one arrives for a production PARK at how frequent the load cannot be met, etc.

There are a number of problem areas when using this analytical method. For instance, the number of system states becomes very large, and, most of all, load is a function of time. Not all failures & failure duration is important. Failures can be postponed to the weekend, starting of reserve plant takes time. The solution: simulation. It was thought to use the SPAR program. However, the development took to long (> 2 weeks).

Therefore a special spreadsheet was written using Monte Carlo analysis. It simulates failure of power plants and counts load failures, taking spinning reserve into account. However, it could not take start time of reserve plant in to account. In combination with @Risk it will tell you:

which uncertain inputs will be dominating the overall uncertainty a good estimate of mean value including the effect of very assymmetrical distributions (do not use a normal distribution for unavailability!) all data can be stored to file correlation of variables is possible (for instance costs of unavailability AND costs of operation are coupled by fuel price)

While working for a Dutch power production company, it was possible to check the results using confidential information on actual bidding behaviour, load uncertainty, etc. As a by-product, the spreadsheet gave (modified) operating hrs and starts as a result of plant failures in the production park, optimum spinning reserve and costs of failures when solved with own plant versus market.

Since a spreadsheet uses a lot of calculation time, a dedicated program since has been realized taking into account spinning reserve as well as start-up time of reserve units.

Remark: All items are strictly to be considered from a technical point of view only with regard to the support of the strategic work of EURELECTRIC.