Dealing with uncertainty in the asset replacement decision

Abstract

To prevent dangerous situations from gas leaks, operators of the distribution grids in the Netherlands are required to inspect all pipelines for leakages once every 5 years. Leaks are generally fixed when encountered. However, given that the distribution grid has been in use for many years, it may be wiser to replace leaking sections of the grid. The right choice depends on the costs of replacement versus the expected costs (including the monetized risks) of future repairs. The number of future repairs to expect is uncertain, especially given that the asset is ageing and the failure rate may be rising. In the paper, decision making on a representative case is explored. In the basic approach (using past performance for future failures) repairing is marginally better than replacing. However, assuming an increasing failure rate because of ageing will tip the balance in favour of replacement at some (uncertain) moment in future. The decision problem is thus transformed from a choice between alternatives into timing one specific alternative. To find the optimal moment given the uncertain development of the failure rate both sensitivity analysis and real options analysis are applied. Different assumptions and different decision methods result in different optima. However, from a total cost of ownership perspective the differences are relatively small and they hardly justify the analysis effort: any choice will be acceptable. The paper thus reaches the (surprising) conclusion that it may be better to flip a coin than to try to find the best solution.