Optimizing Maintenance for Multi-Unit Industrial Gas Turbine Installations

High levels of availability and reliability are essential in many industries where production is subject to high costs due to downtime. Examples where gas turbines are used include the mechanical drive in natural gas pipelines and power generation on oil platforms, where it is common to use redundant gas turbines to mitigate the effects of service outage. In this paper, component-level maintenance of parallel multi-unit systems is considered, allowing production at a reduced level when some of the units are not operational. Units are themselves assumed to be composed out of components in a serial configuration; maintenance of one component implies shutdown of the unit. Parallel installations allow maintenance to be performed on one or a few gas turbines without taking down the entire installation. This allows maintenance to be optimized even further than in a serial system. However, the maintenance optimization process is made more complicated, since there now exist both positive and negative grouping effects. The positive grouping effects come from shared setup activities and costs, and the negative effects come from resource limitations, in this case the limited number of gas turbines which can be maintained at the same time. In the approach presented in this paper, each component has its individual preventive maintenance schedule, which is updated at inspections, changes in production and when indicated using remote condition monitoring. A minimal repair model for noncritical routine inspections and service tasks is assumed, which does not affect component state. In addition, previously developed procedures for estimating and measuring residual component lifetime for individual components during operation are used. The procedures are based on a Retirement For Cause (RFC) approach where components are not replaced until a potential failure has been detected. To maximize revenues for an operator, the available information is evaluated using software where scenario analysis and optimization is performed. To show the possible economic effects, gas turbine operation data is used together with maintenance and operator requirements as input for optimization of a production line consisting of a natural-gas compressor station having three SGT-600 gas turbines. Savings can be substantial compared to a traditional preventive maintenance plan.