Improving VIGV predictive monitoring by developing a failure mode virtual sensor

Aero derivative gas turbine which drive a gas compressor is a very critical equipment in oil and gas topside facilities. This study will focus on the unwanted shutdown of aero derivative gas turbine causes by variable inlet guide vane (VIGV) issue that affects the production reliability. In modern days, the health of gas turbine is monitored by complex predictive instrumentation and control system, which normally focus direct on individual physical sensor operational alarm limit such as “Variable inlet guide vane deviation above hi alarm limit” and not on failure mode such as “Variable inlet guide vane system drifted outside normal operating envelope”. Hence, understanding the relationship between independent variables or predictor versus dependent or response variables and having a failure mode virtual sensor is the key factor to improve the variable inlet guide vane predictive monitoring by developing a failure mode virtual sensor. Based on the study, the failure mode virtual sensor is capable to predict the expected operating condition of the variable inlet guide vane at a steady state condition. This is measured by stability and normality analysis and error simulation using offline data. From the results analysis, which supported by hypothesis testing, it can be determined that failure mode virtual sensor could be used to predict the potential failure of variable inlet guide vane at the steady state, by comparing the deviation between the expected value from the virtual sensor model versus the actual value from the control panel. The detected deviation, which provides an early detection or a warning mechanism, will help to improve the maintenance planning and operability actions ahead. This early warning will lead to topside production reliability improvement due to minimization of unwanted production platform shutdown and its times to recovery.