Applications of statistical process control in the management of unaccounted for gas

The phenomenon of Unaccounted-for-Gas (UAG) in natural gas transmission networks can be summarised as the failure to account for a percentage of network throughput on either side of the balancing equation - typically around 0.3% of throughput per annum, a cost directly absorbed by the transporter. This article presents a case study of historical UAG data from the UK's National Gas Transmission System, compromising two instances of significant under-reads amounting to 1617 GWh and 1141 GWh respectively. Statistical process control techniques are applied and the usefulness of such analysis in relation to drawing inferences about the system state is assessed. Practical difficulties relevant to the gas industry in particular are discussed. Whilst we focus on transmission in the UK, the techniques and analysis presented herein should be pertinent to most grids around the world. It is demonstrated that significant practical advantages can be gained by implementing statistical monitoring in the balancing process. The importance of using multiple such measures is highlighted, in addition to a thorough analysis of the UAG time series focusing on the presence of seasonality. Furthermore, we discuss issues regarding the quantification of UAG sources, and present practical numerical limits to online systematic error detection with gas time series data via a simulation study.