Set of indicators for dependability evaluation of gas compression units

The paper is dedicated to the improvement of the evaluation methods of one of the most important operating characteristics of gas compression units (GCUs), i.e. dependability, under the conditions of decreasing pipeline utilization rate. Currently, the dependability of units is characterized by a set of parameters based on the identification of the time spent by a unit in certain operational state. The paper presents the primary findings regarding the dependability coefficients of GPA-Ts-18 units, 41 of which are operated in multi-yard compressor stations (CSs) of one of Gazprom’s subsidiaries. The dependability indicators (technical state coefficient, availability coefficient, operational availability coefficient) identified as part of the research are given as well. GCUs were classified into groups depending on the coefficient values. The feasibility of using integral indicators in the analysis of GCU groups’ dependability was examined. It was proposed to use confidence intervals for identification of the integral level of dependability of the operated GCU stock and the ways of maintaining the operability of units under the conditions of decreasing main gas pipeline utilization rate. The Gini index was suggested for the purpose of generalized estimation of GCU groups’ dependability. It is shown that the advantage of the Gini coefficient is that is allows taking into account the ranks of the analyzed features in groups. The graphic interpretation of the findings was executed with a Lorenz curve. The paper implements the sigma rule that characterizes the probability of the actual coefficient value being within the confidence interval, i.e. prediction limits (upper and lower), within which the actual values will fall with a given probability. The confidence intervals were identified by the type of coefficients distribution and a standard deviation, ć. A histogram of an interval range of technical utilization coefficient distribution is given as an example. Testing of the hypothesis of the distribution type at confidence level 0.95 showed that the distribution of coefficients is normal. Using the moment method, the mathematical expectation and mean square deviation for the distribution of the values of each type of dependability indicators were established. Using the sigma rule, all extreme outliers among the GCUs in terms of the level of factor attribute were excluded from the body of input data. All units whose factor attribute value does not fall in the interval were excluded. According to the three sigma rule, 3 and 2 GCUs did not fall in the confidence interval (µ±3σ) in terms of the utilization factor and availability factor respectively. The performed analysis of causes of low availability coefficients of the above GCUs showed that the systems had been long in maintenance. The paper sets forth summary data on the maximum allowable value of the Gini index of dependability coefficients (C TU , C A , C OA ) depending on the sample size (the complete sample of 41 units and samples with the interval of 1, 2, 3 sigma). In case of higher values of Gini index it is recommended to adopt measures to individual units in order to improve the dependability of the operated GCU stock.