Estimating system indices for short interruptions from data obtained by a limited number of monitors

•Probability density functions are presented for the number of short interruptions experienced by individual customers.•It is shown that the average value (MAIFI) alone does not gives sufficient information to describe the quality of supply.•Estimations are presented of the error in system indices that is made when monitoring for short interruptions at a limited number of locations.•Recommendations are given for the number of monitors needed to obtain a rough and an accurate estimation of system average and high percentiles.•Those recommendations are compared with recommendations given by the European Energy Regulators. This paper presents an estimation of the number of monitoring locations needed to get an accurate estimation of system indices for short interruptions. The study uses data from 11 Swedish distribution network operators, where the number of short interruptions is available for every single costumer. The error range in the estimated system parameters is obtained as a function of the number of monitor locations by obtaining mean and spread of the estimations from 10000 randomly-selected sets of monitor locations. To obtain a rough estimation of the system average or 95-percentile (error margin less than 50%), about 200 monitors are needed. For system 99-percentile, about 400 monitors are needed. For a reasonably accurate estimation, with an error margin less than 10%, a few thousand monitors are needed for both system average and system 95-percentile. Several thousand monitors are needed for the system 99-percentile. The paper also presents some system statistics for the 11 distribution network operators and compares the requirements with recommendations given by the Council of European Energy Regulators.