Monte Carlo simulation of load profiles for low-voltage electricity distribution grid asset planning

J. Slootweg,W. Du,Pwg Bots

Abstract:This paper presents a Monte Carlo simulation method for generating stochastic load profiles for models of low voltage (LV) electricity grids to support middleand long-term strategic asset planning processes. Models that calculate aggregated loads in a deterministic way using a coincidence coefficient (simultaneity factor) do not give insight in the probability of an overload and eventual asset failure. Analysis of minute-to-minute load data obtained with Monte Carlo simulation, based on the characteristics and behaviour patterns of different household types, can provide more accurate probabilities of peak loads, especially for subordinate grids where individual consumption behaviours have relatively high impacts. Calculating this indicator for different future scenarios can help improve LV distribution grid capacity planning down to the component level. INTRODUCTION: WHY MONTE-CARLO? In electricity grid asset management, the minimum required grid capacity is typically determined on the basis of an estimated peak load. This peak load is usually calculated with the Rusck algorithm [1], i.e., by summing over all connected consumers the maximum consumption per individual consumer, multiplied by a so-called “simultaneity factor” [1, 2] or “maximum coincidence coefficient” [3]. This simultaneity factor depends on the number of connected consumers n, and may be computed by a function. Gwisdorf et al., for example, used f (n) =  + (1–) / n  with parameter values  = 0.028 and  = 0.75 for German distribution nets [3]. When aggregated peak loads are estimated using simultaneity factors (e.g., f = 0.89 for n = 5, f = 0.80 for n = 20, f = 0.75 for n = 50, and so on), the result is a single number. This provides no insight in the probability of the occurrence of a peak load that is higher than the capacity of the affected assets. If such a probability were known, this might help avoid investments in overcapacity [4]. The method and software presented in this paper aims approximate the probability distribution of peak loads, and especially the probability of a peak load at some point in the low-voltage grid that exceeds the affected assets’ capacity (cf. Figure 1). An important limiting factor for stochastic approaches that represent consumer loads as probability distributions (see e.g., [5, 6]) is that the detailed load data at the household level that is needed to empirically determine this distribution is usually not available, and costly to obtain. To overcome this problem, load profiles of individual consumers are generated with a simulation model based on the characteristics and behaviour patterns of households (see also [6, 7]).

Mathematics,Environmental Science,Engineering