A methodology for modelling synthetic daily sequences of hourly power demand for villages and small towns, based on stochastic processes

A model to generate daily sequences of hourly power demand values is described. Inputs are the daily values of energy consumption and load-factors (the ratio between mean and peak load). The whole model covers three independent first-order autoregressive models to generate data sequences of, respectively, daily energy consumption, daily load-factor and hourly power demand. The analysis of a power demand data set reveals that energy consumption and load-factor are independent variables; two independent time series of daily values of energy consumption and load-factor are built; their statistical and sequential properties can be described with first-order autoregressive models. Assuming a villlage’s consumption-structure as characterized by load curves with a peak load at night, the load-factor is taken as a shape-factor for these curves. A frequency distribution is built on daily load factors. The data sequence is, then, sorted into groups of daily curves, each one characterized by a load-factor-class. A daily average load curve is estimated for each class, along with its daily standard deviation curve. An analysis of each group of daily curves shows that the statistical and sequential properties of each one can be also described with first-order autoregressive models. For modelling purposes, the autocorrelation coefficient is determined for each load-factor class. Thus, energy and power relate to each other under different load-factors. Application examples are offered, for design purposes.