Linear programming models used for optimisation of various energy systems have received increased interest during the last ten years. One reason for this is the use of personal computers. Models with thousands of variables and constraints can now be rapidly optimised. If integers are introduced, which are necessary when increments or steps in cost functions are part of the model, the computing power is of even higher interest. However, many scientific authors do not discuss in detail how the model is designed and what basic data lie behind this design. This paper presents an attempt to study municipal thermal and electricity loads, and further how to divide data for one year into useful segments for linear and mixed integer programming purposes