ESTIMATION OF EXTREME MASS FLOW-RATE IN DISTRICT-HEATING SYSTEMS

This paper focuses on how extreme mass flowrate conditions, i.e. T yea r events, in district heating systems can be estimated by extreme valu e analysis, namely the method based on annual maximum values and the p eaks over threshold (POT) method. Only a few years of mass flowrate da ta are available, which is far too short a time to be of any practical use for extreme analysis. However, 42 years of climate data are avail able, which are used to generate daily mass flow data using an autoreg ressive external (ARX) model. The simulated mass flow contains uncerta inty which must be considered in the analysis. This is not possible us ing a traditional analysis without some modifications. Furthermore, it is also of interest to consider the uncertainty in the climate. A boo tstrap technique is applied in order to answer these questions and a c omparison is made of the traditional analysis and the bootstrap approa ch. The results show that when using the bootstrap method the estimate d T year events and their uncertainties are always higher compared to traditional methods, whether based on the annual maximum or the POT me thod. The difference is, however, quite moderate in most cases. The PO T method yields somewhat higher estimates than the annual maximum meth od. It turns out that the predicted load in severe cold spells is not drastically higher compared to, for example, what has been experienced in Reykjavik over the last ten years when the climate has been quite normal. For instance, the 100 year event is estimated to be approximat ely 20 per cent higher than the average maximum mass flowrate during t he last ten years.