Search eceee proceedings

Forecasting and technoeconomic optimization of PV-battery systems for commercial buildings

Panel: 5. Smart and sustainable communities

This is a peer-reviewed paper.

Authors:
Jonathan Fagerström, IFE, Norway
Alexander Severinsen, Storekeeper AS, Norway
Josefine Selj, Institute of Energy Technology, Norway
Kari Aamodt Espegren, Institute of Energy Technology, Norway

Abstract

The cost structure of electricity tariffs is being modified from a two-part tariff, where the cost is divided between a fixed installation cost (EUR/installation) and a cost for consumed electricity (EUR/kWh), to a three-part electricity tariff where customers additionally pay demand charges for capacity usage (EUR/kW). To combat demand charges, commercial customers are looking into supplementing PV installations with batteries to more efficiently reduce peak electricity demand, i.e. peak shaving.

A crucial part of the complete energy system is also the energy management, where forecasting improves the efficiency and economics. The objective with this work was to investigate the profitability with peak shaving in Norway for a commercial building. A forecasting algorithm for load prediction was developed, and the economic value of forecasting was determined for a PV-battery system.

The load forecasting was developed using component-wise gradient boosting and the results from the model was verified against a renowned benchmarking load forecasting model. The economic value of forecasting was determined through simulations with Homer Energy Software that optimizes the net present cost of the systems. The results showed that battery storage was only economically beneficial when forecasting was deployed. Moreover, the cost savings came mainly from reduced demand charges, not from increased self-consumption of PV electricity. It was also discussed that the application of forecasting in an energy management system could be divided into three phases. One phase where forecasting is deployed to dimension energy system components in an early stage, one monthly forecast overview that identifies height and frequency of maximum peaks, and finally one high-resolution forecast that operates the battery on an hourly basis. Altogether, such an energy management system could additionally also be used by utility grid owners to schedule demand response actions for power quality control.

Downloads

Download this paper as pdf: 5-342-19_Fagerstrom.pdf

Download the display as pdf: 5-342-19_Fagerstroem_display.pdf