Conservation Voltage Reduction i Sverige - en simuleringsstudie

Sammanfattning: Conservation Voltage Reduction, or short just CVR, is a technique in which the reference voltage at a substation is lowered a few percent in order to conserve energy and decrease peak demand in an electrical distribution grid. Thus, CVR can be used when the energy or power capacity can't be satisfied by the feeding grid. Energy capacity has been a recurring topic in Sweden for a long time with the ongoing decommission of nuclear power plants, while increasing power demand and lacking power capacity is an urgent matter. This thesis has through modelling and simulation studied the impact of this technique at a typical Swedish distribution grid. To measure the effectiveness of CVR, CVR Factor has in previous studies been defined as the ratio between the reduction of energy consumption or peak demand in percent and the reduction of voltage in percent. Typical values for CVR Factors in north american studies and implementations range from 0,3 and 0,8. While energy conservation and peak shaving are the main goals of CVR, one also needs to consider the voltage level requirements on the distribution grid. These vary all over the world and from one grid to the next, but typical limits are +/- 10 % of the nominal voltage. OpenDSS, in conjunction with Matlab, has been used in this thesis to model and simulate two systems. One system was based on the IEEE 13-bus test system, and the other was based on a real distribution grid in southern Sweden. Load modelling is a crucial part in estimating credible CVR factors through simulation. What model, and which corresponding be modelled and simulated is hence of utmost importance. The exponential load model was chosen to model the loads of the Swedish system. The parameters were retrieved from a PhD thesis originating from 2005, where the most critical parameter for CVR-studies is the steady state active load-voltage dependence, α_s, or just α for static load models. That thesis examined a method for estimating parameters for the exponential load model, and used long-term measurements from the same Swedish grid that this thesis has studied. The simulations for the Swedish grid model showed that CVR Factors between 0,36 and 0,64 could be achieved during the winter months, and between 0,61 and 0,77 during the summer months without violating the voltage level requirements. This corresponded to up to 0,5 MW of peak shaving out of a 25,7 MW peak demand and 4 MWh energy reduction of a total of 510 MWh energy consumption. However, due to higher load during the winter months the margin for potential voltage decrease was more narrow during that period. Even though both peak demand and consumed energy were reduced with CVR implementation, higher losses due to increased currents in the grid was observed in most simulation cases. Only when the total aggregated load of a system corresponded to α>1 (constant current load) the losses also decreased when CVR was implemented. This fact could however be used to only activate CVR in a system when the load circumstances are beneficial, provided one could predict or estimate load model parameters in real time.