Feature-conserving gradual anonymization of load profiles and the impact on battery storage systems
(2023)
Electric load profiles are highly relevant for battery storage research and industry as they determine system design and operation strategies. However, data obtained from electrical load measurements often cannot be shared or published due to privacy concerns. This paper presents a methodology to gradually anonymize load profiles while conforming to various degrees of anonymity. It segregates the original load profile into base and peak sequences and extracts features from each of the sequences. With the help of the features, a synthetic, anonymized load profile is created. Different levels of anonymization can be selected, which transform the original profile to the desired extent. A random permutation of the peak sequences or base sequences is used to achieve this transformation. Exemplary profiles from a household and an electric vehicle charging station are used to demonstrate the functionality of the anonymization. The indicators of the anonymized load profiles are compared with the original ones in both time and frequency domains, and the effects of load profile anonymization on the operation of battery storage systems in two scenarios are analyzed. While the anonymized load profiles retain the time-invariant indicators from the original profile, the permutation causes a loss of regularity in the load profiles. As a result, relevant indicators of battery storage systems subjected to these anonymized profiles deviate to a greater extent in time-dependent applications such as self-consumption increase. This is reflected in the overestimation of equivalent full cycles by up to 6% and underestimation of self-sufficiency by up to 9 percentage points. In time-independent applications such as peak shaving, however, the indicators can be well reproduced with deviations of up to 3% despite the lost regularity. In order to make the anonymization methodology usable for everyone, we present the open-source tool LoadPAT, in which users can anonymize their load profiles and choose their desired level of anonymization. This work is intended to further encourage the dissemination of open-source data.
Electric vehicles (EVs) can participate in various markets through a vehicle-to-grid (V2G) interface. Aggregators can combine the individual contributions of EVs to offer them, for example, on the frequency containment reserve (FCR) market or to use them for arbitrage trading. A simple approach is combining EVs in random fashion until the pool is able to reach the threshold for a service of choice. Alternatively, aggregators can compose their pools in smart fashion and include only EVs that contribute significantly to the pool's performance. In a previous publication, we have shown that optimizing the aggregated pools of commercial vehicles for the provision of FCR or arbitrage trading can increase revenues by up to 7-fold. In this work, we analyze the optimally composed pools and show that large vehicle batteries in the order of 80 kWh are particularly useful for arbitrage trading, while FCR provision is also possible with medium-sized EV batteries in the range of 30 kWh due to the small cycle depths. The inclusion of EVs with very small vehicle batteries around 20 kWh in aggregated pools is neither economically optimal for arbitrage trading nor for FCR provision. An analysis of the economic sectors of the commercial EVs selected for the optimal EV pools shows that some economic sectors are more suitable for V2G than others: In particular EVs of the sector “human health and social work activities” are unsuitable for V2G provision due to regular and long travel times during the day. In contrast, EVs from the "manufacturing" sector are particularly well represented in all applications and the "transportation and storage" sector in the arbitrage application. In addition to these analyses of the optimized pools, we reveal that a reduction in the required minimum power and increments would make the FCR market even more attractive to EV pools by increasing revenues by 50% to 66%. It would also better exploit the potential of EVs, as increments could be better utilized than they are in the current 1 MW minimum power requirement in central Europe.
The market ramp-up of electromobility is shifting vehicle-to-grid (V2G) issues into the focus of research and industry. Electric vehicles (EVs) have the potential to support the trend towards renewable energies in their role as storage units during idle times. To participate in balancing power and energy markets, EVs are pooled via aggregators. Instead of a random composition, aggregators can smartly compose their pools and add only those vehicles that actually contribute to the pool’s performance, gaining advantages over competitors. The optimization methods presented in this paper form optimized pool combinations based on the power and energy capability profiles of commercial EVs. Genetic algorithms are used to determine the revenues of the possible pools per participating EV. The use cases analyzed are the provision of balancing power on the frequency containment reserve (FCR) market of Central Europe and energy arbitrage trading on the European power exchange intraday continuous and day-ahead auction spot markets. The results show that through smart pool composition, an aggregator can increase revenue per vehicle by up to seven-fold across the markets compared to randomly assembled pools. In the Central European market, for example, the potential V2G revenues on the FCR market (380 €) exceeded those of arbitrage trading (28 € − 203 €) in 2020. In a simulation, we show the increased degradation of the vehicle battery in V2G operation compared to sole use for mobility with a smart charging strategy. However, the additional revenue can make V2G financially worthwhile, depending on costs for measuring equipment, bidirectional charging stations, and aggregator costs.