Mr Joachim Geske, Imperial College, London, United Kingdom
Prof Richard Green, Imperial College Business School, United Kingdom
Overview: Storage has the technical potential to increase efficiency of electrical systems significantly – especially in the context of integrating intermittent renewable technologies. This is achieved by shifting energy from periods of low demand to periods of high demand. Thus, the utilization of medium load power plants is increased and the utilization of peak load power plants is reduced. The full extent of efficiency gain is achieved if generation capacity is adapted to the “equilibrated” load situation – with a higher base load and lower peak load share. In this case, the installed fossil generation capacity falls below peak load level. Since the amount of energy stored is generally limited, there is a risk of outages in cases of prolonged demand peaks. This problem does not occur in perfect foresight based analyses that are still the paradigm of electrical system analysis. The subject of this analysis is to show how storage is operated optimally under renewable and load uncertainty in the system context.
Methodology: We estimate a homogeneous Markov Chain representation of the load in GB on an hourly basis and design a very simple dynamic stochastic electricity system model with fossil generation technologies and storage (a Markov Decision Process). This model is solved for a stationary state using numerical methods (Linear Optimization) and the optimal storage strategy is presented. This optimal strategy is then compared to optimal strategies derived under perfect foresight of explicit drawings of the stochastic load process. Thus the perfect foresight “error” can be quantified.
Results: It is shown that under uncertainty at high demand an increasing share of the storage is “frozen” in its charged state to avoid lost load (outages). Therefore the “buffer share” of the storage is not used for arbitrage any more. Furthermore, this buffer state of charge is established, if necessary, even in periods of high demand, so that the storage operation stresses the system.
Conclusions: To implement the full efficiency potential of storage, generating capacities have to be reduced. Peak load can then exceed installed capacity. If this is the case, in the optimum under load uncertainty a storage buffer is created and maintained. This strategy is not required in optimal storage operation under perfect foresight assumption. Therefore efficiency gains caused by the usage of storage are overestimated in analyses based on perfect foresight. We will quantify the extent of this overestimation.
When using storage to prevent outages it has to be decided whether it is operated “inefficiently” with respect to “full” capacity adjustments, or “efficiently” and peak load capacity is not decommissioned “one for one”. This analysis refines the assessment of the economic potential of electricity storage, thus contributing to more effective planning of energy systems of the future, where outages are avoided.GeskeGreen-Optimal-Storage-Management-Under-Uncertainty1.pdf 893.03 KBGeskeGreen-Optimal-Storage-Management-Under-Uncertainty.pdf 687.17 KB