Several recent works have proposed the simultaneous provision of multiple services in order to optimize the exploitation of distributed battery storage systems, as well as other distributed con- trollable resources, such as thermostatically controlled loads [26, 33, 34, 37, 60, 61, 92, 94–96, 113, 126, 135, 142–144].
4.1 Introduction 51
For instance, [92] proposes the usage of distributed storage to provide simultaneously primary fre- quency control and minimize PV curtailment. In [144], the authors formulate a stochastic dynamic programming model to schedule the operation of distributed energy storage devices simultaneously participating energy and regulation markets. Similarly, [143] proposes a rolling horizon optimization scheme and considers an energy storage device generating revenue from energy arbitrage, balanc- ing service, distribution system deferral and outage mitigation. Table 4.1 summarizes the existing literature, highlighting the service synergies analyzed by the works cited above.
Independently of the specific control scheme or objective, the shared conclusion of these works is that the provision of multiple services simultaneously results in an operation that is more effective than the single service provision either in terms of higher revenues or satisfaction of technical objectives.
Most of the aforementioned works aim at optimizing the revenue of the considered resource. For instance, [61] propose the joint participation to the wholesale electricity market (i.e. doing energy arbitrage) and the ancillary services market (e.g. providing primary reserve), while [126] aims at maximizing the revenue of a battery providing primary reserve and jointly performing peak- shaving in a tariff scheme including a charge for the peak demand. Such schemes are subject to market rules that may change as the penetration of storage devices increase (notably, invalidating the assumption that such resources can be considered as price-takers). On the other side, BESSs and other distributed resources are most often installed in distribution networks to provide local services that satisfy technical objectives. Among these, there are energy management (e.g. peak shaving [109] or load levelling [97]), voltage control for active distribution networks [28, 140, 141] or congestion management [55].
In this chapter, we aim at designing a scalable and general model-based method to formally char- acterize the amount of local (e.g. congestion management of a MV network, peak shaving for a residential building) and shared services (e.g. SFC to the upper grid layer) that a set of distributed resources, otherwise employed individually and each for a single service, can provide when controlled in a coordinated fashion.
Moreover, we observe that in most of the aforementioned literature the proposed control schemes focus on determining only the active power schedules and/or real-time injection of the controlled devices, although BESSs, as well as other distributed controllable resources, are interfaced to the grid through power electronics capable of injecting reactive power as well. The works [95] and [34] do schedule the reactive power injections; however, the objective in these works is limited to constraining the apparent power injection and voltage levels at the connection point of the storage system, rather than managing the voltage levels across the considered distribution network, which is in fact a more ambitious goal attainable through the control of distributed controllable resources (see e.g. [30, 129]). With respect to such literature, we propose a scheme to control both active and reactive power injections of such devices, so as to take advantage of their reactive power capabilities.
Author, year Considered services Methods, contributions Main outcomes Wu 2015, [143] Energy arbitrage, balanc-
ing services
Receding horizon opti- mization, includes outage mitigation in the revenue streams
Multiple revenue streams can be captured simultaneously
Kazemi 2017, [60, 61] Simultaneous offering in day-ahead energy, spin- ning reserve, and regula- tion markets
Robust optimization con- sidering uncertainties re- lated to market prices and energy deployment and, in [60], the battery lifetime
Participation in multiple mar- kets increases profits
Drury 2011, [33] Operating reserves in addi- tion to energy arbitrage
Based on a heuristic method, specific to com- pressed air energy storage (CAES)
Providing operating reserves simultaneously to arbitrage in- creases annual net CAES rev- enues
Cheng 2016, [26] Energy arbitrage and fre- quency regulation
Multi-scale dynamic pro- gramming
Focuses on assessing the algo- rithm performance rather than its economic benefits
Megel 2015, [92] Frequency regulation and load smoothing or mini- mization of PV curtailment
MPC scheduling, considers transformer overheating
Multitasking can almost dou- ble a storage system’s profits as compared with a single-service approaches
Shi 2017, [126] Peak shaving and fre- quency regulation
Joint optimization consid- ering battery degradation, operational constraints, and uncertainties in cus- tomer load and regulation signals
The saving from joint opti- mization is often larger than the sum of the optimal savings for the two individual applica- tions, with the gain being su- perlinear
Xi 2014, [144] Energy arbitrage and fre- quency regulation, backup service vs outages
Stochastic dynamic pro- gramming, market and sys- tem uncertainties
Batteries can achieve much larger economic benefits than previously thought if they jointly provide multiple ser- vices
Moreno 2015, [95]; Perez 2016, [113]
Distribution network con- gestion management, en- ergy price arbitrage and various reserve and fre- quency regulation services
Mixed integer linear pro- gramming, constraints the power flow at the electrical substation; [113] considers as well battery degradation
Distributed storage revenues associated with frequency con- trol services are significantly more profitable
Namor 2018, [96] Dispatch of a MV feeder and primary frequency reg- ulation
Day-ahead robust opti- mization accounting for uncertainties in the load and regulating signal forecasts; experimental validation of the control framework
The simultaneous provision of the two services allows to fully exploit the BESS energy capac- ity
Engels 2017, [37] Frequency control and maximization of self- consumption for a PV- storage installation
Chance-constrained robust optimization
optimally combining the two services increases value from batteries significantly
Trovato 2016, [135] Frequency services and en- ergy arbitrage
Controls aggregates of thermostatically controlled loads, linear optimization model
Clustering of appliances with similar capabilities can signif- icantly enhance the flexibility available to the system Dutrieux 2013, [34] Energy arbitrage, removal
of grid constraints on a re- active power management
Two stages approach defin- ing a priori a BESS operat- ing domain respecting grid constraints
Demonstrates the feasibility of providing simultaneously the mentioned services
Moreira 2016, [94] Energy arbitrage, peak de- mand shaving and various balancing services
Rather than focusing on control, it assesses the syn- ergies and conflicts among possibly concurrent ser- vices
Services interact differently de- pending on markets and sys- tem operating conditions
4.1 Introduction 53
In particular, this is exploited to guarantee the satisfaction of network constraints across the hosting distribution grid, as well as performing other local (peak shaving) and global (frequency regulation) services. We do so by integrating, in the proposed scheduling problem, a set of constraints based on the power flow (PF) equations of the grid which are linearized around the current operating condition (as detailed in Section 4.6) to preserve the tractability of the planning problem.