In the context of modern power systems, the integration of renewable energy sources has significantly increased the demand for flexible and fast-responding frequency regulation resources. My research focuses on the application of battery energy storage systems to enhance both the dynamic performance and economic efficiency of power system frequency regulation. Unlike conventional studies that treat energy storage as a single aggregated unit, I address the critical issue of coordinating multiple distributed battery energy storage systems to work synergistically with traditional generation units. I propose an improved control framework that combines a two-layer distributed strategy with algorithm optimization, aiming to achieve superior frequency response speed, reduce battery degradation, and minimize overall operational costs. The work is published in Instrumentation Users (Vol. 31, No. 7, 2024) and presents a comprehensive methodology validated through MATLAB/Simulink simulations.
Introduction
The increasing penetration of intermittent renewable energy sources, such as wind and solar, has introduced significant challenges to power system stability, particularly in terms of frequency regulation. Traditional generation units, such as thermal and hydro plants, have limited ramp rates and response times, which are often insufficient to counteract rapid load fluctuations. Battery energy storage systems have emerged as a promising solution due to their fast response, high efficiency, and scalability. They can provide primary and secondary frequency regulation services, enhancing grid resilience. However, the economic viability of battery energy storage systems depends on their operational strategy, which must balance performance with battery lifetime degradation. My review of existing literature reveals that most studies consider a single large-scale storage system or treat distributed units as a homogeneous group, neglecting the need for coordinated control among multiple distributed storage devices. This gap motivates my work to develop a control strategy that optimizes the coordination of multiple distributed battery energy storage systems for grid frequency regulation, while simultaneously improving economic outcomes.
Load Frequency Control Model with Battery Energy Storage Systems
Control Principle
The primary objective of load frequency control (LFC) is to maintain the system frequency within acceptable limits by balancing generation and load. In a conventional LFC system, automatic generation control (AGC) signals are sent to thermal or hydro units to adjust their output. However, these units have inherent delays due to mechanical inertia and ramp constraints. By integrating battery energy storage systems, we can provide an instantaneous power injection or absorption that mitigates frequency deviations much faster. The functional principle is illustrated in the following: when a load disturbance occurs, the frequency deviation (Δf) is sensed, and the control center dispatches a regulation signal to the energy storage system. The battery responds almost instantly with power output PB, which compensates for the imbalance and reduces the settling time. The overall effect is a faster return to the steady-state frequency equilibrium point.
Mathematical Model of Battery Energy Storage Systems
To represent the dynamic characteristics of battery energy storage systems in frequency regulation, I adopt a first-order inertial transfer function that captures the time delay of the power electronics and battery response:
$$
G_{B}(s) = \frac{1}{1 + s T_{B}}
$$
where \(T_{B}\) is the time constant of the battery energy storage system, typically in the range of 0.1 to 0.5 seconds for modern systems. This simple model adequately represents the output dynamics for LFC studies. The state of charge (SOC) of the battery is also a critical parameter, as it determines the available energy and the constraint on continuous operation. I define the SOC dynamics as:
$$
SOC(t) = SOC_{0} – \frac{1}{E_{rated}} \int_{0}^{t} \eta P_{B}(\tau) d\tau
$$
where \(E_{rated}\) is the rated energy capacity, \(\eta\) is the round-trip efficiency (typically 0.85–0.95), and positive \(P_{B}\) indicates discharge. The SOC must be maintained within safe limits (e.g., 20%–90%) to prevent accelerated aging and ensure availability for future regulation.
Two-Layer Distributed Control Structure
My proposed control architecture consists of an upper layer (regional dispatch level) and a lower layer (distributed energy storage level). The upper layer allocates the total regulation requirement between the aggregated battery fleet and conventional units, based on their techno-economic characteristics. The lower layer uses graph theory and multi-agent consensus algorithms to distribute the battery regulation signal among individual battery energy storage systems in a way that minimizes operational costs and SOC divergence. This structure ensures efficient coordination and scalability. The two-layer control diagram can be represented conceptually as follows:
- Upper layer: ACE signal processing, regulation zone classification, initial power split between batteries and generators.
- Lower layer: Consensus-based power sharing among distributed batteries considering SOC, capacity, and degradation costs.
Regional Dispatch Layer Control Strategy
ACE Signal Processing and Regulation Zones
The area control error (ACE) is the key input for AGC. It is defined as:
$$
ACE = \Delta P_{tie} + B \Delta f
$$
where \(\Delta P_{tie}\) is the net tie-line power deviation and \(B\) is the frequency bias factor. Based on the magnitude of ACE, I classify the regulation into three zones: dead zone (negligible ACE), normal regulation zone, and emergency regulation zone. Within each zone, different control objectives are applied:
- Dead Zone: No action is taken unless necessary for SOC recovery of batteries. The control center may allow self-restoration of battery SOC without causing ACE to exit the dead zone.
- Normal Regulation Zone: The fast response of battery energy storage systems is used to quickly eliminate ACE, while conventional units adjust slowly. Simultaneously, SOC of batteries can be gently restored without worsening frequency deviation.
- Emergency Regulation Zone: All available regulation resources, including both batteries and fast-responding generators, are deployed to reduce ACE rapidly and bring the system back to the normal zone.
Optimal Power Allocation Between Batteries and Generators
To minimize the cost of frequency regulation, I formulate an optimization problem that considers the operational costs of battery energy storage systems (including degradation), startup costs of generators, and the ramp rate constraints. The objective function for each control interval is:
$$
\min \sum_{i} C_{B,i}(P_{B,i}) + \sum_{j} C_{G,j}(P_{G,j})
$$
subject to:
- Power balance: \(\sum_{i} P_{B,i} + \sum_{j} P_{G,j} = P_{reg}\)
- Battery capacity limits: \(P_{B,i}^{min} \le P_{B,i} \le P_{B,i}^{max}\)
- Generator ramp limits: \(r_{G,j}^{min} \le \Delta P_{G,j} \le r_{G,j}^{max}\)
- Battery SOC constraints: \(SOC_{i}^{min} \le SOC_{i} \le SOC_{i}^{max}\)
The cost function for battery energy storage includes a term proportional to power output (energy cost) and a term representing battery aging. I use a polynomial approximation of the degradation cost based on cycle depth:
$$
C_{B,i}(P_{B,i}) = a_i P_{B,i}^2 + b_i |P_{B,i}| + c_i
$$
where the coefficients are derived from battery lifetime models. The solution of this optimization yields the reference power for the aggregated battery fleet, \(P_{B,ref}\), and for conventional units, \(P_{G,ref}\).
Allocation to Individual Battery Energy Storage Systems
Once the total battery reference power is determined at the regional level, the lower layer distributes it among individual distributed battery energy storage systems using a consensus-based method. The goal is to share the regulation burden proportionally to the remaining capacity and SOC, while minimizing the divergence in SOC among units. The consensus algorithm updates the power setpoint for each battery \(i\) as:
$$
\Delta P_{B,i}^{(k+1)} = \Delta P_{B,i}^{(k)} + \sum_{j \in N_i} w_{ij} \left( \lambda_j^{(k)} – \lambda_i^{(k)} \right)
$$
where \(\lambda_i\) is a local marginal cost that includes both energy cost and SOC penalty. The weights \(w_{ij}\) are chosen according to the Laplacian matrix of the communication graph. The algorithm converges when all units have the same marginal cost, ensuring economically optimal dispatch. The SOC of each battery is also regulated to avoid deep cycling.
Simulation and Results
Parameter Settings
I implemented the proposed control strategy in MATLAB/Simulink using a regional power system model with multiple distributed battery energy storage systems. The system base was 4000 MVA, with a single thermal unit rated at 1000 MW. The battery energy storage systems had a total capacity of 200 MWh distributed among 10 units with varying ratings (from 5 MW to 30 MW). The nominal parameters are listed in Table 1.
| Parameter | Value |
|---|---|
| System frequency (f) | 50 Hz |
| Speed regulation coefficient (R) | 0.05 |
| Frequency bias factor (B) | 21 |
| Governor time constant (tg) | 0.08 s |
| Turbine time constant (tch) | 0.3 s |
| Generator inertia constant (M) | 10 s |
| Load damping coefficient (D) | 1 |
| Battery response time (TB) | 0.2 s |
| Optimal SOC setpoint (SOC0) | 50% |
| Weight coefficient τ1 | 0.02 |
| Weight coefficient τ2 | 1.55 |
| AGC sampling interval | 4 s |
| Convergence tolerance ε | 0.01 |
Simulation Scenario
At t = 5 seconds, a step load disturbance of 0.2 p.u. (800 MW) was applied to the system. I compared three strategies: (1) no battery participation (only thermal unit), (2) battery participation with fixed proportional power sharing among batteries, and (3) my proposed two-layer distributed control strategy. The frequency deviation and the outputs of generators and batteries were recorded for 60 seconds.
Frequency Deviation Response
The results show that the proposed strategy significantly reduces the maximum frequency deviation and the settling time compared to the fixed proportional sharing and the no-battery case. Table 2 summarizes the key performance metrics.
| Strategy | Maximum frequency deviation (Hz) | Time to first zero crossing (s) | Settling time (to ±0.005 Hz) |
|---|---|---|---|
| No battery | 0.285 | 6.2 | 22.4 |
| Fixed proportional sharing | 0.192 | 4.8 | 16.1 |
| Proposed two-layer control | 0.142 | 3.5 | 11.3 |
Furthermore, the rate of change of frequency (RoCoF) was improved by 35% compared to the fixed sharing method, demonstrating faster initial response due to better coordination.
State of Charge and Battery Life
Figure 4 (inserted earlier) shows the SOC trajectories of three representative battery units under the fixed proportional strategy and the proposed strategy. Under fixed sharing, the SOC of the smallest battery (5 MW) dropped below 20% within 30 seconds, forcing it to stop regulation and potentially causing a secondary disturbance. In contrast, my proposed strategy maintained all SOC values between 40% and 60%, indicating balanced usage and reduced depth of cycling. The resulting cycle life estimation (using a standard degradation model) indicates a 28% reduction in average capacity fade per cycle compared to fixed sharing. This translates into lower replacement costs and improved economic benefits for battery energy storage systems.
Economic Analysis
I also evaluated the operational cost over a 24-hour simulated period with random load fluctuations. The total cost included energy cost, battery degradation cost, and generator fuel cost. The proposed strategy achieved a 12.3% reduction in total cost compared to fixed proportional sharing, and a 41.7% reduction compared to the no-battery scenario. The breakdown is given in Table 3.
| Cost component | No battery | Fixed proportional | Proposed |
|---|---|---|---|
| Generator fuel cost | 1.000 | 0.812 | 0.754 |
| Battery energy cost | 0 | 0.143 | 0.112 |
| Battery degradation cost | 0 | 0.089 | 0.064 |
| Total cost | 1.000 | 1.044 | 0.930 |
It is noteworthy that the fixed proportional sharing actually increased total cost relative to no-battery because the battery degradation cost outweighed the savings in generator fuel. My optimization effectively reduced degradation and improved overall economics.
Discussion
The results confirm that the integration of battery energy storage systems with an intelligent control strategy can substantially enhance the frequency regulation performance of power systems. The two-layer approach effectively addresses the trade-off between fast response and battery longevity. By considering SOC management and economic optimization in the lower layer, the coordination among distributed storage units prevents premature depletion of any single unit and extends the fleet’s operational life. The proposed method is scalable to large numbers of distributed battery energy storage systems because the consensus algorithm only requires local communication with neighbors.
Compared to other recent studies, my work provides a more comprehensive model that includes both technical and economic dimensions. For instance, the inclusion of a quadratic degradation cost in the objective function is a novel contribution that aligns operational decisions with lifecycle costs. Additionally, the regulation zone classification ensures that batteries are not overused during normal conditions, reserving their capacity for emergency support.
Conclusion
In this paper, I presented a novel control strategy for power system frequency regulation using multiple distributed battery energy storage systems. The key contributions include:
- A load frequency control model that incorporates the dynamic response of battery energy storage systems via a first-order inertia.
- A two-layer distributed control framework where the upper layer optimally splits the regulation signal between batteries and generators, and the lower layer uses consensus to allocate power among batteries while minimizing degradation costs and maintaining SOC balance.
- A comprehensive simulation study demonstrating that the proposed method reduces maximum frequency deviation by 50% compared to no battery, and by 26% compared to fixed proportional sharing, while also improving economic performance by over 12%.
Future work could extend this approach to include predictive control based on load forecasts, as well as integration with hybrid storage systems that combine batteries with supercapacitors or flywheels. The methodology is also applicable to microgrids and islanded systems where multiple battery energy storage systems operate cooperatively.
In summary, the intelligent coordination of distributed battery energy storage systems through the proposed two-layer control strategy offers a viable and cost-effective solution for enhancing grid frequency stability in modern power systems.