The global energy landscape is undergoing a profound transformation, driven by the imperative to mitigate climate change and ensure sustainable development. This transition is characterized by the rapid and large-scale integration of renewable energy sources (RES), primarily wind and solar photovoltaic (PV) systems, into power grids worldwide. While offering a clean and abundant alternative to fossil fuels, these resources introduce significant operational challenges due to their inherent intermittency and stochastic power output. The fundamental task of any power system is to maintain a continuous, real-time balance between electrical generation and load consumption. This balance is directly reflected in the system frequency. A sudden increase in load or a drop in generation causes the frequency to fall, while a decrease in load or a surge in generation causes it to rise. Traditional power systems rely on the rotational inertia of synchronous generators and their governor-controlled primary frequency response (PFR) to arrest frequency deviations following a disturbance. However, most inverter-based RES units are decoupled from the grid frequency and do not inherently provide inertia or damping, effectively reducing the system’s overall resilience to imbalances.
This “inertia deficit” necessitates new, fast-responding assets to support grid stability. Among various solutions, the cell energy storage system, particularly battery energy storage systems (BESS), has emerged as a paramount technology due to its exceptional characteristics: near-instantaneous response (millisecond to second timescales), high ramp rates, precise power controllability, and flexible siting. A cell energy storage system can be dispatched to inject or absorb power within its rated capacity, providing both synthetic inertia and primary frequency regulation services. This capability makes the cell energy storage system an ideal companion for renewable-rich grids, smoothing fluctuations and ensuring reliability.
While the principle of using a cell energy storage system for frequency regulation is established, the control strategy governing its power output is critical for maximizing effectiveness, ensuring economic viability, and preserving the health of the storage asset. A naive approach, such as a fixed-gain droop control, can quickly deplete or overcharge the batteries during sustained disturbances, leading to state-of-charge (SOC) limits being reached. Once a cell energy storage system hits its SOC boundary, it must withdraw from service, potentially exacerbating the frequency event if other resources are insufficient. Furthermore, in large-scale deployments, a cell energy storage system installation often comprises multiple parallel units or strings. Variations in initial SOC, aging, and internal resistances can lead to uneven SOC trajectories during coordinated operation. Without proper management, some units may be over-utilized while others are under-utilized, accelerating the degradation of the overworked units and reducing the overall available capacity and lifespan of the cell energy storage system fleet.
Therefore, an advanced control paradigm is required. This paradigm must be dual-objective: firstly, to meet the grid’s frequency regulation needs by providing appropriate and timely power support; and secondly, to intelligently manage the internal energy state of the cell energy storage system, maintaining SOC within a healthy range and ensuring balanced utilization across multiple units. This article delves into the design and analysis of such an optimized, integrated control strategy for a cell energy storage system participating in primary frequency regulation. The strategy synthesizes grid-frequency requirements with real-time battery SOC, enabling adaptive power dispatch and coordinated unit operation.
Grid Frequency Regulation Requirements and BESS Control Framework
The primary frequency control loop is the first automated response to a generation-load imbalance. Its objective is to stabilize the frequency at a new steady-state value, albeit with a non-zero deviation, known as the static droop characteristic. The response of a traditional synchronous generator can be modeled by its swing equation and governor-turbine dynamics. The power-frequency relationship is governed by the system’s natural characteristics and the combined droop response of all participating generators.
For a cell energy storage system to emulate this behavior, a virtual control layer is implemented. The most common and effective method combines virtual inertia emulation with droop control. The total reference power adjustment for the cell energy storage system, \(\Delta P_{\text{ref, BESS}}\), is calculated as:
$$ \Delta P_{\text{ref, BESS}} = K_{\text{BESS}} \cdot \Delta f + K_{\text{ine}} \cdot \frac{d(\Delta f)}{dt} $$
where:
\(\Delta f = f – f_{\text{nom}}\) is the system frequency deviation (Hz).
\(K_{\text{BESS}}\) is the droop coefficient (MW/Hz), typically negative.
\(K_{\text{ine}}\) is the virtual inertia coefficient (MW·s/Hz), also typically negative.
\(\frac{d(\Delta f)}{dt}\) is the Rate of Change of Frequency (RoCoF).
The second term, proportional to RoCoF, provides an immediate power response akin to inertial energy release/absorption, helping to slow down the initial rate of frequency change. The first term provides sustained power support proportional to the frequency deviation itself, mimicking governor action.
To design an efficient control strategy, it is essential to segment the frequency deviation into distinct operational zones, each triggering a specific control action from the cell energy storage system. This segmentation prevents unnecessary wear from responding to negligible noise and ensures maximum effort during critical events.
| Frequency Deviation Zone | Description | BESS Action | Droop Coefficient (\(K_{\text{BESS}}\)) |
|---|---|---|---|
| Dead Band: \( |\Delta f| \leq \Delta f_{\text{db}} \) | Normal minor fluctuations. No action required to minimize cycling. | No response. \(\Delta P_{\text{ref, BESS}} = 0\). | \(K_{\text{BESS}} = 0\) |
| Normal Regulation I: \( \Delta f_{\text{db}} < \Delta f \leq \Delta f_{\text{norm}} \) | Frequency above nominal. System has excess generation. | Charge (absorb power). | \(K_{\text{BESS}} = K_{\text{ch}}\) (Negative) |
| Normal Regulation II: \( -\Delta f_{\text{norm}} \leq \Delta f < -\Delta f_{\text{db}} \) | Frequency below nominal. System has a generation deficit. | Discharge (inject power). | \(K_{\text{BESS}} = K_{\text{disch}}\) (Negative) |
| Emergency Regulation: \( |\Delta f| > \Delta f_{\text{norm}} \) | Large disturbance. Critical for system stability. | Charge/Discharge at maximum capable power. | \(K_{\text{BESS}} = K_{\text{max}}\) (Negative, largest magnitude) |
Where typical values might be \(\Delta f_{\text{db}} = 0.033 \ \text{Hz}\) (equivalent to ~2 rpm for a generator) and \(\Delta f_{\text{norm}} = 0.2 \ \text{Hz}\). The coefficients satisfy \( |K_{\text{max}}| > |K_{\text{disch}}|, |K_{\text{ch}}| \). This zonal approach forms the first layer of our optimized strategy, ensuring the cell energy storage system response is appropriate for the severity of the grid event.
SOC-Aware Adaptive Droop Control and Unit Balancing
The fixed droop coefficients \(K_{\text{ch}}\) and \(K_{\text{disch}}\) in the normal regulation zones, while effective for grid support, disregard the internal energy state of the cell energy storage system. Continuously operating with a high-power coefficient when the SOC is already very high (during charging) or very low (during discharging) pushes the battery towards its limits prematurely. To address this, we introduce SOC feedback to adaptively modify the droop coefficient, creating a variable droop control. This strategy intelligently reduces the power contribution of a cell energy storage system as its SOC approaches undesirable limits, thereby preserving its operational availability and longevity.
The real-time SOC is calculated as:
$$ \text{SOC}(t) = \text{SOC}(t_0) + \frac{1}{E_{\text{rated}}} \int_{t_0}^{t} P_{\text{BESS}}(\tau) \ d\tau $$
where \(E_{\text{rated}}\) is the rated energy capacity of the cell energy storage system. We define critical SOC thresholds to implement the adaptive logic:
| Parameter | Description |
|---|---|
| \(\text{SOC}_{\text{max}}\) | Upper safety limit. Operation must stop above this. |
| \(\text{SOC}_{\text{min}}\) | Lower safety limit. Operation must stop below this. |
| \(\text{SOC}_{\text{high+}}, \text{SOC}_{\text{high-}}\) | Upper warning thresholds (\(\text{SOC}_{\text{high-}} < \text{SOC}_{\text{high+}} < \text{SOC}_{\text{max}}\)). |
| \(\text{SOC}_{\text{low+}}, \text{SOC}_{\text{low-}}\) | Lower warning thresholds (\(\text{SOC}_{\text{min}} < \text{SOC}_{\text{low+}} < \text{SOC}_{\text{low-}}\)). |
The adaptive droop coefficient \(K_{\text{BESS, adapt}}(\text{SOC})\) is defined as follows:
1. Discharge Mode (\(\Delta f < 0\)):
– If \(\text{SOC} \leq \text{SOC}_{\text{min}}\): No operation (shut down).
– If \(\text{SOC}_{\text{min}} < \text{SOC} < \text{SOC}_{\text{low+}}\):
$$ K_{\text{disch, adapt}} = \frac{1}{2} K_{\text{max}} \left( 1 – \sqrt{ \frac{\text{SOC} – \text{SOC}_{\text{low+}}}{\text{SOC}_{\text{min}} – \text{SOC}_{\text{low+}}} } \right) $$
The coefficient is reduced significantly as SOC approaches the minimum limit.
– If \(\text{SOC}_{\text{low+}} \leq \text{SOC} \leq \text{SOC}_{\text{low-}}\):
$$ K_{\text{disch, adapt}} = \frac{1}{2} K_{\text{max}} \left( 1 + \sqrt{ \frac{\text{SOC} – \text{SOC}_{\text{low+}}}{\text{SOC}_{\text{low-}} – \text{SOC}_{\text{low+}}} } \right) $$
The coefficient ramps up from a reduced value to the maximum.
– If \(\text{SOC}_{\text{low-}} < \text{SOC} \leq \text{SOC}_{\text{max}}\): \(K_{\text{disch, adapt}} = K_{\text{max}}\) (full capability).
– If \(\text{SOC} > \text{SOC}_{\text{max}}\): No discharge (should charge).
2. Charge Mode (\(\Delta f > 0\)):
The logic is symmetric, reducing the charge coefficient \(K_{\text{ch, adapt}}\) as SOC approaches \(\text{SOC}_{\text{max}}\).
– If \(\text{SOC}_{\text{high-}} \leq \text{SOC} \leq \text{SOC}_{\text{high+}}\):
$$ K_{\text{ch, adapt}} = \frac{1}{2} K_{\text{max}} \left( 1 + \sqrt{ \frac{\text{SOC} – \text{SOC}_{\text{high+}}}{\text{SOC}_{\text{high-}} – \text{SOC}_{\text{high+}}} } \right) $$
– If \(\text{SOC}_{\text{high+}} < \text{SOC} < \text{SOC}_{\text{max}}\):
$$ K_{\text{ch, adapt}} = \frac{1}{2} K_{\text{max}} \left( 1 – \sqrt{ \frac{\text{SOC} – \text{SOC}_{\text{high+}}}{\text{SOC}_{\text{max}} – \text{SOC}_{\text{high+}}} } \right) $$
This adaptive mechanism ensures that a cell energy storage system with a critically high or low SOC contributes less to frequency regulation, allowing other units or systems to take more share, thus preserving its own health.
For a system with \(N\) parallel cell energy storage system units, SOC imbalance is a major concern. To coordinate them, we introduce a multiplicative balancing factor, \(G_i\), for each unit \(i\). The power reference for unit \(i\) becomes:
$$ \Delta P_{\text{ref, BESS}_i} = G_i(\text{SOC}_i) \cdot \left[ K_{\text{BESS, adapt}}(\text{SOC}_i) \cdot \Delta f + K_{\text{ine}} \cdot \frac{d(\Delta f)}{dt} \right] $$
The balancing factor \(G_i\) is defined as:
For Discharge Mode:
$$ G_i = \exp\left( k_{\text{SOC}} \cdot (\text{SOC}_i – \overline{\text{SOC}}) \right) $$
For Charge Mode:
$$ G_i = \exp\left( -k_{\text{SOC}} \cdot (\text{SOC}_i – \overline{\text{SOC}}) \right) $$
where \(\overline{\text{SOC}} = \frac{1}{N} \sum_{i=1}^{N} \text{SOC}_i\) is the average SOC of the fleet, and \(k_{\text{SOC}} > 0\) is a tuning gain.
This design ensures that during discharge, units with a higher-than-average SOC (\(G_i > 1\)) are assigned a proportionally larger share of the total power demand, draining them faster. Units with a lower-than-average SOC (\(G_i < 1\)) are assigned a smaller share, conserving their energy. The opposite occurs during charging. Over time, this feedback loop drives the SOC of all units towards the fleet average, promoting balanced wear and maximizing the aggregate available capacity of the cell energy storage system cluster.
Integrated Control Strategy for a Cell Energy Storage System
The complete, optimized control strategy synthesizes the grid-frequency zonal logic with the SOC-aware adaptive and balancing algorithms. The decision flowchart and final power calculation for each cell energy storage system unit are as follows:
Step 1: Grid-Frequency Assessment. Measure \(\Delta f\) and \(\frac{d(\Delta f)}{dt}\).
Step 2: Zonal Mode Selection.
a) If \(|\Delta f| \leq \Delta f_{\text{db}}\): Set \(\Delta P_{\text{ref, BESS}_i} = 0\). Return.
b) If \(|\Delta f| > \Delta f_{\text{norm}}\): Set operational mode to Emergency. Proceed to Step 3 with \(K_{\text{BESS, mode}} = K_{\text{max}}\).
c) If \(\Delta f_{\text{db}} < \Delta f \leq \Delta f_{\text{norm}}\): Set mode to Normal Charge.
d) If \(-\Delta f_{\text{norm}} \leq \Delta f < -\Delta f_{\text{db}}\): Set mode to Normal Discharge.
Step 3: Unit SOC & Adaptive Coefficient Calculation. For each unit \(i\):
a) Check if \(\text{SOC}_i\) is within \([\text{SOC}_{\text{min}}, \text{SOC}_{\text{max}}]\). If not, set \(\Delta P_{\text{ref, BESS}_i} = 0\).
b) For Normal modes: Calculate \(K_{\text{BESS, adapt}}(\text{SOC}_i)\) using the formulas in Section 2.
c) For Emergency mode: Set \(K_{\text{BESS, adapt}} = K_{\text{max}}\).
Step 4: Balancing Factor Calculation. Calculate the fleet’s \(\overline{\text{SOC}}\) and then each unit’s \(G_i(\text{SOC}_i)\) based on the operational mode (charge/discharge).
Step 5: Final Power Reference Computation.
$$ \Delta P_{\text{ref, BESS}_i} = G_i \cdot \left[ K_{\text{BESS, adapt}} \cdot \Delta f + K_{\text{ine}} \cdot \frac{d(\Delta f)}{dt} \right] $$
This power reference is then sent to the inner-loop power converters of the respective cell energy storage system unit.
This integrated strategy ensures that the aggregate response of the cell energy storage system fleet is always aligned with grid needs (via the zonal and virtual inertia terms), while internally managing the energy state of each unit through adaptive droop and cooperative balancing. It prevents any single unit from becoming a weak link due to SOC extremes and extends the useful service life of the entire cell energy storage system asset.
Simulation Analysis and Performance Evaluation
To validate the proposed integrated control strategy, a simulation model of a power system region with integrated cell energy storage system support is developed. The model includes a traditional thermal generator (representing the bulk system inertia and governor response) and a BESS cluster. The key simulation parameters are listed below:
| Component | Parameter | Value (pu where applicable) |
|---|---|---|
| Power System | System Base Power (\(S_{\text{base}}\)) | 100 MVA |
| Inertia Constant (\(H\)) | 5.0 s | |
| Load Damping Coefficient (\(D\)) | 1.0 pu/Hz | |
| Governor Droop (\(R_G\)) | 0.05 pu (5%) | |
| Thermal Generator | Governor Time Constant (\(T_G\)) | 0.08 s |
| Turbine Time Constant (\(T_{CH}\)) | 0.30 s | |
| Reheater Time Constant (\(T_{RH}\)) | 10.0 s | |
| High-Pressure Power Fraction (\(F_{HP}\)) | 0.5 | |
| Cell Energy Storage System | Rated Power (\(P_{\text{BESS, rated}}\)) | 0.02 pu (2 MW) |
| Rated Energy (\(E_{\text{BESS, rated}}\)) | 0.02 pu (2 MWh) | |
| Max Droop Coeff. (\(K_{\text{max}}\)) | -0.12 pu/Hz (equiv. to ~8.33% droop) | |
| Virtual Inertia Coeff. (\(K_{\text{ine}}\)) | -0.01 pu·s/Hz | |
| Number of Units (\(N\)) | 2 | |
| Control Parameters | Deadband (\(\Delta f_{\text{db}}\)) | ±0.033 Hz |
| Normal/Emergency Threshold (\(\Delta f_{\text{norm}}\)) | ±0.2 Hz | |
| SOC Limits: \(\text{SOC}_{\text{min}}, \text{SOC}_{\text{max}}\) | 0.2, 0.9 | |
| SOC Thresholds: \(\text{SOC}_{\text{low-}}, \text{SOC}_{\text{low+}}, \text{SOC}_{\text{high-}}, \text{SOC}_{\text{high+}}\) | 0.4, 0.3, 0.7, 0.8 | |
| Balancing Gain (\(k_{\text{SOC}}\)) | 2.0 |
Case Study 1: Step Load Increase. A sudden load increase of \(\Delta P_{\text{load}} = 0.015\) pu (1.5 MW) is applied at t=1s. Four scenarios are compared: (a) No BESS support; (b) Fixed Droop Control (\(K_{\text{BESS}} = K_{\text{max}}\), no SOC adaptation/balancing); (c) SOC-Adaptive Droop only (no balancing); (d) Full Integrated Strategy (with balancing). The cell energy storage system units have initial SOCs of 0.5 and 0.35, respectively.
The system frequency nadir is deepest in the “No BESS” case. The “Fixed Droop” case provides the fastest and strongest initial frequency support, resulting in the highest nadir. However, it aggressively discharges both units equally. The unit starting at SOC=0.35 hits its lower limit quickly and is forced to shut off, causing a secondary dip in frequency as the remaining load imbalance must now be handled solely by the slower thermal generator. The SOC-Adaptive only strategy reduces the discharge rate of the low-SOC unit, preventing its early shutdown and avoiding the secondary dip. The frequency recovery is slightly slower than Fixed Droop but more stable. The Full Integrated Strategy achieves a similar frequency trajectory to the SOC-Adaptive case. Crucially, the balancing action is evident: the high-SOC unit (0.5) discharges more aggressively, while the low-SOC unit (0.35) discharges more conservatively. Over the simulation period, their SOCs converge, demonstrating effective fleet energy management. The cell energy storage system’s role in stabilizing the frequency is clearly effective while also maintaining its own operational health.
Case Study 2: Continuous Stochastic Disturbance. To simulate a period of high renewable variability and load fluctuations, a 10-minute profile of random power imbalances (acting as a net load change) with a standard deviation of 0.008 pu is applied. The performance of the Full Integrated Strategy is evaluated.
The cell energy storage system fleet successfully mitigates the frequency deviations, keeping \(|\Delta f|\) predominantly within the normal regulation band and preventing any excursions beyond the emergency threshold. The SOC trajectories of the two units, starting again at 0.5 and 0.35, show coordinated action. During net discharge periods, the higher SOC unit takes a larger share; during net charge periods, the lower SOC unit absorbs more power. Over the 10-minute window, their SOC values steadily move towards each other, moving from a gap of 0.15 to a gap of less than 0.05. This demonstrates the strategy’s robustness in managing the cell energy storage system fleet under realistic, persistent disturbances, ensuring all units contribute fairly and remain within their safe operating envelopes.
Conclusions
The integration of a cell energy storage system into modern power grids is a critical enabler for maintaining frequency stability amidst growing renewable penetration. However, to unlock the full potential of this technology, control strategies must evolve beyond simple grid-following approaches. This article has presented a comprehensive, optimized integrated control strategy that harmonizes the external requirement for fast and reliable primary frequency regulation with the internal necessity of prudent battery energy management.
The strategy’s effectiveness stems from its multi-layered design. First, it respects the practical requirements of grid operation by implementing a zonal frequency response, differentiating between deadband, normal, and emergency regulation. Second, it incorporates a sophisticated adaptive droop mechanism where the power contribution of a cell energy storage system is dynamically scaled based on its real-time state of charge. This prevents damaging over-charge or over-discharge cycles, thereby preserving the asset’s lifespan and ensuring its availability for future disturbances. Third, for multi-unit deployments, it introduces a balancing factor that promotes equitable load-sharing among units, driving their SOC levels towards a common average. This eliminates the risk of individual unit failure due to SOC extremes and optimizes the use of the total installed capacity of the cell energy storage system cluster.
Simulation studies under both step-change and continuous stochastic disturbance scenarios confirm the advantages of the proposed strategy. It provides superior frequency support compared to a system without storage, and it offers more sustainable and reliable performance compared to a fixed-droop control strategy by avoiding premature unit shutdown and managing internal energy states. The cell energy storage system, governed by this intelligent control paradigm, thus transitions from being a mere power buffer to a smart, self-managing grid asset that enhances system resilience while optimizing its own operational economics and longevity. Future work may involve the real-time coordination of the cell energy storage system with other fast-responding assets like demand response or the optimization of the adaptive parameters for specific battery chemistries and market structures.