Reducing electricity costs is a critical operational objective for industrial and commercial consumers. As a flexible resource, battery energy storage cell systems offer a compelling solution. By implementing “charge-low, discharge-high” strategies, these systems can shave peak loads and shift energy consumption, thereby reducing both the maximum demand charge (a key component of the basic electricity fee) and the total energy cost by arbitraging time-of-use (TOU) price differentials. The user-side application scenario, with its well-defined load profile and relatively stable influencing factors, presents a favorable environment for energy storage cell deployment. Supported by favorable policies and potentially widening peak-valley price gaps, optimizing the investment and operation of user-side energy storage cell systems is paramount to unlocking their economic potential and accelerating adoption. This work proposes an integrated framework encompassing optimal sizing, pre-month planning, and intra-day rolling-horizon dispatch to maximize the economic benefits for the end-user.
Economic Benefits and Operational Constraints of Energy Storage
The economic benefit for a consumer installing an energy storage cell system stems from two primary revenue streams: demand charge reduction and energy arbitrage. The demand charge is typically calculated based on the maximum average power (in kW) drawn by the consumer in a billing period (e.g., monthly). The energy storage cell system can discharge during high-load periods to lower this measured maximum, thus generating demand charge savings. Simultaneously, by charging during low-price (off-peak) periods and discharging during high-price (on-peak) periods, the system captures revenue from the TOU tariff differentials.
The monthly demand charge savings \( R_d \) and the daily energy arbitrage revenue \( R_{e,i} \) for day \( i \) can be calculated as:
$$ R_d = m ( P^{L}_{max} – P_{max} ) $$
$$ R_{e,i} = -\sum_{t=1}^{k} \lambda_t \cdot P^{CS}_{i,t} \cdot \Delta t $$
where \( m \) is the unit price for the demand charge (\( \text{\$/kW} \)), \( P^{L}_{max} \) is the original monthly maximum load (kW), \( P_{max} \) is the target maximum demand (defended value) after storage dispatch (kW), \( \lambda_t \) is the TOU electricity price at time interval \( t \) (\( \text{\$/kWh} \)), \( P^{CS}_{i,t} \) is the charge/discharge power of the energy storage cell system at time \( t \) on day \( i \) (positive for charging, negative for discharging, in kW), \( k \) is the total number of time intervals in a day, and \( \Delta t \) is the duration of each time interval (hours).
The operation of the battery energy storage cell system is subject to physical and performance constraints:
- Power and Capacity Limits: The charge/discharge power cannot exceed the power rating \( P_{N}^{CS} \), and the state of charge (SOC) must remain within safe bounds.
$$ -P_{N}^{CS} \le P^{CS}_{i,t} \le P_{N}^{CS} $$
$$ SOC_{min} \cdot C_N \le S_{i,t} \le C_N $$
where \( C_N \) is the rated energy capacity (kWh) of the energy storage cell system, and \( S_{i,t} \) is the stored energy at time \( t \).
- System Coupling Constraint: The net load seen by the grid (user load plus storage power) must be non-negative and should not exceed the original maximum load \( P^{L}_{max} \) to maintain the demand defense functionality.
$$ 0 \le P^{L}_{i,t} + P^{CS}_{i,t} \le P^{L}_{max} $$
- Performance and Longevity Constraints: Frequent cycling between charge, discharge, and idle states accelerates energy storage cell degradation. To mitigate this and extend operational life, a constraint on the daily energy throughput \( Q_i \) is introduced. This effectively limits the number of deep cycles and state transitions.
$$ Q_i = \sum_{t=1}^{k} |P^{CS}_{i,t}| \cdot \Delta t \le Q_{max} $$
Here, \( Q_{max} \) is the maximum allowable daily throughput, typically set as a multiple of the system’s capacity (e.g., \( 2C_N \)), considering the battery’s cycle life and the specific load-shaving profile.
Integrated Optimization Strategy and Predictive Modeling
The proposed optimization framework consists of two sequential stages: 1) optimal sizing and 2) operational dispatch with rolling-horizon adaptation.
Stage 1: Optimal Sizing. This stage determines the most economical power (\( P_{N}^{CS} \)) and energy (\( C_N \)) ratings for the energy storage cell system. The objective is to maximize the total net present value or average monthly net benefit over the system’s lifetime. The model incorporates the capital cost, annualized operation and maintenance (O&M) cost, and the projected revenues from demand charge management and energy arbitrage based on historical load data. The optimization is subject to all operational constraints listed above. A key output of this stage is the selection of the optimal battery technology (e.g., Lithium Iron Phosphate, Lead-Acid) based on its techno-economic parameters.
| Energy Storage Cell Type | Power Cost ($/kW) | Energy Cost ($/kWh) | Energy/Power Ratio (h) | Cycle Life | Round-Trip Efficiency |
|---|---|---|---|---|---|
| Lithium Iron Phosphate (LFP) | 600 | 1,600 | 2.0 | > 4000 | ~95% |
| Lead-Acid | 600 | 800 | 3.5 | ~1500 | ~80% |
| Sodium-Sulfur | 420 | 1,400 | 7.0 | > 3000 | ~85% |
Stage 2: Operational Dispatch with Rolling-Horizon Optimization. Once sized, the system’s operation is optimized in a hierarchical manner:
- Pre-Month Optimization: Using a forecast of the monthly load profile, this step calculates an optimal target maximum demand \( P_{max} \) to defend for the coming month, balancing the potential demand charge savings against the required energy arbitrage dispatch.
- Intra-Day Rolling Optimization: This is the core real-time dispatch mechanism. For each dispatch day, a schedule is initially created using day-ahead load forecasts. However, to counteract forecast errors, a rolling-horizon approach is implemented. At each time step \( t \):
- The optimization is re-run for the remaining part of the day, using actual load data for past intervals and updated short-term forecasts for future intervals.
- The objective function for intra-day optimization is designed to encourage desirable behavior. During off-peak periods, it rewards a higher SOC to ensure sufficient energy is stored for later peak periods:
$$ \max F_{intra} = \begin{cases}
-\sum \lambda_t P^{CS}_{i,t} \Delta t + \omega S_{i,t}, & \text{if } t \in \text{Off-Peak} \\
-\sum \lambda_t P^{CS}_{i,t} \Delta t, & \text{otherwise}
\end{cases} $$
where \( \omega \) is a weighting coefficient.
- If the solver fails to find a feasible dispatch plan that respects the current \( P_{max} \) due to a higher-than-expected load, the energy storage cell dispatch and the target \( P_{max} \) itself are co-optimized and updated in real-time to the best feasible value. This adaptive mechanism ensures grid-friendly operation (no inadvertent export, feasible SOC trajectory) while still maximizing economic benefit under the new conditions.
Load forecasting is a critical input. A Long Short-Term Memory (LSTM) neural network model is typically employed for both monthly and day-ahead predictions, providing the necessary accuracy for effective planning and dispatch.
Case Study and Numerical Results
An industrial park case study demonstrates the effectiveness of the proposed framework. The park operates under a two-part tariff with a demand charge of $40/kW/month and a TOU energy rate. The optimization was performed for different energy storage cell technologies.
Sizing Results: The net monthly benefit (revenue minus monthly amortized cost) was calculated for various configurations. The results for the optimal size of each technology are summarized below:
| Optimal Technology | Power Rating (kW) | Energy Capacity (kWh) | Total Investment (k$) | Avg. Monthly Net Benefit (k$) | Return on Investment |
|---|---|---|---|---|---|
| Lithium Iron Phosphate | 1,365 | 2,730 | 540 | 4.28 | 0.71 |
| Lead-Acid | 1,200 | 4,200 | 412 | 3.80 | 0.55 |
| Sodium-Sulfur | 1,440 | 10,080 | 1,490 | 5.40 | 0.52 |
The Lithium Iron Phosphate (LFP) energy storage cell system offered the highest return on investment, providing the best balance between cost, performance, and longevity for this specific load profile.
Impact of Performance Constraint: Enforcing the daily throughput limit \( Q_{max} = 2C_N \) significantly reduced the number of charge/discharge state transitions compared to an unconstrained schedule. This leads to smoother operation and reduced degradation stress on the battery energy storage cell, directly contributing to extended system life and lower long-term O&M costs.
Dispatch Optimization Results: Comparing the proposed rolling-horizon method against a conventional day-ahead scheduling method for a sample month revealed clear advantages. The rolling-horizon method successfully adapted to load forecast errors, dynamically adjusting the target \( P_{max} \) and dispatch schedule. This resulted in a lower final monthly maximum demand and a higher total monthly revenue. For a representative day, the rolling-horizon approach increased the combined daily benefit (from demand and energy) by over $300 compared to the rigid day-ahead schedule. Over the entire month, the cumulative benefit increased by approximately $9,000.
The mathematical models were implemented and solved using the YALMIP toolbox with the CPLEX solver in MATLAB, demonstrating computational tractability for practical application.
Conclusion
This study presents a comprehensive, two-stage optimization framework for user-side battery energy storage cell systems. The framework seamlessly integrates economic sizing with an adaptive, rolling-horizon dispatch strategy. Key contributions include: 1) the incorporation of a performance constraint that reduces detrimental cycling and extends the operational life of the energy storage cell; 2) the formulation of an intra-day rolling optimization with a segmented objective function that ensures efficient charging during off-peak windows; and 3) a real-time mechanism to update the target maximum demand, thereby robustly managing load forecast uncertainties.
The case study validates that the proposed methodology enables consumers to select the most cost-effective energy storage cell technology and configuration. More importantly, it demonstrates that the rolling-horizon dispatch strategy can yield significantly higher economic returns than static scheduling by dynamically responding to real-time conditions. This work provides a practical and effective tool for maximizing the economic value of user-side energy storage cell investments, facilitating lower electricity costs and promoting the wider adoption of this valuable grid resource.