⚡ electrical engineering

Data-Driven Sequential Market Optimization for Front-of-the-Meter Battery Energy Storage Systems

This paper introduces a data-driven, sequential optimization framework for Front-of-the-Meter Battery Energy Storage Systems that realistically models market gate closures and rolling forecasts to derive opportunity-cost-based bidding strategies, thereby enhancing operational feasibility and profitability across multiple electricity markets.

Original authors: Steffen Kortmann, Hannah Sanders, Andreas Ulbig

Published 2026-07-28

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Steffen Kortmann, Hannah Sanders, Andreas Ulbig

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the electricity grid as a giant, high-stakes game of musical chairs, but instead of people, it's electrons dancing between power plants and your toaster. The goal is to keep the music playing perfectly in tune; if there are too many electrons, the frequency spikes, and if there are too few, it crashes. To keep the rhythm steady, we need "flexible" helpers who can instantly grab extra energy or let go of some when the music gets chaotic. Enter the Battery Energy Storage System (BESS), a massive digital battery that acts like a super-athlete, ready to sprint in or out at a moment's notice. But here's the tricky part: this athlete doesn't just play one game. It has to juggle multiple markets simultaneously—selling energy when it's cheap, holding it back to help stabilize the grid, and trading it again later. The problem is that the rules for these markets change at different times, and the prices are like weather forecasts: they change as you get closer to the event. If you plan your whole day's strategy at 6:00 AM based on a guess, you might miss a sudden storm (or a sudden price spike) that happens at noon.

This paper tackles the headache of how to manage these giant batteries without getting caught in a trap of bad guesses. The authors, Steffen Kortmann, Hannah Sanders, and Andreas Ulbig, realized that most existing methods are like trying to plan a whole road trip in one go, ignoring that traffic updates every hour. They argue that to make money and keep the grid safe, you need a "sequential" approach—a step-by-step strategy that updates its plan every time a new market closes its doors. They built a smart, data-driven framework that acts like a seasoned coach, constantly re-evaluating the battery's moves as new price forecasts roll in, ensuring the battery doesn't overcommit or run out of juice at the wrong moment.

The Game of "Opportunity Cost"

Think of the battery's energy like a backpack full of gold coins. You have a limited number of coins, and you want to spend them where they buy the most happiness (or profit). But you can't spend the same coin twice. If you use a coin to buy a ticket to a concert (selling energy now), you can't use that same coin to buy a ticket to a VIP party later (providing emergency power). The "Opportunity Cost" is the value of the party ticket you didn't buy because you spent the coin on the concert.

The paper introduces a clever way to calculate this cost. Instead of just guessing, the framework asks: "If I commit this battery power to Market A right now, how much potential profit am I losing from Markets B, C, and D that haven't happened yet?" It then uses this calculation to set a "minimum price" for bidding. If the market isn't offering enough money to cover the lost opportunity of the other markets, the battery simply says, "No thanks," and saves its power for a better deal. This ensures the battery never makes a deal that looks good in isolation but is actually a bad move for the whole day.

The Step-by-Step Strategy

The authors designed a system that mimics the real-world timeline of electricity markets in Germany. Imagine a series of doors closing one by one:

Frequency Containment Reserve (FCR): The first door closes early. This is for immediate emergency help. The battery locks in a plan here.
Automatic Frequency Restoration Reserve (aFRR): The next door closes. This is for slightly slower, but still urgent, grid balancing.
Day-Ahead Auction (DAA): The big door for selling energy the next day.
Intraday Auction (IDA): The final doors, opening up closer to real-time for last-minute adjustments.

The magic of their framework is that it doesn't just plan for the first door and forget the rest. It solves the puzzle for the current door, then immediately re-solves the puzzle for all the remaining doors, using the latest price forecasts. It's like playing a video game where the map updates every time you level up. If the price of electricity spikes in the afternoon, the system notices, re-calculates the "opportunity cost," and might decide to hold back some energy for the Intraday market instead of selling it all in the Day-Ahead market.

What They Found (and What They Didn't)

The researchers tested their idea using a simulation of a single day (April 1, 2024) with a battery rated at 3.65 MW and 7.3 MWh capacity. They didn't just guess; they ran the numbers through a complex mathematical model that respects the physical limits of the battery (it can't charge and discharge at the same time, and it can't go below 0% or above 100% charge).

The results showed that their step-by-step, data-driven approach works. When they compared the "Day-Ahead" plan (made early in the morning) with the "Intraday" plan (made later with better info), they saw that the later plan was much closer to what actually happened. The framework successfully adapted to updated forecasts, reducing the gap between what was planned and what was actually done. This means the battery operator can make more money and avoid the risk of breaking the rules or running out of power.

However, it's important to note that this is a simulation. The paper proves the math works and the logic holds up under realistic conditions, but it hasn't been tested on a real battery in a live, chaotic market yet. The authors suggest that their method is a solid foundation, but future work will need to add even more complexity, like using Artificial Intelligence to trade in the final "continuous" market (Intraday Continuous) where prices change every second.

Why This Matters

For anyone who cares about clean energy, this is a big deal. As we switch to wind and solar power, the grid becomes more unpredictable. We need batteries to be the shock absorbers of the system. But if battery operators are using outdated, rigid planning tools, they might miss out on profits or, worse, fail to help the grid when it's needed most. This paper offers a new playbook: a way to be flexible, smart, and data-driven. It turns the battery from a static storage unit into a dynamic player that can navigate the complex, shifting landscape of modern energy markets, ensuring that the lights stay on and the wallet stays full.

Technical Summary: Data-Driven Sequential Market Optimization for Front-of-the-Meter BESS

Problem Statement

The integration of Battery Energy Storage Systems (BESS) into electricity markets has increased the need for coordinated participation across energy and balancing services to maximize operational flexibility. However, existing revenue-stacking models frequently rely on simplified assumptions, such as simultaneous bidding across multiple markets and the neglect of rolling forecast impacts. These simplifications lead to unrealistic scheduling, infeasible operational plans, and overestimated revenue projections. Operators currently lack transparent methods to determine feasible bid quantities and prices that align with the sequential nature of market operations, specific Gate Closure Times (GCT), and updated price forecasts.

Methodology

The paper proposes a sequential, data-driven optimization framework designed to mirror actual market operations for Front-of-the-Meter (FTM) BESS. The methodology structures the scheduling process into consecutive market stages, updating decisions at each Gate Closure Time (GCT) based on the most recent market results and forecast revisions.

1. Sequential Optimization Framework

The framework models the chronological sequence of markets: Frequency Containment Reserve (FCR), automatic Frequency Restoration Reserve (aFRR), Day Ahead Auction (DAA), and Intraday Auction (IDA).

State Propagation: The State of Charge (SoC) resulting from a cleared market becomes a fixed boundary condition for the subsequent optimization stage.
Rolling Forecasts: Price forecasts are updated at each stage, incorporating information from already cleared markets to refine the optimization for remaining markets.
Constraints: The model enforces rigorous technical and regulatory constraints, including:
- Battery Physics: Energy balance equations, power limits, efficiency losses, and mutual exclusivity of charging/discharging.
- Market Rules: Fixed bid increments (e.g., 1 MW for reserves, 0.1 MW for energy), block consistency, and specific energy-to-power ratios ( $\gamma$ ) and blocked power factors ( $\phi$ ) required for reserve products (e.g., FCR requires 1.25 MW of power capacity per 1 MW of reserve in intraday stages).
- Capacity Blocking: Constraints ensure that reserved balancing capacity blocks the necessary portion of usable storage capacity for subsequent energy trading.

2. Opportunity-Cost-Based Bidding Strategy

A core component of the methodology is the derivation of bid volumes and minimum prices based on opportunity costs (OC).

Residual Capacity Optimization: For a specific market stage, the framework solves a "residual" optimization problem for all future markets. This determines the maximum feasible bid quantity ( $L_b$ ) and the value ( $V_b$ ) of the remaining markets if the current market is not participated in.
OC Calculation: The opportunity cost is calculated as the foregone profit from future markets caused by reserving capacity for the current market. This accounts for the "blocking" effect of reserve commitments on future trading capabilities.
Bid Price Formulation: The minimum bid price is derived by distributing the opportunity cost over the offered quantity and time duration.
- FCR: Uses a single price covering the block's OC (pay-as-cleared).
- aFRR: Employs a "ladder" bidding strategy, splitting capacity into multiple bids with prices ranging from a conservative floor (adjusted by a risk factor $\theta$ ) to aggressive prices based on price forecasts, to balance acceptance probability and margin.
- DAA & IDA: Determines trading direction (buy/sell) and sets prices based on the opportunity cost floor, adjusted for market granularity and settlement rules (pay-as-bid vs. pay-as-cleared).

3. Market Clearing and Settlement

The framework simulates market clearing where bids are accepted or rejected based on the Market Clearing Price (MCP).

Allocation: Accepted bids become fixed parameters for subsequent optimization steps.
Re-optimization: If allocations result in open positions or infeasibilities (e.g., due to imbalance energy requirements), the model re-optimizes with slack variables to derive a feasible target schedule for the next stage (e.g., intraday).

Key Contributions

The paper identifies four primary contributions:

Sequential Multi-Market Formulation: A unified framework that explicitly models the chronological sequence of GCTs and product durations for FCR, aFRR, and energy markets, moving beyond simultaneous optimization assumptions.
Data-Driven Market Coupling: The integration of rolling price forecasts allows for adaptive re-optimization at each stage, utilizing real-time information from cleared markets.
Opportunity-Cost-Based Bidding: The derivation of feasible bid volumes and minimum bid prices from residual capacity optimization, ensuring economic consistency and preventing infeasible commitments.
Realistic Validation: Demonstration of the framework's consistency and feasibility through a single-day case study, confirming alignment between day-ahead plans and realized dispatch.

Results and Validation

The framework was validated using a representative operating day (April 1, 2024) for a BESS with 3.65 MW power and 7.3 MWh capacity.

Schedule Evolution: The study showed that schedules evolve significantly between optimization stages (e.g., 8:00, 9:00, 12:00, 15:00) as forecasts are updated and markets clear.
Feasibility: The framework successfully generated consistent and technically feasible schedules. The State of Charge (SoC) trajectories remained within technical limits required for FCR provision throughout the simulation.
Deviation Reduction: Comparisons between planned and realized dispatch indicated that intraday re-optimization substantially reduced deviations compared to day-ahead plans. The closer the schedule was to the actual dispatch time, the more accurate it became.
FCR Impact: The results highlighted how FCR allocation blocks specific SoC ranges (preventing operation near 0% or 100% SoC), which the model correctly accounted for in subsequent aFRR and energy market bids.

Significance

The paper claims that its primary significance lies in addressing the gap between idealized optimization models and the practical realities of sequential market participation. By explicitly modeling Gate Closure Times, rolling forecasts, and opportunity costs, the framework provides a method for operators to determine feasible bid quantities and prices that are consistent with real market timing.

The authors assert that this approach enhances the realism and profitability of BESS operations by minimizing deviations between planned and realized dispatch and avoiding the overestimation of revenues common in single-stage or non-data-driven models. The work establishes a foundation for coordinated participation across balancing and energy markets without requiring complex learning-based models that are sensitive to data shifts, instead relying on robust optimization and opportunity-cost pricing.