Making Aggregations Reliable: Realizability Guarantees for Battery Fleets with Heterogeneous Power and Energy Limits
Mazen Elsaadany, Mads R. Almassalkhi · 2026
Read Making Aggregations Reliable: Realizability Guarantees for Battery Fleets with Heterogeneous Power and Energy Limits by Mazen Elsaadany, Mads R. Almassalkhi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-01562.txt.
Introduction
Elsaadany and Almassalkhi make fleet battery plans that can really be run. A fleet of batteries can join grid planning and power markets as one unit. Planning each unit on its own is too slow at scale. Past aggregate models use convex relaxations to stay tractable. They skip the rule that a unit cannot charge and discharge at once. Plans then fail when split back onto real units, wasting the work.
Their composite model keeps plans feasible at the unit level. Simple linear rules say when an aggregate plan splits safely. Each unit's power and energy limits are respected in the split. A priority-based controller carries out that split in practice. Tests in a unit-commitment setting give feasible plans. Solve times stay near flat as the fleet grows.
Catalog record
| CREATOR | Mazen Elsaadany, Mads R. Almassalkhi |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.01562
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Making Aggregations Reliable: Realizability Guarantees for Battery Fleets with Heterogeneous Power and Energy Limits Mazen Elsaadany, Mads R. Almassalkhi Abstract Aggregated battery energy storage systems (BESS) enable large fleets of heterogeneous battery elements to participate in system-level optimization and electricity markets. Scheduling each element independently is computationally impractical at scale. While many aggregate battery models rely on convex relaxations, they often ignore element complementarity constraints, leading to dispatch solutions that may be infeasible when implemented on individual battery elements. This paper develops a realizable composite battery model for parameter-heterogeneous BESS fleets that guarantees feasibility at the element-level while preserving computational tractability. We derive simple linear conditions under which aggregate charging and discharging trajectories can be safely disaggregated while respecting individual power limits, energy limits, and complementarity constraints under a priority-based controller. Numerical experiments in a unit-commitment setting demonstrate that the proposed realizable composite battery formulation produces feasible dispatch solutions. Solve times are effectively independent of system size, unlike micro-model mixed-integer formulations. Solutions obtained from the proposed formulation converge to the optimal benchmark as control granularity is refined. Additional studies illustrate the robustness of the framework to moderate violations of key modeling assumptions, including heterogeneous power-to-energy ratios. I Introduction FERC Order No. 2222 enabled aggregated battery energy storage systems (BESS) fleets to participate in wholesale markets and provide energy and ancillary services [ 1 ] . However, explicitly scheduling and dispatching each element in a large fleet is computationally impractical [ 2 ] . This motivates aggregating and coordinating geographically distributed batteries as virtual power plants (VPPs) that reduce problem complexity while still producing dispatch decisions that can be physically delivered by the underlying fleet. Most aggregate modeling approaches characterize each battery element’s flexibility set and define the fleet-level flexibility as the Minkowski sum of these individual sets [ 3 , 4 , 2 , 5 ] . Since exact Minkowski sums are intractable at scale, the literature focuses on tractable inner or outer approximations. Examples include zonotope-based approximations, which scale well but can be conservative for heterogeneous populations [ 6 , 3 ] , homothetic and union-based constructions that improve accuracy by combining affine images of a base set [ 7 ] , and methods that derive the exact aggregate feasible set but require simplifications for long time horizons [ 4 ] . Recent vertex-based methods construct compact inner approximations from a structured subset of the vertices of the Minkowski sum [ 5 ] . Finally, multi-battery models represent heterogeneous fleets using a small number of representative devices to trade off accuracy and complexity [ 2 ] . A key challenge is that most aggregate models rely on convex representations and therefore sidestep the non-convex complementarity constraint that prevents simultaneous charging and discharging at the element level. Ignoring this constraint can produce schedules that appear feasible at the aggregate level but are infeasible when implemented on individual elements [ 8 , 9 ] . In particular, convex relaxations can underestimate state of energy (SOE), leading to dispatches that cannot be realized without violating element constraints [ 9 ] . Existing approaches handle complementarity in several ways. Mixed-integer formulations enforce it exactly but scale poorly with fleet size and horizon length [ 10 , 8 ] . Other work derives conditions under which convex relaxations don’t violate complementarity, typically under strong assumptions on costs, prices, or parameter values [ 11 , 12 , 13 , 14 , 8 ] . Penalty-based formulations discourage simultaneous charging and discharging by assigning a penalty term in the objective; however, the resulting guarantees are typically tied to a specific application (e.g., DCOPF) and do not readily generalize to other optimization settings [ 15 ] . Robust convex restrictions can guarantee feasibility by bounding worst-case energy evolution, often at the expense of increased conservatism over longer horizons [ 9 ] . In summary, there is a lack of aggregate battery models that both (i) scale to large heterogeneous fleets and (ii) guarantee realizable dispatch in the presence of charging–discharging complementarity constraints. Other work addressed this challenge for homogeneous fleets by introducing a linear composite model with a priority-based disaggregation strategy that guarantees feasibility at the element level while preserving convexity [ 16 ] . Extending such guarantees to heterogeneous fleets remains an open problem. This paper addresses this gap by developing a realizable aggregate modeling framework for heterogeneous BESS fleets. The contributions are: 1. We extend the composite battery framework to heterogeneous elements and derive a linear aggregate model that captures heterogeneous power and energy limits. 2. We generalize the priority-based disaggregation strategy and provide sufficient conditions under which any feasible aggregate dispatch is realizable while respecting element-level power limits, energy limits, and complementarity. 3. We characterize the resulting aggregate feasible region and show how accuracy improves with finer control granularity while maintaining the computational tractability needed for large-scale optimization. The remainder of the paper is organized as follows. Section II presents the element-level model and the composite battery formulation. Section III introduces the priority stack controller and derives sufficient conditions for realizable dispatch. Section IV evaluates the proposed model using numerical case studies, including scalability and the effects of relaxing key assumptions. Section V concludes and outlines future research directions. II Composite Battery Modeling To enable system-level optimization without explicitly modeling every battery element, we introduce a composite battery model that aggregates the power and energy states of a fleet of N independently controllable battery elements. We first define the element-level model, then derive the corresponding composite model and discuss the challenges that arise when mapping aggregate charging/discharging power to individual battery elements. II-A Battery Elements Each element i\in\{1,\ldots,N\} is operated in discrete control time steps indexed by l , with duration \delta t . The charging and discharging powers are denoted by P^{i}_{\text{c}}(l) and P^{i}_{\text{d}}(l) , with efficiencies \eta^{i}_{\text{c}} and \eta^{i}_{\text{d}} , respectively. The SOE of element i at time step l is E^{i}(l) . Each element has power limits P^{i}_{\text{c,max}} , P^{i}_{\text{d,max}} , energy limit E^{i}_{\max} , and an initial SOE E^{i}_{0} . The element constraints are summarized by \displaystyle E^{i}(l+1)=E^{i}(l)+\delta t\eta^{i}_{\text{c}}P_{\text{c}}^{i}(l)-\delta t\frac{1}{\eta^{i}_{\text{d}}}P_{\text{d}}^{i}(l) (1a) \displaystyle E^{i}(0)=E^{i}_{0} (1b) \displaystyle 0\leq E^{i}(l+1)\leq{E}_{\text{max}}^{i} (1c) \displaystyle 0\leq P_{\text{c}}^{i}(l)\leq{P}_{\text{c,max}}^{i} (1d) \displaystyle 0\leq P_{\text{d}}^{i}(l)\leq{P}_{\text{d,max}}^{i} (1e) \displaystyle 0=P_{\text{c}}^{i}(l)P_{\text{d}}^{i}(l), (1f) for all i=1,\ldots,N and l=0,\ldots,L-1 . II-B Composite Battery The composite battery aggregates the N elements and is operated on coarser scheduling time steps of duration \Delta t=M\delta t , where M\in\mathbb{N}^{+} . We index scheduling steps by [k] , where k=0,\ldots,K-1 and L=MK . Let P_{\text{c}}[k] and P_{\text{d}}[k] denote the composite charging and discharging powers. A piecewise-constant composite schedule \{P_{\text{c}}[k],P_{\text{d}}[k]\}_{k=0}^{K-1} is realized by element-level actions if \displaystyle\sum_{i=1}^{N}P_{\text{c}}^{i}(l)=P_{\text{c}}[k]\quad\land\quad\sum_{i=1}^{N}P_{\text{d}}^{i}(l)=P_{\text{d}}[k], (2) for all l with k=\lfloor\tfrac{l}{M}\rfloor . The composite SOE is defined as the sum of element SOEs at the start of each scheduling interval: \displaystyle E[k]=\sum_{i=1}^{N}E^{i}(Mk). (3) Under the assumption of homogeneous charging and discharging efficiencies ( \eta_{\mathrm{c}}^{i}=\eta_{\mathrm{c}} and \eta_{\mathrm{d}}^{i}=\eta_{\mathrm{d}} ), the composite SOE evolves as \displaystyle E[k+1] \displaystyle=E[k]+\Delta t\eta_{\text{c}}P_{\text{c}}[k]-\Delta t\frac{1}{\eta_{\text{d}}}P_{\text{d}}[k] (4a) \displaystyle E[0] \displaystyle=\sum_{i}E^{i}_{0}. (4b) according to the following dynamics, which follow directly from ( 1a ),( 1b ) and ( 3 ). Similarly, the composite power and energy limits follow from ( 1c )-( 1e ), ( 2 ) and ( 3 ): \displaystyle 0\leq \displaystyle P_{\text{c}}[k]\leq\sum_{i}P_{\text{c,max}}^{i} (5a) \displaystyle 0\leq