Mixed-Integer Optimization for Integrated Energy Systems

38 concepts40 papers1 proposal

Modern energy grids are evolving from simple, unidirectional electricity delivery networks into complex Integrated Energy Systems (IES). These systems simultaneously manage multiple energy carriers—such as electricity, heat, hydrogen, and synthetic natural gas—across microgrids, smart buildings, and regional networks.

The core challenge in managing an IES lies in its hybrid nature: it is governed by both continuous physical laws (such as fluid dynamics, thermodynamics, and power flow) and discrete operational logic (such as turning a turbine on or off, routing a vehicle, opening a valve, or making an infrastructure investment decision). Mixed-integer programming provides the mathematical bridge required to co-optimize these disparate domains.

Why Mixed-Integer Formulations are Essential

To find the most cost-effective or lowest-carbon operational strategy, engineers cannot rely on standard linear programming. Continuous optimization models assume that every variable can vary smoothly. However, physical energy systems are full of non-convexities and mutually exclusive states:

  • A battery cannot simultaneously charge and discharge.
  • A combined heat and power (CHP) plant has a minimum stable generation level and cannot operate at $1\%$ capacity; it is either off or running above a specific threshold.
  • A district heating network has physical valves that are either open or closed.

Mixed-Integer Programming (MIP)—which includes Mixed-Integer Linear Programming (MILP), Mixed-Integer Quadratic Programming (MIQP), and Mixed-Integer Nonlinear Programming (MINLP)—solves this by introducing binary variables ($0$ or $1$) to represent these discrete states.

By combining these binary variables with continuous variables representing energy flows, the optimization problem can explicitly model the combinatorial complexity of the system. Commercial solvers then use branch-and-bound algorithms to systematically prune the search space, guaranteeing mathematical optimality or providing a known bound on how close a solution is to the global optimum.

Bridging Thermal and Electrical Networks

One of the most powerful applications of this framework is decoupling the rigid dependencies between electricity generation and heating. In traditional systems, a CHP plant must produce heat when residential demand spikes, even if electricity prices are low or renewable generation is abundant, leading to inefficiencies.

By formulating the network topology as a mixed-integer problem, operators can dynamically reconfigure the system to unlock operational flexibility. For example, Yixun Xue et al. demonstrated in Reconfiguration of District Heating Network for Operational Flexibility Enhancement in Power System Unit Commitment that integrating district heating network valve configurations into a unified MILP framework allows operators to re-route heat in real-time. This structural flexibility significantly reduces wind power curtailment by utilizing the thermal inertia of the network as a buffer.

Furthermore, the temporal resolution of the underlying data dictates the accuracy of these models. In A Study on Operational Pattern and Heat-storage Capacity of Micro Co-generation System for House Use Based on Minute-order Hot Water Load, Yoshitsugu Wakazono et al. showed that modeling high-resolution, minute-by-minute hot water loads within a mixed-integer framework is essential. Traditional models using hourly averages smooth out demand spikes, severely underestimating the required heat storage capacity needed for real-world residential deployment.

[Discrete Decisions: On/Off, Valve Open/Closed]
                       │
                       ▼
┌──────────────────────────────────────────────┐
│       Mixed-Integer Optimization Engine      │◀─── [Weather & Demand Forecasts]
└──────────────────────────────────────────────┘
                       ▲
                       │
[Continuous Variables: Power Flow, Temperature]

Managing Storage and Demand-Side Flexibility

As intermittent renewables dominate the grid, storage assets must be precisely scheduled. MIP formulations excel at managing the discrete charging states and degradation boundaries of these assets.

In Cost Minimization of Charging Stations with Photovoltaics: An Approach with EV Classification, Wayes Tushar et al. utilized a mixed-integer framework to optimize energy trading schedules for electric vehicle (EV) charging stations. By categorizing EVs based on their willingness to discharge back to the grid, the model balances local solar generation and grid pricing against individual vehicle state-of-charge requirements.

To capture more complex physical realities, such as non-linear component degradation, researchers shift toward MIQP. For instance, Junzhe Shi et al. developed a health-aware energy management system in Health-aware energy management for multiple stack hydrogen fuel cell and battery hybrid systems. Their MIQP approach uses binary variables to track individual fuel cell stack on/off statuses, preventing rapid cycling and significantly reducing component degradation compared to collective control strategies.

Flexibility does not just come from physical batteries; it can also be harvested from consumer behavior. By treating building air conditioning and EV schedules as a "Joint Virtual Energy Storage" system, Xu Zhu 0007 et al. formulated an MILP in Optimal Scheduling Method for a Regional Integrated Energy System Considering Joint Virtual Energy Storage that minimizes operational costs and carbon taxes while explicitly penalizing deviations from user comfort boundaries.

Long-Term Planning and Multi-Period Transitions

Beyond daily economic dispatch, mixed-integer formulations are crucial for long-term infrastructure planning under uncertainty. Investment decisions are inherently discrete—an operator either builds a transmission line or does not.

In Optimal Grid Layouts for Hybrid Offshore Assets in the North Sea under Different Market Designs, Andreas Themelis et al. applied a multi-period, stochastic MILP to optimize offshore wind grid layouts. Their model balances long-term capital investments against daily operational welfare across different market designs, proving that planning under nodal market assumptions maximizes overall social welfare.

On an industrial scale, Kaliaperumal Karthikeyan et al. leveraged multi-period MILP in Pareto optimal solutions for decarbonization of oil refineries under different electricity grid decarbonization scenarios to map out decades-long decarbonization pathways for oil refineries. The model optimizes the exact schedule for adopting carbon capture and green hydrogen technologies, demonstrating how investment timelines must shift in response to evolving carbon taxes and grid decarbonization rates.

Overcoming Computational Complexity

While MINLP models offer the highest fidelity for capturing non-linear chemical and thermodynamic relationships—such as those in hydrogen-electricity coordination (Xiaofeng Li et al. in Optimization of Integrated Energy System Considering Electricity and Hydrogen Coordination in the Context of Carbon Trading) or waste-to-energy systems (Junjie Zhong et al. in Optimal and Sustainable Scheduling of Integrated Energy System Coupled with CCS-P2G and Waste-to-Energy Under the “Green-Carbon” Offset Mechanism)—they are notoriously NP-hard and computationally expensive.

To scale these solutions for real-time operations, the research direction has shifted toward hybrid or distributed architectures:

  1. Bi-Level Hybrid Frameworks: In Co-Optimization of Capacity and Operation for Battery-Hydrogen Hybrid Energy Storage Systems Based on Deep Reinforcement Learning and Mixed Integer Programming, Tiantian Qian et al. combined Deep Reinforcement Learning (DRL) with MIP. The DRL agent handles the high-level capacity configuration, while the inner MIP loop provides fast, precise operational feedback, drastically reducing computation time without sacrificing solution quality.
  2. Distributed Optimization: For residential applications where data privacy is paramount, the authors of A Reliability-aware Distributed Framework to Schedule Residential Charging of Electric Vehicles decoupled the global optimization problem into localized MIQP subproblems. By utilizing the Alternating Direction Method of Multipliers (ADMM), individual agents optimize local EV charging intervals based on time-of-use pricing, using iterative message-passing to ensure grid reliability without centralizing sensitive consumer data.

Key Foundational Work

A Study on Operational Pattern and Heat-storage Capacity of Micro Co-generation System for House Use Based on Minute-order Hot Water Load (2001) — contributed 1 concept elements to this direction.
Supply Chain Optimization in Continuous Flexible Process Networks (2000) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_mixed_integer_linear_programming_formula_e05f06c7,
  title = {mixed-integer linear programming formulation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/mixed-integer-linear-programming-formulation/rdr_e05f06c73bf84f25b1cb2ec401898285},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_mixed_integer_linear_programming_formula_e05f06c7_v1,
  title = {mixed-integer linear programming formulation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/mixed-integer-linear-programming-formulation/rdr_e05f06c73bf84f25b1cb2ec401898285/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake