It has been said that although the power sector is only 7% of the US economy, it’s the first 7%. Every home, hospital, factory, data center, and transportation system depends on reliable electricity. At the same time, the industry is undergoing one of the most significant transformations in its history. Renewable generation is growing, electricity demand is increasing rapidly in many regions, and utilities are under pressure to improve reliability, affordability, and sustainability simultaneously. This transformation is a massive infrastructure challenge, which also makes it an enormous decision-making challenge.
Every day, utilities, grid operators, and planners must determine how to operate an increasingly complex energy system. The quality of those decisions can have a direct impact on operating costs, reliability, emissions, and customer outcomes. As complexity grows, many traditional approaches struggle to keep pace, building a stronger case for modern mathematical optimization.
A Problem at the Heart of Power Systems Operations
Two of the most important and continuous operational decisions in the electric grid are known as unit commitment and economic dispatch. In simple terms, grid operators need to decide which generating units should be online and when. In particular, for each hour of the day, they must determine which plants should start up, shut down, or continue operating so that electricity supply matches expected demand. The goal is to find a plan that satisfies all physical, regulatory, economic, and load service requirements simultaneously. Solving this problem to provable optimality may sound impossible, and for many years, it was considered to be so. It was only in 1999 that mixed integer programming (MIP) was shown to be able to solve the 7-day ahead unit commitment problem to proven optimality in an acceptable time period.
The variables and constraints involved are numerous. Generators have technical and economic limitations. Some units require significant time and costs to start. Others cannot increase or decrease production too quickly. Operators must maintain reliability reserves and account for expected demand patterns. Increasingly, they must coordinate conventional generation with storage systems and renewable resources whose output depends on weather conditions.
Since the adoption of MIP as the de facto standard for the day-ahead market engines, unit commitment has served as a benchmark for how the industry approaches optimization, making it a useful lens through which to examine the evolution of decision-support technology.
The Evolution of Solution Methods
Historically, power system operators relied on approaches such as dynamic programming and Lagrangian relaxation to solve unit commitment problems. These methods represented important advances and helped support power systems for many years. However, they were developed for a grid that looked very different from today’s. Power generation was more centralized, renewable penetration was lower, and operational requirements were often less complex.
As market rules, environmental regulations, and operational constraints evolved, a key challenge emerged: every new requirement increased the complexity of the solution process. Incorporating additional business rules often requiredsignificant modifications to the underlying algorithms. This is one reason why MIP has become the dominant framework for many modern unit commitment applications.
Why MIP Changed the Game
The appeal of MIP is not that it can solve large optimization problems. Its greatest strength is flexibility. When new operational requirements arise, they can often be incorporated by adding or modifying mathematical expressions within the model. The overall solution methodology does not need to be redesigned from scratch. This allows organizations to adapt more quickly as regulations, technologies, and market conditions evolve.
MIP also introduces a useful separation of responsibilities. Domain experts can focus on developing and maintaining the mathematical model that captures how their system operates. The optimization engine handles the complex search process required to identify high-quality solutions. This separation allows organizations to improve models without needing to continuously redesign optimization algorithms.
Another important advantage is transparency. MIP solvers provide quantitative measures of solution quality, such as the MIP gap, which indicates how close a solution is to the best-known bound. While no optimization technology can guarantee that every real-world problem will be solved to proven optimality within a fixed time limit, the MIP gap provides valuable information about solution quality and helps decision makers balance runtime and accuracy.
Finally, MIP does not force organizations to abandon decades of operational knowledge. Many practical applications combine advanced optimization technology with industry-specific heuristics to find feasible solutions, allowing organizations to draw on both mathematical rigor and domain expertise.
Navigating Complexity in the Evolving Grid
While modern optimization provides a framework for supporting decision making in power systems, utilities and grid operators continue to face a range of challenges that influence how these tools are deployed and used in practice. Some stem from the evolving nature of the grid itself, while others arise from regulatory and operational realities that have long characterized the industry. Together, they help define the environment in which power system decisions are made today.
One challenge is uncertainty. As renewable resources such as wind and solar account for a larger share of electricity generation, operators must make decisions using forecasts that are inherently imperfect. A plan that appears optimal in one scenario may require adjustment if weather conditions change. This has increased the value of optimization technologies that can evaluate higher numbers of scenarios and support decision making under uncertainty.
Companies like encoord use mathematical optimization, powered by Gurobi, to solve the large-scale mixed-integer programs at the core of integrated generation and transmission planning—where generation unit commitment, economic dispatch, and network feasibility constraints must be evaluated simultaneously across thousands of operating conditions. Utilizing Gurobi’s capabilities, encoord enables users to identify optimal solutions for resource allocation, system expansion, and risk mitigation, even in the face of rapidly changing market and technical conditions.
A second challenge is the complexity of underlying physics. Although many planning and operational problems can be expressed effectively as mixed-integer linear models, some applications involve highly nonlinear relationships arising from power flows, network constraints, and equipment behavior. Advances in optimization technology continue to expand the range of nonlinear problems that can be addressed, but these applications remain among the most demanding in the industry.
The industry must also navigate organizational and regulatory constraints. Utilities operate critical infrastructure, and changes to operational software often require extensive testing, validation, and approval processes. While these requirements help ensure reliability and compliance, they can also slow the adoption of algorithmic improvements and new modeling capabilities available in modern optimization platforms. It is incumbent on optimization technology providers to deliver solutions and features that enable adherence to these requirements in a timely fashion.
Finally, resilience has become a growing priority. Extreme weather events, wildfires, hurricanes, and other disruptions can damage critical infrastructure and require rapid restoration of service. In these situations, optimization can help decision makers determine how to design the grid for resiliency, prioritize repairs, deploy crews, and coordinate restoration activities.
Research by Gurobi’s Head of AI Innovation, Dr. Pascal Van Hentenryck, and others has shown how optimization and AI can speed up economic dispatch, unit commitment, planning, decentralized optimization, real-time risk assessment, topology switching, and security-constrained optimal power flow, to name a few use cases.
Taken together, these challenges illustrate an important reality: the future of power systems will require more than efficient algorithms. It will require optimization frameworks that can adapt to uncertainty, capture increasingly complex system behavior, evolve with changing regulations, and support resilient operations under both normal and extraordinary conditions.
A Core Technology for the Future of Energy
The energy transition is fundamentally increasing the number and complexity of decisions that utilities and system operators must make. They must determine how to integrate renewables, manage uncertainty, maintain reliability, invest in infrastructure, and recover from disruptions, all while controlling costs.
Traditional methods played a critical role in the development of modern power systems. However, as the grid becomes more dynamic and interconnected, flexibility, transparency, and scalability become increasingly important. Mathematical optimization—and MIP in particular—provides a framework for addressing these challenges in a systematic and adaptable way. Combined with state-of-the-art optimization engines such as Gurobi, it enables organizations to transform complex operational questions into data-driven decisions.
The transition to a cleaner, more resilient grid will require significant investment and innovation. Equally important is the ability to make effective use of these resources. As power systems continue to evolve, optimization is becoming a critical capability for turning complexity into operational and strategic advantage.
Want to be part of the conversation shaping the energy transition? Join us in Zurich this November for the Energy Innovation Summit, where we’re bringing together energy practitioners, researchers, software providers, utilities, asset operators, industrial energy users, and decision-makers from across Europe to discuss how optimization can help turn complexity into trusted decisions.
