Applications › Dispatch, allocation and grid behaviour
Applications · Energy & smart grids
Dispatch, allocation and grid behaviour
Optimisation and simulation for power systems with distributed generation, storage and demand that responds.
The problem
A grid with rooftop solar, batteries and flexible demand is no longer a system with a few large controllable generators and passive load. It is a large number of small agents, each responding to price and to each other, under physical constraints that must hold at every instant.
That makes it two problems at once: a hard constrained optimisation (what should be dispatched when) and a complex adaptive system (what will the participants actually do). We work on both sides, and they need different methods.
Unit commitment and dispatch cast as constrained optimisation and formulated for annealing or hybrid solvers; agent-based simulation of participant behaviour; and materials simulation feeding into storage technology from the chemistry side.
Mature. Utilities run large MILP dispatch models that work. The opening is in problem classes that scale badly — high penetration of distributed resources, many discrete decisions, short re-optimisation windows.
Emerging to active. Methods are established in the other pillars; grid-specific application work is at an early stage and would benefit from a utility partner.
Distribution network planning; dispatch under high renewable penetration; storage sizing and siting; demand-response programme design.
Network and load data at usable resolution; a defined operating objective and constraint set; and a partner able to compare against their existing dispatch tooling.
Evidence
- “Modeling thermal dynamics of lithium-ion batteries via fractional action memory effects.” Journal of Energy Storage 160, 121982 (2026). doi
- “A fractal electrochemical porous model for lithium-ion batteries.” Journal of Energy Storage 141, 118960 (2026). doi
- Optimisation methods are shared with logistics & optimisation.
Talk to us about this
This is an area where we would welcome a utility or regulator partner — the methods are ready and the domain data is what is missing.
Last updated 14 September 2026.