Context
Millions of distributed assets such as EVs (electric vehicles), heat pumps, batteries, and smart appliances could provide electricity system flexibility, but only if they are coordinated in real time across tariff structures, grid constraints, and customer preferences.
Challenge
The operational need is to coordinate large volumes of distributed assets in real time while accounting for tariff structures, grid constraints, and customer preferences, so that flexibility can be made available and dispatched when needed.
AI Solution
Kaluza uses a machine learning driven optimisation engine to continuously forecast flexibility availability across millions of devices and dispatch flexibility in real time to: absorb excess renewable generation, shifting demand away from peak periods, and maximising value for consumers.
Impact & evidence
Impact is assessed using metrics including flexibility dispatched in TWh (terawatt-hours) per year with curtailment avoided, customer value delivery in £ (pounds) per customer per year via dynamic tariffs and demand response payments, dispatch reliability as a percent of dispatched events executed without customer discomfort, and carbon intensity reduction measured in gCO₂/kWh (grams of carbon dioxide per kilowatt-hour).
Implementation / Adoption
Replication requisites include smart meter infrastructure and device connectivity with IoT (Internet of Things) integration, real-time grid signal integration, customer opt-in and consent management, a regulatory framework supporting demand-side flexibility programmes, and OEM (original equipment manufacturer) device integrations.
Deployment would be phased: a proof of concept lasting 3 to 6 months in shadow run, a pilot lasting 3 to 9 months with a small cohort, and a scale-up lasting 12 to 24 months expanding to tens of thousands of devices with multi-asset orchestration; prerequisites also include standard integration using APIs (Application Programming Interfaces), SDKs (Software Development Kits), or connectors, significant client-side data preparation, and custom development or joint engineering effort.
Data
Customer-provided and externally sourced inputs include real-time smart meter data, device telemetry such as EV state of charge, heat pump operation, battery status, and solar generation, grid signals such as balancing mechanism dispatch, frequency, and wholesale prices, weather forecasts including temperature, solar irradiance, and wind speed, and customer preferences and tariff structures.
Future enhancements
Ongoing enhancements include continued iteration on device-level optimisation by category, such as EV, battery, and HVAC (heating, ventilation, and air conditioning), alongside multi-device co-optimisation to support whole-home and VPP (virtual power plant) or portfolio-level performance.