Product

The Grid Coordination Engine.
Market infrastructure, not a dashboard.

Three composable layers — feasibility, scarcity, fairness — operating continuously between retail commitments and physical network constraints.

Live demo → Use cases
Engine architecture

Inputs in. Coordination out.

The Grid Coordination Engine takes wholesale schedules and retail commitments as given inputs and produces local balancing signals as outputs. It never overrides wholesale; it operates in the space wholesale cannot see.

Inputs

What flows in

Wholesale schedules

Planned positions, dispatch targets, and forward commitments. Treated as constraints, not overridden.

Retail commitments

Flexibility requests with time windows, power ratings, priority tiers, and maximum willingness to pay.

Network state

Feeder voltage, thermal limits, congestion indices, dynamic operating envelopes — in real time.

Core mechanism

AMM coordination kernel

Feasibility layer

Computes the admissible action set — what local corrections are possible within wholesale commitments and network limits.

Scarcity layer

Maps composite scarcity to bounded local price signals. Convergent within each interval. No unbounded spikes.

Fair Play layer

Stateful fairness memory. Tracks historical service delivery and adjusts future priority. Compatible with convergence guarantees.

Outputs

What flows out

Local price signals

Bounded buy/sell prices per feeder per interval. Consumable by DERMS, aggregators, and smart devices.

Access allocations

Who gets served, when, and how much. Auditable. Traceable. Regulation-ready.

Settlement artefacts

Cleared quantities, schedules, fairness ledger, and cost allocation records for billing and reporting.

Feature set

Six primitives. One mechanism.

Each feature is a component of the same underlying control logic — not a separate module bolted on.

Pricing

Dynamic AMM pricing

Continuous scarcity inference drives buy and sell prices per feeder per interval. Prices are computed from physical state, not discovered through bid-based clearing. Bounded by design — no VoLL dependency.

BP(t,n) = BP_base + F(Δ, α_stability) · [p̲, p̄]
Allocation

Fair Play algorithm

Shortage-budget state tracks cumulative under-service per participant. Under-served participants receive higher allocation priority in subsequent intervals. Service ratios converge toward parity over time.

z_{i,t+1} = Π[0,z_max][(1−β)z_{i,t} + ℓ_{i,t}]
Network

Voltage shadow pricing

Measured feeder voltage acts as a direct physical input to scarcity inference. Undervoltage triggers higher buy signals; overvoltage recruits flexible demand. No centralised OPF required for local coordination.

α_network = exp(−θ·ΔV) · exp(−φ·cong)
Cost recovery

Shapley cost allocation

Non-fuel costs (CapEx, OpEx) are recovered through a Shapley-based channel linked to each generator's marginal contribution to served demand. Revenue anchored to physical contribution, not market position.

φ_i = Σ [v(S∪i) − v(S)] / |S|!(n−|S|−1)! / n!
Products

Subscription service tiers

Participants enroll into service tiers that define their access rights, priority levels, and exposure to scarcity signals. Retail products can be layered on top without changing the underlying coordination logic.

QoS tier → Fair Play weight → allocation priority
Governance

Audit and traceability

Every allocation decision is logged with its inputs — scarcity state, fairness weights, feasibility constraints. Regulators can verify outcomes. Participants can audit their treatment. Designed for regulated infrastructure.

Full decision trace per interval · regulation-ready
Integration

Works with existing infrastructure.

The coordination layer integrates into existing operational stacks — it does not require replacing SCADA, DERMS, or wholesale systems.

01

DNSP constraint ingestion

Consumes feeder topology, voltage time series, thermal limits, and DOE profiles via API. DNSP-specific adapters handle schema differences. Output: normalised constraint envelope per feeder.

02

Wholesale schedule interface

Ingests AEMO / system operator dispatch targets as upstream constraints. Wholesale commitments define the admissible action set for the coordination kernel — not overridden, respected.

03

DERMS and aggregator API

Publishes local price signals and access allocations via REST + webhooks. DERMS consumes signals as economic inputs to dispatch decisions. No direct asset control — coordination by price.

04

Retail and settlement

Produces cleared quantities, fairness ledger entries, and Shapley cost allocation records in settlement-ready format. Compatible with existing billing and reporting infrastructure.

Signal interface — example output
feeder_idNW-047
interval_start14:35 UTC
composite_scarcity0.61
buy_price_GBP_MWh87.40
sell_price_GBP_MWh112.20
available_export_kW142
fair_play_activetrue
convergence_q0.946
confidencehigh

See it running.

The live demo shows market clearing, pricing, and dispatch logic in action on a real system model.

Live platform → Use cases Book a walkthrough