Spatial network model
A topology-aware representation connects substations, feeders, distributed assets, loads, geography, weather exposure, ownership, and operating zones in one versioned context.
Energy systems
Synqara
A governed control architecture for adaptive energy grids.
Research thesis
A grid becomes adaptive only when every decision can be traced from observation through constraint to outcome.
Synqara is a research programme for coordinating distributed energy systems through a common operating context. The proposed system joins field telemetry, topology, state estimation, forecasting, constraint-aware optimisation, operator authority, and replayable evidence.
The programme is not presented as an autonomous production grid or a demonstrated quantum-advantage system. Its immediate engineering task is to establish trustworthy data, classical baselines, safe control boundaries, and testable acceptance criteria before progressively increasing operational responsibility.
06Control stages
Observe through verified outcome
04ABQS responsibilities
Intelligence, records, optimisation research, spatial context
03Validation environments
Replay, simulation, bounded pilot
02Operating modes
Advisory and authorised closed loop
The operating deficit
Grid data is abundant. Grid context is not.
A useful decision requires more than a demand forecast. It requires a current topology, qualified telemetry, equipment limits, market rules, weather context, operator authority, and a safe path back when the model is wrong.
Distributed generation, storage, flexible loads, electric vehicles, and local markets increase the number of states a network operator must understand. Measurements arrive at different rates and levels of quality. Assets may be incorrectly mapped, communications may be delayed, and a locally efficient action may violate a system-level constraint.
Synqara treats this as an observability and control problem. Forecasting is one component inside a larger evidence chain that must show what the system believed, which constraints were active, who authorised the action, what was dispatched, and what happened next.
Proposed control loop
Move from raw telemetry to a replayable operating decision.
Observe the network
Ingest telemetry, weather, market signals, asset status, protection state, and operator inputs without assuming that every source is complete or current.
Qualify the evidence
Check timing, identity, units, calibration, missing values, communication delay, and topology assignment before a measurement is allowed to influence control.
Estimate and forecast
Reconstruct the best available network state and forecast demand, renewable output, congestion, and uncertainty across the relevant operating horizon.
Optimise under constraints
Compare feasible actions against classical optimisation baselines. Quantum methods enter only as bounded research candidates with declared instances and comparison criteria.
Authorise and execute
Present the action, assumptions, risk, responsible actor, and fallback route. Execution remains governed by operational roles, protection systems, and market authority.
Verify the outcome
Compare the observed result with the expected state, preserve overrides and exceptions, and feed measured error back into models and operating policy.
System architecture
Five responsibilities, separated so each can be tested.
Data quality and state estimation
Measurement validation, bad-data detection, time alignment, observability analysis, and state estimation establish whether the system knows enough to recommend an action.
Forecast and anomaly intelligence
Models estimate demand, renewable output, congestion, asset behaviour, and uncertainty while retaining the training scope, baseline, version, and conditions under which a result is valid.
Constraint-aware optimisation
Classical solvers remain the reference path for dispatch, storage, curtailment, and flexibility studies. Experimental quantum routines must be benchmarked against the same declared problem and constraints.
Authority and evidence layer
Permissions, approvals, model versions, input evidence, recommended actions, execution records, settlement context, and measured outcomes form one replayable decision record.
Acceptance framework
Evaluate the complete control chain, not one model score.
A forecasting gain has little value if data quality is unknown, network constraints are violated, the action arrives too late, or the result cannot be reconstructed after an event.
- Forecast quality
- Error and calibration
- Network safety
- Constraint compliance
- Operational value
- System outcome
- Control behaviour
- Latency and fallback
- Evidence quality
- Replay and lineage
Report absolute and normalised error by horizon, asset class, weather regime, and operating condition, together with uncertainty calibration and baseline comparison.
Measure voltage, thermal, reserve, protection, stability, and market-rule violations in normal cases, degraded telemetry, and contingency scenarios.
Track losses, curtailment, unserved energy, flexibility delivered, switching burden, and the distribution of benefit rather than presenting one aggregate efficiency claim.
Record decision latency, failed integrations, operator overrides, protection interventions, safe-state transitions, and recovery time.
An independent reviewer should be able to reconstruct inputs, topology, model and solver versions, constraints, authority, action, and observed outcome.
Bounded deployment contexts
Different grids require different authority, safety, and evidence models.
Transmission and system operators
Planning, congestion, reserve, and system-state studies where recommendations must fit established control-room procedures, protection responsibilities, and reliability rules.
Distribution utilities
Feeder observability, distributed-energy coordination, loss analysis, voltage management, and maintenance prioritisation under incomplete instrumentation.
Microgrids and campuses
A contained environment for validating forecasting, storage dispatch, resilience modes, operator handoff, and fallback behaviour before broader network responsibility.
Aggregators and market participants
Flexibility and settlement workflows that require explicit consent, metering quality, market eligibility, counterparty rules, and auditable delivery evidence.
Research stage
Prove observability and advisory value before autonomous control.
Synqara is developing prediction models and a spatial digital twin against declared grid models, datasets, and classical baselines. Production dispatch, live market settlement, and autonomous control remain outside the present validation scope.
The next validation package combines historical replay, simulated contingencies, integration results, operator review, and a bounded pilot with pre-agreed safety and acceptance criteria.