Real-time source selection is useful only after the owner has decided which obligations take priority. A controller cannot resolve a conflict between the tenant's emergency reserve, the landlord's bill-reduction target and a third party's dispatch rights. It can only implement the rules it is given. The development work is to make those rules technically feasible and commercially consistent before the software starts optimizing them.
The campus problem
Illustrative scenario, not a claimed client result.
The campus battery is nearly full at noon. A price forecast suggests discharging now is profitable. A protected tenant needs a fixed reserve, afternoon cooling demand may set the month’s charge, and a second tenant has bought demand-response flexibility. The owner has three claims on one asset. A software procurement will not resolve which promise has priority.
The cheapest immediate action can increase the total cost
An energy price signal is one input, not the complete objective. The controller may need to preserve storage for a later demand peak, retain outage reserve, respect equipment limits and avoid a costly recharge period. Forecast errors matter because the best decision with hindsight is not necessarily available to the operating system. The owner should judge dispatch against the information available at the time and the service obligations it was required to protect.
Asset ownership adds another constraint. A third-party battery owner may have performance guarantees or usage limits. A generator may have dispatch and fuel provisions. A tenant may prohibit certain interruptions. The control specification needs to translate those contracts into allowed actions, priorities and exception handling. If two counterparties have conflicting dispatch rights, the remedy is a contractual decision, not a more advanced optimization algorithm.
The Texas context
Distributed-energy planning tools can optimize asset size and dispatch, while ERCOT has distinct demand-response participation routes. Neither replaces a site operating agreement or a qualified control system. The actual tariff, interconnection permissions and tenant service commitments determine what a Texas campus controller may do.
Allocate the reserve before counting the trading opportunity
For illustration, assume an operating battery has 20 MWh of usable energy at the relevant point, of which 12 MWh is reserved for a protected duty. At most 8 MWh remains for other uses at that moment, before any additional constraints. The example does not establish a system design. It shows why the commercial model cannot count all 20 MWh toward arbitrage while separately promising the reserve. A discharge may also leave insufficient time or power to restore the reserve before the next protected period.
Sitebraid would coordinate a priority schedule with the operator, tenants, asset owners and qualified controls team. The system should be tested against forecast error, communication loss, asset unavailability and competing requests. Define when the operator overrides the optimizer and how those events are recorded. Evaluate net savings after degradation, fuel, fees and changed billing peaks. An attractive simulation is only a candidate operating plan until the actual constraints, measurement and failure behavior have been demonstrated.
A normal-day trace should show forecast, constraint, action and resulting bill effect. A stressed-day trace should add low renewable output, a grid event, an asset trip and the recovery period. If the controller saves money before the event but cannot restore the protected reserve afterward, the daily optimization has failed the campus operating objective.
Solutions and their tradeoffs
Conservative rule-based dispatch
Use a transparent import limit and protected reserve for initial operations.
Conditions to resolve: Test edge cases; simple rules still need technical approval.
Forecast-based optimization
Schedule assets over a defined forward horizon with operating constraints.
Conditions to resolve: Measure forecast error and show a safe fallback before enabling autonomous changes.
Delegated asset operation
Let a specialist operator optimize an asset within campus limits.
Conditions to resolve: Resolve conflicts between asset-owner revenue and campus service obligations.
Dispatch hierarchy to agree before commissioning
Swipe or scroll to compare all columns.
| Priority | Constraint | Evidence |
|---|---|---|
| 1. Safe operation | Protection, permitted topology and equipment limits | Qualified design and accepted controls |
| 2. Contracted continuity | Tenant load, cooling and protected reserve | Operator-approved service envelope |
| 3. Firm external obligation | Qualified response or export commitments | Compatible dispatch and settlement contracts |
| 4. Economic optimization | Avoided cost net of losses and wear | Back-tested tariff and price model |
This is the development problem behind our energy cost and commitments challenge. Our Power & energization capability explains the scope Sitebraid can take on.
Putting the analysis to work
Establish the billing interval, price exposure, export limit, battery state-of-charge limits, degradation warranty, generator duty and fuel inventory. Record tenant service envelopes and market commitments. Identify communications latency, forecast errors, manual overrides, cybersecurity ownership and the mode to use if telemetry fails. The control authority must match the physical meter and contracts, including which parties can change setpoints.
How Sitebraid would support this decision
Write the operating policy
Define protected demand, discretionary actions and escalation authority. An economic signal cannot override an operator’s safety constraint or an unavailable export permission. Put the priority into contracts as well as software requirements.
Replay difficult operating days
Test simultaneous high demand, low solar, an asset trip, missing prices and a failed tenant reduction. Examine recovery, not only the event. The most informative result may be the value of keeping reserve unused.
Commission modes and boundaries
Qualified teams test normal, constrained and outage modes, telemetry loss, manual control and restoration. Islanding and resynchronization require specific engineered capability; they are not implied by an energy-management platform.
Reconcile performance to the bill
Compare measured imports, asset duty and settlement with the model. Sitebraid brings the operator, tenant, provider and asset-owner decisions together and closes the development interfaces. The operating team owns live control and subsequent optimization.
How we protect the decision
Define who benefits from arbitrage and who pays for battery replacement. Do not count the same dispatch twice as avoided demand and an incompatible external service. Preserve audit logs of constraints and overrides. Changes in tenant duty or import limits should trigger an operating-model review, not just a new optimization parameter.
The result the owner should require
The result is a tested dispatch policy with approved authority, telemetry, fallback and commercial allocation. Report performance against a documented counterfactual and all material costs. A dashboard showing price-responsive operation is not evidence of lower total electricity cost or improved resilience.
How to measure progress
- Constraint violations and unapproved dispatches.
- Net avoided cost after losses, degradation and fees.
- Protected reserve and recovery demand after events.
- Forecast error, override response and telemetry availability.
The decision to take forward
Use automation to execute an agreed hierarchy of service, reserve and cost objectives. Establish control rights and test exception behavior before monetizing the asset's flexibility. The successful result is lower supported operating cost without quietly spending the reliability another party has already paid for.
If your microgrid dispatch is chasing the cheapest source each hour, our Power & energization work sets a dispatch plan around what your tenants need.
Continue with Diversify the supply, not the unpriced risks. to examine the connected decision.