ERCOT identifies large electronic loads such as data centers as materially different from conventional aggregate load. Their power-electronic front ends, controls, ride-through settings and fast recovery can affect voltage and frequency response. The model therefore has to represent both the network that supplies the campus and the behavior behind the meter.
Use the current ERCOT process, not an old checklist
ERCOT's large-load requirements have continued to evolve. Its current Large Load Integration page contains the applicable process guides, forms, submission routes and Dynamics Working Group survey. The page also directs projects to the current Planning Guide applicability rules rather than treating every site the same.
ERCOT's Large Load Modeling program explains why electronic loads need deeper dynamic representation and publishes informational generic EMT resources. Those examples are not substitutes for plant-specific controls, equipment data or the interconnecting provider's study instructions.
Document the current ERCOT Planning Guide path, the interconnecting TSP or DSP requirements, requested energization stages, study software versions and dynamic/EMT deliverables before fixing the model boundary.
Model the project from the POI into the campus
The project model should begin at the agreed POI and follow each material impedance to the load. A typical case may include the utility equivalent, 138 kV or 345 kV interconnection bus, parallel main transformers, medium-voltage distribution buses, tertiary windings, cables, capacitor banks, harmonic filters, on-site generation and BESS.
Keep redundant paths explicit
Data centers are often designed with A/B supply paths, tie breakers and standby equipment. These paths should not be merged into one equivalent if the operating configuration, N-1 duty or restoration sequence matters. Use separate equipment records and status values for each circuit, then create study scenarios for the intended normal and contingency states.
- Represent each parallel transformer with its own circuit ID and rating.
- Use manufacturer pairwise impedances for three-winding transformers.
- Model normally open ties in the case with their normal status preserved.
- Include cable charging and transformer reactive consumption where material.
- Define the utility source with the supplied short-circuit and sequence equivalents.
Break the demand into behavior, not just end use
At steady state, the total campus MW and MVAr must reconcile with the load forecast and auxiliary demand. For dynamics, the important question is how each portion responds to voltage and frequency disturbances.
| Data-center component | Useful model distinction | Data to request |
|---|---|---|
| Servers, UPS rectifiers and IT supplies | Power-electronic load with voltage/frequency ride-through, current limiting and recovery logic | Active/reactive control, undervoltage behavior, trip thresholds, recovery ramp and minimum operating voltage |
| Cooling fans and VFDs | Electronic drive and motor fraction; may not behave like constant power through a disturbance | Motor type, inertia, drive ride-through, restart logic and load torque |
| Pumps | Motor load with hydraulic torque characteristic | Motor rating, inertia, contactor/protection behavior and restart sequence |
| Chillers and compressors | Motor or drive-connected compressor block | Motor/drive model, stalling limits, protection and staged recovery |
| Lighting and miscellaneous | Static ZIP or electronic fraction | Constant impedance/current/power fractions and voltage dependence |
Do not assign 100 percent of the facility to a single constant-power load unless that behavior is supported. Constant-power demand can become numerically and physically severe as voltage falls, while real controls may current-limit, ride through, transfer, shed or recover in stages.
Keep operating scenarios separate from model parameters
The same campus may need commissioning, initial energization, partial build-out, ultimate load, emergency generation and maintenance configurations. Preserve the equipment model, then change status and dispatch through named scenarios so assumptions remain traceable.
Represent BESS, generation and reactive support by operating role
A BESS installed for power-quality support can be set to 0 MW in the normal steady-state case while retaining active-power capability for a dynamic event. Its steady-state MVAr can be zero, fixed, power-factor controlled or voltage controlled, subject to the inverter MVA capability and the project's control philosophy.
On-site gas generation, standby generators and renewable resources require their own status, dispatch, P/Q limits, source impedance and dynamic controls. A standby unit that is normally disconnected should remain out of service until a switching or dynamic event starts it. It should not silently reduce the normal grid import.
Capacitor banks and transformer LTC controls should identify the controlled bus, setpoint, deadband, delay and discrete limits. Avoid multiple devices fighting to regulate the same voltage without a defined hierarchy.
Dynamic studies need events and differential behavior
A credible dynamic case begins from a converged power flow and initializes without material mismatch. It then integrates machine, motor, inverter and controller states while solving the network algebraic equations at each time step. Step changes alone are not enough to reproduce electromechanical oscillation, control delay, current limiting or staged load recovery.
Representative data-center events
- Three-phase and unbalanced faults at the POI or an internal bus, cleared by the expected protection time.
- Voltage and frequency excursions applied through the utility equivalent for ride-through assessment.
- Loss and restoration of one transformer or distribution path.
- Transfer of UPS or BESS operating mode and active/reactive current priority.
- Trip, delayed restart and staged reconnection of cooling or IT load blocks.
- Loss of a shunt step, generator or plant-level controller.
Monitor POI MW and MVAr, selected bus voltages, frequency, device current and control output. Report MW and MVAr in absolute units and voltage/frequency in per unit where that makes the traces comparable.
Prepare a package that can be reproduced
The project-side submission should connect assumptions to files. A useful package includes the solved power-flow network, supported dynamic records, sequence data, one-line, operating scenarios, model documentation, event definitions and a validation summary.
- Export the RAW file in the required PSS/E revision and read it back.
- Re-solve and compare POI import, bus voltages, transformer flows, taps and losses.
- Load DYR records and resolve every undefined or unmatched model.
- Initialize dynamics and document any initial-condition warnings.
- Load sequence data and compare representative fault levels to the design basis.
- Reconcile the one-line, equipment schedule and model IDs.
- Identify which values are manufacturer-provided, utility-provided or preliminary.
A strong project-side tool should get the topology, ratings, controls and model identities right before PSS/E validation. Utility testing may refine parameters, but it should not reveal that the electrical architecture was represented incorrectly.
How HV Power Flow supports this workflow
HV Power Flow can represent the POI-to-campus network, solve and inspect the steady-state case, run screening studies and prepare PSS/E-compatible RAW, DYR and sequence-data outputs. It is independent software and is not affiliated with ERCOT or Siemens. Current ERCOT requirements, the official utility case, manufacturer models and professional review remain authoritative.