Two inverter model defects, caught in benchmarking

Inverter models arrive from the manufacturer and go into interconnection and compliance studies. A defect in a model passes into every study built on it, and is often found only when the operator rejects the model months later. This owner was evaluating about 20 plants against the ride-through standard and needed to know its models could be trusted first.

Operators expect a plant model to behave like the equipment. During a fault an inverter should inject reactive current to support voltage, on balanced and unbalanced faults alike, and a plant with negative-sequence capability should have it switched on.

We benchmarked each plant's PSCAD model against its PSS/E model and its as-left settings, every case from one model and one simulation setup. A consultant who submits one PSCAD file for the leading case and another for the lagging case can hide a defect; one model cannot.

Two inverter models injected no reactive power during voltage ride-through. In one, the root cause was a low-side voltage and a power-module rating set for a different inverter variant. The second model's defect was found and documented. In the owner's settings, the negative-sequence gain was zero even though the hardware supports it, which the manufacturer confirmed. All three would have passed a casual review.

The owner sent the model findings to the manufacturer for correction and took the negative-sequence setting as a field change. The evidence for each sits on the plant's record.