Validation of the Inertia Constant of a Synchronous Generator

Are you testing the system, or just the generator?

Hello there.

I work for an utility in Madeira island and we have wind, solar, natural gas and fuel generators.

We have the generators manufacturers inertia figures but not the fuel prime-movers.

I want to study the synchronous inertial response of the generators but I cannot isolate them from the grid. Typically a fuel generator produces around 8 MW and the top load of the grid is around 115 MW.

Is it possible to get results with this scenario?

Thanks.

Hello and thanks for your reply.

The machine is installed and is one of several in the utility I work for in Madeira island.

You´re right, the goal is to study the generator dynamic response to system disturbances.

The prime-mover is a fuel engine and the alternator typically produces around 8 MW for a top load of 115 MW.

That's my main problem: I cannot isolate one machine for testing. Any generation loss that I perform will have the load acceptance response of all the other synchronous generators connected to the grid at that moment. I can measure the response of one of those generators but how can I reach a plausible inertia figure result?

I have just printed it and I will study the CIGRE document you have provided.

Thanks for your help.

After reading about your isolated grid and it's multiplicities of generation resources, I would suggest that you run a simulation of the dynamics first, long before you attempt any type of live testing, then you can determine if it's really necessary to perform any real-time tests.

First, the prime mover is a multi-fuel diesel engine, so not only is its mass rotating, it is also reciprocating, and this makes its moment of inertia dynamic, not static like the alternator rotor.

Second, your 8MW machine represents only 5% of the system load, so its response is like the tail wagging the dog.

Third, you have an interesting mix energy sources, including solar, wind, battery, hydro, thermal, and modeling some of them is highly dependent on their manufacturer.

I would start by reading the technical papers and contacting the authors that have already done studies similar to yours, and then there's Google, CIGRE, and IEEE for source documents.

So, really you are interested in the engine-generator response, it is their combined inertia which matters.

I think a load rejection test is most likely to give an accurate measure of inertia, if you do not have any maker's figure. But if the maker used CAD design system and got it to compute mr² bit by bit & sum them, you will not get a more accurate result by test (even if mr2 was computed by "hand" integration).

The real issue is how the system will react to loss of a generator or putting on a big load. So when you reject load on one generator, it is of even more interest to record the response of the system at several points, perhaps generator sites. You may have to set up a system to trigger or mark all recorders at same time or synchronise their clocks by cable or radio link/standard time transmission, so you have accurate phase differences.

There is also the issue of losing a big load for test purposes. I have heard of single shaft gas turbines being motored to provide a test load, they draw more than they generate - may be more problems with "pogo-stick" engines - in any case, I think their motoring losses are a small part of the output rating.

As already posted by RAMconsult, you must use all available data to make a system model including the probable step load increase/decrease possible & simulate load loss/addition on model to assess what is tolerable for test. Load reject on smallest genset first - previous experience of accidental or fault trip-outs may give a real check on what is possible and the probable system inertia/MW versus Hz [MW versus Volt] response of loads.

Usually, generator excitation response to add/cut VArs or get a rotor swing can be faster than prime-mover kW response. Some AVRs have a datum-step facility for setting response or a potentiometer can be switched -in/out of the resistive volt-sensing chain by relay.

Why?