PLC based steam turbine PID governor

Hi,
I normally work on small hydro-electric turbines between 600 and 1500 rpm and am able to use PLC based PID control on these. My next project is a small steam turbine which operates at 4800 RPM.

On the end of the STG shaft are 12 magnets providing a 960Hz speed signal to a magnetic pick up unit (MPU). A woodward governor expert tells me that this signal will not provide sufficient resolution for control. Perhaps not even optimal control, but any control.

I was also told that generally 1800Hz speed signal is appropriate for control of diesel generators and 3-4kHz was appropriate for steam turbines.

A PLC with scan time 10ms is only 100Hz. Suppose the PLC gets the speed signal through a speed transducer device which converts the 960Hz-4kHz signal into a 4-20mA. The governor man said already this adds too much delay.

I suppose it is beneficial to be measuring changes in speed faster than PLC scan time of 100Hz (PID loop execution frequency), but I wouldn't have thought changes in the control valve position could be made fast enough for the PLC to be too slow.

The analog input card says sample time is 5ms default, configurable to 1ms for 1 input. resolution 24 bit. so an analog speed input is sampled at 200-1000Hz.

PLC can also have counter modules which sample at 10kHz or 60kHz. Sampling a 3kHz signal at 60kHz doesn't seem fast enough, but using the speed transducer solves this problem.

Interested in everyone's experience with control of steam turbines.

960 counts per second is not very fast.
The closed loop rate must be faster ( shorter ) than one tenth time constant of the generator. A 10 millisecond loop isn't very fast either but it may be fast enough. It depends on the generator. The problem with using a PLC on high performance jobs is that the speed sampling must be done at precisely equal intervals. PLCs suffer from sample jitter. Even if you close the loop every 10 milliseconds you will see 9.6 counts per scan on the average. Most of the time the PID will see 10 counts and the rest of the time the PID will see 9 counts. That is a change of 10%. That isn't real good because the PID needs to have an accurate velocity for the derivative time constant or gain. More resolution is required.

Slow analog cards don't help either.

I would use a high resolution encoder that generates about 100,000 counts revolution. That is 8 million counts per second. That is 8000 per millisecond or 2000 every 250 micro seconds. Now small changes in speed can be detected. You will probably need a real motion controller for that.

Thanks for the reply. I was just reading some older posts about cascaded loops and inner/outer loop rates.

The PLC has an interrupt routine every 2.5ms which could run the PID function block thereby calculating the D, presuming the PLC can process the PID loop in less than 2.5ms and have enough time to process the rest of the program. This could address the jitter. even though the PID would be running at 400Hz the outputs are still only processed once per program scan, so back to 10ms and the time constant of the generator.

We are going to use a woodward 2301-ST (manual 26184) governor for the isochronous control with a 100 tooth gear on the 1500 RPM shaft.

The generator is going to run islanded with load as well as on the utility.

When the island is to be re-joined with the utility, one of two things has to happen when the breaker closes:
1) the 2301 switches from isochronous to droop mode, but without changing the load on the generator. With the setpoint at 60Hz in isochronous, the speed at 60Hz, and then enabling 5% droop the governor will want to go to no-load.

2) the 2301 stays in isochronous mode and is manipulated by the synchronizing bias input which is directly into the summation point of speed and speed reference.

Thus a signal proportional the speed error can be injected to cancel the error and keep the 2301 at the current output.


My problem with 2 is still the PLC scan time.
The PLC PID loop process variable would be kW from a watt transducer and the control variable would be a +/-V to the 2301.

In the MAST task scan time is ~10ms. Fast task I could set to 5ms and have the IO assigned to FAST task as well, and I don't think this will cause overruns but not sure on that one without trying it.

The PLC has to counteract any change in generator output caused by the governor reacting to speed error. Since the generator is on utility speed error should always be very small, so the governor PID shouldn't be taking strong action which would be difficult for the PLC to keep up with.

The alternative is to buy a load share unit such as the SPM-D11.

When the alternator is islanded then you can control speed with throttle control.
HOWEVER when paralleled to grid the grid will be your speed and you will NOT be able to change the speed of your unit. As you open (close) the throttle you will increase (decrease) load on your unit.

One other thing I did not see you mention is reactive load control. You will need to ensure voltage regulator has this feature.

I do not think you want to try handling all this control stuff as a do it yourself project. Also your utility may not recognize it and may not allow you to parallel with them.

NO they do NOT have to allow you to parallel and THEY have final approval on your equipment.

Recommend you talk to your utility AND a good controls outfit familiar with your application.

Dan Bentler

Hi Dan,
Thanks for the reply.

I understand when paralleled I am not able to adjust the speed - however I have the synchronizing with load conundrum of either 1) enabling droop and the governor goes to no-load or 2) staying in isochronous mode, in which case the governor will try to correct any speed error from 60Hz, and not being able to do so will either go to 100% or 0%. In this situation I have to fool the governor into maintaining the same load using the syncrhonizing bias input to the summation point so that it sees no error. Will the PLC be able to keep up with the governor and turbine?

The voltage regulator in fact does not have reactive power control. It has droop to allow the voltage to sag as reactive power increases. I was going to program power factor control and a limiter using the AVR analog input for reactive power control but you do raise a good point regarding the utility. For a small machine I don't think they are as stringent with the equipment: while they ask for protection settings and a single line they don`t ask for a dynamic model.

Thanks again for your thoughts

Von

You said the turbine is designed for 4800 RPM, and you are generating 60 Hz. Do you have a gear reduction to get alt to either 3600 or 1800 or are you just going to slow the turbine and run it a bit less efficient and at reduced load??

What RPM do you run hydro turbines at? Are they grid paralleled?

I can see what you are thinking ie PLC can control hydro so it should be able to control steam turbine. I do not pretend to know if this "leap of faith" is correct or not. Interesting thought though.

I agree that your speed sense is going to need more resolution -- consider that 1 Hz change is 60 RPM on 2 pole 3600 RPM alternator.

Dan Bentler

Hi Dan,

yes there is a reduction gear to 1800 RPM.

The hydro turbines I've worked on in the past have been 900 RPM, 720 RPM.

9/10 plants I've worked on are grid paralleled, 1 is on an islanded distribution network with only another diesel for backup, another has the ability to run islanded.

I've come to the conclusion that the PLC wouldn't be fast enough for control of the steam turbine so have got the woodward 2301-ST coming. Unfortunately it can't handle the transition from operating in an island to paralleled. So I need another box to help it operate in isochronous mode against the utility ... but the PLC isn't fast enough to be that box either as the governor and turbine can react much faster than the PLC can.

Von

What is "generic" or maybe range of power on units you have worked on so far? What is power of the steam unit?

I would like to stay updated on all this. I suppose you could PM me but I think others would be interested in this.

Dan

Hi Dan,
STG is 300kw, hydros from 250kw to 20 mw

I manufacture steam turbines from 1 kw to 100 kw suitable for both 50 & 60 hz operations. Up to 20 KW is at 3000 / 3600 RPM and others at 1500 / 1800 RPM.

I had designed a mechanical governors (flying weight)but above 12 KW, this was not efficient. I switched to Woodward TG13. The PLC based governors in the market are too expensive for such small units. Can someone suggest any low cost solutions?? Do reply to [email protected].