We are recomissioning liquid fuel after years out of service. The site has a single shaft 9FA. We can move the LF by pass valve,there are no LVDTs. Two simple questions:

1.- Can we move the valve to percentages somehow simulating flow? When we move it saturates fully open.We move it forcing the output in the MkVI.The valve has a pointer so you can easily see its position.

2.- Would you recommend dry fire? I thought about firing and looking for leaks at firing speed and after if all stable check at FSNL and at different loads.

This post could be a good one for CSA.

Thanks.
Regards.

 

The easiest way would be a dry fire. Run a trend on all the liquid fuel parameters at the same time and it should give you a good idea of the system integrity. If everything looks good you can then try a real fire and stay in "Fire" mode until you get a good system check done, you can then progress through the start-up to FSNL, all of the time keeping a good watch on the system.

Before you start the above, I would force stop valves etc. open and give your system a good purge all the way through to the HP filters, this can be done at stand still just using liquid fuel forwarding pressure.

 

Thanks for the answer.

Regarding testing LFBV,is there a way to simulate flow and be able to check accordingly the LFBV behaviour? Besides what we did forcing the MkVI putput.

Regarding dry fire,what is your opinion safetywise? Isn't there a risk to fill the unit with LF? Will permissives allow to force it,I guess at crank speed? Do you have any procedure for dry fire?

Thanks in advance.

 

You can simulate flow using a frequency generator connected in parallel to each of the liquid fuel flow divider speed pick-up inputs (after disconnecting the inputs so as to not excessively load the frequency generator output).

I've never tried this with speed pick-up inputs but since you say you have a Mark VI, you might be able to force the flow divider speed pick-up inputs to simulate a frequency generator....

I believe what Bob Johnston said is best; use the liquid fuel forwarding system pressure to purge all the way to/through the high-pressure liquid fuel pump (by forcing open the liquid fuel stop valve) and venting all filters as best as possible.

Then, perform a false fire or two, if necessary, because if there is still air in the line it might trip when the air goes through the flow divider followed by a slug of liquid fuel on excessive flow. You want to monitor the false start drains immediately after firing is complete to be sure the liquid fuel is actually being drained from the combustors/wrapper and exhaust.

You should be using Trend Recorder to monitor liquid fuel flow reference, valve position, and liquid fuel flow divider feedback, as well as speed, at a minimum.

A successful false start is one where the actual liquid fuel flow rate was very nearly equal to the liquid fuel flow-rate reference for most of the firing timer period, the liquid fuel bypass valve position was relatively stable once the flow-rate was equal to the reference, liquid fuel was observed to be coming from the false start drains, AND it should be possible to observe light white vapours coming from the exhaust stack after the firing period is over.

The vapours will NOT be smoke; they will be atomized liquid fuel that has passed through the turbine and exhaust. You won't be able to see them as soon as firing starts and liquid fuel is flowing because it takes some time for them to travel through the combustor, turbine, exhaust transition and HRSG (if present) and up the stack. But, they should be visible a short period after firing has completed for a brief period of time.

Be sure to allow the unit to continue to crank to purge all the vapours out of the unit prior to actually initiating a fired start attempt.

Be sure to write back to let us know how the re-commissioning of the liquid fuel system went!

 

Thanks for the reply.

In order to be able to perform a successful dry fire,from what I understand,we should somehow force the flow reference as at crank speed the MKVI won't give any flow reference on its own,could you give some advice on this? We will remove the spark plug connectors in the PDM,that is clear.

Venting the system and all the rest is feasible but I have some doubts regarding dry fire from the controls side of it,flow reference.

Thanks.
Regards.

 

Sorry,ignore my last post,I don't know what I was thinking...

We have to hit fire with the MkVI Spark plugs PDM connectors disconnected and see flow reference and feedback during the fire timer,see leaks,etc.

Nothing at crank,sorry for my useless question,everything at fire speed normal removing spark plug connectors.
Regards

 

We checked LFBV and with a frequency of 300000Hz with 10 as fqrout in the MkVI managed to control the LFBV.

Just as feedback.

Still checking rest of components both mechanically and instrumentation.

Regards.

 

Javier,

That's a <b>LOT</b> of zeroes.

Way too many, in fact.

You should have a 'Control Specification' document which was provided with the Mark VI at the time the unit was installed and commissioned. In Sect. 05.01.nn of the document, you should find a table which lists 'Expected Liquid Fuel Flow Characteristics' (if I remember the title of the section correctly). It should list some speeds and loads and the <b>expected</b> FSRs and liquid fuel flow-rates and probably the frequency feedback from the liquid fuel flow divider for each of the listed speeds/loads.

When using a frequency generator to simulate speed inputs to a Speedtronic turbine control panel the output of the frequency generator must cross the zero volt axis. It doesn't matter if the output waveform is a sine wave or a square wave or a sawtooth wave--but it <b>must</b> cross the zero volt axis, preferably with no offset (in other words, the waveform has the same magnitude above and below the zero volt axis).

Many less expensive frequency generators do NOT have the capability to output a zero-crossing waveform. These frequency generators output a waveform that has zero volts as its base, and the waveform (usually a square wave) increases above zero volts then decreases back to zero volts then increases above zero volts, and so on--but it does not cross zero volts. (Sometimes these frequency generators have what's called a TTL output, which is 0-5 Volts, but, again does not cross the zero volt axis.)

Speedtronic turbine control panels measure frequency (speed feedback; liquid fuel flow divider feedback) by looking at the zero volt crossings and calculating the time between zero crossings (in the same "direction", I believe). So, if the frequency being simulated does not cross the zero volt axis, then the Speedtronic cannot accurately determine frequency (speed; flow).

The magnitude of the simulated frequency does not have to be very large; you can set the frequency generator to approximately 5-7 Volts, peak-to-peak, and the Speedtronic will read the frequency just fine--<b>as long as the frequency waveform crosses the zero volt axis</b>. Some better quality frequency generators have a know for "offset" which means the waveform can be shifted about the zero volt axis by a variable amount--but for a Speedtronic panel, no offset is required nor should be used.

I'm not familiar with any liquid fuel flow divider that has a frequency output of more than 8,000 when the absolute maximum amount of liquid fuel is flowing through the flow divider. That's 8000 (8 x 10^3), at approximately 5-7 Volts pk-pk, with zero offset (meaning the waveform, nearly sinusoidal, crosses the zero volt axis).

When you are using a frequency generator to simulate liquid fuel flow divider feedback, you will need to disconnect the wires at the Speedtronic input terminal board for all liquid fuel flow divider inputs, and jumper all the even-numbered flow divider input terminals together and jumper all the odd-numbered flow divider input terminals together, and then connect one lead of the frequency generator to an even-numbered terminal and the other lead of the frequency generator to an odd-numbered terminal, so the frequency generator is powering all of the flow divider inputs in parallel. (I'm speaking of the actual flow divider inputs that are used on that Speedtronic turbine control panel. I don't have access to a Mark VI System Guide to provide example terminals.) You need to disconnect the speed pickup input wires because sometimes the extra load (reactance) of the speed pick-ups will prevent some frequency generators from putting out a sufficient voltage/frequency.

Now, let's say the Control Specification said that at FSNL the liquid fuel flow divider feedback was 1000 Hz and the corresponding FSR was approximately 20%. If you used a reference for the LFBV of 20% before you applied the frequency to the Speedtronic panel the LFBV would move to the fully closed position--because without any feedback the Speedtronic would close the LFBV to try to increase the feedback.

But, if you apply the 1000 Hz frequency BEFORE you increase the 20% LFBV reference, the LFBV will not move from the full open position--because the feedback is greater than the reference. As you decrease the frequency to some value less than 1000 Hz, the LFBV would start to close, and if you increased the frequency back to 1000 Hz the LFBV would stop at whatever position it was at--because at that position the feedback would be equal to the reference.

Now, don't go reading a lot into this test--because you will drive yourself (and us) crazy. When the feedback equals the reference, the LFBV will stop moving. If that happens when the valve is at 14.7% or 35.9% or 63.0% or 88.4% (you said the valve doesn't have LVDTs; this is just for exemplary purposes)--it doesn't matter! Even if the 'Expected Liquid Fuel Flow Characteristics' says that at 20% FSR and 1000 Hz the valve position would be 18.1%--that's assuming there is liquid fuel flowing through the system and the check valves and the fuel nozzles at the expected back-pressures and flow-rates and with the design supply pressure, etc. In other words, under ideal conditions--and you are just simulating feedback, <b>and only to demonstrate the LFBV will maintain position when the feedback is equal to the reference</b>. So, remember: This is a test. This is only a test. And it's being done without flowing liquid fuel (which would be occurring during a false fire--when conditions would be more "normal").

Also, the values you should find in the Control Specification are <b>expected</b> and the actual values will likely differ, sometimes greatly (unfortunately) but not usually (unless the engineer who did the calculations used incorrect information, or a liquid fuel flow divider with a very different frequency-per-kg/sec characteristic was provided or substituted at some point).

You said all you wanted to do was to see if the Speedtronic could control the LFBV. We suggested you perform a false fire. You wanted to use a frequency generator. There are things you can't simulate when using a frequency generator. And, there are real-life values versus expected values.

So, there you have it. It's your choice; just don't expect a high degree of accuracy or that any data will match some table very accurately, either.