Dual Fuel 7EA DLN1 - Matching FSR1&2 for fuel transfers

The control specification for our dual fuel DLN1 7EA units discusses how to match oil and gas FSR but does not discuss how closely they should match. It simply states "It may be necessary to match the FSR settings between the fuels......."

Question: In your experience how close should the oil and gas FSR be matched to achieve reliable fuel transfers? Is 1-2% a reasonable expectation? If not what is?
Reliable fuel transfers? That's a VERY BIG ask, especially if you want the fuel transfer to occur once a year or so and automatically when loss of gas fuel is detected.

GE recommends frequent fuel transfers in order to exercise the liquid fuel system components (MANY of which ARE NOT controlled by the Mark* but all of which have to work correctly for a fuel transfer to be successful). And that means about once per week--maybe more often if the transfer is critical to the plant and their purchase power agreement(s).

And anything in the Control Specification about how to accomplish fuel matching is pretty old and has probably never been updated since Mark* IV days, or maybe Mark* V days. It's really a very detailed and complicated process, and because the process usually involves a few trips during transfer attempts it's usually "aborted" after one or two attempts. It's really kind of a seat of the pants, trial-and-error thing; there's no real procedure for this (never was, really, except back in the analog control system days, maybe).

Matching fuels (FSR1 and FSR2) is primarily done to try to prevent load swings during fuel transfers--and THAT presumes all of the liquid fuel system components (including any fuel purge AND any water injection systems and their components!) are all working properly. If the liquid fuel purge check valves are spring-loaded poppet style valves they are usually the weakest link in the system(s). Especially if they don't get "exercised" regularly to know if they are leaking or not (because there's no other way to know).

But the other big culprits of poor/unsuccessful fuel transfers are air in the liquid fuel piping (between the liquid fuel storage tanks and the liquid fuel stop valve, and even beyond that, on the turbine (and accessory) base, and problems with the liquid fuel forwarding system (pressure regulator, if used).

And when air is present in liquid fuel piping for long periods of times (and even liquid fuel vessels) bacteria and algae of sorts can grow and cause problems with check valves and fuel nozzles. Another reason frequent transfers--if only for 30 minutes or so--is recommended. AND, when operating on liquid fuel people need to know what and where to be looking for possible problems. I know of sites that run liquid fuel forwarding for a couple of hours per day (whether they are running gas fuel or not) and go around and vent piping and fuel vessels--just to try to keep air out of the lines. (But most liquid fuel piping wasn't built properly to have high-point vents so some sites have had to install them.) It helps, but transfers can still be problematic, especially if not done regularly and proper visual checks aren't made during liquid fuel operation.

Even if a fuel transfer is (more or less) successful and doesn't have large load swings during and/or after the transfer is complete, the water injection system has to start and run without tripping the turbine (usually because a huge slug of water puts the flame out--also can be caused by air in the water injection piping, as well as leaking water injection purge valves causing excessive uneven water injection flows which can cause high exhaust temperature spread trips...).

In my experience, what you want is for the liquid fuel flow rates at a couple of load points (say, 50% of rated load and as close to Base Load as possible--or whatever load the plant usually runs at when a loss of gas fuel might occur) to be within 5% or less of each other (so FSR1 might be at 41% and FSR2 might be at 45% for example--at the same load). If they differ by more than 10% that's a real recipe for disaster--because at some point the loads will be so different AND if the unit is operating on Pre-Selected Load Control (which it SHOULD NOT be!!!) the Mark* will be trying to bring the load back to the setpoint and that can cause even more load swings....

With DLN-I it's even a little more difficult to match fuels--because the easiest way to match fuels BEFORE DLN-I was to use P2 pressure as the "knob" to turn to get FSR1 and FSR2 to match. BUT, you DON'T WANT to do that with DLN-I because the gas fuel flow-rates through the nozzles have to remain within very tight ranges to keep emissions in compliance and maintain combustion stability. So, that "knob" isn't available with DLN-I.

That leaves adjustments to liquid fuel flow-rate measurement--basically the liquid fuel flow divider scaling. And, that is a pretty big trial-and-error "knob." Heavy emphasis on the error part of trial-and-error. AND, it can also affect starting reliability.... AND, let's NOT forget, water injection flow-rate references are based on liquid fuel flow-rate--so making changes to the liquid fuel flow-rate feedback can cause emissions to go out of compliance.

Again, fuel matching was something that was done back in the days when GT plants were used primarily as peaking power plants and many of them were unmanned (believe it or not). So, a reliable automatic fuel transfer was pretty desirable. And achievable. BUT, many turbines back then didn't have liquid fuel purge systems that were as complicated as today, and for some reason the poppet check valves seemed to last longer than newer ones (even from the same manufacturer--I think it's really caused by the higher compressor pressure ratios newer GTs run at now than they used to). And, back then GTs didn't usually have water injection, either....

In my experience, 1-2% is desirable--but not really achievable without a lot of "tuning" with DLN-I combustion systems and water injection. 1-2% will give a small load swing from the start of the transfer to the end of the transfer, but if everything else is working correctly it will really help. Usually, it's getting to that point that's hard. Unless you're fortunate enough to be close to that now--and then you are REALLY lucky. (Do you know what the two FSRs are now at various loads (say, 50% and 90%)?)

And remember, running at Part Load with Pre-Selected Load Control enabled and active is really NOT advisable. I know it's done all the time, but that's because Droop Speed Control is so poorly understood and Pre-Selected Load Control is so easy for operators.
We are in commissioning of a MK-5-to-Ovation upgrade. At FSNL, FSR2(gas)=about 11%. FSR1(oil)=about 18%. At base load on gas FSR is 44.8% (77 degF ambient). Have not gone past FSNL on oil yet. Do you know of a better time to match FSRs than during commissioning of a new system?.......I believe it best to eliminate unmatched FSR's as a cause or contributor to transfer problems. Am I reading your response correctly that if it were you, you would advocate for the new system to be tuned to achieve FSR match to within 1-2%?
Yes. 1-2% difference would be pretty ideal.

Do you have any data from the Mark* V for FSRs at FSNL and at similar loads? Because a 7% difference at FSNL is rather large—and an FSR2 of 11% at FSNL is rather low. Also, an FSNL of 44.8% at Base Load is pretty low. Most Mark* control systems are usually set up for Base Load FSRs of around 70-75% at GT nameplate conditions. Unless the natural gas composition has changed significantly since original commissioning (which has been known to happen).

The liquid fuel FSR at FSNL seems more normal for a GE-design Frame 7EA. So why the big difference on gas fuel? It would be very interesting to know what FSR2 is on liquid fuel with the new control system

Do you have any data/experience with fuel transfers with the Mark* V?

Typical FSNL FSR (for gas and liquid fuels) is around 20% for GT nameplate conditions.Typical FSR (for gas and liquid fuels) is around 70% at FSNL for GT nameplate conditions. That means no load to full load is an FSR difference of approximately 50% at GT nameplate conditions, which translates to about 12.5% FSR per percent Droop for a machine with 4% Droop setpoint. Those are typical values for fuels which meet the expected fuel characteristics used during the ordering of a specific GE/design heavy duty gas turbine with a Mark* control system. There are machines which operate fine at different values of FSR but that’s GE’s typical design criteria for a new unit—and that factors into account the fuel nozzle orifice sizing and in most cases the fuel control valve sizing as well.

That’s about all I can offer. I would be trying to get FSR2 (gas fuel) to more closely match FSR1 (liquid fuel) but, again, DLN-I combustion stability is very important and pressure drops across gas fuel nozzle orifices is very critical (no I don’t have or know of any spec’s on that). And I don’t know how the new control system calculates gas fuel FSR, or what the gas fuel composition is compared to what it was during original commissioning.

It would probably be best to get some data at load on liquid fuel before starting fuel matching. And, looking at any data from the Mark* V on FSRs at similar loads. I would be surprised (but not shocked) if the two FSRs were so different at similar loads (and at FSNL). But if they were, I would imagine fuel transfers were a wild and woolly bull ride with the Mark* V (unless there was some very unusual sequencing in the Mark* V).

Let us know how this goes, please.
To answer your question " Do you have any data/experience with fuel transfers with the Mark* V? "

In April, 2021 a transfer was performed and data collected:

Unit in preselected load control with setpoint 53 MW on gas, FSR=36.8%.
After the transfer complete unit controlling at 53 MW on oil, FSR=47.0%.
During the transfer, load oscillated between a low of 45 MW and a high of 64 MW. TTXM oscillated between a low of 1,080 and a high of 1,133 DegF.
Spread is not included in the data collected.

This is one of a group of 8 identical units, all with the same behavior. Higher load transfers have always been about 50% successful, especially when auto-initiated by low gas pressure.
Eight units? Indiana or Ohio?

Anyway, the mismatch in fuel FSRs is responsible for the load swings. And, if the units were in Pre-Selected Load Control that was making the problem worse (operators are SO terrified they will lose their job if the unit drifts half a MW (and the swings when on Pre-Selected Load Control can be more than +/- 1 MW--but they REFUSE to cancel Pre-Selected Load Control, even for 5 minutes ("We've never done THAT before!")). That's pretty unusual (in my experience)--the large mismatch in fuel FSRs. But it also explains why the transfers are not reliable.

I'm just not sure how I'd tackle matching such large differences with DLN-I combustion systems. Whatever you do can have serious knock-on effects, and yet if you do nothing--as you noted--the transfers ain't going to be very smooth ('specially if Pre-Selected Load Control is enabled and active! sorry; now I sound like CSA--and I don't want THAT). If you mess with gas fuel P2 pressure, you can screw up flame stability and possibly burn up fuel nozzles and combustion liners. And screw up emissions compliance. If you mess with liquid fuel flow feedback scaling you can screw up starting and shutdown reliability, AND cause emissions problems because water injection is based on liquid fuel flow-rate. What you describe is a lose-lose situation (in my opinion). Sheesh.

The OEM ain't gonna help you because you gone and got rid of the Mark* turbine control, so they ain't gonna do nuthin' to make that NIH control system better. You really need combustion engineering and maybe some help with the fuel control valves. (I don't remember if units of that vintage have IGCVs or still had the GSV and GTV arrangement. If I recall correctly, the switch was somewhere in the late Mark* V days/early Mark IV days, but that's a LONG time ago--both.) I can't recall--and don't quote me on this (it's been a LOOONG time) but the fuel nozzle pressure drops have to be sub-sonic or super-sonic--I can't remember which, or if that's even the right terminology. Screwing with that can have bad consequences; BAD--and I don't know how to know when one is near the danger point, or has gone too far. I was ALWAYS admonished NOT to screw with P2 pressure on DLN machine when trying to match fuels. As I wrote, it was the easiest way to match fuels (that weren't too mismatched to begin with (less than 10%, usually))--much easier than messing with liquid fuel flow feedback scaling. I used to try to get one of the DLN combustion gurus to help me come up with a simple method of fuel matching--but he recognized the potential for trouble and every time we talked about it (several times--I could be a pest about things like that) he was ADAMANT not to screw with P2 pressure on DLN machines. "Just DON'T do it!" he would always say. But he never had a suggestion for making fuel matching easier--it was always his contention that the two should be reasonably close, and while that was generally true, there were (are) machines like yours.

I always thought those big "bundle" purchases during the bubble would have some issues--I just didn't think about this particular issue. If the gas fuel systems (control valves and fuel nozzles) were sized for a particular gas fuel but the buyer put them in locations where the gas fuel was significantly different than the specification, well, that could cause problems. (If you're reading the Control Specification, and it sounds like you are, in Sect. 05, Fuels, there is a sub-section which lists the fuel constituents (or should be) the machine was designed for. (At the least there should be a chart which shows expected FSRs at FSNL and various loads, for both gas and liquid fuels. But if I recall correctly, there was a paragraph/description of MMBTU/scf or something like that for gas fuel; maybe just above the gas fuel expected FSR table. MIND YOU--those were just estimates based on calculations which were generated by the OEM's 'Gas Program' which was very, very good--if the actual fuel matched the expected fuel the buyer gave during the ordering/requisition process.) You MIGHT look at the current gas fuel make-up and see if there's a substantial difference--that might explain the mismatch of FSRs. Liquid fuel is gonna (usually) be pretty consistent for distillate fuels, so that's why I believe the liquid fuel FSNL FSR is close to what I experienced for most machines which burnt liquid fuel (around 20% for FSR1 at FSNL). But, then you gotta decide is it the gas control valve(s) or the gas fuel nozzles, or both. Only the Gas Program and the OEM would know for sure. And you would have to get LOTS of data, like gas control valve part numbers and fuel nozzle part numbers.

Not something I would want to be doing (fuel matching on DLN machines with this kind of mismatch to begin with)--or trying to get the OEM's help with given the circumstances.

I think you're kind of ... stuck. Yeah, that's the (PC) word I would use on this forum. Stuck. ;-)

Let us know how it goes. Best of luck. (Glad I'm retired! In more ways than one....)
So nice to see the big hitters CSA and WTF being able to take a few minutes, well more time than that for sure, to provide some great information.
Update: They ended up matching oil-to-gas FSR, within 2% all the way up, actually closer to 1% than 2%. Its a lot easier in Ovation (or any modern control system) than it would have been in MK-5. We did an actual "Auto Transfer" from gas to oil in premix SteadyState at 50 MW (Bleed heat unit) by throttling down on the manual gas inlet valve. This is a cogen unit and spends most of its time around that load. 92 DegF ambient temperature.

At 50MW preselected load and premix steady-state during the auto transfer to oil including water injection coming on: Load swung between 43-54.5 MW. TTXM: 1053-1118 DegF. FSRgas before transfer: 36.7% FSRoil after transfer: 35.5%.

The only directly comparable data I have is a transfer from oil (Without water injection) to gas at 20 MW. Both before and after matching FSR, the load swing was about 15 MW and TTXM 1071-1010 DegF.

Did transfers at FSNL also and those seemed rougher, with speed swings of about 55 rpm and TTXM 778-830.
Thank you for the feedback, and it sounds like "they" did a good job of it. I've always found fuel transfers at FSNL, particularly on DLN-I units, to be a little ragged. I attributed it to the low fuel flow-rates, particularly liquid fuel, during the transfer. The Mark* is also on more or less "permanent" droop speed control, even when the generator breaker is open so there's no integral action to hold the actual speed closer to the speed reference. At least that's my personal experience.

Personally, I am interested in the nuts and bolts of how "they" massaged the liquid fuel control, if you could share more details of that it would be helpful (at least to me). Did "they" muck with the liquid fuel flow feedback scaling and then massage the liquid fuel start-up and shutdown parameters? In the non-modern Mark* V control system the start-up and shutdown are very tightly integrated to the fuel scheme and so making changes to flow-rate feedback can have knock-on effects that can be difficult to smooth out after fuel matching is made smoother. Just a high-level description would be very helpful.

Thanks in advance!
I am interested in the nuts and bolts of how "they" massaged the liquid fuel control
Very much in the manner you describe. Scaled their feedback such that less oil FSR was required for the same load. As you suggest, they changed the firing and warmup FSR to achieve proper lite-off and warmup flow.

Overview of process, which I would not think you would find surprising:

1. They gave us new liquid firing and warmup constant values.
2. They instructed us to add a gain in the fuel control loop in a specific location. This is done without impacting the indicated oil flow. (The individual providing the direction was not on site, but what was done is straight forward).
3. We did a false fire and got the same range of oil flow as previous lite-offs. We had already run 2/3 times on oil, once to baseload for data gathering.
4. On the first firing attempt, one ignitor flickered and flame went out within seconds when fuel demand transitioned to warmup. Firing and Warmup FSR were increased slightly and the next firing attempt was successful.
5. Went to base on oil, tuned the gain to match FSR exactly at base. After that we found FSR matching within less than 2% throughout the load range.

The "They" is an Ovation control, designed, supplied by Emerson and installed by their contractors. It was not a 3rd party designed and installed Ovation system which I know some out there have done. Included an "Ovation" excitation system interfacing to the existing KATO exciter.

It was a lot of effort, but it was the absolute smoothest retrofit I've ever been a part of (I was working for the owner) but as I am sure you understand, a lot of any success depends on the owner - a supplier can't be successful with an absent/uninvolved/distracted owner.
Also glad your Ovation experience was so good. I know of other Mark* V-to-Ovation conversions (Ovation-supplied and engineered) where gas valves didn't work correctly, and the excitation didn't work correctly and other niggly problems. The installation was smooth; the loop-checking and commissioning was less than stellar. And, the person at one site who fought for the Ovation system has since moved on and they are having a good deal of difficulty keeping the units running reliably are going to go to an OEM solution. (I think that person was working hellaciously long hours and days to keep the system running whilst always saying it wasn't the Ovation's fault, or the engineering of the Ovation wasn't at fault. In the end, he couldn't be persuaded and couldn't persuade the site of his feelings (which were contradicted by the facts).)

I'm also happy to hear there is a control system alternative to the OEM that is worth considering. The problem arises when the control system for DLN combustion system-equipped turbines isn't the OEM, the site is left without much in the way of OEM support in the event of problems. Which can be very frustrating. A lot of smoking fingers get pointed at others and very little gets solved in a timely manner. A bad taste is left in everyone's mouth. That's when the owner needs to step in and get everyone to work together to resolve the problem(s).

It's also why a lot of sites aren't allowed to use non-OEM control systems. (Don't ask me how I know....) Corporate gets a quote from the non-OEM supplier (because they need more than a quote from the OEM) for a control system retrofit/upgrade but they will not choose anyone other than the OEM because of the potential for lack of support from the OEM in the event of problems (which may or may not be related to the control system, but the big, bad control system is ALWAYS the problem, right?).

You've probably heard that somewhere before, though, right.?.?.? (The control system is ALWAYS the problem!)