Frame 7EA High Gas Flow


Thread Starter


We just performed a HGP on our 1993 vintage 2055F Frame 7EA with steam injection (ratio set at 1.4) and Mark V controls in combined cycle. Coming out of the outage the machine is roughly 1.5MW down on power compared to its pre-outage operation yet it is consuming roughly 10% more gas. CPD and CPT are nearly the same, exhaust temps have risen 3F at 90MW. We have compared the fuel gas flow signal in the Mark V to our site utility meters and they are with 1% of each other. We have walked the entire site several times trying to identify gas leaks between the flow meter and the GT and haven't found anything. With this much extra gas going to the machine we should be overfiring and seeing a large increase in power yet we are actually down on MW's. During the outage we installed new GE fuel nozzles, 3rd stage buckets, first stage nozzle and combustion liners. The rest of the items were overhauled/new by a third party (LONG story...). The biggest part change was that the transition pieces we removed did not have the extra window pane cooling holes yet the new ones we installed did. Also wheelspace temps have barely changed and if anything are slightly lower. Any help would be greatly appreciated.

Judging from your past posts, you seem to find yourself in this situation quite frequently--problems with performance after a maintenance outage.

Did the gas turbine fuel flow measurement sensors get calibrated during the outage? If so, what were the as-found conditions versus the as-left conditions? In other words, were the gas turbine fuel flow sensors out of calibration before the outage, and then adjusted to make them more accurate after the outage?

Do you have data from before the outage to compare your flow measurements ("revenue" vs. Mk V)? Were they pretty equal prior to the HGPI?

Was the site utility gas flow meter calibrated during the outage? I've been at sites where the utility has replaced their meter, and there was a large difference in the indicated flow before and after replacement. Quite difficult times those were. Of course, the Speedtronic turbine control system is always the usual/first suspect.

GE generally uses an orifice plate to provide a differential pressure, which then gets used to develop a fuel flow-rate value. I've seen the orifice plates installed backwards after maintenance outages and cause similar issues.

While many people would like the GE-provided gas flow measurement systems to be as accurate as the site utility metering, they are not. GE uses mass flow-rate in seconds, not total mass flow. This is in accordance with US CFR (Code of Federal Regulations) regarding NOx emissions reductions methodologies. The sensors GE provides are simply used to generate a reference for NOx steam injection flow-rates (which are calculated in mass per second, pounds per second, US measurement), or as a reference/limit for power augmentation steam injection (also in mass per second). As such, they don't have to be revenue-quality, and any totalization done by the Mark V is suspect. Since NOx steam injection is generally tuned to make NOx emissions slightly less than guarantee after any outage, slight errors in accuracy of the GE sensors are accounted for in the tuning exercise. Many people are quite disturbed, to say the least, when they look at totalized flow-rate data from a Speedtronic and try to compare it to revenue-quality meters (which don't usually measure instantaneous flow-rate, but total flow).

So, we don't know what units you are using to compare to the utility site metering and how the numbers are generated.

We don't know if the steam injection is for power augmentation or NOx reduction. And, the same question could be asked of the steam flow sensors, as well. Were they accurate before the outage? Were they adjusted during the outage?

If the steam injection is for NOx reduction, was tuning done after the outage to make actual emissions just slightly less than guarantee (permit) value? If so, by how much was the steam injection flow reference curve adjusted?

Are you sure one or more of the compressor bleed valves is(are) not passing (leaking)?

Have you let the unit temperatures stabilize at Base Load before taking data? This usually takes as much as four hours after a cold start. "Steady-state" (stable) is usually defined as no wheelspace temperature changing by more than 5 deg F in 15 minutes, which indicates the entire machine is "warm" and stabilized, which is when performance data should be taken.

Also, are you certain of the IGV LVDT calibration?

I think you would need to ask the supplier of the TPs (transition pieces) how they might affect heat-rate/performance versus the ones they replaced.

I would be most concerned about heat rate, a number which is difficult to accurately calculate, but if the same instrumentation is used repeatedly over time and if data can be obtained and/or corrected to ISO conditions, then a quick, crude check can be made of unit performance over time and compared to past values. Heat rate is the measure of efficiency in my book.

If the heat rate is high (lower is better) after an outage compared to before the outage then that's usually an indication of mechanical issues, or instrumentation issues. But, one would need to have data over time (before the outage as well as after the outage) to make the determination of whether or not the heat rate has degraded after the outage. Of course, the calibration of the instrumentation used needs to be checked before data is taken before the outage begins, and then that same instrumentation with the same calibration values/adjustments need to be used after the outage.

And even with reliable heat rate information, trying to pinpoint the cause of a significant change in heat rate can be very difficult. VERY difficult.

Let us know how you progress on this issue.
To clarify on my prior post from a few years back we had a number of issues coming out of an unplanned HGP on another one of our 7EA's. After three months of deliberation it was determined that the throat area of our first stage nozzle was too small due to the fact that the vendor took the measurements which were at the min GE limit before they TBC coated the nozzle which further shrunk its area.

During this outage neither flow meter (utility or orifice plate) was touched or calibrated. Due to our title V permit we are required to pull/swap the orifice plate annually in the late fall so typically we set it and forget it. Before the outage the orifice plate was within 1% of the utility meters and that is still the case today. The orifice plate like you said measures in lbs/s while the utility reads in mcfh. The conversion we used pre outage is the same as what we are using today so I do not think this is a metering problem. Although, I wish it was...

All of the test data we have taken was on the second or third day of baseload operation so the machine was certainly fully heat soaked. The IGV's were recalibrated during the outage due to the fact that we had to move the physical stop up to 86.5 degrees from 86. We did this because while running in AGC the load changes frequently which causes the IGV's to modulate and constantly bump into the physical stop.

The steam injection we use is for NOx control which runs around a 1.4 steam/fuel ratio but can will slightly due to ambient humidity. Our steam injection meters were not touched during the outage but now that we no longer cogenerate we have two IP meters that can be used to determing steam injection flow, both of which are still reading within 2% of each other.

Lastly, the TP's (window pane cooling mod) we put in the machine were actually the ones that should have been put in during the last CI and were not for some reason. However, they were the TP's that were originally installed when the GT was uprated to 2055F in 2001. Currently we are at total loss with this machine because other than fuel flow nothing seems out of the ordinary other that slightly higher exhaust temps and exhaust NOx.

Today we are going to look through the atomizing air system to make sure the pressure is correct and also chase down the tuning pin information which apparently can cause large deviations in performance if incorrect.