We currently possess a total of 10 GE Frame type Gas Turbines in our power plant, consisting of 7 units of Frame 5P and 3 units of Frame 6B. These turbines are interconnected and operate on the Mark VIe system with the Ex2100e excitation system.

The load control method employed is "PRE SELECT," while the PF/VAR control is set to a power factor (PF) value of 0.92, typically. Speed control is achieved through the implementation of a droop mechanism, and the excitation regulation mode is set to "Auto." Moreover, our power plant operates in a simple cycle configuration.

Recently, we experienced a plant blackout due to a tripping incident at the distribution substation. As a result, approximately 150MW of load was rejected at the downstream. Surprisingly, the gas turbines continued to operate at the pre-selected value instead of activating the droop control mechanism to reduce the frequency in response to the excess power generation caused by the load rejection event.

Throughout the event, the frequency of all gas turbines was observed to gradually increase from 50Hz to 51.85Hz over a span of 4 minutes, as recorded in the Historian. The over-frequency protection feature in the G60 Multilin relay was then triggered, leading to the tripping of the 52G breaker and causing all units to run at Full Speed No Load (FSNL) condition. Ultimately, this event resulted in a blackout affecting the stakeholders of the refinery plants.

I am seeking insights into the reason why the droop control mechanism failed to operate during this event. Additionally, I would appreciate any suggestions on preventive measures we can implement to avoid such incidents in the future. One potential approach could involve switching some units to Part Load Control mode and assigning one unit to Isochronous mode to effectively respond to frequency fluctuations perhaps?

 

Do these units interconnected as you wrote on an infinite grid or it a island mode system?

 

Scling, pleaese understand with my response I am trying to be helpful and not trying to be cruel. You need to enlist help from someone that can help you understand the operation of droop control. Droop control is not meant to control frequency!! When you have units operating in parallel with and interconnected to the grid, then droop control works to help control frequency with other units. When your plant is disconnected from the grid it becomes an island. In that case you need to have an overall plant control algorithm that can control the output of each turbine. It sounds easy, but it is not a simple thing to accomplish many times. This has been discussed many times before on this site so try doing some searching of posts related to frequency control and island mode. You have some of the correct ideas, in your site if you plan to operate in island mode then you will have machines in droop mode and at least one in isochronous mode. But getting it all to work correctly takes work. Good luck and let us know how it goes.

 

I second everything MIKEVI wrote!

Pre-Selected Load Control IS NOT the best way to operate a GE-design heavy duty gas turbine. Full stop. Period. Unless GE has modified their Droop Speed Control scheme and the way that Pre-Selected Load Control works (and from your description it DOES NOT sound like they have!) when Pre-Selected Load Control is enabled and active it will FIGHT Droop Speed Control--which will try to reduce load lower/raise frequency to counter the disturbance (but Droop Speed Control DOES NOT control speed/frequency!)--BUT Pre-Selected Load Control will act EXACTLY OPPOSITE and try to keep load at the Pre-Selected Load Control setpoint (reference).

More and more grid system operators are forbidding the use of Pre-Selected Load Control because it makes a grid frequency disturbance worse instead of allowing Droop Speed Control to help to stabilize the disturbance. In some parts of the world operating on "simple" Droop Speed Control is called FGMO, Free Governor Mode of Operation.

It will NEVER happen but you can run a very simple test. Use the RAISE- & LOWER SPEED/LOAD buttons on an HMI to adjust the load of one machine to some value, let's say 15 MW for one of the Frame 5P machines. And, then just monitor the load to see what happens. That's right--DO NOT use Pre-Selected Load Control to try to maintain 15 MW--just use the buttons to get to some value close to 15 MW and watch. In fact, if you have someone (or yourself) that is familiar with the Trender function of ToolboxST use it to monitor load and speed and frequency of the machine. Leave it run like this for as long as you can convince the Operations Manager to let it run (IF you can convince the Ops Manager to perform the test). If the grid/system frequency is stable you will see the load--AMAZINGLY--will be stable too!!! It won't change unless the frequency (speed) changes. And, I predict there will be an added surprise: the load will be much more stable than when the machine is switched back to Pre-Selected Load Control!!! (Quite often, Pre-Selected Load Control is not tuned very well, and so the load swings as much as +/-! MW or +/-1.5 MW, continuously--it's almost never stable at the Pre-Selected Load Control setpoint (reference).

The machine isn't going to scream to max or min load and trip the machine. The machine isn't going to slowly drift to max or min load. It will stay wherever the operator put it and behave very nicely. EVERYBODY thinks without Pre-Selected Load Control the machine will not control load very well, it will drift all over or drift high or drift low.

Pre-Selected Load Control is a LAZY operator's way of (trying to) control load to whatever the Ops. Manager told them to run the machine at. VERY LAZY operator's way of controlling load--AND, as it turns out, it will act to prevent Droop Speed Control from attempting to help stabilize a grid frequency disturbance and will even make the disturbance worse.

But, you WILL NEVER convince anyone at the site to try this test; and, if you did convince someone to do so while recording the operating parameters with Trender and compare that to a Trend of Pre-Selected Load Control you will probably find that the machine operates more smoothly without Pre-Selected Load Control.

If this plant operates independently of a large(r) grid/system and it uses some kind of "power management system" to control load and frequency then that "system" was not working properly. AND, if the machines were operating in Pre-Selected Load Control and the Pre-Selected Load Control setpoint (reference) was coming from the "power management system" then the machines ARE NOT configured properly. Full stop. Period. But, you haven't provided enough information for us to be of any further help--except to say that Pre-Selected Load Control will FIGHT Droop Speed Control (and pretty much win the fight!) during a grid frequency disturbance. (You also didn't say what cause the grid/system frequency disturbance....)

Methinks you got multiple problems, including a poor understanding of Droop Speed Control. (And you/we didn't touch AT ALL on reactive load sharing--and I don't think you reall want to go there at this point in the discussion. Trust me. Please.)

 

scling, you curiously did not say what happened to cause the system frequency to increase. The typical reason for system frequency to increase above normal is for some portion of the system load to (usually) suddenly “disappear.” For example a circuit breaker or some switch opens to separate some loads from the system. In a refinery situation that would be like a portion of the plant being disconnected (losing electrical supply). The immediate effect of that is to cause the system load to be reduced relative to the amount of generation at that time and the system frequency will increase.

The Droop machines that are not running at Base Load will sense the change in frequency (the speed of the prime movers will increase) and Droop Speed Control will reduce the energy flow-rate into the prime movers which will help to reduce the amount of frequency increase BUT it WILL NOT return the frequency to normal. As MIKEVI wrote, that’s NOT Droop Speed Control’s job. That’s the job of either the Isochronous machine operating on the system OR the system operators, OR possibly the “power management system” which is supposed to supplant trained human operators. But it’s NOT Droop Speed Control’s job. Never had been; never will be.

We really cannot tell you what happened without a lot more information about the event—what happened before and during the event (to the system load). We also need to know if this power plant is islanded or connected to a larger grid/system. If it’s islanded why are the machines running in Pre-Selected Load Control? Who or what responds to changes in system load, such as large electric motors starting or stopping, or a problem with one part of the plant, or when some part of the plant that was shut down is brought back online?

There are many control schemes for operating islanded systems, but I, personally, have never seen or heard of one that has all the gas turbines running in Pre-Selected Load Control. Even if the load signals to each of the gas turbines was varying the Pre-Selected Load Control setpoints (references) they would not respond quickly enough to sudden large changes in load to keep system frequency near normal. Without some fancy, non-standard application code.

We would really like to help you work on this, but we need more information.

 

I second everything MIKEVI wrote!

Pre-Selected Load Control IS NOT the best way to operate a GE-design heavy duty gas turbine. Full stop. Period. Unless GE has modified their Droop Speed Control scheme and the way that Pre-Selected Load Control works (and from your description it DOES NOT sound like they have!) when Pre-Selected Load Control is enabled and active it will FIGHT Droop Speed Control--which will try to reduce load lower/raise frequency to counter the disturbance (but Droop Speed Control DOES NOT control speed/frequency!)--BUT Pre-Selected Load Control will act EXACTLY OPPOSITE and try to keep load at the Pre-Selected Load Control setpoint (reference).

More and more grid system operators are forbidding the use of Pre-Selected Load Control because it makes a grid frequency disturbance worse instead of allowing Droop Speed Control to help to stabilize the disturbance. In some parts of the world operating on "simple" Droop Speed Control is called FGMO, Free Governor Mode of Operation.

It will NEVER happen but you can run a very simple test. Use the RAISE- & LOWER SPEED/LOAD buttons on an HMI to adjust the load of one machine to some value, let's say 15 MW for one of the Frame 5P machines. And, then just monitor the load to see what happens. That's right--DO NOT use Pre-Selected Load Control to try to maintain 15 MW--just use the buttons to get to some value close to 15 MW and watch. In fact, if you have someone (or yourself) that is familiar with the Trender function of ToolboxST use it to monitor load and speed and frequency of the machine. Leave it run like this for as long as you can convince the Operations Manager to let it run (IF you can convince the Ops Manager to perform the test). If the grid/system frequency is stable you will see the load--AMAZINGLY--will be stable too!!! It won't change unless the frequency (speed) changes. And, I predict there will be an added surprise: the load will be much more stable than when the machine is switched back to Pre-Selected Load Control!!! (Quite often, Pre-Selected Load Control is not tuned very well, and so the load swings as much as +/-! MW or +/-1.5 MW, continuously--it's almost never stable at the Pre-Selected Load Control setpoint (reference).

The machine isn't going to scream to max or min load and trip the machine. The machine isn't going to slowly drift to max or min load. It will stay wherever the operator put it and behave very nicely. EVERYBODY thinks without Pre-Selected Load Control the machine will not control load very well, it will drift all over or drift high or drift low.

Pre-Selected Load Control is a LAZY operator's way of (trying to) control load to whatever the Ops. Manager told them to run the machine at. VERY LAZY operator's way of controlling load--AND, as it turns out, it will act to prevent Droop Speed Control from attempting to help stabilize a grid frequency disturbance and will even make the disturbance worse.

But, you WILL NEVER convince anyone at the site to try this test. And, if you did so while recording the operating parameters with Trender and compare that to a Trend of Pre-Selected Load Control you will probably find that the machine operates more smoothly without Pre-Selected Load Control.

If this plant operates independently of a large(r) grid/system and it uses some kind of "power management system" to control load and frequency then that "system" was not working properly. AND, if the machines were operating in Pre-Selected Load Control and the Pre-Selected Load Control setpoint (reference) was coming from the "power management system" then the machines ARE NOT configured properly. Full stop. Period. But, you haven't provided enough information for us to be of any further help--except to say that Pre-Selected Load Control will FIGHT Droop Speed Control (and pretty much win the fight!) during a grid frequency disturbance. (You also didn't say what cause the grid/system frequency disturbance....)

Methinks you got multiple problems, including a poor understanding of Droop Speed Control. (And you/we didn't touch AT ALL on reactive load sharing--and I don't think you reall want to go there at this point in the discussion. Trust me. Please.)

Exactly,

(Kindly correct me if I am wrong)

Even if GT Droop control (4% droop setting) is active due to GRID frequency disturbance and 25 % of its active load will drop or increase and once the frequency stable is sensed by the TNRL then the governor will take reference from the Pre-selected load and the load will be increased based on the load set point.
hence there is a chance your frequency may increase again and meet the over-frequency trip point,
1. So, shall we run the unit in Part load (manual Rise or lower) to avoid the over-frequency issue?
2. why does OEM logic not disable the pre-selected load command in case the Droop control is active?

 

1) Yes. In the typical GE heavy duty gas turbine control philosophy when Isochronous Speed Control IS NOT active Droop Speed Control is ALWAYS active from FSNL to Base Load. It’s not only active when there’s a frequency disturbance, it’s active when the machine is at rated speed and below Base Load. Droop Speed Control is primarily the way the machine is loaded and unloaded from generator breaker closure to Base Load and from Base Load to breaker opening. Secondarily, Droop Speed Control is a method for supporting grid stability when a grid frequency disturbance occurs—but that DOESN’T mean returning the grid frequency to normal.

There’s been a LOT written about Droop Speed Control on Control.com. Pre-Selected Load Control assumes the system/grid frequency is stable—and, worse, doesn’t recognize when system/grid frequency is not stable. (GE has a purchased option to fix their original oversight. Yeah, that’s right; the Customer can pay to fix GE’s error.) Because Droop Speed Control is still active when Pre-Selected Load Control is also active when a grid frequency disturbance occurs and Droop Speed Control tries to change load to compensate for the grid frequency change Pre-Selected Load Control tries to over-ride Droop Speed Control and acts in exactly the opposite way of Droop Speed Control.

Pre-Selected Speed Control should only be used to change load and when the actual load reaches the Pre-Selected Load Control setpoint (reference) Pre-Selected Load Control should de automatically disabled—it’s done its job of relieving the operator of manually changing load with the RAISE- and LOWER SPEED/LOAD buttons (which isn’t physically difficult…). But, GE’s control scheme doesn’t do that (even though it should, and could). And all an operator has to do is click on RAISE- or LOWER SPEED/LOAD once the desired load is reached to cancel (deactivate) Pre-Selected Load Control. But EVERYONE is so frightened if they do that (cancel Pre-Selected Load Control) the machine will not hold the load at that point without Pre-Select Load Control active and they will lose their job that NOBODY will ever try it. [REAL NEWS FLASH: That’s how most machines in the world were operated before Pre-Selected Load Control was (poorly) developed.]

Most people probably believe (if they ever thought about it) that when they are clicking on RAISE SPEED/LOAD and LOWER SPEED/LOAD while watching the MW meter they are changing a load reference. But they’re not—they’re changing a speed reference that results in a load change. For a machine with 4% droop, every 1% change in the machine’s speed reference results in a 25% change in load. So, increasing the speed reference to approximately 104% results in rated power output (for a machine in a clean condition operating at rated ambient conditions).

Droop Speed Control is about how much the load changes for a given change in the machine’s speed reference. AND, it’s about how much load changes when the grid frequency deviates from normal. The frequency deviation thing is secondary to stable load control. Droop Speed Control does that just fine without Pre-Selected Load Control; but it doesn’t do the frequency disturbance thing very well at all with Pre-Selected Load Control.

2) Now you’re beginning to understand. The people who design control schemes don’t always consider every possibility and with GE there was no one person or group of people who could look at something like this issue of Pre-Selected Load Control and decide it’s not really compatible in its original configuration with Droop Speed Control’s secondary function of supporting system/grid stability in the event of frequency disturbances. And the person who originated Pre-Selected Load Control isn’t around to consult and so no one is going to fix something that’s been around for decades but wasn’t properly sorted out without consulting the originator. (There are other similar things like this in the GE heavy duty gas turbine control philosophy…. L62TT2 is another glaring example of a scheme that’s been around so long no one going to take responsibility and fix it (eliminate it from manned gas turbine power plants)).

 

We currently possess a total of 10 GE Frame type Gas Turbines in our power plant, consisting of 7 units of Frame 5P and 3 units of Frame 6B. These turbines are interconnected and operate on the Mark VIe system with the Ex2100e excitation system.

The load control method employed is "PRE SELECT," while the PF/VAR control is set to a power factor (PF) value of 0.92, typically. Speed control is achieved through the implementation of a droop mechanism, and the excitation regulation mode is set to "Auto." Moreover, our power plant operates in a simple cycle configuration.

Recently, we experienced a plant blackout due to a tripping incident at the distribution substation. As a result, approximately 150MW of load was rejected at the downstream. Surprisingly, the gas turbines continued to operate at the pre-selected value instead of activating the droop control mechanism to reduce the frequency in response to the excess power generation caused by the load rejection event.

Throughout the event, the frequency of all gas turbines was observed to gradually increase from 50Hz to 51.85Hz over a span of 4 minutes, as recorded in the Historian. The over-frequency protection feature in the G60 Multilin relay was then triggered, leading to the tripping of the 52G breaker and causing all units to run at Full Speed No Load (FSNL) condition. Ultimately, this event resulted in a blackout affecting the stakeholders of the refinery plants.

I am seeking insights into the reason why the droop control mechanism failed to operate during this event. Additionally, I would appreciate any suggestions on preventive measures we can implement to avoid such incidents in the future. One potential approach could involve switching some units to Part Load Control mode and assigning one unit to Isochronous mode to effectively respond to frequency fluctuations perhaps?

I believe its LPS

 

Do these units interconnected as you wrote on an infinite grid or it a island mode system?

Those units are running on infinite grid.

 

Hi MIKEVI, the plant relied on importing power from the grid because the plant GIS connection was linked to the national grid at the time of the incident. Usually, the plant units were operating in PreSelect mode, and manual load sharing was conducted by the operator due to the absence of a Power Management System (or EMS) for automated load dispatch. Although, my suggestion is to have one Isochronous unit configured to operate under frequency control for this plant, however, I have concerns regarding the plant's interconnection with the national grid. Say if there is an external disturbance, the same event might recur because the Isochronous machine will only capable to respond within a narrow frequency range of 49.5Hz to 50.5Hz (let's assume we configure one Frame 6B in the plant as isochronous), which is insufficient to accommodate fluctuations in the national grid frequency.

 

Scling, pleaese understand with my response I am trying to be helpful and not trying to be cruel. You need to enlist help from someone that can help you understand the operation of droop control. Droop control is not meant to control frequency!! When you have units operating in parallel with and interconnected to the grid, then droop control works to help control frequency with other units. When your plant is disconnected from the grid it becomes an island. In that case you need to have an overall plant control algorithm that can control the output of each turbine. It sounds easy, but it is not a simple thing to accomplish many times. This has been discussed many times before on this site so try doing some searching of posts related to frequency control and island mode. You have some of the correct ideas, in your site if you plan to operate in island mode then you will have machines in droop mode and at least one in isochronous mode. But getting it all to work correctly takes work. Good luck and let us know how it goes.

Hi MIKEVI, the plant relied on importing power from the grid because the plant GIS connection was linked to the national grid at the time of the incident. Usually, the plant units were operating in PreSelect mode, and manual load sharing was conducted by the operator due to the absence of a Power Management System (or EMS) for automated load dispatch. Although, my suggestion is to have one Isochronous unit configured to operate under frequency control for this plant, however, I have concerns regarding the plant's interconnection with the national grid. Say if there is an external disturbance, the same event might recur because the Isochronous machine will only capable to respond within a narrow frequency range of 49.5Hz to 50.5Hz (let's assume we configure one Frame 6B in the plant as isochronous), which is insufficient to accommodate fluctuations in the national grid frequency.

 

Hi MIKEVI, the plant relied on importing power from the grid because the plant GIS connection was linked to the national grid at the time of the incident. Usually, the plant units were operating in PreSelect mode, and manual load sharing was conducted by the operator due to the absence of a Power Management System (or EMS) for automated load dispatch. Although, my suggestion is to have one Isochronous unit configured to operate under frequency control for this plant, however, I have concerns regarding the plant's interconnection with the national grid. Say if there is an external disturbance, the same event might recur because the Isochronous machine will only capable to respond within a narrow frequency range of 49.5Hz to 50.5Hz (let's assume we configure one Frame 6B in the plant as isochronous), which is insufficient to accommodate fluctuations in the national grid frequency.

I second everything MIKEVI wrote!

Pre-Selected Load Control IS NOT the best way to operate a GE-design heavy duty gas turbine. Full stop. Period. Unless GE has modified their Droop Speed Control scheme and the way that Pre-Selected Load Control works (and from your description it DOES NOT sound like they have!) when Pre-Selected Load Control is enabled and active it will FIGHT Droop Speed Control--which will try to reduce load lower/raise frequency to counter the disturbance (but Droop Speed Control DOES NOT control speed/frequency!)--BUT Pre-Selected Load Control will act EXACTLY OPPOSITE and try to keep load at the Pre-Selected Load Control setpoint (reference).

More and more grid system operators are forbidding the use of Pre-Selected Load Control because it makes a grid frequency disturbance worse instead of allowing Droop Speed Control to help to stabilize the disturbance. In some parts of the world operating on "simple" Droop Speed Control is called FGMO, Free Governor Mode of Operation.

It will NEVER happen but you can run a very simple test. Use the RAISE- & LOWER SPEED/LOAD buttons on an HMI to adjust the load of one machine to some value, let's say 15 MW for one of the Frame 5P machines. And, then just monitor the load to see what happens. That's right--DO NOT use Pre-Selected Load Control to try to maintain 15 MW--just use the buttons to get to some value close to 15 MW and watch. In fact, if you have someone (or yourself) that is familiar with the Trender function of ToolboxST use it to monitor load and speed and frequency of the machine. Leave it run like this for as long as you can convince the Operations Manager to let it run (IF you can convince the Ops Manager to perform the test). If the grid/system frequency is stable you will see the load--AMAZINGLY--will be stable too!!! It won't change unless the frequency (speed) changes. And, I predict there will be an added surprise: the load will be much more stable than when the machine is switched back to Pre-Selected Load Control!!! (Quite often, Pre-Selected Load Control is not tuned very well, and so the load swings as much as +/-! MW or +/-1.5 MW, continuously--it's almost never stable at the Pre-Selected Load Control setpoint (reference).

The machine isn't going to scream to max or min load and trip the machine. The machine isn't going to slowly drift to max or min load. It will stay wherever the operator put it and behave very nicely. EVERYBODY thinks without Pre-Selected Load Control the machine will not control load very well, it will drift all over or drift high or drift low.

Pre-Selected Load Control is a LAZY operator's way of (trying to) control load to whatever the Ops. Manager told them to run the machine at. VERY LAZY operator's way of controlling load--AND, as it turns out, it will act to prevent Droop Speed Control from attempting to help stabilize a grid frequency disturbance and will even make the disturbance worse.

But, you WILL NEVER convince anyone at the site to try this test; and, if you did convince someone to do so while recording the operating parameters with Trender and compare that to a Trend of Pre-Selected Load Control you will probably find that the machine operates more smoothly without Pre-Selected Load Control.

If this plant operates independently of a large(r) grid/system and it uses some kind of "power management system" to control load and frequency then that "system" was not working properly. AND, if the machines were operating in Pre-Selected Load Control and the Pre-Selected Load Control setpoint (reference) was coming from the "power management system" then the machines ARE NOT configured properly. Full stop. Period. But, you haven't provided enough information for us to be of any further help--except to say that Pre-Selected Load Control will FIGHT Droop Speed Control (and pretty much win the fight!) during a grid frequency disturbance. (You also didn't say what cause the grid/system frequency disturbance....)

Methinks you got multiple problems, including a poor understanding of Droop Speed Control. (And you/we didn't touch AT ALL on reactive load sharing--and I don't think you reall want to go there at this point in the discussion. Trust me. Please.)

WTF, when you input the Presel value and select the "Raise" or "Lower" button on the Cimplicity HMI, the GTG's status switches to "Part Load" operation. We conducted tests on one unit to observe its response when we adjusted the MW output of the other GTG, but the results were not significant when the plant was still connected to the national grid. However, we did notice that the selected unit's MW output varied by approximately +/- 1MW, and its FSR also fluctuated accordingly. Another concern is that the GCV and SRV servo valves may experience accelerated wear and tear if operated in "Part Load" mode since the servos will continuously respond to the grid load frequency. Therefore, the plant was previously decided to operate those machines in "PreSel" mode.
We extracted a trip log from Mark Vie and discovered that during the incident, the MW appeared to drop but was subsequently restored to the desired "Presel" value. It's unclear whether Droop was opposing or counteracted by the "Presel" value.
FYI, the disturbance was caused by an external cable fault (at transmission), which resulting in the substation breaker opening and an instant loss/rejection of 150MW power for residential areas.

 

What was the grid frequency when the machine(s) was(were) tested in Part Load (“simple” Droop Speed Control)? Was it stable or was it drifting just above and below 50.0 Hz? What were the states of L70R and L70L during that test?

Droop Speed Control is about the difference between the turbine speed reference, TNR, and the actual turbine speed, TNH (which is directly related to frequency). When at Part Load and Pre-Selected Load Control is not active the turbine speed reference will be NOT changing, and if the grid frequency is changing then the load will change. But if the grid frequency is pretty stable there will only be minor changes in load--which are entirely attributable to frequency changes.

Pre-Selected Load Control May be able to hold a more stable load if the grid frequency is not stable—but to do that the SRV & GCV will be moving around to do so (as evidenced by the changes of state of L70L and L70R). So, nothing to be gained with this line of thought. And it's not the servos which will experience the wear--it's the hydraulic actuators and valve stems of the SRV & GCV. So, again faulty reasoning here.

If there are multiple machines synchronized together and operating in Droop Speed Control while connected to a grid changing load on one machine while watching another machine's output to check for changes is fruitless. Droop Speed Control is about sharing load--in other words, what one machine does doesn't affect any other machine. Each one develops power strictly based on the difference between TNR and TNH. This has been covered MANY times before on Control.com. Each machine is independent of the others. Droop Speed Control is the governor mode that allows stable operation while other machines are operating stably as well. Multiple machines (as few as two) operating in Isochronous WILL NOT operate together very well, at all.

If the plant is connected to a grid you CAN'T operate ANY machine in Isochronous Speed Control mode. It will try to control GRID frequency, not just the frequency in the plant. You can have one machine set to immediately change to Isochronous Speed Control when the plant bus is completely separated from the grid, but not when the plant bus is connected to the grid. In fact, it would be preferrable to have one machine configured to do that automatically--as long as the machine can handle the expected load change required to maintain system frequency. And, that's not an easy thing to predict--but it's not impossible, either, especially if the amount of power being imported from or exported to the grid is small (just a few MW).

It's not clear what the "tie line" power flow was at the time of the 150 MW loss. If it was about 0 MW, then as long as the refinery load(s) were not changing then the power should not have increased. If, however, the refiner load(s) were decreasing after the plant bus was disconnected from the grid and no one was paying attention to the change in load and the resultant change in plant frequency then it's entirely possible the plant tripped on overfrequency.

 

During the conducted test, all machines were interconnected to the grid with a constant frequency at 50Hz. When we decreased/increased the load of other GTG units, the specific machine being tested(Part Load Mode machine), the L70R and L70L, exhibited a response.

Your point is quite valid, the isochronous unit will continuously adjusts its reference in accordance with the grid frequency. However, it is only feasible if the plant is completely isolated from the national grid and operating in island mode.

As mentioned in my initial post, the frequency gradually increased from 50Hz to 51.85Hz over a span of 4 minutes, triggering the over-frequency threshold and causing all GTGs' 52G breaker to open. Personally, I believe that 4 minutes is a sufficient timeframe for operators to reduce the load and stabilize the machine's frequency. Unfortunately, the incident occurred at 4 am, a time when people are usually asleep and may not be responsive or aware of abnormalities. Hence, we are planning to implement a Plant Management System (PMS) to prevent this type of incident from recurring, which emphasizes the need for an automated system.

 

You mean Power management system PMS.. Not Plant management system

 

Good idea to implement a PMS !

 

I had a strong feeling that "operators" failing to take action (either because they are inexperienced and untrained and poorly supervised, or just plain fearful of doing something wrong (again, because of inexperience and lack of training/supervision)). That's about the only way the frequency could have drifted so high--and again because of conditions in the refinery(s) changing.

It would be helpful to see a SLD (Single-Line Diagram) of the plant power generation scheme and the grid tie configuration. I would be very surprised to see L70R & L70L changing on the Part Load machine when changing the load(s) of other machine(s) IF the grid was indeed stable. 50 Hz is not 50.04 Hz, or 49.83 Hz, or 50.12 Hz. I suggest the resolution of the data being recorded for grid frequency on the Historian is not configured properly (a VERY common problem with GE Historians--that data points are not configured properly to record relevant data with the necessary resolution for troubleshooting).

I have seen a situation similar situation involving two GE-design Frame 9E machines which didn't automatically switch one of the machines automatically to Isoch when the grid tie was lost (a grid switchgear problem), and the units tripped on overfrequency after about 60 seconds and the plant went black and the host facility (a paper mill) also went black. The two operators on shift at the time of the incident (which occurred shortly after commissioning) had not attended the formal operator training and were recently hired and simply did not know what to do. They were fired later the same day for not taking proper action to avert the trip. Their Operations Supervisor on shift was in his office asleep with a prayer rug wrapped around his head to block light and noise from distrubing his sleep. He was not fired, not disciplined in any way. How do I know all this, because I was babysitting the plant for 30 days after commissioning and was working the night shift. I tried to get the operators to take action but their English was poor and my local language was only adequate enough for ordering in restaurants and being polite--and I was not able to touch the controls because the equipment had been handed over to and accepted by the Customer. (It was suggested that I be fired for not jumping in to take action, but that wouldn't have saved the two operators from being fired. And, it was written in the acceptance agreement that the commissioning personnel were prohibited from operating the equipment. All I could do was observe and report--which I did, and it was the Operations Supervisor who suggested I be fired (for reporting he had been asleep at the time of the event). I was trying to wake the Operations Supervisor when the trip occurred, from outside his locked office (which had a window looking out to the Control Room where he could be seen sleeping under his prayer rug). The Customer was trying to blame the Mark* turbine control system for the problem saying it (the "automation") should have prevented the problem, but the Trip Display had been configured to record the state of the grid tie breaker and it clearly showed the breaker hadn't opened and the units were running in Droop Speed Control, because someone had opened a knife switch in the status circuit to the Mark* which during commissioning it was pointed out, in writing, that having a knife switch in the circuit was a not a good idea. It was all documented very clearly, but two unfortunate people lost their jobs.) After only four (4) days, the Operations Supervisor was back to sleeping during his shift with the prayer rug wrapped around his head.

Anyway, it sounds like you have your answer(s). And automation will be the solution.

 

WTF, your feelings is completely true, but we must acknowledge that the events have already occurred, and it is pointless to blame or point at any individual for the responsibility. The most crucial aspect now is to enhance the robustness and reliability of this power station, given that the majority of stakeholders are oil and gas refineries.

In addition to choosing the appropriate GTG operation mode (either a "Presel" or "Part Load"), we are now required to conduct a study on the existing load shedding scheme. Furthermore, power system studies must be performed for the downstream distribution substations, as many of them are newly constructed and may require adjustments to relay settings. In summary, numerous tasks need to be carried out to determine the necessary actions for both the generation and transmission & distribution sides. Simply selecting the preferred load operation mode might not help to resolve this issue alone.

FYI, at this power plant, we do not have a Historian. We previously proposed the idea to the asset owner, but they declined it, considering it is not important for the plant. Moreover, due to their financial regulations, we were unable to proceed with the proposed idea. Therefore, most of the time, we only can refer to the Mark VIe trip log whenever the GTG trips to analyze the underlying cause. In addition to that, we have to refer the event and tripping logs captured in the disturbance recorder at the GIS for further investigation. Make investigation job harder and complicated sometimes.....

Recently we have consulted with the Yokogawa EMS team to seek their expertise in this field, but they asked us to clarify our requirements instead of offering their support and professional insights regarding the PMS operation for this plant. We also approached GE, but they seem less interested about it. We have to engage with more PMS system providers to seek their expertise. By consulting with multiple providers, we can gather a range of perspectives and insights to make an informed decision. Engaging with different providers will allow you to understand their proposed solutions, and determine which provider+proposal can best meet our specific requirements.

We do not have other alternative ideas to address operator negligence aside from implementing the PMS/EMS for the plant. Do you all perhaps possess a better suggestion?

 

You mean Power management system PMS.. Not Plant management system

Yes, we are considering the implementation of a Power Management System (PMS). To successfully integrate this solution, several site modifications will be required beforehand. For example, installation of signal cabling is necessary for the MW, MVAR, kV, GSUT OLTC feedback, setpoints control, and the exchange of operation signals. These modifications will involve hardwiring instead of using a soft-wired approach.

 

I believe its LPS

It appears that we may be residing in the same country