Load sharing during multi-unit island droop operation

Good day,

I have two gas engine generators rated at 5,200kW each, running parallel to the grid.

In parallel mode, both are running in droop mode 5% with kW output control. DCS controls the load sharing between the two generators by sending the kW setpoint to each generator, and the engine control system adjusts the droop curve accordingly.

In Island mode, there is no control from DCS. Both engines change the operation mode to droop mode 5% with speed control. There is provision for speed raise/lower command from DCS at engine control system, but it was not configured.
There is no communication between both engines. The engines are not aware of individual generation and the total house load.

Until recently, the islanding process happened during both generators running at equal load. For example, house load is at 5MW; each engine runs at 2.4MW while importing 200kW from the grid.
When island, each engine takes the load from the grid equally. Due to the stable house load (electrolyzer plant), the generators can maintain the frequency with no intervention from the operator, operating in the droop mode.

However, recently, the grid connection was lost during the generator start-up. During this time, the load sharing was at 75% on generator 1, 25% on generator 2. After one hour without intervention from the operator, the generator trip due to underfrequency. Based on the 1s trending data, it was +/- 2% fluctuation. (maybe higher in finer resolution, as the underfrequency setting is at 5%). The typical island operation with 50/50 load sharing has only +/- 0.5% frequency fluctuation.


Does the imbalance load sharing prior islanding contribute to this frequency instability? In my understanding, the droop curves are different due to the load during the islanding.
How do the engines react from the additional load released by the grid? Are they going to share the load in 75/25 manner?

I tried to visualize this problem with the tandem bicycle analogy, this time it is an e-bike.
With full loaded rack, both riders cycle the same cadence/rpm of the electric motor. Rider 1 inputs more power (75%), while rider 2 inputs 25%. Once the electric motor goes off, I couldn't visualise how the riders cadence will be.

Thanks!
 
wanfauzan,

1) Which unit tripped due to under-frequency: generator 1 or generator 2?

It's VERY difficult for me to understand how the two units can maintain frequency when separated from the grid which was supplying 200 kW. If the two diesels have identical control systems with identical programming and each was previously supplying 2.4 MW prior to the separation then when the separation occurred the load would STILL be 5.0 MW--but the fuel flow-rate would only be capable of supplying 4.8 MW at rated frequency. Once the 200 kW was lost, the two units--operating in Droop Speed Control--would each increase their power output by 100 kW, which would cause the island frequency to decrease. (YOu can do the maths--there a deficiency of 200 kW out of 5.0 MW. It's all percentages.) Unless an operator intervened the frequency would decrease slightly.

It's the same with your e-bike analogy. If the two riders kept applying the same force to the pedals when the electric motor is shut down, there will be a decrease in the speed of the e-bike because of the loss of the torque from the electric motor.

You didn't say what happened to the island load when one of the two generators tripped. If the total load on the island dropped by the 25% the one generator was carrying OR the island load shifted to the other generator which was originally carrying 75% of the island load, then it's conceivable the generator with the smaller amount of load would unload to less than zero load.

Again, my description presumes BOTH diesel control systems are similar--if not identical--and both have the same configuration (programming; set-up; adjustment; tuning). Sometimes, even identical prime mover control systems, configured identically, will drift a little during unattended operation. If the fuel rack is hydraulic, it's possible the hydraulic fluid is not equally clean in the two actuators. It could be one or more of several things that can cause a little drifting, and even small changes in load--which WILL affect frequency if not attended--can also cause load shifting/governor drift (some people call it drift, but it's really just part and parcel of Droop Speed Control and the inherent differences in machines and machine conditions). Ideally, all things being equal, the load should remain stable on the two machines if the island load is not changing or is only changing by 100 kW or so. But if the units are really in Droop Speed Control, and are not being controlled externally or adjusted manually by operators the island frequency will change with any changes in load (of 100 kW or more, probably--that's 0.1 MW, which is not insignificant).

Hope this helps! This isn't rocket science (as a former colleague used to repeatedly say). It's not intuitive, but you are going to get to the right understanding if you keep working at it. When a unit (or units) are operating in parallel (synchronized together supplying a load) AND neither unit is in Isochronous (or some kind of Isochronous load-sharing) and there is no external load control monitoring frequency and the unit (or units) is (are) supplying an islanded load independent of a large grid then any changes in load--any changes in load--are going to cause the island frequency to change. The LOAD changes--not because the fuel flowing to the unit changes, but because motors or electrolyzers or auxiliary equipment start and/or stop--and the unit (units) must provide the load required. But, if the fuel doesn't change as the load changes then the frequency of the island will change. That's AC power generation--how it works and how it's always worked. It takes a certain amount of energy (fuel flow) just to get to and maintain rated frequency. Add some load to the generator and it requires more fuel flow to maintain frequency AND supply the load. If load is added but fuel flow is not changed, then the energy required to supply the additional load is "diverted" away from maintaining rated speed (frequency) and directed to the requirements of the load. There ain't no such thing as a free lunch.

A person, or persons, on a bike (regular or e-bike) trying to maintain a specific speed while transporting a load in the basket have to supply the required torque to move the load at the desired speed. Add load without increasing torque, and the speed will decrease. Or, add load to the basket without increasing torque and the speed will decrease. In the case of the e-bike, shut off the motor without the rider(s) increasing the torque being applied to the pedals and the speed of the bike and the riders and the load in the basket will decrease.

It's as simple as that.

Again, hope this helps!
 
Thanks for the reply CSA!

1) Which unit tripped due to under-frequency: generator 1 or generator 2?
Unit 1 trip first by underfrequency, Unit 2 followed due to lube oil pump suddenly stop (3 seconds later), suspected due to the bus disturbance, after Unit 1 trip.


It's VERY difficult for me to understand how the two units can maintain frequency when separated from the grid which was supplying 200 kW. If the two diesels have identical control systems with identical programming and each was previously supplying 2.4 MW prior to the separation then when the separation occurred the load would STILL be 5.0 MW--but the fuel flow-rate would only be capable of supplying 4.8 MW at rated frequency. Once the 200 kW was lost, the two units--operating in Droop Speed Control--would each increase their power output by 100 kW, which would cause the island frequency to decrease. (YOu can do the maths--there a deficiency of 200 kW out of 5.0 MW. It's all percentages.) Unless an operator intervened the frequency would decrease slightly.
Sorry for the lack of communication. Yes, indeed, the frequency decreased slightly. What I was trying to say, that there is no significant fluctuation of the frequency. I think, I should've expressed the fluctuation -0.1% instead of +/-0.1%.

You didn't say what happened to the island load when one of the two generators tripped. If the total load on the island dropped by the 25% the one generator was carrying OR the island load shifted to the other generator which was originally carrying 75% of the island load, then it's conceivable the generator with the smaller amount of load would unload to less than zero load.
The Unit 1 tripped first by underfrequency, followed by Unit 2 by lube oil pressure LL due to pump trip. The period between those two instances was only 3 seconds.

Again, my description presumes BOTH diesel control systems are similar--if not identical--and both have the same configuration (programming; set-up; adjustment; tuning). Sometimes, even identical prime mover control systems, configured identically, will drift a little during unattended operation. If the fuel rack is hydraulic, it's possible the hydraulic fluid is not equally clean in the two actuators. It could be one or more of several things that can cause a little drifting, and even small changes in load--which WILL affect frequency if not attended--can also cause load shifting/governor drift (some people call it drift, but it's really just part and parcel of Droop Speed Control and the inherent differences in machines and machine conditions). Ideally, all things being equal, the load should remain stable on the two machines if the island load is not changing or is only changing by 100 kW or so. But if the units are really in Droop Speed Control, and are not being controlled externally or adjusted manually by operators the island frequency will change with any changes in load (of 100 kW or more, probably--that's 0.1 MW, which is not insignificant).
The control systems are identical with same tuning parameters. The fuel gas governors are electric driven. Both units are recently commissioned in 2020. So, I think it is safe to say that both units are identical.

I agree that the additional 100kW load to the system will change in frequency, but seems insignificant to the operation.

A person, or persons, on a bike (regular or e-bike) trying to maintain a specific speed while transporting a load in the basket have to supply the required torque to move the load at the desired speed. Add load without increasing torque, and the speed will decrease. Or, add load to the basket without increasing torque and the speed will decrease. In the case of the e-bike, shut off the motor without the rider(s) increasing the torque being applied to the pedals and the speed of the bike and the riders and the load in the basket will decrease.
Thanks for the neatly put analogy. In the event of rider 1 input more torque than rider 2 (let's say 75/25), I wonder how the additional load will be shared once the electric motor shutoff. Since the riders are not talking to each other, and no electric motor to dictate the cadence. Both riders will reduce the cadence as the load increases. Since the droop setting is the same (same gradient), I think the rate of change of cadence and rate of change of load are the same. So, both riders will share the additional load equally, regardless how much torque they put, prior to the electric motor shut off. Is this analogy on the load sharing is correct?



Thanks so much!
 
wanfauzan,

When asking about the load after the separation from the grid, I wanted to know what the load did. Did it remain constant at 5.0 MW, did it increase or decrease, or was it fluctuating slightly? How long were the two units operating in island mode before the trip occurred? It seems you have some kind of once-per-second historical data; this should be easy to answer.

As for the e-bike analogy, I was presuming the bike and riders and goods were riding on a flat, level road at a constant speed--as indicated by some kind of speed indicator (cadence would be a similar indication--so riding at a constant cadence would be akin to a constant speed). If the e-bike motor was providing some of the torque to move at a constant speed (cadence) and then it was shut off then the two human riders would either have to increase the proportion of the force they are applying to their pedals in order to maintain speed (cadence) or the speed of the vehicle and goods and riders would decrease (if they didn't apply more force to their pedals). And we can't presume the riders have identical force-producing muscles in their abdomens, legs and and ankles, nor without some kind of meter (speed-o-meter, or cadence meter) would they be able to maintain a constant speed (cadence) without some kind of coordination (conversation; screaming; berating; etc.). And that is the difference between machine Droop Speed Control and human Droop Speed Control: speed is actually factored into the equation for machine Droop Speed Control. When two generator-sets are synchronized together (I could have written "operating in parallel"--but synchronization is a MUCH MORE descriptive and powerful indication of the condition), operating in island mode supplying an electrical load (or loads) independently of a larger grid with other, potentially larger, generator sets and both gen-sets are operating in Droop Speed Control without any outside adjustment (automatic or manual) the reference for fuel flow to the prime movers is speed. The thing about Droop Speed Control is it's proportional control--the amount of fuel (energy) input to the prime movers is a function of the difference (the error) between the speed reference and the actual speed. When the fuel flow (energy input) to the prime movers is such that the load the unit(s) are carrying equals the requirements of the load being supplied at the desired frequency (and remember: speed and frequency are directly related!) then the island is operating properly. However, if something happens to change that balance of load and speed (frequency), one has to suffer. If the electrical load of the island increases without increasing the fuel flow (energy input) to the prime mover(s) then the frequency will decrease. The load doesn't decrease--it increased, actually--and the generator(s) and their prime mover(s) must supply the load of the island. And, the only way the gen-set(s) can do that is by decreasing the speed (frequency) of the gen-set(s); a portion of the torque being provided to generator(s) from the prime movers (which is a function of fuel flow (energy input)) gets converted into more amps--which means the speed can't be maintained.

Just like on the bike. If the load in the basket of the bike suddenly increases but the forces on the pedals aren't increased the bike and the load it's carrying will slow down. The load increased and if the pedal forces aren't increased then some of the force being applied to the pedal will be used to "carry" the increased weight of the load in the basket, leaving less force for maintaining speed.

Without some kind of speed (cadence) indicator--and without any assistance from the electric motor of the e-bike--the speed of the e-bike and the riders and the loads is probably going to vary. And, if the riders aren't communicating with each other, it's going to get "ugly" (the speed, that is).

AC power generation is not only about load--it's about carrying load (producing amperes) at a relatively constant frequency (speed). It's a balancing act--just like carrying loads in a basket on a bicycle (e- or otherwise) at a constant speed (cadence). It's NOT just about the load; it's also about the speed.

Hope this helps!
 
CSA

When asking about the load after the separation from the grid, I wanted to know what the load did. Did it remain constant at 5.0 MW, did it increase or decrease, or was it fluctuating slightly? How long were the two units operating in island mode before the trip occurred? It seems you have some kind of once-per-second historical data; this should be easy to answer.


It was difficult to determine the total load as it fluctuated once disconnected from the grid.

The two units operated for around 1 hour before both unit trips. I have discovered today that the unit 2 was not tripped due to the lube oil pressure LL. Instead, the operator has manually commanded the 52G CB to open, intending to transfer all the load to unit 1. Consequently, unit 1 tripped due to underfrequency, which led to total blackout.





However, if something happens to change that balance of load and speed (frequency), one has to suffer. If the electrical load of the island increases without increasing the fuel flow (energy input) to the prime mover(s) then the frequency will decrease. The load doesn't decrease--it increased, actually--and the generator(s) and their prime mover(s) must supply the load of the island. And, the only way the gen-set(s) can do that is by decreasing the speed (frequency) of the gen-set(s); a portion of the torque being provided to generator(s) from the prime movers (which is a function of fuel flow (energy input)) gets converted into more amps--which means the speed can't be maintained.


The fact that both generators are synchronized means that they will run at the same frequency. Having the same droop setting, I understand if both generators are runnning with constant frequency and output, provided that the load is stable. However, from my observation, the generator load, current, and frequency fluctuated very much during island operation. However, I couldn't verify whether the plant load was actually stable, since it was just information from the operator. They don't have any indication, except the kW reading from the generators, which fluctuated.



I believe picture speaks thousand words. I attached here the trend screenshots for your reference.

I still couldnt explain the fluctuation in the frequency and kW during the island operation. I can only think of unstable house load, that contributes to the fluctuation. Although both units were running in droop control (no isoch unit), there is no explanation from the droop control point of view, how the frequency can fluctuate, as both units had the same droop setting

Thanks much for your valuable feedback!

ControlcomUnit1.png

ControlcomUnit2.png
 
wanfauzan,

That explains how the more heavily loaded unit tripped on underfrequency (when Unit 2 was tripped).

Is the the typical profile of operation when the two units are synchronized and operating as an island? The frequency changes as quickly as it seems to on these charts? Or, only when the load was unbalanced as it was on this event?

Are you sure the electrolyzer(s) and other loads (I presume pumps and such--motor-driven equipment) were stable? We don't know what kind of governors the units have, and how they have been tuned. It's possible the adjustment ("tuning") of the governors could be changed to make the island more stable.

But without being able to see other operating charts it's really difficult to say. The amount of load appears to have been fairly stable, so it's probably just a governor adjustment (on one or both governors) that needs tuning.
 
wanfauzan...
How old are the units and when were hey last serviced?
Phil Corso
The units were commissioned in Feb 2021. Last service was done at 6000 hours last December.


wanfauzan...
Please provide Engine and Generator Data Sets.
Phil Corso
Gas engine Kawasaki KG-12
Type: Lean combustion, Pre-combustion chamber with spark ignition, 4 stroke
Gen output : 5200kW
Nominal speed: 750rpm
No of cylinder: 12, V type
Cylinder bore: 300mm
Cylinder stroke: 480mm
Turbocharging system: constant pressure turbo charging system

Below is the generator nameplate:
Generator nameplate.png
 
Is the the typical profile of operation when the two units are synchronized and operating as an island? The frequency changes as quickly as it seems to on these charts? Or, only when the load was unbalanced as it was on this event?
This is not the normal behavior. In fact this is the first occurence of this kind of fluctuation. I forgot to mention that the operator tried to backsynch with the grid after the reset the reverse power relay. However, to no avail. I suspected, the backsynch panel was not able to match the generator frequency with the grid, due to heavy fluctuation.

I attached the behavior during island operation commissioning. There is a little frequency fluctuation, but not as much as the previous chart. The correlation between load increase and frequency drop can be clearly seen.


Unit1 island operation - comm.png
Unit 2 island operation - comm.png

Are you sure the electrolyzer(s) and other loads (I presume pumps and such--motor-driven equipment) were stable? We don't know what kind of governors the units have, and how they have been tuned. It's possible the adjustment ("tuning") of the governors could be changed to make the island more stable.
This remains my biggest suspect, that the plant load was not stable. But I am not able to prove this, and the operators don't have any data on the plant load (except the fluctuating kW output reading at generator panel).
 
wanfauzan...
Two additional questions:
1) Ambient conditions where units are located?
2) Can you produce view of engine-speed signal-pulses for several operating conditions:
a) Isolated.
b) Paralleled.
c) Synch'd with Grid.
Phil Corso
 
wanfauzan,

This is most likely what Phil Corso is alluding to:

ONE OUR OF TWO PARALLEL GENRATORS SHOWING REVERSE POWER ALARM | Automation & Control Engineering Forum

And when he writes combustion-driven generator I think he really means combustion engine-driven generator. But, I've been wrong before, and I'll be wrong again (for a few more months).

But, looking at the graphs you sent, it would appear that even when synchronized the frequency isn't really very smooth.... And, it can clearly be seen that as load increases, the frequency decreases--especially when the units are separated from the grid (running as an island). Further, if I interpret the graphs correctly, I don't understand how two generator-sets, SYNCHRONIZED together and operating as an island independent of the grid can operate at different frequencies....?.!.?.!.?.! THAT'S pretty strange. Maybe I'm interpreting your original post incorrectly. Anyway, something doesn't look quite right--perhaps it's just the instrumentation (and I'm not going to spend any more time on this issue).

I don't really have a lot of experience with combustion engine-driven generators but I have to believe that especially larger generators are going to have slightly more irregular frequency waveforms when operating in island mode than, say, a turbine (steam or combustion (gas))-driven generator, because of the pulses from the cylinders. I might also wonder if these pulses and frequency oscillations could be exacerbated if there was a unique situation--akin to harmonics in piping systems--between to combustion engine-driven generators.

I'm thinking digitally, and if anyone had ever seen anything like this--it would be Mr. Phil Corso.

Anyway, best of luck! It seems a poor method of operating in island mode--both machines in Droop Speed Control, without some kind of power management/frequency/load control or even having human operators make some adjustments though in your case it almost seems the operators aren't well trained and may have made a mistake. Early in my power generation career I was sent to site to investigate a generator rotor (field) ground that had caused a 600 MW steam turbine-generator to trip at almost full load. It was a 30 year-old plant in the mid-1980's with very little instrumentation and no kind of chart recorders (that worked!). After two days of collecting data and speaking to operators and measuring resistances and checking shaft-grounding brushes and such, a kindly older gentleman took me aside and said, "Whenever there's an operator-initiated problem in the plant it gets blamed on a generator field ground. And, management brings in a manufacturer's representative to write a report and that's the end of that. The unit gets re-started and the report goes in a file cabinet never to see the light of day again." So, in effect I was being told I wouldn't find anything, to write my report and send it in, and not give this another thought. Once the report is in hand, the unit can be restarted (it takes almost two days to restart a unit that large--VERY big natural gas-fired boilers and drums and pipes which have to be warmed-up and vacuum raised on the condenser before steam can begin to be admitted up to the stop valves). And, no one gets blamed or loses their job and everything goes back to normal (and lots of people get overtime pay as the unit gets restarted.) This happens more than one would think it would, or should. (I think it's one reason why alarm printers are never working at some plants.... no record of anything. Though with electronic alarm history files that's getting harder and harder to hide these things.)
 
wanfauzan...
Two additional questions:
1) Ambient conditions where units are located?
2) Can you produce view of engine-speed signal-pulses for several operating conditions:
a) Isolated.
b) Paralleled.
c) Synch'd with Grid.
Phil Corso
Phil Corso,

There is no instrument installed to measure the ambient conditions. But on average, the ambient temperature is 35 deg C, RH 60%, 10m AMSL.
a) isolated (unit 1 single unit in island)
Unit 1 single island.png

b) parallel (unit 1 & unit 2 parallel without grid) - you can refer to the individual chart in the post above. this chart I combined both units in a single chart.
multiunit island operation - commissioning.png

c) sycnhed with grid
Unit1 synch with grid.png

Thanks!
 
CSA,

But, looking at the graphs you sent, it would appear that even when synchronized the frequency isn't really very smooth.... And, it can clearly be seen that as load increases, the frequency decreases--especially when the units are separated from the grid (running as an island). Further, if I interpret the graphs correctly, I don't understand how two generator-sets, SYNCHRONIZED together and operating as an island independent of the grid can operate at different frequencies....?.!.?.!.?.! THAT'S pretty strange. Maybe I'm interpreting your original post incorrectly. Anyway, something doesn't look quite right--perhaps it's just the instrumentation (and I'm not going to spend any more time on this issue).
When I overlay the graph from unit1 and unit 2, the frequency seems the same for both unit. The minimum deviation I would consider as instrumentation accuracy, as you pointed out.

multiunit island operation - commissioning.png

I don't really have a lot of experience with combustion engine-driven generators but I have to believe that especially larger generators are going to have slightly more irregular frequency waveforms when operating in island mode than, say, a turbine (steam or combustion (gas))-driven generator, because of the pulses from the cylinders. I might also wonder if these pulses and frequency oscillations could be exacerbated if there was a unique situation--akin to harmonics in piping systems--between to combustion engine-driven generators.
I believe this will be an interesting topic to discuss in the future!

That aside, I agree with you on the poor implementation of island frequency control. I could give a recommendation to have frequency control from PMS, although I doubt that this will be implemented, considering the additional cost, and the fact that they can still live with this shortcoming. I still don't understand the load fluctuation so much during that incident. The only explanation I have is that the house load itself fluctuated.

Nice sharing on the troubleshooting during those days. People tend to blame the machine, as the machine cannot speak up for itself. Nowadays, it is getting harder, as most of the modern units are connected and can be monitored miles away.

Thanks much!
 
wanfauzan,

If you're talking about the total load on the two synchronized generator-sets, that would only vary if the island load varied.

If you're talking about the load on the two synchronized generator-sets "shifting" from one to the other, that would likely be a function of the diesel governor and/or governor settings if no one touched the controls....

If someone touched the controls and did so in error, then that would also explain the problem. But, no one touched the controls, right??? (You already know someone did touch the controls....)

Best of luck!
 
wanfauzan,

If you're talking about the total load on the two synchronized generator-sets, that would only vary if the island load varied.

If you're talking about the load on the two synchronized generator-sets "shifting" from one to the other, that would likely be a function of the diesel governor and/or governor settings if no one touched the controls....

If someone touched the controls and did so in error, then that would also explain the problem. But, no one touched the controls, right??? (You already know someone did touch the controls....)

Best of luck!
CSA, thanks for your conclusion.

After multiple similar occasions (total house load fluctuation), I came to a conclusion that the house load varies despite what the operator said. The generator AC output is connected to thyristor to convert to DC for the electrolyzer. I suspect something is off with the thyristor control. Since it is beyond my scope, I pass the ball back to the plant side.

Have a great weekend!
 
Top