Power sharing issue 2

[matlab2008]

Hello Everyone,

This is rather a simulation question, but it needs a practical point of view.

We have two Diesel Engines. One is working normally and the other is just a stand-by. Both engines are the same ratings (35kW). We need to test the power sharing if they are operating in parallel.

Our problem is when we connect the second diesel - while the first one is running - it does not supply real power because the first one is capable of supplying the entire load the moment they are connected. We tried to increase the frequency setpoint of the second one (1.05 pu) so that it shares some of the load, but the result was unstable frequency! (I think, because both engines having the same rating, increasing the frequency of one system will affect the frequency at the common bus)

Just to give a brief background on our system:
- the two diesel engines are 35 kW systems, and they should supply an isolated load. (stand-alone system).
- the governors of both generators are controlled by a simple PI controller which sense the speed and regulate it to 1 pu.

I would be very grateful if anyone can give some advices on how to share the power between gen-sets.

Best regards,
Mike

 

[davidbeach]

You will not have much luck with load sharing of two generators with isochronous controls. Put them both into droop and you should have much better luck.

 

[waross]

Take the "I" out of the PI controller and set the "P" at 3% proportional band. This is another way of saying what davidbeach said.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[davidbeach]

I knew you'd show up Bill. Is that really what I said? ;-)

 

[matlab2008]

Many thanks Davidbeach and Bill, I appreciate your contribution.

Just to make sure I understand what you mean, let me explain what I understood:

Both governors are using PI before and after the connection. So, if the PI is in the form:

PI= kp*e + (ki/s)*e

where kp and ki are the gains, s is integrator, and e is the error between reference speed (1pu) and the measured speed.

Do you mean when the two generators are connected, I have to set kp gain of the incoming generator to 0.03 and ki to 0? How about the reference speed? should it be 1 pu for both before and after the connection?

Please explain more as I don't have much knowledge about the droop?

Once again, many thanks.
Mike

 

[catserveng]

Go to this site,

http://www.woodward.com/pubs/pubpage.cfm

And find these two publications,

Manual 01302 Speed Droop and Power Generation

Manual 26260 Governing Fundamentals and Power Management

These two should cover all your questions regarding droop and load sharing on diesel engine driven generator sets.

 

[matlab2008]

Thanks Catserveng,

I have checked these documents but I still feel there's something missing!

My confusion is about the frequency/load curve, (or droop curve?). To run the generator on the droop mode, after removing the "i" gain from the PI controller, is it just changing the reference point of my speed to, say, 1.05 pu. I have tried this on a simulation package but the frequency changes whenever I change the set point and no power sharing I could achieve!

I've been working on this issue since long time, but no good achievements have been made!

I hope that I can get more explanations on this issue.

Thanks
Mike

 

[waross]

Standard settings of governors;
Proportional band; 3%
Integral: None.
No-load frequency; 61.8 Hz.
Full load frequency: 60 Hz.
If you apply a 35 kW load to two generator with these settings, the frequency should be 61.9 Hz and each machine will carry 17.5 kW.
To synchronize, adjust the speed of the incoming machine to match the loaded machine.
Close the tie breaker.
Advance the governor control until the machines are sharing the load in proportion to their ratings.
Note;
1) Most machines have no indication of the actual frequency set point, but are successfully paralleled.
2) You can't use current to indicate the loading of paralleled machines. You must use a power meter (kw meter). Any disparity in voltage settings will cause unequal division of reactive current so that current is not a valid indicator of kW load sharing. (The sharing of reactive current may be improved with a quadrature circuit.)
These settings work so well that the overwhelming majority of users don't realize that there is a small frequency variation. Some UPSs don't like the frequency shift. The easiest and best solution is to change the UPS for one that is generator friendly.
DON'T set the No-Load frequency to 60 Hz.
The Automatic Voltage Regulator has a feature called Under Frequency Roll Off. (UFRO) Below about 57 Hz or 58 Hz, UFRO will reduce the voltage proportionally to the reduction in frequency. This unloads the set and gives the prime mover an easier time recovering from speed drops caused by block loading. This feature is especially helpful for motor starting. When the no-load frequency is set to 60 Hz on a droop governor, the full-load frequency will be about 58.2 Hz. You can expect to experience more severe voltage dips when block loads are applied if the no-load frequency is set to 60 Hz.


Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[catserveng]

Probably the next best question to ask is what kind of governor are you using?

If you're using an electronic governor then setting the "I" to zero and the "P" to 3 may not be correct.

In most electronic governing proportional equals the GAIN setting, integral equals the RESET, and if equippped, the derivitive equals the COMPENSATION. The droop setting is usually in percent of rated speed, for example an engine that runs at 1800 rpm to produce 60 Hz has a rated speed of 1800. 3% of the rated speed is 53 rpm, to acheive the desired full load speed the high idle speed isusually set to the rated speed plus the droop, in this case the high idle speed would be 1853 rpm. So to parallel these units you should assure each unit has the same droop setting and response. Set the engine speed the 1853 (or 61.8 Hz as Bill explained above), add load to see the speed decrease at full rated load to 1800 rpm (or 60 Hz).

The governor gain and reset adjustments are used to determine speed stability and governor and engine response to load changes.

In greatly simplified terms, gain setting determines how fast the governor responds to a change in sensed speed, reset determines how much of a change the governor tries to make in the fuel system to respond to the speed change, and compensation (if used in your governor) tries to compensate for how fast the engine responds to a control change. A modern diesel engines responds very quickly and generally has a compensation setting at or near zero, a carburated natural gas engine has a fairly slow response and would have a compensation setting at 40-60% (based on the Woodward governing type systems).

A few more details on what you're using can lead to more precise answers. Hope that helps.

 

[matlab2008]

Thank you very much Guys. Your information was absolutely invaluable.

I think I need to make my query more clear to get most of your help. As I said in my first thread, my question was rather a simulation issue.

Currently I'm in a designing stage, so I'm trying simulate the dynamic performance of the one diesel engine when it is connected to another one to share the available load. (at this stage, I don't wan increase the load, just bringing another unit online!). The issue I had is the power sharing as when the second unit is connected, does not take any load. As I understand from you guys, that was simply because I not putting the incoming unit in the droop mode.

The point here is my governors are simulated by simple PI controllers (e(kp + ki/s)) multiplied by the diesel transfer function. Now, if I want the incoming unit to operate in a droop mode, should I put kp = 0.03 and ki = 0 in the simulation? (in fact, I made this but the result was uncontrolled frequency. So what I did precisely is when the two unit are in synchronizm, I reset ki to 0 and kp to 0.03! then closed the circuit breaker). Even though the PI of the other governor has not been changed, it was not able to control the frequency because "I think" the two units having the same rating.

I'm very sorry for this long thread, and am sincerely appreciating the time you spent to respond to my questions.

Respectfully,
Mike

 

[waross]

Proportional band or Gain or Droop. This is the basic controller response to a change in the measured parameter.
3% droop means that a 100% change in load will result in a 3% change in speed or frequency. After the droop has adjusted the frequency to compensate for a change in load, integral or reset detects that the measured parameter (speed or frequency) is no longer at the set point. It then adds a bias to return the measured variable to the setpoint.
Suppose that we are running in isochronous mode at 60 Hz, with droop set at 3%. If the load is now increased from no load to full load. The frequency will initially drop to about 58.2Hz The Reset or Integral will compare the measured variable (the frequency at 58.2 Hz.) with the setpoint (60 Hz.)

The reset function will detect the difference and bias the output to return or reset the measured variable to 60 Hz.
The first response of a PI controller will be a droop response. The Integral or reset will then correct the frequency back to exactly the set point.
This DOES NOT WORK WELL with two systems in parallel. Use droop only.

You will not have much luck with load sharing of two generators with isochronous controls. Put them both into droop and you should have much better luck.

A load control panel is able to run two generators in isochronous, it does so by sending the same control setting to the fuel control mechanism on both generators. If you are having difficulty simulating two generators in droop, you may not be ready to tackle load control panels for a few years.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[davidbeach]

Is this a school project or is it real work?

 

[matlab2008]

Once again, thanks guys.

Davidbeach, No, it is not a school project, it is a real project. This is a remote area hybrid power system project. Part of the project is to implement duel fuel engines. Our engines should work on diesel/hydrogen. We are modeling the engines from scratch and we want test the performance of the engines when they operate on a certain percentage of hydrogen and diesel.

Any suggestion would help.

Thanks,
Mike

 

[davidbeach]

OK, that's fine. What Bill is telling you is true. So, if you don't get load sharing with both machines in proportional only your model is overly simplified.

 

[matlab2008]

Thank you very much guys, your contribution was absolutely helpful.

I finally figured out the missed joint in your discussion. The way I should set the machine to operate on the droop mode is based on the difference between a reference power and the actual power. This difference should be multiplied by the droop factor then by the frequency set point of the PI controller. In other words, the reference speed (frequency) should be changed based on the difference between the desired power the unit should supply and the actual power.

Once again, many thanks to you.

Mike