Gas Engine Governors (DEGOV - Typical Settings??)

[stevesummers]

Hi All,

bit of a how long is a piece of string question... but i have been tasked with looking into some gas engine generators and doing some preliminary stability studies to see what they are capable for in terms of motor starting etc.

In terms of project background, there are 2x 2MW sets running, to act as prime generation for an islanded system, one running as isoch / speed control, the other in droop when its needed. The problem is that the generator supplier is not being helpful at all, and for someone reason don't seem to be able to give any straight answers to me.

I assume that as a gas engine is pretty much identical to a diesel engine in design, we could use a standard DEGOV / DEGOV1 type model based off the Woodward standard. But I also know that gas engines, really don't like load fluctuations and the governors have to be set down pretty tightly to stop it damaging the engine... so my question is does anyone have any ideas what sort of settings to start with - or where to look.

I have attached a grab of a typical DEGOV model - i am guessing that we would wind down the gain factor K10, to 25, or even higher... I have a few load step capability diagrams, so not sure if i could extrapolate backwards from these... any thoughts?

Thanks in advance
Steve

 

[waross]

You will get better answers if you use the term droop.
All generator people are familiar with drop, but if their only exposure to proportional control is governors some may not be familiar with the terms "Gain' nor with "Proportional band".
The proportional band will be 3% for islanded plants and 5% for grid connected co-generation.
A gain of 33.33 or 20.
This is a hard number, not a range.
The proportional band must match on all paralleled sets.
Governors are described in droop rather than proportional band.
x% droop = (x% proportional band plus x% offset).

eg:
At 50 Hz, 3% proportional band results in a no-load frequency of 50 Hz and a full load frequency of 48.5 Hz.
At 50 Hz, 3% droop results in a no-load frequency of 5i.5 Hz and a full load frequency of 50 Hz.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[stevesummers]

Hi Bill,

thanks - but i think we are talking at cross-purposes slightly. To my understanding a DEGOV model, is a speed control / isochronous governor with a fixed droop setting of 0%. The DEGOV1 model, specifically has an additional control loop / mode to allow it to run in droop mode, and has a specific setting where you can enter the droop response in it.

My understanding of the various time constants and gain factors, was to limit the ramp-up rate and load acceptance of the machine without it misfiring or knocking.

 

[FreddyNurk]

To be honest, with modern capability these days, I'm not sure what the benefit would be in running one set as isoch and the other droop, particularly if the two sets are the only source of power for an islanded network. Most of the synchronising gear I'm aware of will also handle load sharing or parallel isoch operation, in which case (if the sets are the same) then its one model and the same behaviour whichever set is running.

I'm aware it doesn't really answer your question though. As far as the topic is concerned I am aware of one entity that actually modelled set behaviour for such purposes, none of the OEMs I am aware of in this size range had any concept of what model parameters are applicable to your sets, which may partly explain the lack of support from the OEM.

Can you advise what you're intending to achieve with the model? Is it compliance for a larger network and the system is now what is known as a microgrid, or is it an islanded network and you're looking at getting an idea of what you can start against tolerable voltage and frequency deviations?

EDMS Australia

 

[stevesummers]

Hi Freddy,

the ultimate aim of the information is to get an accurate representation of the engine capability so we can look at motor starting and stability, in one of the other generating sets trips, so we can identify how much load we can shed.

I have a load step capability curve for the machine, which i have done some simple hand / Excel calculations from - but ultimately we will need a proper power system study in ETAP or DIgSILENT Powerfactory etc.. to demonstrate the machine response, whilst considering inertia, and the other items of plant.

 

[Hoxton]

With modern turbocharged spark ignition gas engines, the load acceptance can be very poor compared to a direct injection compression ignition Diesel engine.

Very often they come with a dedicated engine controller, and do not have a separate governor.

Think of where the power comes from to start motors: with a Diesel engine you squirt some more fuel into the combustion space; with a gas engine the throttle has to open, extra gas has to be sucked into the combustion space by the turbo, which has to wait for this extra gas to be burnt and drive the turbo faster to suck more gas in.

One solution is to start the motors with diesel generators then start the gas sets.

The gas engine supplier is probably disinterested because he can see the problems ahead and wants someone else to get the business!

 

[catserveng]

To get a better idea of what you might be able to model you would need a lot more information about the prime mover, its governing, and fuel/air management systems. And in the end, your model might be close but not really, especially the first time someone make dynamic adjustments to speed or air/fuel dynamics. You will also find that voltage regulation dynamics will come into play, but usually to a lesser extent than dynamic adjustments for the prime mover. Also environmental conditions, such as altitude, humidity and ambient temperature greatly affect some lean burn gas engine transient response characteristics.

So what kind of engine?
Is it operating as lean burn or rich burn?
Do you have site emissions requirements?
What are your steady state stability requirements?
What types of loads are you working with?
What kind of generator end? (in working with units in the field have found units with larger/heavier rotors can impact transient response as well)
What kind of fuel? Is fuel quality stable?

Also in agreement with using modern controls for real and reactive power sharing, and you will find that load control system dynamic adjustments also play a big part in systems ability to accept and reject loads. My own experience using droop or mixed droop/isoch control on multiple engine systems with only gas engine powered units online gets to be an operational nightmare.

In highly turbocharged lean burn engines you will also find that in many cases, load rejection is a bigger problem than load acceptance, and is greatly impacted by air system design, exhaust back pressure and how the engine manages excess air.

Yes, it can be done, but takes more than a computer model since so many variables exist in the actual operation.

Hope that helps, MikeL.

 

[stevesummers]

Hi Mike,

the project is in 'concept' stage so a lot of some details are a bit sketchy, what i am trying to do is get a sense of how viable the approach is, what the gas engines are caapble of and what they are not of, and then put some mitigation in place for the electrical design, so that what we can mitigate the risks. So this might be options like load shedding, a load bank for the engines to chew on in load rejection light load cases, battery storage to stabilize the system during load changes. So yes this won't be the final answer and there will be more development to do, but i am bench marking it for one for a better term.

In order to answer your specifics, its a 2.5MW gas engine, that has been de-rated to 2MW due to altitude and fouling. Generator nominal speed is 1500rpm, 50Hz system, running on a pretty lean gas taken off from an oil extraction well. Gas would be a bit mixed, but we are not particularly looking too closely at that side at the moment. Inertia of the engine is 81kg.m2 and inertia of the alternator is around 95kg.m2, so total inertia constant of 1.1. The generators are running a small islanded Oil & Gas plant, so a mix of induction motors and static loads.

I am not too concerned about the transient / dynamic stability side of things, as that can be managed with protection - its more just the load step (and rejection) that is of interest, and i need to try and get a starting point of figures for the governor so i can see what sort of response we get i.e. do we need to stick all motors bigger than 50kW onto a soft starter, or does it need to be 37kW etc etc.

So to reiterate i am just looking for some typical figures or a starting point on how to model the governor, so its at least in the right sort of response, even if its only about 80% accurate.

In terms of shared isoch vs droop, yes that's fine, i understand and will look at that later. I have models of the AVRs and can also tune them, as i understand their limits reasonably well, its more the mechanical governor side of things i am less confident on.

 

[waross]

Are you familiar with PID controllers? (Proportional, Integral, Derivative)
A droop controller is a Proportional only controller. (With a little offswt.)
An isoch controller is a proportional controller with Integral added.
If you are familiar with control actions you will be familiar with the textbook curves:
A proportional controller (droop) responds to a system upset with a typical overshoot/undershoot curve and settles at a new process variable slightly higher or lower than before the upset.
An isoch controller responds to a system upset with a typical overshoot/undershoot curve and settles at a new process variable slightly higher or lower than before the upset. The integral function then becomes active and corrects the output to remove the offset.
I imagine that it is possible to design a load control panel which controls the fuel to an engine based on the measured load, but such a system would still need a frequency dependent governor for fine tuning.
My point is that it may be worth the time to spend an hour or so exploring the functions of PID controllers with particular emphasis on the Proportional function.


Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[catserveng]

I can't help you with modeling, frankly for all the years I have dealt with SI engines from multiple manufacturers, modeling of SI lean burn turbocharged engines never seemed it get it right. Even with significant amounts of test cell time devoted to trying to predict load step response and best possible control settings, by the time it gets to the site, with environmental, ambient, fuel and operational related impacts, it never seems to quite work out as expected.

Not that it can't be done, I am currently at a site with four 2MW lean burn miller cycle engines that run an island mode hotel and casino complex, including their mechanical plant. It's stability and reliability exceeds the local utility in this area. But it came with a lot of effort.

Many oil and gas sites, including drilling, pumping, salt water reinjection, and site power generation all work pretty good on gas engine driven generators, but issues like black start, differing response characteristics across the load range and changes in engine response as it ages. You also throw in a crappy well head gas fuel source, that can have a wide energy content variation and contain liquids and heavy hydrocarbons that can cause erratic operation and detonation.

Personally I'd forget trying to model response and find operators who are using similar engines as you are attempting to use in similar applications and see what their experiences can provide you.

MikeL.

 

[stevesummers]

Hi Mike,

thanks for the practical insights, its a bit of a tricky one, intuitively i can see all sorts of issues. I just need to frame some sort of report with a bit of empirical data and outputs from ETAP or Digsilent to justify the position.

 

[waross]

You may be barking up the wrong tree.
Governor response to a step change in load is so fast compared to the engine response to a step change in throttle or fuel feed rate as to be almost immaterial.
Forget the governor and concentrate on engine response.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!