Twenty diesel generators in parallel - what problems to expect

[ters]

Case:

Twenty diesel generators are connected through step up transformers to a common MV bus, off grid. Generators are similar but not exactly the same, with power output ranging 1000 – 1200 kVA. Governors and AVR are also different on some units. All units have PM pilot exciters. After all of them are synchronized they have to deal with picking up some large loads. The largest load is a 5 MW motor driving a pump (load at nominal speed is abut 3.5 MW). How likely is that there could be some severe problems starting the largest motor (direct start, inrush current about 4 In). We are attempting to model this using Cyme, but cannot be sure as we do not have all input information (no data on governors and AVRs, such as time constants, curves, etc) and results are contradictory depending what typical AVR and governor specs are selected. Common sense suggests that a 20 MW plant should be able to deal with a 5 MW motor with ease.

 

[waross]

Governor response times should be close. A mix of hydraulic and electronic governors may be problematic.
With hydraulic governors the oil must be approximately the same temperature before large block loads are applied. Putting hydraulic governor sets online immediately prior to starting the large motor before the governor oil is hot may be problematic.
respectfully

 

[catserveng]

How are you doing kW and kVAR share? Do you have active controllers or in droop? With a large number of smaller units you can see a noticable difference in frequency and voltage response to a block load being applied, and with that depending on how well they "share" is going to determine how well they respond to your large applied load.

I have done systems with up to 30 paralleled units, and in some cases with a larger mix of dynamic response we saw poor system response with a 20% system capacity load step applied, resulting in several units tripping and usually dumping the entire plant.

We have also had success in applying large step loads to large systems, both active controlled and in droop. Either way setup of the system will be critical, you will need to assure dynamic response of the units is as close as possible to each other for both frequency and voltage, including same droop slopes (as close as possible) and gain and or stability settings (again as close as possible).

Our best success in large networked systems has been with active controllers like the Woodward DLSC managing both kW and VAR share, but there is a substantial amount of work required to get everything working well together.

You'll also need to make sure whatever protective relays installed on the generators have been set to allow transient response without unwanted trips.

Good luck,

 

[rbulsara]

""Common sense suggests that a 20 MW plant should be able to deal with a 5 MW motor with ease. ""

Yes, I would agree.

 

[rbulsara]

If all alternator pitches and impedances are not matched you will have some serious circulating current issues. At least the winding pitches must be matched, impedances shoudl be close to each other.

 

[TurbineGen]

"Common sense suggests that a 20 MW plant should be able to deal with a 5 MW motor with ease."

I'd agree here so long as your plant is not highly loaded when you try and start it. If your 20 MW plant already has 15 MW of load and you try to start the motor, you may have some issues. Depending on how much is available when you go to start that motor, you might consider a reactor to start it.

------------------------------------------------------------------------
If it is broken, fix it. If it isn't broken, I'll soon fix that.

 

[ters]

As always, this forum is really great. Thank you very much to all. It seems that most of our units will be Cat model 3512 B 1200 kW. We are unsure which AVRs and governors will be furnished. catserveng your nick implies that you may be working with Cat , so would you (or anyone else) know what AVRa and governors are typically supplied for this model and where I can find more technical details for it. The modeling software (Cyme) is asking all sort of questions such as regulator time constants, gain, saturation, min and max values, etc. With some typical AVR characteristic for rotating brashness exciters the software indicates that the voltage drop would be about 20% when a 5 MW motor is started (as the only load initially) but it would recover within about 20 cycles and stabilize after some 150 cycles.

As for the load pickup, the largest load would be picked up first. We do not know (yet) what provisions will be available for load sharing (perhaps only droop), this is a kind of emergency project, no clear definitions yet.

 

[catserveng]

The 3512B is an electronically controlled engine with it's own governor. A dealer technician using CAT ET should go in and check all the gain settings and make sure they are the same.

The voltage regulators can be a mixed bag, usually VR3's and older DVR's, newer or recently repaired units will have VR6's or CDVR's.

If these ar CAT power modules they will likely have CAT switchgear in them with Woodward Loadshare modules, and kW share will be across the load share lines (a shielded twisted pair daisy chained between all the units, you may need to break these into "blocks". You will need to make sure all the load gain settings in the load share modules are correctly set.

VAR share will be a larger concern, as usually power modules are just set for droop, and depending on where the units came from and what they were used for before these settings are likely all over the place. Try to get the voltage droop settings as close as possible, best would be to use a resistive/reactive load bank and load each unit to a constant load, and adjust the load gains and voltage droop all the same, this would give best system performance. I would suggest you do NOT try to use cross current compensation on the CAT AVR's, especially if they are mixed types and a large system. A voltage droop setting
of 3-5% has been my best experience, unless you want to go to the expense of an active VAR controller for all the units, but time consuming and can be pricey for that many units.

You will need to get with your local dealer and get generator performance info based off the serial numbers, this information can be gotten from the TMI system, this should provide the data you are asking for.

How are you dealing with the step up transformers? One per unit or a larger GSU for a block of units?

Depending on what switchgear is in the modules, you will need to review the settings, "utility grade" modules had pretty tight protective settings assuming parallel with grid, and with a large transient may cause a problem. Stanndard modules use the protection seetings in the generator control panel, just make sure they get reviewed and reset so they're all the same. Modules having a way of getting "tweaked" on a per job basis, and gathering up 20 for your project probably means they are coming from a few different sources with different rental histories.

Good luck

 

[ters]

Catserveng

Thank you very, very much for your help. Yes, the equipment is supposed to be a Cat package, so each unit would have its own step transformer and a switchgear with interconnections between units on the MV side. We are not aware where the units would come from and what they were used for before (they are not new in any case).

The plant will eventually tie to an incomer of the exiting MV industrial distribution system presently fed from the grid. Based on your very informative description, and the fact that units would occasionally be on the grid as well (but have to also be set to operate in island mode) seems that tuning up all AVRs, governors and protection will be rather very challenging. Protection relays at the existing MV switchgear will probably need dual settings (for on and off grid) as well. The short circuit level on the low voltage side (when on grid) may also be a potential problem.

This is not a real project yet, the exact number of units is still undetermined as well (anywhere between 20 and 30) but it may materialize soon depending what kind of compromise the Client will feel comfortable with.

 

[itsmoked]

Just curious ters, is this a semiconductor type gig?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[ters]

itsmoked, forgive my ignorance (or English), but I'm unsure that I understand meaning of gig... Could you clarify please.

 

[catserveng]

One of the biggest challenges in a job like yours is dealing with neutral grounding of the generators. The units will likely all come in with a bonding jumper and if it utility grade switchgear in the module, a ground overcurrent device.

If this was strictly a large stand-alone or islanded project, we would usually recommend grounding resistors. If it parallels to grid or large installed system, then we have to go a little further and see if grounding reactors are required. It usually ends up as a compromise since these setups are normally short time, cost and or time constrained.

In general our best success has been in using step up transformers that are Delta connected on the low side and Wye on the high side. I know there are many different opinions on that subject, but there are a number of factors that come into play, and the entire system should be reviewed by a competent electrical engineer with distributed generation experience to come up with the best solutions for your site and operation.

I have run into these people on a few of the large or more complex jobs we have done, the principle was very good to work with and is not associated with any manufacturers, and they have published and presented a number of papers to IEEE on the subject of distributed generation.

http://www.novaenergyspecialists.com/novahomepage.html

You might look at their website and search IEEE for some of their papers to help get you going in the right direction.

 

[itsmoked]

ter; Sorry your English is so good I didn't realize.
gig => job or task.

As in, "Hey, Bob I have a gig delivering pizza".

I was curious what application would be contemplating a twenty generator setup. I know that Hewlett Package feels a power glitch at one of their fabs costs them 35 million dollars per interruptions. I also know that a lot of semiconductor fabs are in vulnerable locations. A few years ago a submarine cable was damaged and an island covered with fabs was off line for a few months with disastrous results.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[catserveng]

We usually set up "generator farms" for utilities suffering from short term capacity issues, for large industrial plants doing upgrades or utility changes, and for government/military projects.

Modular units can be shipped relatively quickly, set up in combination with other units, and provide a fairly large block of "temporary" power for lower cost. Mostly larger diesels 1.5 to 2 MW, and smaller turbine packages. Usually the contraint is package size in this business, as they have to be fairly portable by normal means, like road and ship transport, although I've done a few jobs where they have been airlifted when the need is critical.

Just another piece to the power puzzle.