CAT3616 with Woodward 2301A; paralleling to grid without governor droop enabled 1

[Mechkane]

I recently started working with a 1990s vintage CAT3616 genset(with Kato generator) that includes PLC-5 controls; I'm trying to come up to speed with this system (no pun intended) and all the fancy accessories they added in the 90s. It has a EGB-29P and 2301A (low voltage model); a KCR-760 (Kato) Voltage regulator, a VAR/PF controller (Basler) set to PF control, SPM-A, Excitation limiter, and a Woodward Load pulse sensor unit. Need I say more..
This genset is only used for backup emergency type power and only is tied to the grid for testing periodically. So testing tied to the grid is 99.9% of what we do with this genset.

I'm more familiar with non PLC Gensets with a 2301A that will enable droop when paralleling with the grid. I was shocked to see that the droop contact (14) on this CAT setup will close when the PLC sees we are tied to the grid versus what I'm more familiar with. (PLC sends signal to a Potter/Brumfield socket relay, which one of its contacts is the aux droop contact for the 2301A) (another set of contacts does put the Voltage regulator in droop).

I would have thought being in Isoch tied to the grid would be bad. I first thought the relay contacts for the 2301A were incorrect.
The 2301A terminals 10/11 show to go to a OFE1 (Analog output Voltage type PLC card). I thought initially that this was a useless connection since those terminals 10/11 are disabled in Droop mode.
I then read in the Woodward manual for the 2301A (Manual 82389_CAT) that includes a small section that included "when running a single unit on an infinite bus with a generator loading control or Import/export control, terminal 14 must be connected to terminal 16 (i.e. close the droop contact) to connect the load matching circuit to the load-sharing lines. The load-sharing lines must be wired to the generator loading control or Import/export control". I'm assuming the PLC I have is the "Import/Export" control? The PLC logic for sending a Block trans write to the Analog PLC card...and to the 2301A is very slim...
I saw in other forums that Woodward 26260 manual was a good reference, so I checked it out. But I'm scratching my head on comparing its terms to what I have installed?

Does anyone else have a similar setup/ or experience with this setup?

 

[waross]

My first thought:
Does closing the droop contact put the unit into droop or take it out of droop?
Second thought:
Does the PLC put the unit into base load mode when a buss is detected?


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

 

[Mechkane]

Thanks for the response waross.
Closing the droop contact will take the 2301A unit out of droop.
The PLC logic reads: that once it detects a bus, that is tied to the grid, it will close the droop contact. The logic wording includes "Governor and PF controller in Load Share Mode when active".

Looking at the Woodward manual 26260, there's a section on Isochronous Base Load, it says an auxiliary bias signal can be applied to the load sharing lines. and includes "If we now connect such an isochronous load sharing system to a utility, where the speed/frequency Is fixed by the utility, and we place a fixed bias signal on that system’s load sharing lines, all units in that system will be forced by load
bridge imbalance to carry the load demanded by the bias signal."

This makes me think that the PLC Analog output card OFE1 (that is wired to the 2301A load sharing lines) is this "auxiliary bias signal".

 

[waross]

Does this make sense with what you have found to date?
When the governor is controlling the set, it should be in droop when paralleled with the grid.
When an external load control module is controlling the set, the governor should be in isochronous mode.
Why?
Contrary to popular belief the grid frequency is not always 60 Hz. Abrupt load changes will affect the frequency which will then be corrected by the swing set.
Base load sets are typically running at 5% droop.
With the base load sets in droop, all sets will respond to frequency deviations so that frequency excursions will be much less that if the base load sets were running at fixed throttle.
But, large grid tied sets run at 5% droop and standby sets run at 3% droop.
With a 3% droop set on a 5% droop system, the load variations will be 5/3 times as much as for the 5% droop sets.
But, running in load control mode the set will not be able to support grid frequency response.
On the one hand, the effect of a relatively small standby set on the grid will be negligible.
On the other hand, there are serious possible back-feed issues should the grid go down.
The worst case would be if crew took your circuit out of service while your generator is on-line.
Your set could back-feed lethal potential into the circuit with a real hazard to lives of the service crew.

At this point it becomes a matter for the agreement with your grid operator.
Does the grid operator allow parallel operation for testing?
If allowed, do you have adequate anti-islanding controls in place?
There are more questions that answers here.
Let's hope that someone more familiar with your equipment posts in.

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

 

[davidbeach]

A machine that size, running against a decent sized utility system doesn't need any droop. The governor controls should maintain the desired power output (90% of full nameplate is a nice number) and the AVR should simply maintain the desired power factor (unity is a good value). Operationally it's just that much negative load and the utility won't be adversely affected.

But, as mentioned, that's all after the necessary interconnection application has been filed and the utility has enumerated all of their requirements.

I’ll see your silver lining and raise you two black clouds. - Protection Operations

 

[wcaseyharman]

True isochronous control when connected to the grid is not a pretty sight, we did that once on a 50MW hydro unit and it closed the gates 100% and pulled 6MW as a motor.
I don’t have any experience with the 2301A, I am currently learning about the 2301D on a project I am working on, but I looked at the manual you referenced. It looks like the 2301A uses a load control on top of the isochronous setting, so when it is online the bias is the fixed load control and the unit is not technically running in a “isochronous” mode, but a block loaded mode. (I remember this from the old fly ball hydro governors, you could run the governor up against a “gate limit” and effectively the unit no longer regulated speed and you would control power with the gate limit.

 

[waross]

The governor controls should maintain the desired power output

I prefer to use droop to set a base load speed.
With 3% droop and the governor set to 61.62 Hz, the set will output 90% of rated power and will better ride through slight frequency variations on the grid.
Another point that may be of some concern:
If the unit is intended to run in droop, but is tested with droop disabled, developing issues with the droop circuit will not be discovered be testing.
Is this testing method valid, or is this a comment to save for the report, after the wreck?

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

 

[FreddyNurk]

I am aware of use of a 2301A as a load bias device for Caterpillar generators, as I understand it one can't just use a 4-20mA or similar output, but put a 2301A in load share mode and off it goes.
The context I am aware of was parallel isoch in isolated networks (supervising controller varies the load, or it ends up being a proxy interface for a loadsharing system), but I would expect it'd be doing the same thing here.
I would expect catserveng has some perspective on this.
EDMS Australia

 

[Mechkane]

Thanks all for the responses. Our Operators will contact the dispatcher to verify no grid disturbances are expected and to verify we can parallel this unit to the grid. Our protective relays (not to mention the 14 of them by name) gives standard protection. The PLC logic includes monitoring for if we lose ties to the grid to remove the load sharing feature and put us back to essentially the 2301A in droop mode. I don’t think that’s ideal but, I’ve spot checked this in past testing (about ten years ago) when we were running the unit on an isolated bus. There was an obvious drop in frequency with each raise in KW. Not too severe for the Operators to not be able to recover frequency if they had needed to.
As far as the load sharing, it’s a new concept for me, and interesting at that. Our Operators use an HMI interface (InTouch Wonderware) touch screen that among other things allows them to set a KW demand. It has its own set of logic that communicates with the PLC. The PLC will respond by sending a signal to the 2301A load sharing lines via the analog output OFE1. Apparently an OFE1 card is voltage output and the vintage of PLC (PLC-5) is standard 0 to 10 VDC output. Standard Woodward guidance states load share voltage is 6VDC, so this output from the card makes sense. I’m suspecting the analog card was scaled between 0 to 10vdc to represent the range of KW this unit can provide. 0 to 10 vdc scaled to 0 to 4400KW. Looking at the Woodward document of the load sharing circuit, it appears the Operator choosing a set KW will provide a VDC to the terminals 10/11 load sharing lines of the 2301A, the summing junction will not be equal and the set will respond until it does equal. There’s also a pot in the circuit which I’m guessing is the load sharing pot? Other past testing records I’ve seen shows there was a difference in Operator demand and actual KW output (during parallel load sharing testing) following a replacement of the 2301A. It was resolved by adjusting the load sharing pot. Im guessing this was essentially to resolve a slight scaling issue between the displayed operator demand in KW and the actual KW output. I wonder if the standard Woodward 6VDC is what Caterpillar used or if they originally set up this unit with a different value.

 

[waross]

This genset is only used for backup emergency type power and only is tied to the grid for testing periodically. So testing tied to the grid is 99.9% of what we do with this genset.

If it runs in droop when in standby mode shouldn't it be tested in droop mode?

There was an obvious drop in frequency with each raise in KW. Not too severe for the Operators to not be able to recover frequency if they had needed to.

Not a problem.
For a number of years I was associated with an island plant that supplied about 5000 residences.
"There was an obvious drop in frequency with each raise in KW."
No customer ever had an issue.
Actually, no customer ever even suspected that the frequency varied slightly.
Our plant had 5 sets,
Normal operation was with 1 to 3 sets online.
All sets were in droop.

I suggest that your testing will be more valid if you disable the fancy load sharing and just test in droop.


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

 

[Mechkane]

(Nightshift is getting old)
Waross, thanks for the info.
I had the same question in regard to having droop enabled while isochrounous. From what I can tell, there's no documented basis for the droop pot setting for our setup; CAT may very well have had the droop pot dialed to zero when they provided the design back in the 90's; and in 2013 the 2301A was replaced in-house with no CAT support; in-house guys didn't fully understand the load sharing design we utilize and the droop pot is dialed to something other than zero. I'm considering the simplest option of dialing the droop to zero on the 2301A. It is standard for our setups here for providing safety power to an isolated bus to ensure freq stays relatively stable with each KW load, so no (or very little) droop dialed sounds acceptable for our needs here.

The other option of taking out the fancy load sharing is still an option with a little more design changes. Taking two wires off the load sharing terminals 10/11 is super easy, but it would entail some Operational changes on how we perform testing and training the operations dept, plus how the HMI and PLC logic will respond. and some drawing updates.

 

[waross]

Too little droop and your governor may become unstable.
A proportional controller may go into oscillation with too little proportional band.
A governor is generally the P (Proportional) of a PID controller.
3% droop is 3% proportional band with 3% offset.
A governor with isochronous function is the PI (Proportional plus Integral) of a PID controller.
I have run into real world issues with PID controllers with too little proportional band.
3% proportional band is almost universal for islanded sets.
It works well and no one ever realizes that the frequency is not a perfect 60 Hz.
And, by the way, in a large grid network, all but one of the sets will be set at 5% droop. (That does not imply 5% frequency variations.)
And, isochronous does not keep the frequency at exactly 60 Hz.
The first response of an isochronous governor to a change in load is in droop.
When the isochronous feature detects an error between the process variable (The actual delivered frequency) and the set point (60 Hz) it acts to cancel out the offset.
Load sharing in droop is quite simple:
Use the frequency setting to adjust the speed until you are synchronized with the grid.
Once you have successfully synchronized with the grid, the frequency control becomes the load control.
By varying your frequency set point from 60 Hz to 61.8 Hz, you will vary the load from zero% load (at 60 Hz) to 100% load (at 61.8 Hz.)*
This is for a base frequency of 60 Hz and 3% droop.
Every standby set that I have installed, serviced, repaired or consulted on has been running at 3% droop and no one ever knew or complained.
When I started working on standby sets, I thought that isochronous was a great idea and I wondered why so few small standby sets had isochronous mode.
After working on dozens of sets with only droop mode and seeing how well droop works, I am content with droop control.
The one possible issue with standby frequency is clocks tied to grid frequency.
Grid operators count the actual cycles and if they have been running a little below 60 Hz and loosing cycles, they will run a little above 60 Hz until the cycle count is correct.
A standby set, whether in droop or isochronous will not control the frequency close enough to keep time.
The solution is to replace any clocks using grid frequency as a time base with clocks that don't depend on grid frequency.
Droop control is also the KISS solution.


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

 

[wcaseyharman]

Just a slight comment - at least in the western United States interconnection there are no machines that I am aware of operating in isochronous. All machines are in droop or a non-frequency controlling mode (temperature control on gas turbines for example). Automatic Generation control is used to correct frequency - this area wide function provides setpoint changes to groups of generators to correct frequency after an event, not an isochronous unit.
I believe it is a common misconception due to the use of a slack bus in software modeling tools, as it is mistakenly believed that the slack bus represents an isochronous unit. This is modeling technique and does not necessarily represent how the real system functions.
As an aside I worked on a 100 MW hydro plant that islanded the generation compound. We set it up for 1% droop instead of isochronous and it works great, but the load is quite small relative to the machine size.

 

[waross]

Are swing sets no more?
About 15 years ago, give or take, I saw JungleMux equipment being installed at a new substation.
The system was tying substations together with fibre optics.
The word was that the system would prevent domino type failures when a local outage started overloading adjacent areas which, when they tripped offline, caused overloading and tripping to spread over a wide area.
Is a single similar FO system now in use to control set-points system wide?
Thanks.

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

 

[wcaseyharman]

I can only speak in the grid area I am familiar with (united states western interconnection) but here’s how I understand system frequency is managed - by the way, NERC has some good documentation for further reading:
After an event there are two stages of frequency recovery - primary frequency response and secondary frequency response. Primary frequency response is generally all the generator governors that operate in droop. When the frequency decays the machines respond according to their droop settings. Within the US western interconnection that is required to be from 3% to 5%. Generally the droop response is complete within a minute or two after the event. It primarily arrests the initial frequency decay, but stabilizes the frequency at a lower value than nominal (assuming this is a generation loss event and the frequency is declining, which is the most common)
Secondary frequency response is via Automatic Generation control. This function within my company is built into the SCADA system. AGC works to minimize the Area Control Error, which is a parameter that is principally calculated based on the deviation of scheduled interchanges with neighboring utilities and the difference of frequency from nominal. AGC sends dispatch signals to each generator to increase or decrease their output to reduce the area control error. AGC working among multiple utilities will bring the frequency back up to nominal (or slightly above nominal to correct the average frequency back to 60 Hz). This generally occurs 5-10 minutes after the event.
Hope this provides some clarity on how grid frequency is managed, at least in the United States.

 

[waross]

Thanks for the explanation.
That was how isochronous resonded.
First in droop, second with an error correction by the swing set.
Unfortunately, a load change greater than the spare capacity of the swing set, required manually (originally by phone calls to individual generation operators) increasing or decreasing the outputs of the base load sets.
Do you know when the new control system was implemented?

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

 

[wcaseyharman]

I don’t, AGC was present when I started working for the utility in the mid 2000s. I’ll ask around and see if anyone still has the history.

 

[waross]

As an aside I worked on a 100 MW hydro plant that islanded the generation compound. We set it up for 1% droop instead of isochronous and it works great, but the load is quite small relative to the machine size.

Not to doubt your good experience but a couple of comments come to mind.
It may not be valid to compare a 100 MW set with a 1.5 MW standby set.
Most of the major standby set manufacturers have standardized on 3% droop for standby sets.
And generally;
Running a set in parallel to export power may not be appropriate mode for testing a standby set.
My preferred test:
Pull the main switch and prove that the set will start properly and pick up the block loading that is common to standby sets.
Use the normal building load as the load bank.
That will test all normal aspects of the standby set.
That will also highlight any concerns or problems with running on standby power.
Will there be objections? Probably, but regulations mandate regular testing and part of full testing is an outage of a few seconds.
There may be special equipment that does not like an outage of a few seconds.
Warn everyone before the first test.
Try to identify equipment that may have problems and find a remedy before hand.

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

 

[Hoxton123]

hi waross

one problem is that esting often takes place at night time and / or weekends when site load is low. Diesel standbys need to be run at full power to prove all systems. Solution: load banks so the set can be run at full load.

 

[waross]

I would suggest dropping the mains and picking up what load there is, and if needed, then sync and use the grid to load up.
Paralleling first leaves a lot more untested.
If it runs in droop, test in droop.
Paralleling with the grid in droop is easy.

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