I have been reading through a number of the generator islanding posts here on control.com, which have been very helpful (especially input from CSA!) however I believe the setup at my work is slightly different, and would appreciate an opinion on how the system would respond in the event of becoming islanded.

First, some details of our system. We have our own 33kV distribution system, to which one 140MW (third party owned and operated), three 17MW and two 28MW generators are connected. We also are connected to the National Grid via two 275/33kV transformers. Usual site operation is that we export ~90MW to the National Grid, importing ~50MW when the 140MW generator is offline. The site is therefore self sustaining when the 140MW generator is online and we are separated from the grid.

The five smaller generators generate at 11kV and are stepped up to 33kV by dedicated unit transformers. They are all driven by steam turbines, backpressure machines - the electricity output is effectively a 'byproduct' of them being used to let steam pressure down from 130bar to 14bar. They were installed in 1965, however they all have new AVR systems (the 17MW machines utilise Brush A50 AVR's, the 28MW use Alstom DX21's).

As far as I am aware, the generators are run up in speed governing mode, then set to pressure governing when synchronised to the network. These machines often operate in parallel - usually one or two 17MW machines run with one 28MW in service.

The operators adjust the MW setpoints to maintain the pressure in the 14bar steam system. The AVR's generally operate the 17MW machines overexcited to produce around +5MVar, and the 28MW around 21MVar.

The AVR's do not have communication between each unit, or 'islanded' detection. They are set to have a voltage droop of 5% which I believe should reduce the reactive power flow between the machines in the event of becoming islanded (both National Grid connections out of service).

My question is how the 17/28MW generators will likely respond upon loss of both grid connections, with the 140MW generator in service (thus site is net exporting power), and out of service (net importing). The 140MW generator does detect loss of grid, and goes into frequency control mode should this happen.

Could anyone provide any advice? The backpressure element of the generators is a concern for me, as I don't think they can be operated on-line in speed governing mode.

Please let me know if I can provide any clarification / more information!

 

BairdUK,

I can address the "real" power output side of your scenario. I'm hoping another contributor, Bruce Durdle, will address the reactive power/island system voltage side of your scenario.

As you described your scenario where the 140 MW unit is acting as the Isochronous unit when the plant is islanded from the National Grid, it will adjust it's energy input (fuel or steam) to vary it's load (MW) to maintain frequency. Because it is so large with respect to the other units they will "be along for the ride" and will continue to run just as before the separation. As long as they are not switched to Isochronous Speed Control mode, they will operate just fine; if they were in Inlet Pressure Control mode before the separation, they will continue to control inlet pressure after the separation.

In my view of the reactive side of your scenario, I believe what will happen is that the island system voltage will either increase or decrease, and I believe if the plant was exporting MVArs before the separation I believe the island system voltage will be higher than desired after the separation.

Real power (MW) and reactive power (MVAr) & system voltage are two relatively distinct things. Fuel, or steam, input to the prime mover driving the generator controls the real power output (MW) of the generator. Excitation controls the generator terminal voltage, and as that increases or decreases as the reactive load increases or decreases the system voltage will vary. That's how I view things; and, no, I don't have the maths to prove my theory--only personal experience, and most of my experience was in trying to tune the "system" to maintain frequency when separated from the grid. Others were looking after the reactive load and island system voltage!

Hope this helps--and I hope Mr. Durdle will add to this discussion, and correct me if necessary!

 

How can I resist a challenge like that! Some further information would be useful -

- what is the "normal" plant load (MW and MVAr)?

- what is the "normal" plant power factor? Is there any on-site power factor correction?

- what is the normal power factor or MVAr of the 140 MW set?

I'll address the case where the 140 MW set is off-line and you are importing power first - that's the easiest. In this case, there will be some restrictions or financial penalties based on the imported power factor. To keep it simple, I'll assume that you want to keep the power factor of the incoming feed at 1.0 - in that case, you must generate all the plant's MVAr requirements on-site. If the plant load is 200 MW and the power factor is 0.75, that will be about 176 MVAr. With on-site generation of 150 MW, the power factor of the generators will be 0.65. This is fairly low but should be within the capability of the generators if they are running at reduced load. Any on-site static power factor correction will help and reduce the requirements for the generators to produce MVAr.

With the 140 MW set on-line and exporting, there may well be commercial requirements for the site to export MVAr as well as MW. But again there will be a balance between the total MVAr produced and the sum of export and local load requirements. Generally, the requirement will be for a net export of MVAr.

You can treat your generators as a fixed internal voltage (proportional to excitation) in series with a reactance, with the difference between the bus voltage and the internal voltage due to the reactive volt drop. The load represents a fixed impedance. So in effect you have a voltage divider with the top leg the reactance of the alternator winding and the bottom leg the impedance of the connected load - the feed to the voltage divider is the excitation voltage and the voltage across the load is the bus voltage.

If the site is exporting and then becomes islanded, the MW requirement will be reduced and the governors will act to match total generation to the total site MW. This will not significantly affect the internal excitation voltage. However, the load power will fall and the effective load impedance will increase - so the voltage divider will have a much higher impedance in its lower leg and the bus voltage will rise. To bring the voltage back to normal, excitation must be reduced.

With an islanded plant, as well as a MW balance, the total MVAr generated will match the MVAr required by the load - so using MVAr or power factor to control excitation won't be effective. Some sort of bus voltage regulation will be needed - if the 140 MW set has this, the other machines can be left on MVAr or pf control.

Hope that helps.

Bruce.

 

Thanks CSA and Bruce for the very useful replies.

The operators of the 140MW generator deal with the energy trading side of things - I'm not aware of the terms of their connection agreement with the DNO, or any financial penalties associated with this.

However, I can provide the extra information you asked for Bruce. The site load is around 110MW, 70 MVAr. Power factor can vary from 0.85 to about 0.9 - there is no power factor correction on site.

The 140MW generator always operates very close to unity power factor (overexcited to export around 6-10 MVAr). Thus we import around 40 MVAr from the National Grid at all times (140MW generator online or not). The remaining ~25 MVAr are produced from the one running 28MW generator and 1 or 2 running 17MW generators.

One possible concern that was highlighted to me was regarding reactive power sharing between the running 17MW / 28MW machines upon islanding - am I correct in thinking this is controlled by the voltage drop compensation of the AVRs? I have visions of large reactive power swings between machines as each tries to correct for a step change in site demand.

I am slightly less concerned with the state of the system upon islanding with the 140MW generator in service, due to its size I appreciate it will maintain system frequency and presumably make up for the shortfall in MVAr's - I will speak with the operators in the morning. My main concern is when it is offline, and we have no machine operating in speed governing mode. However, all the AVR's operate in voltage regulation mode (none in PF or field current regulation).

 

Bruce Durdle,

Thanks very much for your reply! I wasn't issuing a challenge; just expressing a hope that you could use your real-island experience and shed some light on this reactive power situation when operating as an island.

I have a couple of questions, if you will entertain them, please. First, when operating in island mode, I presume the reactive load (VArs) is fixed, except for changes caused by switching motors and fluorescent lights and the like on an off. So, presuming the real- and reactive loads are stable and presuming the smaller units in this thread are at unity power factor and the 140 MW unit is supplying all of the reactive power required to keep the system voltage at desired, what happens when someone increases the excitation of one of the smaller units? Does the system voltage increase? Presuming there's no PMS (Power Management System) controlling system voltage by adjusting the 140 MW unit's excitation and the exciter regulator is in AUTO mode, what happens to the island reactive load and system voltage?

Again, I've spent all of my time fighting (people) to control frequency when trying to operate in island mode, and haven't had time to focus on reactive load control. (You wouldn't believe how many people think a machine operating in Droop Speed Control should be able to automatically maintain frequency very tightly when separated from a larger grid with some load (a "load acceptance" or "load throw-on" test they like to call it, and they will NOT allow the unit to be switched to Isoch control when isolated with some load. Such is the life of a start-up person, eh?) So, any information you could provide would be greatly appreciated.

Thanks in advance!

 

BairdUK,

there is an aspect of your post which is somewhat contradictory to me. You indicated that the 17 and 24 MW sets are put to pressure control mode once synchronised, but later you state the operators control MW to maintain the 14 bar pressure? Does pressure control mode maintain the throttle pressure or the back pressure? How is the MW setpoint super-imposed on pressure control?

My expectation, bearing in mind the vintage of the sets, is that they will have the hydro-mechanical governors (unless they have been retro-fitted with Electronic Governors). If the original flyball governors are still in place then they will always operate with an element of speed / droop control, but with a super-imposed MW or pressure controller adjusting the speeder gear to maintain MW or pressure. Even if Electronic Governors have been retrofitted, this may still be the case depending on the type of Electronic Governor fitted.

Could you clarify the above as it will have an impact on how the system as a whole will respond to a grid disconnection.

 

PeterM,

Upon checking, I've discovered that the 'pressure governing' mode looks at the pressure of the speed adjusting oil, not the steam pressure. The governors are oil-pressure operated which acts against a large spring.

The MW setpoint is set by our DCS, which has a control function which raises / lowers the governor oil pressure (which in turn opens / closes the governor valve to let more / less steam into the turbine) to reach the required MW output.

The governor valves have never been replaced (although they are maintained on a regular basis).

Hope this helps - I'm much more aware of the electrical system and AVR operation than the mechanical side of things!

 

The actual behaviour when islanded will depend on what is used as set-point and PV for the AVRs. I'm a bit confused as to whether this is purely terminal voltage or if it is MVAr - you say in your original post that the machine produce around +5 MVAr and +21 MVar. Is this just a result of how things are set up or is it some sort of operator-adjusted set point?

With simple load compensation, the internal set-point will be based on a notional value which is reduced by the compensation system. So if a machine increases its reactive current draw, the internal voltage setpoint will be reduced and the AVR will reduce excitation. Whether the different AVRs will fight will depend on settings of time parameters in the regulators and the time constants of the excitation system components - an older machine with several stages in the excitation train will be slower to react than one using static excitation or rotating rectifiers. As with any control system, time delays or cascaded time-constant effects are the major cause of instabilities.

If a slow system has a faster setting than necessary (perhaps for historical reasons when it was the only one in service) then adding a newer system without modifying the old one is likely to cause problems as the older one will initially attempt to over-excite - the new one will do its job - then the adjustments to the older system will kick in and the new one will back out. Whether this is a problem or not depends on your system.

In general it would be better to have one machine looking after the bus voltage with the others running on reactive power control of some sort. However, the details will depend on the actual installation, the overall system requirements, and the tolerance of the electrical load to voltage swings. Whether or not any system alterations can be justified will depend on how often the islanding is likely to occur.

 

BairdUK,

it sounds like you have fairly conventional mechanical governors. I suspect if you look more closely you will find that the DCS actually sends raise / lower commands to the "speeder gear" motor. Its a bit difficult to describe in text, but the "speeder gear" motor then adjusts the position of a multi-port valve which is also influenced by the position of the flyball governor, which in turn is determined by the shaft speed. The exact valve port arrangements and flyball governor will determine the overall governor droop (traditionally 4% in the UK).

The upshot of this, is that in the event that your site is exporting power and then becomes disconnected from the grid, all the machines (140, 28 and 17 MW) will initially speed up and shed load. Assume we start at total generation of 185 MW (140 + 28 + 17 MW) with site load of 110 MW, there needs to a reduction of 75 MW to achieve balance. Assuming we start at 50 Hz, all the machines have 4% droop and the governors / speeder gear at not "over-wound" (to ensure the governor valves are wide open), all three machines will shed a proportionately similar amount of load 75 / 185 = 40%. 140 MW machine will drop to 84 MW, 28 MW machine will drop to 17 MW and 17 MW machine will drop to 10 MW. This will happen at +1.6% speed/frequency (4% * 0.4) = 50.8 Hz.

However, while this is the built-in reaction of the turbine governors, you have a super-imposed level of control on top. For the smaller machines, the DCS will see the MW reducing and send raise commands to restore the output back to the setpoint. Depending on the tuning in the DCS, the re-loading could be so quick you may only see a dip of a few MW on the 17 and 28 MW machines.

In the case of the 140 MW machine, the behaviour will depend upon exactly how the machine responds to the grid disconnection. If the machine drops into a pure P / droop mode, the frequency will rise even higher as the 140 MW machine will have to accommodate the restoration by the DCS of the load on the 28 and 17 MW machines. The 140 MW machine will need to reduce load by 75 / 140 = 54%, equating to a speed increase to 51.1 Hz.

If however, the 140 MW goes into an "island mode" (I believe the GE terminology is "Isochronous Speed Control" mode) typically an Integral term is switched in which will endeavour to maintain speed / frequency to 50 Hz. The 140 MW machine will shed load to maintain frequency without the frequency needing to rise to 51 Hz (as required with pure droop speed control).

Is the 140 MW machine a steam turbine or gas turbine?

 

Hi PeterM,

You're spot on with your interpretation of the speeder gear - I recently one of those very motor and oil manifolds refurbished on one of the 17MW machines. I have also seen the interposing relays sending raise / lower pulses to the governor from the DCS, based on the setpoint as you mention.

The 140MW machine is a gas turbine, part of a CHP plant. I am assured it does drop into frequency control mode on loss of both grid connections.

Thanks for the clarification regarding the likely response of the turbine governors - I conclude that in the case the 140MW machine is offline when islanding occurs, we will have a shortfall of generation of around ~65MW which cannot be taken up by the running machines as they will already be at rated output. In this case we would have to have some automated load shedding facility to prevent the 17 / 28MW generators from tripping on overload/over current. The cost to the site from losing the 33kV network is so high that measures like this could be considered.

With regard to reactive power, I believe that the 140MW generator AVR would act to produce 35-40MVAr which were previously imported from the grid. As the 17 / 28MW machines setpoint (DCS) has not been adjusted, they would try to maintain this output. If the 140MW generator was offline, the 17 & 28MW AVR's would try to increase up to their over excitation limits - again would result in a shortfall requiring load shedding.

 

Only the gas turbine (GT) is capable of reacting (shedding load) fast enough to maintain the 50 Hz frequency.

The steam turbines do not react fast enough to be of any significant help during the transient when the loss of load is detected, although the governor may have a 5 to 6% droop built in.

That is why when doing National Grid Code compliance testing in a combined cycle facility the burden of the test is carried by the GT's.

Whatever load was lost will be shed by the gas turbine. If [load lost > GT load] , the GT will continue to unload and eventually would open the generator breaker on reverse power.

If the load lost < GT load, the GT will unload very fast (in seconds) to reach the new equilibrium between load and frequency and the rest of the units will not notice the change.

Once in island mode, the only loads available to the units are the "domestic" loads. The isochronous unit is maintaining the frequency and everything is fine and dandy... until the operators decide to synchronize again with the grid: before they do that it is imperative to remove the "isochronous mode control", i.e. go back to "droop control" otherwise it will not be possible to synchronize (a crap shoot: if the frequencies match, then it will be possible to do it)

 

Abeltio,

I believe you are incorrect. A ST is inherently capable as fast as reacting as a GT to frequency changes. Otherwise there would have been no control of frequency prior to large scale use of GT's for generation. Typically ST's are capable of and can contribute to rising frequency response in Grid Code tests. Typically what ST's in a CCGT cannot do is provide a fast increase in power as they are usually operated with Valves Wide Open / Sliding Pressure. The limitation is actually the HRSG which will not instantly generate additional steam to pick up the ST load. To overcome the delay in the HRSG, both Alstom and Siemens offer an option for throttling the ST in on a CCGT to provided enhanced Frequency Response.

Also although Euro and other Grid Codes allow droop >5%, GB Grid Code requires droop to be in range 3-5%. Traditional UK value was 4%.

 

BairdUK

I would agree to cater for a loss of grid connection with the 140 MW machine offline you will need some sort of load shedding scheme.

If all your house load was directly connected motors (not connected by VSD's) you would get some natural load shedding effects (as the frequency dropped the motors and connected pumps etc would also slow down reducing the mechanical and electrical load). However, VSD's etc are so prevalent now that this effect is likely to be minimal.

However, a load shedding scheme alone will not be enough. You still would need to match generation to the remaining load. As your ST's are normally operated in MW control (by the DCS) there will be no speed / frequency regulation for the new island. The ST's would probably therefore need to be dropped out of MW control in the event of a loss of grid connection with the GT offline. One of the things to check is that the ST's have no restriction on frequency. I would guess the ST's were produced to 60's UK standards which means they will be designed to operate over a relatively wide frequency range. Many of the newer European designed ST's have a much narrower operating window and include low and/or high frequency trips.

As an aside do you still have enough steam generation to run all the ST's if you are GT is offline? I presume much of the steam is normally produced by the GT exhausting to an HRSG.

I wouldn't claim to quite as familiar with the likely reactive power behaviour. However, your capability would depend on where your STs' Generators typically operate and what their rated Power Factors are. If they are rated to PF 0.85 (again typical UK / CEGB standard) I reckon you could get a total of 36 MVAr with 2 x 17 MW machines (10 MVAr each) and 1 x 28 MW (16 MVAr) machine running. By the time you shed some load you may well be OK on reactive power.

 

Agree, regarding ST response. I was considering the CC case and it is not clear in my posting.

 

ST can respond quite fast. Typical loading ramp rate is 10-15% per second if and only if it has steam drum to provide reserve energy.

The idea of fast response of ST to react during system transient instability is to arrest frequency decay over the first three seconds. Later the slow but immediate response of prime movers such as hydro plans will take the lead over 3-15 seconds. Frequency decay shall be stopped within less than 15 seconds to allow secondary response which is called AGC- load & frequency to take place.

Note: I don't exclude the possibility that some of us here have never experienced with the 3-10 transient load-frequency instability. Thanks for your very big and stable grids. For some of the countries in the world they are not that lucky.