1. Tell me one thing, i have heard that power system frequency is unified and same everywhere. but is it really the same at the terminals of all generators output when we consider them present in a single area? Suppose my area consists of one hydro and 3 steam units, will the energy be sensed same everywhere at a particular time of load pick up?

Also,i know areas are determined by coherent behavoiur of its residing units, i.e., the units behave similarly to a step load. in that respect would it be practical to consider the hydro unit as one area and the 3 steam units on separate area?

2.Dynamic response is the period (30-40 seconds) when governor acts after the inertial response (2-10 secs phase determined by rotating masses kinetic energy + frequency sensitive loads damping coefficient).

You are telling that speed setpoint of all droop governors are to be changed after the end of this dynamic response period. but i wonder whether the isochronous unit has already brought back the frequency to its setpoint ->50 Hz fully, within the end of this dynamic response period?

Now, I STATE MY DIFFERENT UNDERSTANDING ON TWO SEPARATE PARAS .PLZ LOOK AT THEM INDEPENDANT OF THE OTHER ONE 2A AND 2B.

2a. IF SO, i.e., the iso already does its job at the end of dynamic response phase, the droop units will see that frequency returns to its old value. and therefore it will fully release whatever load it has previously taken out of frequency deviation at the end of this dynamic period, as the frequency deviation due to iso unit is now zero. Now the steady state mode begins and now operator changes droop setpoints either manually or as per some auto load control program so that the droop units can again increase their generation (it has previously increased). and thus relieve the iso unit, which now again decrease its load to offset over frequency on account of change in droop setpoints.

Is that how our power system operates? here the droop units are increasing generation due to sensing negative speed deviation. then decreasing generation due to zero speed deviation, and then again increasing generation as the speed setpoint increase (so again a negative error). So the droop units are doing the same job for twice-thrice the time.

Won't it be easier if we control droop units' speed setpoints automatically within the dynamic response period (assuming the ramping rates of droop units are comparable) before the iso units finish its job of regulating frequency? so that the iso unit on account of its quick (P+I) action gives a much larger share to the step load increase, but simultaneously the droop units also gives a lesser share to this load relatively slow as compared to iso unit.

2b. If NOT SO, i.e., THE ISO UNIT DOESN'T FULLY REGULATE THE FREQUENCY AT THE END OF THIS DYNAMIC RESPONSE. so its integral control is still not finished acting. so now still some frequency deviation remains. and the droop units hold on to its last increased value during dynamic primary governor response and ryt at this moment, the speed setpoint of droop units are changed. So beginning from this moment the iso and droop units are again parallelly acting to regulate the already improved but slightly deviated frequency, back to its scheduled 50 HZ value. and after the order of 2-3 minutes (depending on the step load increase), the frequency returns back to 50hz. and we conclude that droop control + their setpoint changing in one hand and an isochronous unit on the other hand does the same job. but while iso unit is faster (large integral gain), it takes a larger share of step load, and since the integral gain associated with auto setpoint control of droop units is lesser or it is manual change, they take lesser percentage of the step load.

NOw tell me which according to my above two paras 2A and 2B actually goes on the power system w.r.t a step load increase?

Lastly,
>This is different from a system where the dynamic response of a >number of sets is controlled by a single controller, trying ..disturbances.

How this 'number of sets' is decided? Is it by checking the frequency responses? i,e droop coefficients and ramping rates of those 'no. of sets'? or by some other economical consideration? If droop speed setpoint changing, is all about load scheduling and not dynamic response? the i guess the single AGC controller does some kind of droop coefficient control (R control) of the droop units and does not affect the iso unit in any way. Is that it?

 

> 1. ..i have heard that power system frequency is unified and same everywhere. but is it really
> the same at the terminals of all generators output when we consider them present in a single area?

Yes - the frequency of all generators is the same, and what is more they are all in phase. This is not confined to an "area" but applies to all interconnected machines - even across a continent. (There will be some slight differences in phase which are required to give power flow through line reactances, but these will not change unless the power flow changes.)

> Also,i know areas are determined by coherent behaviour of its residing units, i.e., the units behave similarly to a
> step load. in that respect would it be practical to consider the hydro unit as one area and the 3 steam units on separate area?

In my view, an area is a group of closely interconnected generators, with little line reactance between them. Such a group can be replaced by a single equivalent machine if you are doing dynamic studies. For instance, in my local area, there was a thermal power station with 5 x 120 MW reheat steam-powered generators and a gas turbine station with 3 x 50 MW generators, located about 30 km apart. These were connected to the main grid by a 200 km 220 kV transmission line. While the machines in the local area were tightly coupled and would remain more or less in synchronism and phase during a disturbance, the area could swing in phase relative to the main grid (with an increased risk of pole slipping if the upset was severe).

> 2.Dynamic response is the period (30-40 seconds) when governor acts after the inertial response (2-10 secs phase
> determined by rotating masses kinetic energy + frequency sensitive loads damping coefficient).

The governor and inertial responses are closely tied - on a load change, the inertial effects cause the frequency change, and the governor system reacts to these as they occur. On a load increase, the inertial effects will cause the frequency to fall steadily with no self-regulation - if we were to wait till the inertial effects had settled out before the governors reacted, the system would not respond at all.

> You are telling that speed setpoint of all droop governors are to be changed after the end of this dynamic response
> period. but i wonder whether the isochronous unit has already brought back the frequency to its setpoint ->50 Hz
> fully, within the end of this dynamic response period?

It would be possible for the isochronous unit to react during the dynamic response, but this requires careful adjustment. Consider the following sequence-

- a load increase occurs.
* in response, the droop elements in the governors all act to increase speed. However, because a speed error is needed to trigger the change in generation, and there are transfer lags in the governors and turbines, the overall response is slow and frequency falls more than is needed. With a droop response, too low a frequency means that the governors will be calling for more than the required increase in generation.

* as the generator outputs increase, the system frequency levels off and then starts to increase again because of the excess generation. At this point, the frequency is still below the target of 50 Hz, and an isochronous element would be calling for more power.

* once the frequency starts to increase, the droop elements in the governors will be calling for a cut-back in generation.

* the overall behaviour will depend on how fast the isochronous response increases the generation when compared to the rate at which the droop elements reduce it. If the isochronous response increases too quickly, the frequency will rise to 50 Hz and beyond while the droop elements are still reacting to the drop in frequency.

Since the isochronous response includes an integral component, it has the potential to make the system unstable. While it can be tuned to give a required response, this tuning will depend on the behaviour of the rest of the system, including for example the number of machines connected. Turbine response can also depend on loading - so if a machine is loaded to 25 %, its overall response may be quite different than the response for a load of 75 %. The easiest way to overcome this is to set the isochronous element so that it takes effect as the other dynamic responses settle out.


> 2a.
> Won't it be easier if we control droop units' speed setpoints automatically within the dynamic response period
> (assuming the ramping rates of droop units are comparable) before the iso units finish its job of regulating frequency?
> so that the iso unit on account of its quick (P+I) action gives a much larger share to the step load increase, but
> simultaneously the droop units also gives a lesser share to this load relatively slow as compared to iso unit.

P+I action is not going to be quicker that pure P action. The P term response to a step change in error will be a P step, whose magnitude depends on the gain (droop). The I term in response to a step is a steady ramp which will reach a magnitude equal to the P step after the Integral Action Time (hence doubling the response). The I term will continue to increase. In standard approaches to tuning, introducing an I element requires that the proportional gain is reduced to compensate for the increased overall effect due to the I element.

> 2b. If NOT SO, i.e., THE ISO UNIT DOESN'T FULLY REGULATE THE FREQUENCY AT THE END OF THIS DYNAMIC RESPONSE. so its
> integral control is still not finished acting. so now still some frequency deviation remains. and the droop units hold
> on to its last increased value during dynamic primary governor response

** this is correct **

> and ryt at this moment, the speed setpoint of droop units are changed.

** no - the isochronous element will have a small effect during the dynamic period, but will only become significant when the dynamic responses due to droop action are more or less settled out **

> So beginning from this moment the iso and droop units are again parallelly acting to
> regulate the already improved but slightly deviated frequency, back to its scheduled 50 HZ value. and after the
> order of 2-3 minutes (depending on the step load increase),

** the response time shouldn't depend significantly on the magnitude of the step increase in load **

> the frequency returns back to 50hz. and we conclude that droop control + their setpoint changing in one hand and an
> isochronous unit on the other hand does the same job. but while iso unit is faster (large integral gain), it takes a
> larger share of step load, and since the integral gain associated with auto setpoint control of droop units is
> lesser or it is manual change, they take lesser percentage of the step load.

> NOw tell me which according to my above two paras 2A and 2B actually goes on the power system w.r.t a step load increase?

** If there is a load increase -
- all machines will initially react through droop action to increase their outputs. Faster-responding machines will increase output by more than than slower-responding ones: as the slower-responding machines react, frequency will increase again and the faster machines will reduce load.

- the machines settle out at a frequency below 50 Hz, with total generation increased by the amount of the load increase. The isochronous element takes over and gradually increases the output of the relevant machine. This lifts the total power generation and frequency will increase slowly back towards 50 Hz. As a result, the machines without an isochronous element will reduce their generation. As frequency closes on 50 Hz, the error in frequency falls and the rate of change of power in the isochronous machine falls.

- eventually, the frequency will settle back to 50 Hz. The droop machines will be at their initial loads, and the isochronous machine will have picked up all the added load.

- at this point, the load scheduling system can take over and adjust the speed settings of one or more machines to reduce the load on the isochronous machine and load up the most appropriate of the others.

> How this 'number of sets' is decided? Is it by checking the frequency responses? i,e droop coefficients and ramping
> rates of those 'no. of sets'?

** no - it is not generally relates to the dynamic responses of any of the sets **

> or by some other economical consideration? If droop speed setpoint changing, is all
> about load scheduling and not dynamic response? the i guess the single AGC controller does some kind of droop
> coefficient control (R control) of the droop units and does not affect the iso unit in any way. Is that it?

** Ultimately, in a single interconnected power system, all machines will be part of some sort of load scheduling arrangement. On a large-scale country- or continent-wide system, this obviously becomes very complex with one of the factors being to maintain generation so that it tends to match local loads as much as possible (to minimise transmission losses and the risk of a line tripping on overload). Other factors will include trying to anticipate demand where possible so that there is always sufficient reserve generation to match expected load increases - e.g. at half-time during a football match, there is usually an increase in power demand.

In some cases, there may be other concerns that mean that it makes sense to pass control down to an area level. For instance, if there are several hydro stations on the same river with the discharge of one feeding into the storage of another, it makes sense to allocate a block of generation to these stations as a block or area. Within the area, load scheduling can be arranged to take account of local issues such as maintenance, expected flood inflows, or water shortages. A single automatic load scheduling system may act on the speed settings of the machines in an area, but it is usually easier to apply to smaller more integrated systems such as a single factory with 3 or 4 generators as well as an external power supply. In this case, load scheduling will usually be based on economic factors such as the total cost of power and the likelihood of load increases - for instance, if a plant has a total generation capability of 20 MW and a 1 MW motor is to be started, something like 4 MW of reserve must be available to allow the motor to start.

There is another factor to add to the complexity on a national level. Many clocks use the power frequency as their time base. If the load increases are sudden, and load shedding is relatively slow, the frequency will be lower for longer than it is high, and the clock system will eventually run slow. An isochronous machine may have to be run a bit faster than 50 Hz for a time to bring the clocks back into step with a time reference based on a crystal or similar standard.

 

Bruce...

Hopefully you aren't suggesting bypassing the furnace-purge part of the boiler's restart sequence.

Regards,
Phil

 

Hi Phil,

I'm not sure what triggered off this comment, but no. My comments are generally to do with machines that are already running. And there are a lot of issues associated with starting systems that need to be factored into the load scheduling approach.

> Hopefully you aren't suggesting bypassing the furnace-purge part of the boiler's restart sequence.

 

BruceD...

So basically what i infer:-
there are 4 phases;

1) inertial phase

2) dynamic response phases during droop units and iso unit increase as per droop control. though i understand the governors do not wait until the end of inertial phase to respond as u rightly have pointed out..

3) iso-integral phase where iso unit regulates the drop fully with the result that iso fully takes up the whole step load increase and droop units fully relinquish whatever generation it increased and goes back to its previous value before load step increase. in this regard,i have a question. if say all droop units are just synchronised before load pick up, and they picked up some finite value in dynamic response, will all of them become zero now after the end of dynamic response?

4) the fourth phase is load scheduling phase where all or some droop units setpoint is changed to relieve the iso unit.

that's it .ryt???

Now, regarding the last 3 paras of urs where u told the selection factors for the 'no. of units’. This selection factor is for the auto load scheduling or manual load scheduling. i.e., the fourth phase.ryt? But I want to know the selection factors of those units which are on AGC. .which u intimated are the set of phenomenon which starts and ends with the dynamic response phase.(between the 2nd and 3rd phase). plz comment. they surely takes into account the ramp rate limits of its selected units as they are used for dynamic response. isnt it?

Now i am working on an restoration plan purely for academic purpose. So i have to design a MATLAB model of generation control where all rotating masses, if present at a particular time, is present in the model. The no. of units in my model gradually increase as more units come online. Here i have to calculate the time between two restoration steps. The way i view it is that if a restoration step is associated with generation and load increase, then total time should be the summation of time for all four phases. i.e.,first 3 phases at end of which frequency recovers fully, and the 4th phase where load is scheduled and setpoint changed to load droop units and relieve iso unit. that is the total time.

I am using ieee 39 bus system where final load will be around 6000 MW and here for each load increase up to 6000 MW. i am allowing a transient frequency drop of 4 percent and designing my load steps in such a way that the frequency error doesn’t go below this 4% drop value. Here i saw,initially when system inertia is low, a step load increase of 10-30 MW is good for regulating frequency. In the last and middle stages of restoration, i even saw a step load increase of 150-200 MW is possible w/o a drop of frequency outside 4% as by then many generation as come online (Inertia improved). also many transmission lines are reenergised and available. So dynamic response has improved.

Now u tell me:-Is that large step load pick up (around 200 MW) practicable in a real system if say, one has one hydro and 9 steam units at disposal with sufficient spinning reserve available?

If here,there is a way of uploading image files or matlab models i could have shown u glimpse of a particular step of my generation control model and thus could verify its practicality. Can i send it to ur email instead?

<b>Moderator's Note:</b> Use a site like Dropbox to upload image files for others to see.

 

Bruce,

I responded to your 03-Mar-'15 (2:19pm) post.

Phil

 

BruceD...

> So basically what i infer:- there are 4 phases;
> 1) inertial phase
> 2) dynamic response phases (snip)
> 3) iso-integral phase (snip)
> 4) .. load scheduling phase w
>that's it .ryt???

Pretty well - although I would not separate out the "inertial" and "dynamic response" phases - this is a single event in the sequence.

> l I want to know the selection factors of those units which are on AGC.
(snip)

I would very strongly suggest that you find a system somewhere in your local area or country where what you are interested in is actually done, and find out what they do and why they do it. There are a number of different possible scenarios, which may be complicated by other factors (such as having to coordinate steam generation and power generation in large plants with Combined Heat and Power installations).

> Now u tell me:-Is that large step load pick up (around 200 MW) practicable in a real system if say, one has one hydro
> and 9 steam units at disposal with sufficient spinning reserve available?

To get a large load pickup is not very usual in large power systems - more common is the loss of a generator or block of generation. It is more significant as well, as the power limits on transmission lines can be checked. The amount of frequency change on a load change is dependent on the running generation at the time, and can be estimated by looking at the overall droop. So if a system has 1000 MW of active generation and gains 100 MW of generation (or loses 100 MW due to a machine trip) the change in frequency with 4% droop will be about 0.4 %. The change will be similar with a change of 500 MW when 5000 MW of generation capacity is on-line.

In New Zealand at one stage, the total installed capacity was around 6000 MW in the North Island, 3500 MW in the South Island. A DC link between the two islands can carry up to 1000 MW. So a possible scenario is that up to 16 % of the supply could be lost in the north, while nearly 45 % of the load is lost in the south. This system is made more complicated as there is no inherent droop in DC transfer.

Again to look at what is a feasible loss or pick-up of load, you really need to study one or more actual installations to see what can happen and what their other constraints or limitations are.

> If here,there is a way of uploading image files or matlab models i could have shown u glimpse of a particular step of
> my generation control model and thus could verify its practicality. Can i send it to ur email instead?

I can handle Dropbox ok.

 

i am satisfied as of now and i think a lot of webs cleared after reading the responses here and an IEEE journal "understanding AGC 1992" by L.H Fink et al...

Thank you all