Simulating a Turbine Generator


Thread Starter


Hello, hopefully this question is suitable for this forum.

I'm an engineer at a consulting company. One of the things we do is create customer specific simulations for operator training purposes. We have modeled many processes in the past, but we are now doing several turbine generator simulations. The mechanical side is working well, but I have some questions about how to model the electrical side.

I will preface the questions by saying I’ve done a lot of research on the internet, as well as on this forum, and have gained a much better understanding of the subject than before I started, but I still have some fundamental questions.

My main questions revolve around reactive power, how its “generated” and how it’s dealt with by turbine generator operators. I’ll tell you what I think (mostly based on info from this site) and hopefully someone can set me straight.

VArs are a function of the type of load a generator “sees” and need to be dealt with by changing the excitation on the rotor.

Controlling VArs at the generator (done by varying excitation) is necessary, and VAr creation must match VAr load or the system voltage will change, just like power (watts) creation must match power load, or the frequency will change.

Increasing excitation above what is required to maintain generator voltage equal to grid voltage will cause lagging VArs to flow out of the generator.

Decreasing excitation below what is required to maintain generator voltage equal to grid voltage will cause leading VArs to flow into the generator.

Leading VArs and Lagging VArs of the same magnitude will “cancel” each other out.

Now for some basic questions:

There is a number on the DCS screen for MVAR. How does that number come to be? Is it a directly measured quantity? Is it back calculated using the power triangle from other quantities that can be directly measured? (Remember, we need to simulate this).

Is the only way a particular plant and operations team “see” and react to VArs by measuring and reacting to small differences in the grid voltage vs. the voltage the generator is currently producing?

Are large capacitor banks there to cancel out some of the lagging load seen by a particular generator, or group of generators at a specific plant? Say you had a capacitor bank that had an effective capacitance of 3 MVAR, and the breaker to that bank was open. Also say that you had a 7.5 MVAR reading on your DCS, and the power factor was .95 in the lagging direction. If you were to suddenly close the breaker to the capacitor bank, would the MVAR reading drop to 4.5, and the power factor would increase toward unity?

I realize that everyone has a different opinion on and way of understanding VArs. What I’m trying to accomplish is to model how a plant “sees” and reacts to VArs in a real world way. These simulations don’t need to deal with extremely complex math models, nor do they need to work in every possible scenario, but if we get the fundamentals right, it may work as expected,

I also have an additional question that arose during my research. It seems there is a governing body that has said it is NOT OK for power plants to run in automatic VAr Control mode. I have no opinion one way or the other, I’m just curious about the reasoning behind why plants should not be allowed to do this.

Hello, Dave,

(Wow--that was a deja vu moment! Like something out of '2001: A Space Odyssey.')

I agree with most of your understanding except:

> Leading VArs and Lagging VArs of the same magnitude will “cancel” each other out.

Lagging VArs <b>"feed"</b> a lagging load--<b>from a generator perspective</b>. When talking about VArs, you always have to consider "perspective." It's different in a switchyard, for example.... Which is what I think you are referring to.

On an infinite grid, a synchronous generator can be "made" to produce or consume VArs--or, by keeping the generator terminal voltage equal to grid voltage, it can be made to operate at 0 VArs (unity power factor). The amount of VArs is pretty much a factor of how the plant responds to changes in grid voltage. (That's the MVARs value you are referring. It's function of the excitation versus grid condition. There is usually a VAr transducer which supplies that value to the control system for display.)

For example, if the power output of a synchronous generator is fairly constant throughout the day and if the operator made NO changes to excitation from a 0 VAr condition at the beginning of the day, the VAr meter would change <b>as the grid voltage changed.</b> Again, with a constant load (watts) and excitation (generator terminal voltage), as grid voltage changes so will the VAr "flow."

So, operators must respond to grid voltage changes (evidenced by the VAR meter movement) to keep the VAR meter at the desired setpoint. (And, by VAr meter, I'm referring to the MVARs value on the screen you are referring to.)

Or, they can use automatic VAr control (if the unit is so equipped) to maintain a VAr setpoint, and it will adjust generator terminal voltage as required to keep a constant relationship between generator terminal voltage and grid voltage <b>as evidenced by the VAr meter/value.</b>

As for some entity forbidding automatic VAr control, I would think that might be possible, but it's probably more like a purchased power agreement contract thing. It would seem that if a power plant was operated at a fairly constant load and fairly constant VAr output that the grid regulator would be pretty happy about that for the most part.

It might be that at certain times of the day or year automatic VAr control is not permitted?

On an infinite grid, there may be power factor correction capacitors at various locations on the distribution system. However, they can only "switched in" or "switched out" as a block--meaning they can't be smoothly "increased" or "decreased" as needed. And, if you stand near to a large bank of power factor correction capacitors when they are switched in or out, you can usually feel it in your feet when it happens! It can shake the ground. And, the VAr meter usually jumps when it happens (as does the voltmeter).

Again, most power plants and operators don't really monitor grid voltage, or even generator terminal voltage once they are on line (MOST plants, anyway--some definitely do monitor one or both values). The only thing they monitor is the VAr meter (the MVARs value you referred to). The "evidence" of a change in grid conditions versus generator conditions once the generator is on-line and at a relatively constant real power output (watts) is a change in VArs (MVARs; VAr meter). Operators (though they don't usually realize it) are responding to differential changes by changing excitation.

Just like most operators don't understand when they are changing watts as they are loading and unloading the machine, they are using Droop speed control to change the prime mover speed reference! They don't monitor prime mover speed or generator speed; they only monitor the wattmeter (MWATTs) and don't know/realize they are changing the speed reference to change the wattmeter.

Hope this helps!
Thanks CSA for your quick and detailed response.

I’m glad that my understanding of it is at least passable. Remember that we are trying to model this behavior, so the direction of the VAr meters’ movement is important as are the simulated conditions that cause the meter to move. The assumptions below are very generalized, and I’m sure dozens of factors come into play, but I’m just trying to get the gross behavior correct. So let me know if these assumptions are correct:

If a generator is synced with a large grid, small increases in grid voltage with respect to the generators’ output voltage would be seen by the plant as a lagging load, and the VAr meter value would increase and the PF meter would move away from unity in the lagging direction. The operator can respond to this by increasing excitation on the rotor and can move the VAr meter back toward 0. If in fact the excitation was increased to exactly match the grid voltage, the VAr meter would be at 0, and power factor would be 1.0.

If grid voltage drops below the generators output voltage, the plant sees that as a leading load, and the VAr meter would move away from 0 and the PF meter would move away from unity in a leading direction. The operator can respond by decreasing excitation and can move the VAr meter back toward 0.

The only parameter that controls what the VAr meter reads is the difference in voltage between the grid and the generator. If the voltages match exactly, the VAr meter is 0 and the power factor is 1.0. This is independent of the magnitude of the actual voltage values, and the power being produced by the generator.

Increasing the inductive loads on the grid tends to increase grid voltage, which needs to be responded to by increasing excitation.

Increasing the capacitive loads on the grid tends to decrease grid voltage, which needs to be responded to by decreasing excitation.

If a generator is running in a stable state with positive number on the VAr meter, this means that the grid and generator voltages are not the same, and that the grid in fact has a higher voltage reading.

This is a question, not an assumption: Do VAr meters typically read negative numbers when there are leading VArs? The generator capability graph we have plots MW vs MVARs, and the MVAR (y) axis goes from -15 to 25 and the curves on the plot all go into the negative region on the y axis.

Thanks again for the information.

Here's how I would explain your scenario, from a slightly different perspective. When a synchronous generator's output (watts) is constant and its terminal voltage is exactly equal to the grid voltage, the VAr meter and the VAr flow is zero, and the power factor meter reads 1.0 (unity).

From the above condition (stable watts output), when the generator excitation is increased causing the generator terminal voltage to attempt to increase above grid voltage the VAr meter moves in the Lagging direction (from a generator's perspective, Lagging VArs are seen as a positive VAr flow, as they are (or should be) indicated as positive on the generator reactive capability curve you have). And, the power factor meters moves to some value less than 1.0 in the Lagging direction.

This is akin to the grid voltage decreasing while the excitation is being held constant--making the generator terminal voltage higher than the grid voltage.

If, while generator output is held constant, the excitation is decreased below that required to keep the generator terminal voltage equal to grid voltage then the VAr meter will move away from zero in the Leading direction (from a generator perspective, Lagging VArs are considered to be negative, as they are (should be) on the generator capability curve).

This is akin to grid voltage increasing while the excitation is held constant--making the generator terminal voltage lower than the grid voltage.

An analog VAr meter is zero-centered. It receives PT and CT signals and determines the magnitude and "direction" (Leading or Lagging) of the reactive current. The VAr meters I have worked with didn't have "+" or "-" indications on either side of zero, they just indicated Lagging or Leading, respectively.

Analog power factor meters are similar, except they are centered at 1.0 (not 0).

When digital displays became commonplace, it wasn't cost-effective to display "Lagging" or "Leading", so "+" and "-" were used, respectively.

I'm no electronics (analog or digital) expert. My best SWAG is that the analog VAr meter actually senses the angle between VA and Watts to calculate the direction of movement of the needle away from zero, and in addition it calculates the magnitude of the VAr component, and that determines how much the needle moves away from zero. (I'm sure transducers and digital meters work similarly.) It's not really about the voltage difference (though it is...), it's also about the angle between VA and real power. And voltage differentials do impact that angle.

When one is synchronizing a synchronous generator to a grid, one of the things that is most commonly done is to make the generator terminal voltage equal to or slightly greater than grid voltage before closing the generator breaker. That ensures that when the generator breaker closes that VAr flow will be zero or just slightly lagging, which is "positive" reactive current. (That's also why the prime mover speed is made equal to or just slightly greater than grid frequency--to ensure that when the generator breaker is closed the real power flow is positive, that watts are being "sent" out to the grid, and that the generator is not being "motorized" by real current from the grid.)

Increasing reactive loads (inductive or capacitive) tends to change the angle between the voltage and current sine waves. ELI the ICE man, if you know the little memory trick/mnemonic). This also has the effect of reducing or increasing the voltage sine wave magnitude slightly (respectively). When the European or North American grids are experiencing their summer load peaks because of air conditioning loads (which are highly inductive because of the refrigeration compressor induction motors and the air handler induction motors) the grid voltages usually decrease as the air conditioning load increases and as the inductive load increases.

When a synchronous generator is synchronized to a large or infinite grid, generator terminal voltage and grid voltage are essentially the same. In the same way it's not possible to change the speed of a synchronous generator when it's synchronized to a grid, it's not really possible to change the generator terminal voltage by very much. A few volts or a few tens of volts is all that's really possible, and that out of 11KV or 13 KV.

In fact, increasing excitation which tends to increase generator terminal voltage was (in the "old" days) referred to "boosting" grid voltage, and decreasing excitation which tends to decrease generator terminal voltage was referred to a "bucking" the grid voltage. On a very large or infinite grid there are so many generators synchronized together that one generator can't really have much of an effect at all on the grid voltage--in the same way the prime mover can't have much of an effect on grid frequency. There's just so much "voltage" (and real current) out there that one generator can't have much of an effect or grid voltage or frequency.

Now, in some locations on a grid, a single generator can have much more of an effect on the grid voltage in that area/region, but that's not always the case. And, it depends on a LOT of factors, which we aren't considering in this discussion--which is just about phenomenon and effects and what an operator sees when this or that happens, or what an operator can generally affect.

Hope this helps!
Thanks again, this helps explain how we should model the behavior of our machine when connected to our simulated grid.

Since we are modeling loads and generators, I would like to ask if this is correct. Suppose you have a balanced (at any given time) grid with a given load and a given amount of power generation. We know that in order to maintain frequency, the generation (of MW) must exactly equal the load at any given time or the frequency will change (higher if the generation exceeds the load and lower if the load exceeds generation). Is it also true to say that each given plant is contributing to the generation in its own amount, and it knows how much its contributing in order to keep the grid balanced. Can an analogy be drawn to the previous situation with regard to Reactive power and voltage? In other words, the aggregate load in our scenario consists of a “need” for power and a “need” for reactive power. As the mixture of needs changes (different types of loads are turned off and on) the generation plants must feed this need in the correct ratio of real and reactive power. So just as the amount of real power generated must match the amount needed, so must the amount of reactive power “generated” match the amount needed ( at a given time). Each plant in the grid contributes to the generation of this reactive power, and it knows how much it’s contributing to the overall need, by its local VAr meter. It does not know how much the total need is, it only reacts (to the VAr need) by changing its excitation to keep the voltages between the generator and the grid equal. If more lagging VArs are “generated” than needed, grid voltage will tend to increase, If less lagging VArs are “generated” than needed, the voltage will tend to decrease.

Thanks again for your time. From researching this before I posted my questions, I know this has been dealt with many times by you and others on this site.


Yes; this has been dealt with before on this site, though there hasn't always been "agreement" on terminology and underlying physics/facts. Which shouldn't seem possible, but, then man doesn't really know what "electricity" is! Is it a mass with a charge, or a charge with a mass? But we sure know how to use it to transmit torque over long distances using wires!

Real power (watts) is the instantaneous product of voltage times current times power factor (times the square root of 3 for a three-phase system). The power factor is a measure of the efficiency of the system at transmitting real power by looking at how much of the total energy/power is being converted to reactive power (remember the power triangle).

This "load sharing" thing you're talking about with respect to watts is called Droop Speed Control. Each machine will respond to changes in frequency by changing its real power output proportionately to the change in frequency. (This is ONE aspect of Droop Speed Control--only one.) Droop Speed Control is what makes all of the generators (up to their rated power output) respond to changes in frequency in a stable and proportionate manner to help maintain grid frequency stability.

Don't get too hung up on the whole "generator terminal voltage versus grid voltage" thing. The reactive load on a grid is what it is--the sum of all the inductive and capacitive loads. People (consumers; users of motors and computers and air conditioners and such) don't control the reactive component of the total load they place on the system with some knob or switch. It just is what it is.

The <b>effect</b> of the reactive component of a load is to shift the voltage and current sine waves out of phase with each other. Since real power is an instantaneous measurement of voltage times current times power factor, as the two sine waves shift out of phase with each other the real power transmitted by the grid changes. Left unchecked, this leads to a very inefficient electrical transmission system and eventually brownouts and even blackouts.

Power factor is one way to measure the "shift" and to factor it into the real power formula.

Synchronous generators (and synchronous motors and synchronous condensors and power factor correction capacitors) can "produce" VArs, to counter the effects of the VArs "consumed" to try to shift the voltage and current sine waves back closer into phase with each other--which improves the power factor and real power capability.

I hate to even bring this up--but excitation control systems have Droop.... So, excitation systems also respond to voltage changes in a proportionate manner. But, please--let's NOT go there.

So, there is some analogy between real and reactive power, except that in the case of reactive power if exactly the same amount of VArs isn't "produced" as are being consumed there is no effect on frequency, just on the relationships between the voltage and current sine waves, which does have an effect on voltage....