Chapter 2.

Stability in electricity supply systems has to be maintained even during disturbances such as a major short circuit, generator failure or losing a large load. In keeping the system stable, the role of rotating inertia is essential. When integrating renewables with no or low inertia to the system, the balancing becomes more difficult. To avoid risks of frequency collapses and blackouts, new solutions are needed for the fuel-based backup generation.

The main task of an electric power supply operator is to continuously match electricity generation with electricity demand. Continuous matching is needed since the supply system as such cannot store electrical energy. If electricity demand systematically exceeds the power delivered by the machines that drive the generators, the generating units will respond by decreasing their rotational speed. Consequently, the grid frequency will drop and the system will collapse in a matter of seconds, resulting in a blackout. Fortunately, blackouts won’t occur if the unbalance in demand and supply is short-term, since generating units and electric motor drives have energy stored in their rotating mass, the so-called inertia. That buffer limits the rate of change in frequency in the case of an unbalance between generation and demand. The energy stored in this inertia creates time for the engines or turbines that drive the generators to adjust their output in order to restore balance.

The stability of the frequency in alternating current sys- tems is a good measure of balance. Frequency is by definition the number of times that a full sine wave occurs per second in the grid. The international unit denoting frequency is hertz (Hz). The rotational speed of the generators determines this frequency, and the so-called nominal value of the fre- quency depends on the global location. America and Japan operate using 60 Hz, while most other areas of the world have 50 Hz. In reality, the grid frequency varies somewhat around the desired value. Figure 2.2 shows the grid frequency in The Netherlands during a short time span of 5 minutes after 5.00 am on April 12, 2013. Customer demand is never fully constant and generators also have some variability in their output. Yet, on average the frequency has to match the desired value.

2. Balancing the electricity supply in case of calamities-image-2.1
Figure 2.1.
An illustration of the delicate balance between electricity demand and production, with rotating inertia as a buffer with some energy stored.

If the actual frequency deviates for just a small fraction from the desired value, no action is taken to change the output setting of the generators. There are a number of reasons for that. Each measurement system is afflicted with some inaccuracy, while control systems also have some insensitivity. Endeavours to keep the frequency within very narrow limits, meaning real isochronous operation, would result in overactive control of the machines that drive the generators, which in turn would lead to unnecessary wear. Figure 2.2 shows an example of a system with a permitted measurement error range of +/– 10 mHz, with an additional zone of +/– 10 mHz where no action from the generator is required. Consequently, the result is a total dead band of +/– 20 mHz. Nevertheless, the average frequency over a prolonged time span should be exactly 50.000 Hz, and grid operators take action if the cumulative deviation from this desired value becomes excessive.

2. Balancing the electricity supply in case of calamities-image-2.2
Figure 2.2.
Example of the rules for frequency control, with an example of actual frequency variations (frequency data source: TENNET)

In Figure 2.2, the grid frequency exceeds the dead band at 20 seconds after 5.00 am on April 10, 2013. At that moment, the so-called primary control reserve power plants start to automatically slightly reduce their output settings. Some one hundred seconds later, the frequency is back within its allowed limits and the control action of the generators ceases. This frequency regulation action is handled automatically by the so-called primary reserves. Automatic action is the only option because of the fast response required to keep the frequency within its narrow band.

The primary control reserves are also known as frequency containment reserves (FCR). Substantial changes in frequency will occur if a large customer disconnects, or if a large power station suddenly fails. In modern power supply systems, many generators are interconnected via the transmission grid. The number of online generators should be sufficient to ensure that a failure of the largest unit can be absorbed, to a large extent, by the spare capacity of the other generating units.


It is interesting to analyse what happens when a power plant in an electricity supply system fails. To simplify such an analysis, we presume a supply system with ten power plants of the same power capacity. Each of the ten power plants has a nominal power capacity of 500 MW, and they are all running at 90% of their capacity to provide 10% of primary control reserves. That would, at least in theory, be sufficient to compensate for a failure of one of the ten power plants. Nominal power means the nameplate power of the generating unit, while nominal speed means the generating unit’s normal amount of revolutions per minute. In this example, the electricity demand that the ten power plants supply amounts to 90% of 10 · 500 MW = 4500 MW. It will now be shown what hap- pens if, for example, power plant number 7 suddenly fails and the system immediately lacks 450 MW of the required power supply. This example may appear to be somewhat exaggerated since a sudden loss of 10% of the dispatchable generation is not common.
However, with much renewable capacity in a system, such occurrences are becoming increasingly realistic. In addition, the effects of unbalance in a system can be clearly shown with this example.

2. Balancing the electricity supply in case of calamities-image-2.3
Figure 2.3.
Ten power plants initially supplying the required electricity demand when one of them, number 7, fails.

After the failure of one plant, the nine remaining power plants cannot instantaneously ramp up the power output of the machines that drive the generators from the initial 450 MW to the newly required 500 MW in order to supply the total system demand of 4500 MW. Power plants need some time to react to a newly desired output value. Therefore, if no energy was available from the rotating mass (the rotational inertia) in the system, the unbalance would immediately stop all generators with a resulting blackout. The amount of energy stored in a rotating generating set is linearly proportional with the moment of inertia I, and the square of the running speed n. Inertia is a property characterising the flywheel effect of
the rotating mass. The running speed n gives the number of revolutions per minute of the generator rotor. In a 50 Hz system, a generator with a single pole pair runs at 3000 rpm. In this case, the frequency f equals n/60.

2. Balancing the electricity supply in case of calamities-image-2.4
Figure 2.4.
The energy stored within the power supply system as a result of the flywheel effect of the spinning generating units.
(f = rotational speed in revolutions per second, Ir = moment of inertia).

The amount of energy Er stored in the rotating inertia of a generating set is generally expressed as a fraction of the nominal power capacity Pnominal of that generating set. This fraction is called the inertia constant τI of a generating set:


*Equation 2.1.

The dimension τI of the inertia constant is the same as that of time, and is expressed in joule/watt = J/(J/s) = s (second). The inertia constant of a large generating unit lies in the range between 5 and 10 s. This means that when a 500 MW generating unit is running at its nominal speed, its rotating parts have 2500 MJ of rotational energy in the case of a τI of 5s, and 5000MJ for a τI of 10s. This is also the amount of energy that has to be transferred to the rotating parts when the generating unit is started up and accelerated to its nominal speed. If the 5000 MJ of energy during this acceleration is supplied to the rotating inertia with a machine that uses natural gas for fuel, we can calculate the amount of gas required. If the fuel efficiency of the driving machine is 40% and the natural gas has a lower heating value of 36 MJ/m3, it requires 5000/36 · 100/40) · 350 m3 of gas to bring the rotor up to nominal speed. Such an amount of gas provides enough energy to heat up 35000 litres of water from 20 °C to 100 °C, for preparing 280000 cups of tea. A cluster of 3000 car batteries can also deliver the required amount of energy. This illustrates that although the energy stored in the rotating inertia of a generating set is not electrical, it is nevertheless an impressive amount of energy.


The inertia constant τI is very helpful in getting a first impression as to how fast a generating unit will change speed in case of an unbalance between the power supply to the generator and power demand. Unbalance occurs when the electrical load changes while the power supplied to the generator shaft from its prime mover, i.e. the driving engine or turbine, remains the same. For an inertia constant of 5 s, the amount of energy in the rotating parts is enough to supply the nominal load of the generator for 5 s without any energy input from its prime mover. After that time span, the generator will come to a standstill. Furthermore, if the prime mover were to supply 90% of the nominal load to the generator, while the load of the generator equals the nominal load, the rotational energy would be enough to cover the unbalance for 50 s. Thus, it takes ten times longer than in the case of no power supply from the prime mover before standstill is reached. However, even in the latter case, the frequency will rapidly reach a value outside the permissible range.

Equation 2.1 reveals that the rotational energy of the generating set is proportional with the square of the instantaneous running speed. This means that at higher speeds, there is much more energy in the inertia than at lower speeds.Therefore, the frequency will not decrease linearly with time if there is a fixed unbalance between the power supply to the generator and the generator load. Figure 2.5 shows how the frequency of the grid served by the nine remaining generators (Figure 2.3) decreases if each generator receives 450 MW from its prime mover while the combined load remains 4500 MW. Each of the nine running generators should receive 500 MW to avoid an unbalance. Therefore, each generating unit ‘feels’ a power supply deficit of 50 MW. The frequency curve in Figure 2.5 is a representation of equation 2.2. The derivation of equation 2.2 requires some considerable mathematical manipulation. The interested reader can find the derivation of equation 2.2 in Appendix 1.

2. Balancing the electricity supply in case of calamities-equation-2.2

*Equation 2.2.

2. Balancing the electricity supply in case of calamities-image-2.5
Figure 2.5.
The decline in speed for a generator where the generator load constantly exceeds its driving shaft power by 50 MW (inertia constant τI = 10 s, constant load presumed).

The drop in frequency during the very first seconds following a major contingency event is a perfect indicator of the amount of unbalance. In the beginning, the rotating frequency of the generator set is still very close to the nominal frequency (50 Hz or 60 Hz) and the approximation can, therefore, be made that the initial drop in frequency per unit of time df/dt equals:

2. Balancing the electricity supply in case of calamities-equation-2.3

*Equation 2.3.

Equation 2.3 reveals that in our example of a τ<sub>I</sub> of 10 s, a power deficit of 50 MW per generator with a nominal power of 500 MW at a frequency of 50 Hz, results in a change in frequency of exactly 0.25 Hz per second at the start of the occurrence. This is indicated in Figure 2.6 by the brown line. Would the power deficit per gen- erator have been only 25 MW, the decline in frequency would have been halved to 0.125 Hz/s. This simple relationship between unbalance size and the initial frequency change is very convenient, especially in island operation where just a few generators have to maintain grid stability. Based on the value of the inclination in frequency versus time, the fuel supply to the machine driving the generator can be immediately and adequately adapted so that no time is lost in restoring the generator frequency. The horizontal red lines in Figure 2.6 give the maximum allowed dynamic frequency limits in the Continental Europe synchronous area. The green line in Figure 2.6 represents the nominal grid frequency of 50.000 Hz.

2. Balancing the electricity supply in case of calamities-image-2.6
Figure 2.6.
A close-up of the first 4 seconds of figure 2.5; the thick brown line gives the decline in grid frequency for a 10% unbalance in the system due to the tripping of a power plant (τI = 10 s).


If the grid frequency decreases, electricity demand automatically goes down slightly. This is primarily caused by the synchronous electric motors in the grid demanding less power, since their load declines along with their running speed. This is called the self- regulating power of the supply system. The power requirement of synchronous motor- driven pumps can decrease by 6% per Hz frequency drop. For motor-driven applications with a constant torque, power demand can decrease 2% per Hz. Since synchronous motors form just a fraction of the total load, the self-regulating power of the system in industrialised countries is generally presumed to be 1% per Hz deviation from the nominal frequency. In areas with minor industrial activities, the self-regulating power of the grid will be close to zero.

The positive effect of self-regulating power should not be overestimated. If the grid frequency drops from the desired value of 50 Hz to 49.8 Hz, the self-regulating power lowers demand by only 0.2 · 1% = 0.2%. This 0.2% is only a 9 MW reduction from the 4500 MW total load in our ten generator system example. If, however, the grid frequency would drop from 50 Hz all the way down to 40 Hz, the decrease in electricity demand because of the self-regulating power would be 10 Hz · 1% / Hz = 10%, i.e. 450 MW in our example. This renders a 50 MW reduction in demand for each of the nine generators that remained online following the trip of one machine. Consequently, at a frequency of 40 Hz, generation and demand are matched again if the nine power plants keep their power output setting at the initial 450 MW. In reality, it will be difficult for the generators to keep their power output constant when the frequency decreases so much. The power output of the prime mover that drives the generator is proportional with the product of torque M and running speed n. If the grid frequency decreases from 50 Hz to 40 Hz, the driving torque M has to increase by a factor of 50/40 = 1.25 in order to deliver the same power to the generator. This can easily result in mechanical overload. At the same time, turbo machinery in particular is burdened with natural frequencies of the rotor system that limit its range in running speeds. Generators suffer when running at low frequencies because of the so-called magnetic over fluxing, which results in possibly harmful overheating. In practice, the self-regulating power of a grid hardly offers any help in balancing.

2. Balancing the electricity supply in case of calamities-image-2.7
Figure 2.7.
Mitigation of the decrease in frequency due to self regulation of grid load in the case of a 50 MW initial unbalance between generator output and load.

A decrease in grid frequency from 50 Hz to 40 Hz is excessive and unacceptable for many applications and generators. The maximum deviation from the nominal frequency during dynamic events, such as the loss of a generator or the loss of major load is, therefore, set at +/– 800 mHz in the Continental Europe synchronous control area. If self-regulation of the load alone would be present, this frequency range would be heavily exceeded in the case of calamities.
Therefore, each electricity supply system has generators offering primary control reserves that are activated automatically if the grid frequency exceeds its tightly defined limits. The change in output from the primary control reserves depends on the extent of the deviation in grid frequency from the nominal frequency. In other words, the desired output of a generator acting as primary reserve depends on the actual grid frequency; the lower the frequency, the higher the output of the primary control reserves. This dependence of the output set point on the grid frequency is generally called the droop sgenerator (Equation 2.4). The minus sign in equation 2.4 indicates that a decrease ∆f in grid frequency results in an increase ∆P in the power output from the generator providing primary reserves.

2. Balancing the electricity supply in case of calamities-equation-2.4

*Equation 2.4.

This is often re-written in terms of the regulating power Pregulating of the system:

2. Balancing the electricity supply in case of calamities-equation-2.5

*Equation 2.5.

in which:

2. Balancing the electricity supply in case of calamities-equation-2.6

*Equation 2.6.

The droop value setting of the control system lies generally within a range of 2 to 8%. A droop s<sub>generator</sub> of 4% means, for instance, that the power output of the generator increases from 0% to 100% if the frequency decreases by 4%, say from 50 Hz to 48 Hz. The relationship between output change and frequency deviation is linear. In measurement and control technology language, this is called a proportional action, indicated with the letter P. It means that the extra power from the primary control reserves is only present as long as the deviation from the desired nominal frequency exists.


As explained above, primary reserves are intended to avoid excessive deviations in frequency during a major occurrence affecting the balance in the electricity supply system. However, the primary reserves only arrest the frequency temporarily. They cannot return the frequency to its nominal value. Figure 2.8 illustrates how the power output from a 500 MW generator running at 250 MW, and acting as a primary reserve, varies with frequency for three different droop settings. For a grid frequency of exactly 50Hz, the output of the generator equals 250 MW. When the grid frequency decreases, the output from the generator will increase linearly along with it. Conversely, if the grid frequency increases, the output from the generator will decrease by following the droop line. Figure 2.8 reveals that the lower the droop percentage is, the heftier the reaction of the generator will be on deviations from the nominal grid frequency.

At first sight, opting for a very low droop value might be the best option to keep the grid frequency as close as possible to the desired 50 Hz. However, if the droop is very small, meaning that the gain factor is very high, the primary reserves may react fiercely to deviations in frequency. This results in extra wear of the machinery, while increasing the risk of oscillations and system instability. Some traditional power plants suffer heavily from rapid changes in output. High temperature steam boilers can experience cavitation and thermal shock during sudden load changes. Moreover, all generators acting as primary control reserves do not have the same dynamic prop- erties in practice. Each unit has its own delay time when reacting to an increase in the output set point, and each unit will have its own typical ramp up rate. Until now, a close to stepwise increase in output was not possible. Nevertheless, some modern generating techniques can react much faster than traditional units. The typical ramp up rate for primary control reserves in the Continental Europe synchronous area system is 100% within 30 seconds (Figure 2.9). After a short delay of, say 2 seconds, the primary control reserves are supposed to increase their output at a fixed rate.

2. Balancing the electricity supply in case of calamities-image-2.8
Figure 2.8.
Output from a 500 MW generator acting as a primary control reserve, running at 250 MW at a nominal grid frequency of 50 Hz, with three different droop settings.

2. Balancing the electricity supply in case of calamities-image-2.9
Figure 2.9.
Typical ramp-up rate of primary control reserves.

Let us now return to our example of the electricity supply system illustrated in Figure 2.3, whereby ten identical generators were each carrying a load of 450 MW when suddenly one generator tripped. Without any action being taken with respect to the set point of the prime mover that drives each generator, the grid frequency would drop to 40 Hz, as shown in Figure 2.7. However, if each of the nine remaining generators is also used partly for primary frequency control, each with a droop of 4% as depicted in Figure 2.10, the grid frequency will not decrease all the way down to 40 Hz. Due to the droop setting, the set point for full output of each power plant will be reached already at a grid frequency of 49.8Hz, as shown in Figure 2.10. Figure 2.6 reveals that after the contingency of one failing power plant, this 49.8 Hz will already be reached in about 0.5 seconds after the trip of power plant number 7. Therefore, the new set point that asks for full output of the generators can be presumed to be present almost immediately after the loss of one of the ten generators in the system of our example.

However, not with standing the quick change in their set points, the nine power plants that use their additional available capacity for primary control reserves will not immediately reach their full output. If we presume that the output of these power plants follow the prescribed ramping up as shown in Figure 2.9, the load from the grid and the power supplied to the generators will equal each other after about 25 seconds (see Figure 2.11). At that point, the grid frequency reaches its minimum.

According to Figure 2.9, the additional 50 MW needed per generator that was lost when one of the original ten generators tripped is fully available from the primary reserves after 30 seconds. This 50 MW per generator is slightly more than the grid load requires for staying balanced at the mentioned minimum in frequency. This is because of the reduction in load created by the self-regulating power. The excess energy delivered is then used to accelerate again the rotating masses, the ‘flywheels’, in the system. Nevertheless, the nominal 50 Hz frequency cannot be reached with primary control reserves following a droop line. In our example, as soon as the grid frequency exceeds 49.8 Hz, the output of the primary reserves once again decreases (see Figure 2.10). That would create another mismatch between power supply and load. Therefore, a deviation between the actual frequency and the nominal grid frequency of slightly less than 0.2 Hz will remain where only primary control reserves are used to restore balance. This deviation between the frequency ultimately reached with the help of primary reserves and the desired frequency of 50.000 Hz is called the quasi steady state deviation. Transmission system operators define the permissible minimum and maximum of this deviation. Extra capacity, the so-called secondary control reserve, is needed for providing the extra power in the system so as to restore the grid frequency to the required narrow band around the nominal frequency.

As mentioned earlier, this narrow frequency band equals only +/– 20 mHz around 50 Hz in the Continental Europe synchronous area. The application of secondary control reserves increases the frequency further, so that the primary reserves can reduce their output and return to their initial load setting at 50 Hz. In other words, secondary reserves release the primary reserves from their duty. This enables primary reserves to be ready for the next major occurrence, such as the sudden loss of a generator or the losing or receiving of a large load.

2. Balancing the electricity supply in case of calamities-image-2.10
Figure 2.10.
The droop function of a generator having a 500 MW nominal power, again with a droop setting of 4%, but now running at 450 MW at 50 Hz.

The previous example whereby primary control reserves equalled exactly the amount of lost generating capacity, is obviously an exception. In reality, the power plants in a system do not all have the same capacity. Primary reserves should be large enough to compensate for at least the loss of the largest power plant in a system. In practice, power plants not dedicated as primary reserves will also provide some compensating power for restoring the grid frequency to its nominal value. Nevertheless, our example illustrates the way contingencies are handled in a grid system.

2. Balancing the electricity supply in case of calamities-image-2.11
Figure 2.11.
Primary control reserves prevent the system from decreasing too much in frequency after a loss in generating capacity (inertia constant 10 s, 1%/Hz self regulation, 50 MW unbalance for a generator with a nominal capacity of 500 MW).

Transmission system operators specify the maximum allowed dip in frequency, officially called the minimum instantaneous frequency after loss of generation. In the Continental Europe synchronous area, this value equals 49.2 Hz. If the grid frequency drops below this minimum allowed frequency, load shedding will be used to avoid too deep deviations from the nominal frequency. This means that a group of consumers will have no access to electricity for a while. The example with the 10 power plants of equal size, where one of the units trips, shows that losing 10% of generating capacity gives a frequency dip way below 49.2 Hz. Hence, there should be enough power plants in a system to ensure that the loss of one power plants does not reduce the online generating capacity by more than about 3%. In particular, systems having a large fraction of renewable electricity sources need dispatchable power plants of a limited size, and consequently more of them than in the case of large power plants only.


Should the inertia constant of the combined generators in a system decrease, for example, due to the introduction of a large amount of renewable energy sources that are indirectly connected to the grid via frequency converters, the grid frequency can drop to quite low values during a contingency. The red line in Figure 2.12 shows a deep dip in frequency if the inertia constant of the system is 5 s instead of 10 s. The other conditions are the same as in Figure 2.11, with 1%/Hz self regulation, 50 MW unbalance per generator and a nominal generator capacity of 500 MW.

The 46 Hz minimum in the red curve occurs 18 seconds after the calamity that tripped one of the 10 power plants in the example. At that point, the output of the primary reserves has not yet reached its maximum, since that occurs 30 seconds after the trip. After 18 seconds, the output of the primary reserve per generator is therefore only 18/30 · 50 MW = 30 MW. However, the self-regulating power of the grid equals 4 · 1/100 · 500 MW = 20 MW for a frequency drop of 4 Hz and a self regulating power sensitivity of the load of 1% per Hz. This means that power demand and power supply are fully matched again at 46 Hz so that the frequency will not decrease further.
The output of the primary reserves continues to increase after the minimum in frequency has been reached. This additional power supply will again accelerate the inertia. This will go faster for an inertia constant of 5 s than for one of 10 s, since less energy is needed to bring the rotors back to nominal speed in case of lower inertia. The deep dip in frequency observed for the lower inertia value is unacceptable is most cases. If it is not possible to increase the inertia, the amount of primary reserves has to be increased or the primary reserves have to be made faster with a smaller initial delay.

2. Balancing the electricity supply in case of calamities-image-2.12
Figure 2.12.
The effect of lowering the inertia constant on the frequency dip in case of a calamity (further conditions as in Figure 2.11.).


Increasing the amount of primary control reserves and having a higher ramp rate in the reserves help mitigate the effects of disturbances. Both measures will reduce the undesired deep dip in frequency, and will shorten the time needed to return to 50 Hz following the loss of a generator.

Doubling the power capacity of the primary control reserves, which results in twice as much capacity as the lost output of a failing generator in our previous example, reduces the dip in frequency by almost a factor of two. This is illustrated by the dark red line in figure 2.13. The reason that the reduction in dip is not exactly a factor of two is that the self-regulating power decreases the load to a lesser extent when grid frequencies come closer to 50 Hz. Another observation is that with a higher amount of primary reserves, the ultimate quasi steady state frequency will be closer to 50 Hz than in the case where the primary reserves equal just the loss in
power from the tripped generator. This is because even above 48.8 Hz, where the output reduction of the primary reserves kicks in because of the chosen droop curve, more power than the power lost from the failing generator is available now.

An additional advantage of higher primary reserves is the shorter time that the grid frequency substantially deviates from the nominal frequency of 50 Hz. Figure 2.13 clearly illustrates that the cumulative, or integral, loss in grid frequency over time is much lower for the red line than for the blue line. This cumulative loss is the area between each curve and the 50 Hz line in figure 2.13. The dimension of cumulative loss is Hz s, because it is the result of a deviation in Hz during a given number of seconds. For the blue line, the cumulative frequency deviation equals 146 Hz s. Doubling the primary reserves brings the cumulative frequency deviation in the example of Figure 2.13 down to only 41 Hz s. Therefore, doubling the primary reserves decreases the cumulative frequency deviation by a factor of 3.5 in this case. A grid operator has to ensure that, at the end of the day, the average frequency is back to 50.000 Hz, to avoid for instance, deviations of synchronous clocks. This means that after a frequency dip, all generators have to run at a higher frequency than 50 Hz for a while to compensate for the dip. With a lower cumulative frequency deviation, less effort is required to compensate for this deviation.

Should more primary reserve capacity be made available than the nominal output of the largest generator in a system in order to avoid excessive frequency dips in case of contingencies, a larger amount of generating capacity must be run below nominal load. This has a negative impact on capital costs, fuel consumption, and operation and maintenance costs.

2. Balancing the electricity supply in case of calamities-image-2.13
Figure 2.13.
Frequency dip in the case of primary reserves with two different power capacities, otherwise the same conditions as in Figure 2.11. apply.

Doubling the power-up ramp rate of the primary control reserves, rendering full output in 15 s instead of 30 s, results in almost the same positive effect on the frequency dip as doubling the capacity of primary control reserves. This is illustrated in figure 2.14. The only slight difference is that the ultimate frequency before the secondary control reserves step in will not be slightly above 48.8 Hz.
This is the logical consequence of the 4% droop in the example: if the grid
frequency rises above 48.8 Hz, the power output of the primary control reserves automatically decreases again. However, if the droop setting of the more agile primary control reserves is changed to 2%, the exact same positive curve as for doubling the primary control reserves will result.

2. Balancing the electricity supply in case of calamities-image-2.14
Figure 2.14.
The almost identical effect of a faster ramp up rate of primary control reserves compared with a doubled primary control response capacity (otherwise same conditions as in Figure 2.13).

A major conclusion here is that faster primary control reserves offer an effective option for reducing the frequency deviation caused by a contingency. If the relative amount of the system’s rotating inertia decreases, such as when much indirectly coupled renewable electricity sources that do not add to the inertia are introduced, the decline in frequency after a major loss in generation will be faster. In that case,
more primary control reserves are required to keep the system stable, or the primary control reserves need to have higher ramping rates and a shorter initial delay.


Nowadays, agile and flexible generator sets are available that have a very fast response to stepwise changes in the desired power output. Such smart generators can change a certain amount of their output rapidly, with a delay of less than 1 s, the actual response time depending to some extent on their running speed. Naturally, the allowed increase in output depends on the power output set point before the request for a change. As an example, a machine that runs already at 80% load can never accept a load increase of 40% of the nominal output.

2. Balancing the electricity supply in case of calamities-image-2.15
Figure 2.15.
Response of different generating techniques to a stepwise change in the power output set point.

Figure 2.15 shows the response of a number of different power generating
techniques to a stepwise change in desired output. The data are based on best-in-class machines. Less agile types might be slower by more than a factor of two. The response curve for the combustion-engine-driven power plant applies for smart power generation systems based on turbocharged engines with electromagnetic gas injection valves per cylinder. Faster primary reserves result in smaller deviations from the nominal grid frequency during calamities, even where there is less inertia in the system.

2. Balancing the electricity supply in case of calamities-image-2.16
Figure 2.16.
Control areas and control blocks interconnected with high-voltage transmission lines in a synchronous area. The blue blocks represent power plants.

One important aspect of primary reserves is their location in the system. If a large power plant of 1500 MW fails, the primary reserves cannot be located 1000 km away since it would cause the interconnecting transmission lines to overload. This is why large synchronous areas are split up into control blocks and control areas.
Each control area should be able to resolve most of the consequences of its own contingencies. Neighbouring control areas in the same block are allowed to offer some support, provided the interconnectors can carry the load. As with most things in life, local problems should preferably be solved locally.


Occurrences such as the loss of an active power plant or the loss or arrival of a major load, will always create a disturbance in frequency. The initial change rate in frequency is determined by the size of the unbalance between power supply and demand and by the inertia constant of the system. Primary control reserves, i.e. power plants running at part-load, adapt their output automatically when the grid frequency changes since their desired output is determined by the grid frequency via their droop curve. Primary reserves are preferably allocated in such a way that a control area can resolve a large part of its own contingencies.

It is clear that opting for just a few large power plants to supply the required electricity for an area is not ideal. In a system having a large number of generators in a single control area, it is easier to compensate for the failure of one generator with the other generators. The output of an individual unit is then just a small fraction of the combined output. With a large number of generators active as primary reserves, there is also no need to operate them at a relatively low load. In addition, the failure of one of them has only a minor effect on the combined reserve capacity. In other words, multiple generators in a system improve the reliability of primary reserves and reduce the impact of a failing unit.

Fast responding primary reserves can compensate for less rotating inertia without the need of having more primary reserves available. Without fast primary reserves, more power plant capacity has to operate at part-load, i.e. below its rated output. Operating at part-load increases the fuel consumption (MJ/kWh), as well as the capital and maintenance costs per kWh (Figure 2.17). Running a generating unit at 50% load doubles the maintenance and capital costs per kWh.


After activation of the primary control reserves, secondary and tertiary reserves (also known as frequency restoration reserves (FRR), and replacement reserves (RR), respectively) are needed. There are two reasons for this. Firstly, when primary reserves have been fully activated, no spare frequency control capacity is available for another tripping power plant or a loss of major load. The risk of another power plant failing or load disturbances taking place is always higher during a major event in the system than when everything is running smoothly.
Secondly, due to the droop characteristic of the deployed primary reserves, a deviation from the nominal frequency of 50 Hz will remain. Activation of the secondary control reserves will supply extra power to the system so that the grid frequency can return to its nominal value. As a consequence, the droop-based primary control reserves will automatically return to their original set point and be released from their action until the next disturbance occurs.

2. Balancing the electricity supply in case of calamities-image-2.17
Figure 2.17.
Example of fuel consumption, and capital and maintenance costs of a 400 MW power plant according to load.

In the Continental Europe synchronous area, the secondary control reserves in the relevant control area automatically commence delivering output within 30 seconds following a major disturbing occurrence. This happens when the primary control reserves are fully active. After 15 minutes, the full capacity of the secondary control reserves has to deliver its power to the system. This approach is illustrated in figure 2.18. Until recently, such a short deployment time required all secondary control reserves to be spinning all the time. Large power plants can never provide full output from standstill within 15 minutes. As a consequence, secondary reserves based on such power plants would always be running at a level below their nominal output. This again causes higher fuel consumption and higher capital and operational costs.

2. Balancing the electricity supply in case of calamities-image-2.18
Figure 2.18.
A possible setup for
primary control reserves
and secondary control
reserves in the system.


Nowadays some engine-driven power plants are able to deliver full output from standstill within about 5 minutes (Figure 2.19). This opens up possibilities for non-spinning secondary control reserves. Reaching full speed following the start command and reaching synchronisation with the grid takes about 30 seconds for such plants. This more than complies with the Continental Europe synchronous area’s requirement for secondary reserves to start 30 seconds after the event that triggered the primary reserves, and to ramp up to full output within 15 minutes. The ability to reach full output in 5 minutes is faster by a factor of 3 than what is demanded in the Continental Europe synchronous area.

2. Balancing the electricity supply in case of calamities-image-2.19
Figure 2.19.
The fast ramping up in
power output from
standstill to full load of a
smart power generator.

A quick-starting power plant has to be constantly preheated to provide the fast performance shown in figure 2.19. However, such power plants generally consist of multiple identical generators operating in parallel. Running one of the multiple units online for electrical energy production releases sufficient heat to keep at least 30 identical generating units preheated. Another advantage of having multiple secondary control reserve units in parallel is the low risk of losing much allocated reserve capacity. In case a single unit fails, only a fraction of the allocated power for secondary reserves is lost.

In some control areas in competitive markets, power plant operators can bid in the ahead markets for offering secondary reserves. The power plants offering the lowest price for their service will normally be selected to provide the reserves. In other systems, the power balance in the relevant area is continuously measured with energy flow meters, and secondary balancing is activated by a computerised control system that sends out set-point changes to selected power plants. In these cases, fast non-spinning secondary reserves also offer substantial advantages. Non-spinning means using no fuel, suffering no wear, and producing no emissions.

Tertiary control reserves are activated to free the secondary reserves for the next contingency. This can be done automatically (directly activated) or manually (schedule activated) by the transmission system operator. Part of the tertiary control reserves can be non-spinning. This is possible when the power plant has the rapid response capability as depicted in figure 2.19.

2. Balancing the electricity supply in case of calamities-image-2.20
Figure 2.20.
Example of the sequence in utilising primary, secondary and tertiary control reserves following a major occurrence.


This chapter has explained the delicate balance between electricity generation and demand and the consequences of rapid disturbances in the system. Rotating inertia and the dedicated control reserves play an important role in keeping the system balanced. With the introduction of a substantial amount of renewable electricity sources, balancing becomes more challenging. Agile, fast-reacting generators appear to offer excellent balancing duties during contingencies, even in the case of less inertia and reserve capacity in the system.