Economic Operation of Power Systems

Overview
A good business practice is the one in which the production cost is minimized without
sacrificing the quality. This is not any different in the power sector as well. The main aim
here is to reduce the production cost while maintaining the voltage magnitudes at each bus.
In this chapter we shall discuss the economic operation strategy along with the turbinegovernor control that are required to maintain the power dispatch economically.
A power plant has to cater to load conditions all throughout the day, come summer or
winter. It is therefore illogical to assume that the same level of power must be generated at
all time. The power generation must vary according to the load pattern, which may in turn
vary with season. Therefore the economic operation must take into account the load
condition at all times. Moreover once the economic generation condition has been
calculated, the turbine-governor must be controlled in such a way that this generation
condition is maintained. In this chapter we shall discuss these two aspects.

Section I: Economic Operation Of Power System

Economic Distribution of Loads between the Units of a Plant

Generating Limits
Economic Sharing of Loads between Different Plants

In an early attempt at economic operation it was decided to supply power from the most
efficient plant at light load conditions. As the load increased, the power was supplied by this
most efficient plant till the point of maximum efficiency of this plant was reached. With
further increase in load, the next most efficient plant would supply power till its maximum
efficiency is reached. In this way the power would be supplied by the most efficient to the
least efficient plant to reach the peak demand. Unfortunately however, this method failed to
minimize the total cost of electricity generation. We must therefore search for alternative
method which takes into account the total cost generation of all the units of a plant that is
supplying a load.
Economic Distribution of Loads between the Units of a Plant to determine the economic

distribution of a load amongst the different units of a plant, the variable operating costs of
each unit must be expressed in terms of its power output. The fuel cost is the main cost in a
thermal or nuclear unit. Then the fuel cost must be expressed in terms of the power output.
Other costs, such as the operation and maintenance costs, can also be expressed in terms of
the power output. Fixed costs, such as the capital cost, depreciation etc., are not included in
the fuel cost.
The fuel requirement of each generator is given in terms of the Rupees/hour. Let us define
the input cost of an unit- i , fi in Rs./h and the power output of the unit as Pi . Then the input
cost can be expressed in terms of the power output as
Rs./h

(5.1)

The operating cost given by the above quadratic equation is obtained by approximating the
power in MW versus the cost in Rupees curve. The incremental operating cost of each unit is
then computed as
Rs./MWh

(5.2)

Let us now assume that only two units having different incremental costs supply a load.
There will be a reduction in cost if some amount of load is transferred from the unit with
higher incremental cost to the unit with lower incremental cost. In this fashion, the load is
transferred from the less efficient unit to the more efficient unit thereby reducing the total
operation cost. The load transfer will continue till the incremental costs of both the units are
same. This will be optimum point of operation for both the units.
The above principle can be extended to plants with a total of N number of units. The total
fuel cost will then be the summation of the individual fuel cost fi , i = 1, ... , N of each unit,
i.e.,
(5.3)

Let us denote that the total power that the plant is required to supply by PT , such that
(5.4)

where P1 , ... , PN are the power supplied by the N different units.

The objective is minimize fT for a given PT . This can be achieved when the total
difference dfT becomes zero, i.e.,
(5.5)

Now since the power supplied is assumed to be constant we have

(5.6)

Multiplying (5.6) by and subtracting from (5.5) we get

(5.7)

The equality in (5.7) is satisfied when each individual term given in brackets is zero. This
gives us

(5.8)

Also the partial derivative becomes a full derivative since only the term fi of fTvaries
with Pi, i = 1, ..., N . We then have

Generating Limits
It is not always necessary that all the units of a plant are available to share a load. Some of
the units may be taken off due to scheduled maintenance. Also it is not necessary that the less
efficient units are switched off during off peak hours. There is a certain amount of shut down
and start up costs associated with shutting down a unit during the off peak hours and
servicing it back on-line during the peak hours. To complicate the problem further, it may
take about eight hours or more to restore the boiler of a unit and synchronizing the unit with
the bus. To meet the sudden change in the power demand, it may therefore be necessary to
keep more units than it necessary to meet the load demand during that time. This safety
margin in generation is called spinning reserve . The optimal load dispatch problem must
then incorporate this startup and shut down cost for without endangering the system security.
The power generation limit of each unit is then given by the inequality constraints
(5.10)

The maximum limit Pmax is the upper limit of power generation capacity of each unit. On the
other hand, the lower limit Pmin pertains to the thermal consideration of operating a boiler in a
thermal or nuclear generating station. An operational unit must produce a minimum amount
of power such that the boiler thermal components are stabilized at the minimum design
operating temperature.

Example 5.2
let us consider a generating station that contains a total number of three generating units. The
fuel costs of these units are given by

Rs./h

Rs./h

Rs./h

The generation limits of the units are

The total load that these units supply varies between 90 MW and 1250 MW. Assuming that all the
three units are operational all the time, we have to compute the economic operating settings as the
load changes.
The incremental costs of these units are

Rs./MWh

Rs./MWh

Rs./MWh
At the minimum load the incremental cost of the units are

Rs./MWh

Rs./MWh

Rs./MWh
Since units 1 and 3 have higher incremental cost, they must therefore operate at 30 MW each. The
incremental cost during this time will be due to unit-2 and will be equal to 26 Rs./MWh. With the
generation of units 1 and 3 remaining constant, the generation of unit-2 is increased till its incremental
cost is equal to that of unit-1, i.e., 34 Rs./MWh. This is achieved when P2 is equal to 41.4286 MW, at
a total power of 101.4286 MW.
An increase in the total load beyond 101.4286 MW is shared between units 1 and 2, till their
incremental costs are equal to that of unit-3, i.e., 43.5 Rs./MWh. This point is reached when P1 =
41.875 MW and P2 = 55 MW. The total load that can be supplied at that point is equal to 126.875.
From this point onwards the load is shared between the three units in such a way that the incremental
costs of all the units are same. For example for a total load of 200 MW, from (5.4) and (5.9) we have

Solving the above three equations we get P1 = 66.37 MW, P2 = 80 MW and P3 = 50.63 MW and an
incremental cost ( ) of 63.1 Rs./MWh. In a similar way the economic dispatch for various other load
settings are computed. The load distribution and the incremental costs are listed in Table 5.1 for
various total power conditions.

Table 5.1 Load distribution and incremental cost for the units of Example 5.1
PT (MW)

P1 (MW)

P2 (MW)

P3 (MW)

(Rs./MWh)

90
101.4286
120
126.875
150
200
300
400
500
600
700
800
906.6964
1000
1100
1181.25
1200
1250

30
30
38.67
41.875
49.62
66.37
99.87
133.38
166.88
200.38
233.88
267.38
303.125
346.67
393.33
431.25
450
500

30
41.4286
51.33
55
63.85
83
121.28
159.57
197.86
236.15
274.43
312.72
353.5714
403.33
456.67
500
500
500

30
30
30
30
36.53
50.63
78.85
107.05
135.26
163.47
191.69
219.9
250
250
250
250
250
250

26
34
40.93
43.5
49.7
63.1
89.9
116.7
143.5
170.3
197.1
223.9
252.5
287.33
324.67
355
370
410

At a total load of 906.6964, unit-3 reaches its maximum load of 250 MW. From this point onwards
then, the generation of this unit is kept fixed and the economic dispatch problem involves the other
two units. For example for a total load of 1000 MW, we get the following two equations from (5.4) and
(5.9)

Solving which we get P1 = 346.67 MW and P2 = 403.33 MW and an incremental cost of 287.33
Rs./MWh. Furthermore, unit-2 reaches its peak output at a total load of 1181.25. Therefore any further
increase in the total load must be supplied by unit-1 and the incremental cost will only be borne by
this unit. The power distribution curve is shown in Fig. 5.1.

Fig.5.1 Power distribution between the units of Example 5.2.

Example 5.3
Consider two generating plant with same fuel cost and generation limits. These are given by

For a particular time of a year, the total load in a day varies as shown in Fig. 5.2. Also an additional
cost of Rs. 5,000 is incurred by switching of a unit during the off peak hours and switching it back on
during the during the peak hours. We have to determine whether it is economical to have both units
operational all the time.

Fig. 5.2 Hourly distribution of load for the units of Example 5.2.
Since both the units have identical fuel costs, we can switch of any one of the two units during the off
peak hour. Therefore the cost of running one unit from midnight to 9 in the morning while delivering
200 MW is

Rs.
Adding the cost of Rs. 5,000 for decommissioning and commissioning the other unit after nine hours,
the total cost becomes Rs. 167,225.
On the other hand, if both the units operate all through the off peak hours sharing power equally, then
we get a total cost of

Rs.
which is significantly less that the cost of running one unit alone.

Economic Sharing of Loads between Different Plants

So far we have considered the economic operation of a single plant in which we have discussed how
a particular amount of load is shared between the different units of a plant. In this problem we did not
have to consider the transmission line losses and assumed that the losses were a part of the load
supplied. However if now consider how a load is distributed between the different plants that are
joined by transmission lines, then the line losses have to be explicitly included in the economic
dispatch problem. In this section we shall discuss this problem.
When the transmission losses are included in the economic dispatch problem, we can modify (5.4) as
(5.11)

where PLOSS is the total line loss. Since PT is assumed to be constant, we have
(5.12)

In the above equation dPLOSS includes the power loss due to every generator, i.e.,

(5.13)

Also minimum generation cost implies dfT = 0 as given in (5.5). Multiplying both (5.12) and (5.13)
by and combining we get

(5.14)

Adding (5.14) with (5.5) we obtain

(5.15)

The above equation satisfies when

(5.16)

Again since

from (5.16) we get

(5.17)

where Li is called the penalty factor of load- i and is given by

(5.18)

Example 5.4

Consider an area with N number of units. The power generated are defined by the vector

Then the transmission losses are expressed in general as

(5.19)

where B is a symmetric matrix given by

The elements Bij of the matrix B are called the loss coefficients . These coefficients are not constant
but vary with plant loading. However for the simplified calculation of the penalty factor Li these
coefficients are often assumed to be constant.
When the incremental cost equations are linear, we can use analytical equations to find out the
economic settings. However in practice, the incremental costs are given by nonlinear equations that
may even contain nonlinearities. In that case iterative solutions are required to find the optimal
generator settings.

Section II: Automatic Generation Control

Load Frequency Control

Automatic Generation Control

Electric power is generated by converting mechanical energy into electrical energy. The rotor mass,
which contains turbine and generator units, stores kinetic energy due to its rotation. This stored kinetic
energy accounts for sudden increase in the load. Let us denote the mechanical torque input
by Tm and the output electrical torque by Te . Neglecting the rotational losses, a generator unit is said
to be operating in the steady state at a constant speed when the difference between these two
elements of torque is zero. In this case we say that the accelerating torque
(5.20)
is zero.
When the electric power demand increases suddenly, the electric torque increases. However, without
any feedback mechanism to alter the mechanical torque, Tm remains constant. Therefore the
accelerating torque given by (5.20) becomes negative causing a deceleration of the rotor mass. As
the rotor decelerates, kinetic energy is released to supply the increase in the load. Also note that
during this time, the system frequency, which is proportional to the rotor speed, also decreases. We
can thus infer that any deviation in the frequency for its nominal value of 50 or 60 Hz is indicative of
the imbalance between Tm and Te. The frequency drops when Tm < Te and rises when Tm > Te .
The steady state power-frequency relation is shown in Fig. 5.3. In this figure the slope of the Pref line
is negative and is given by

(5.21)

where R is called the regulating constant . From this figure we can write the steady state power
frequency relation as

Fig. 5.3 A typical steady-state power-frequency curve.

(5.22)

Suppose an interconnected power system contains N turbine-generator units. Then the steady-state
power-frequency relation is given by the summation of (5.22) for each of these units as

(5.23)

In the above equation, Pm is the total change in turbine-generator mechanical power and Pref is the
total change in the reference power settings in the power system. Also note that since all the
generators are supposed to work in synchronism, the change is frequency of each of the units is the
same and is denoted by f. Then the frequency response characteristics is defined as

(5.24)

We can therefore modify (5.23) as

(5.25)
Example 5.5

Example 5.5
Consider an interconnected 50-Hz power system that contains four turbine-generator units rated 750
MW, 500 MW, 220 MW and 110 MW. The regulating constant of each unit is 0.05 per unit based on
its own rating. Each unit is operating on 75% of its own rating when the load is suddenly dropped by
250 MW. We shall choose a common base of 500 MW and calculate the rise in frequency and drop in
the mechanical power output of each unit.
The first step in the process is to convert the regulating constant, which is given in per unit in the base
of each generator, to a common base. This is given as

(5.26)

We can therefore write

Therefore

per unit
We can therefore calculate the total change in the frequency from (5.25) while assuming Pref = 0,
i.e., for no change in the reference setting. Since the per unit change in load - 250/500 = - 0.5 with the
negative sign accounting for load reduction, the change in frequency is given by

Then the change in the mechanical power of each unit is calculated from (5.22) as

It is to be noted that once Pm2 is calculated to be - 79.11 MW, we can also calculate the changes in
the mechanical power of the other turbine-generators units as

This implies that each turbine-generator unit shares the load change in accordance with its own
rating.

Load Frequency Control

Modern day power systems are divided into various areas. For example in India , there are five
regional grids, e.g., Eastern Region, Western Region etc. Each of these areas is generally
interconnected to its neighboring areas. The transmission lines that connect an area to its neighboring
area are called tie-lines . Power sharing between two areas occurs through these tie-lines. Load
frequency control, as the name signifies, regulates the power flow between different areas while
holding the frequency constant.
As we have in Example 5.5 that the system frequency rises when the load decreases if Pref is kept at
zero. Similarly the frequency may drop if the load increases. However it is desirable to maintain the
frequency constant such that f=0 . The power flow through different tie-lines are scheduled - for
example, area- i may export a pre-specified amount of power to area- j while importing another prespecified amount of power from area- k . However it is expected that to fulfill this obligation, areai absorbs its own load change, i.e., increase generation to supply extra load in the area or decrease
generation when the load demand in the area has reduced. While doing this area- i must however
maintain its obligation to areas jand k as far as importing and exporting power is concerned. A
conceptual diagram of the interconnected areas is shown in Fig. 5.4.

Fig. 5.4 Interconnected areas in a power system.

We can therefore state that the load frequency control (LFC) has the following two objectives:

Hold the frequency constant ( f = 0) against any load change. Each area must contribute to
absorb any load change such that frequency does not deviate.
Each area must maintain the tie-line power flow to its pre-specified value.

(5.27)
The first step in the LFC is to form the area control error (ACE) that is defined as

where Ptie and Psch are tie-line power and scheduled power through tie-line respectively and the
constant Bf is called the frequency bias constant .
The change in the reference of the power setting Pref, i , of the area- i is then obtained by the

(5.28)
feedback of the ACE through an integral controller of the form

where Ki is the integral gain. The ACE is negative if the net power flow out of an area is low or if the
frequency has dropped or both. In this case the generation must be increased. This can be achieved
by increasing Pref, i . This negative sign accounts for this inverse relation between Pref, i and ACE.
The tie-line power flow and frequency of each area are monitored in its control center. Once the ACE
is computed and Pref, i is obtained from (5.28), commands are given to various turbine-generator
controls to adjust their reference power settings.

Example 5.6
Consider a two-area power system in which area-1 generates a total of 2500 MW, while area-2
generates 2000 MW. Area-1 supplies 200 MW to area-2 through the inter-tie lines connected between
the two areas. The bias constant of area-1 ( 1 ) is 875 MW/Hz and that of area-2 ( 2 ) is 700 MW/Hz.
With the two areas operating in the steady state, the load of area-2 suddenly increases by 100 MW. It
is desirable that area-2 absorbs its own load change while not allowing the frequency to drift.
The area control errors of the two areas are given by
and

Since the net change in the power flow through tie-lines connecting these two areas must be zero, we
have

Also as the transients die out, the drift in the frequency of both these areas is assumed to be
constant, i.e.,

If the load frequency controller (5.28) is able to set the power reference of area-2 properly, the ACE of
the two areas will be zero, i.e., ACE1= ACE2 = 0. Then we have

This will imply that f will be equal to zero while maintaining Ptie1 =Ptie2 = 0. This signifies that area2 picks up the additional load in the steady state.

Coordination Between LFC And Economic Dispatch

Both the load frequency control and the economic dispatch issue commands to change the power
setting of each turbine-governor unit. At a first glance it may seem that these two commands can be
conflicting. This however is not true. A typical automatic generation control strategy is shown in Fig.
5.5 in which both the objective are coordinated. First we compute the area control error. A share of
this ACE, proportional to i , is allocated to each of the turbine-generator unit of an area. Also the
share of unit- i , i X ( PDK - Pk ), for the deviation of total generation from actual generation is
computed. Also the error between the economic power setting and actual power setting of unit- i is
computed. All these signals are then combined and passed through a proportional gain Ki to obtain
the turbine-governor control signal.

Fig. 5.5 Automatic generation control of unit-i.