SYLLABUS

UNIT IV UNIT COMMITMENT AND ECONOMIC DISPATCH

Statement of economic dispatch problem – cost of generation – incremental
cost curve co-ordinationequations without loss and with loss, solution by direct
method and λ-iteration method. (Noderivation of loss coefficients). Statement of
Unit Commitment problem – constraints; spinningreserve, thermal unit constraints,
hydro constraints, fuel constraints and other constraints. Solutionmethods -
Priority-list methods - forward dynamic programming approach.
Numericalproblems only in priority-list method using full-load average production
cost.
UNIT IV UNIT COMMITMENT AND ECONOMIC DISPATCH
4.1. INTRODUCTION:
A power system has several power plants. Each power plant has several
generating units. At any point of time, the total load in the system is met by the
generating units in different power plants. Economic dispatch control determines
the power output of each power plant, and power output of each generating unit
within a power plant, which will minimize the overall cost of fuel needed to serve
the system load.
 We study first the most economical distribution of the output of a power
plant between the generating units in that plant. The method we develop also
applies to economic scheduling of plant outputs for a given system load
without considering the transmission loss.
 Next, we express the transmission loss as a function of output of the various
plants.
 Then, we determine how the output of each of the plants of a system is
scheduled to achieve the total cost of generation minimum, simultaneously
meeting the system load plus transmission loss.
4.2. INPUT – OUTPUT CURVE OF GENERATING UNIT

Power plants consisting of several generating units are constructed investing

huge amount of money. Fuel cost, staff salary, interest and depreciation charges
and maintenance cost are some of the components of operating cost. Fuel cost is
the major portion of operating cost and it can be controlled. Therefore, we shall
consider the fuel cost alone for further consideration.
To get different output power, we need to vary the fuel input. Fuel input can
be measured in Tonnes / hour or Millions of Btu / hour. Knowing the cost of the
fuel, in terms of Rs. / Tonne or Rs. / Millions of Btu, input to the generating unit
can be expressed as Rs / hour. Let Ci Rs / h be the input cost to generate a power of
Pi MW in unit i. Fig.1 shows a typical input – output curve of a generating unit.
For each generating unit there shall be a minimum and a maximum power
generated as Pimin and Pimax.
If the input-output curve of unit i is quadratic, we can write
Ci   i Pi   i Pi   i Rs / h
2
(1)
A power plant may have several generator units. If the input-output
characteristics of different generator units are identical, then the generating units
can be equally loaded. But generating units will generally have different input-
output characteristic. This means that, for particular input cost, the generator power
Pi will be different for different generating units in a plant.
4.3. INCREMENTAL COST CURVE
As we shall see, the criterion for distribution of the load between any two
units is based on whether increasing the generation of one unit, and decreasing the
generation of the other unit by the same amount results in an increase or decrease
in total cost. This can be obtained if we can calculate the change in input cost ΔCi
for a small change in power ΔPi.
dCi Ci
Since 
dPi Pi

dCi
We can write Ci  Pi
dPi

dCi
Thus while deciding the optimal scheduling; we are concerned with
dPi

INCREMENTAL COST (IC) which is determined by the slopes of the input-

output curves.
dCi
Thus the incremental cost curve is the plot of versus Pi. The dimension
dPi

dCi
of is Rs / MWh.
dPi

The unit that has the input – output relation as

Ci   i Pi   i Pi   i Rs/h…………………………………………………(1)
2
Has incremental cost (IC) as
dCi
= 2 i Pi   i ……………………………………………………………...(2)
dPi

Here ɑi , βi, γi are constants.

This figure shows that incremental cost is quite linear with respect to power
output over an appreciable range. In analytical work, the curve is usually
approximated by one or two straight lines. The dashed line in the figure is a good
representation of the curve. We now have the background to understand the
principle of economic dispatch which guides distribution of load among the
generating units within a plant.
4.4. ECONOMICAL DIVISION OF PLANT LOAD BETWEEN
GENERATING UNITS IN A PLANT
Various generating units in a plant generally have different input-output
characteristics. Suppose that the total load in a plant is supplied by two units and
that the division of load between these units is such that the incremental cost of one
unit is higher than that of the other unit.
Now suppose some of the load is transferred from the unit with higher
incremental cost to the unit with lower incremental cost. Reducing the load on the
unit with higher incremental cost will result in greater reduction of cost than the
increase in cost for adding the same amount of load to the unit with lower
incremental cost. The transfer of load from one to other can be continued with a
reduction of total cost until the incremental costs of the two units are equal.
This is illustrated through the characteristics shown in Fig. 3

Initially, IC2 > IC1 . Decrease the output power in unit 2 by ΔP and increase
output power in unit 1 by ΔP. Now IC2 ΔP > IC1 ΔP. Thus there will be more
decrease in cost and less increase in cost bringing the total cost lesser.
 This change can be continued until IC1 = IC2 at which the total cost will be
minimum. Further reduction in P2 and increase in P1 will in IC1>IC2 calling for
decrease in P1 and increase in P2 until IC1 = IC2. Thus the total cost will be
minimum when the incremental costs are equal.
The same reasoning can be extended to a plant with more than two
generating units also. In this case, if any two units have different incremental costs,
then in order to decrease the total cost of generation, decrease the output power in
unit having higher IC and increase the output power in unit having lower IC. When
this process is continued, a stage will reach wherein incremental costs of all the
units will be equal. Now the total cost of generation will be minimum. Thus the
economical division of load between units within a plant is that all units must
operate at the same incremental cost. Now we shall get the same result
mathematically.
Consider a plant having N number of generating units. Input-output curve of
the units are denoted as C1 (P1), C2 (P2)…CN (PN). Our problem is, for a given
load demand PD, find the set of Pi s which minimizes the cost function
CT = C1 (P1) + C2 (P2) +……. + CN (PN)……………………………… ………. (3)
Subject to the constraints
PD – (P1 + P2 + P3+ ……. + PN) = 0 …………………………………… ……..... (4)
And Pi min ≤ Pi ≤ Pimax i=1, 2 ….N………………………………………… …….. (5)
Omitting the inequality constraints for the time being the problem to be solved
becomes
N
Minimize CT   Ci ( Pi ) ………………………………………………………… (6)
i 1

N
Subject to PD   ( Pi ) = 0 ………………………………………………............. (7)
i 1
 This optimizing problem with equality constraint can be solved by the
method of Lagrangian multipliers. In this method, the Lagrangian function is
formed by augmenting the equality constraints to the objective function using
proper Lagrangian multipliers. For this case, Lagrangian function is
N N
L (P1, P2… PN , λ) = CT   Ci ( Pi ) + λ ( PD   ( Pi ) )………………………(8)
i 1 i 1

Where λ is the Lagrangian multiplier. Now this Lagrangian function has to

be minimized with no constraints on it. The necessary conditions for a minimum
are
L L
0 i=1, 2…, N and 0
Pi 

For a plant with 3 units

L (P1 , P2, P3, λ) = C1(P1) + C2(P2) + C3(P3) + λ(PD – P1 – P2 – P3)
Necessary conditions for a minimum are

L C1
   (1)  0
P1 P1

L C 2
   (1)  0
P2 P2

L C3
   (1)  0
P3 P3

L
 PD  P1  P2  P3  0

Generalizing the above the necessary conditions are
C i
  i=1,2,….N ……………………………………………………………(9)
Pi

And
N
PD   ( Pi ) = 0 ……………………………………………………………(10)
i 1
 C i
Here   is the change in production cost in unit i for a small change in
Pi

unit i . Since change in generation in unit i will affect the production cost of this
unit alone we can write

Ci Ci
 ……………………………………………………………………… (11)
Pi Pi

Using equation (11) in eqn (9)

We have
C i
  i=1,2……….N ……………………………………………………… (12)
Pi

Thus the solution for the problem

Minimize
N N
CT   Ci ( Pi ) Subject to PD   ( Pi )  0
i 1 i 1

is obtained when the following equations are satisfied.

C i
  i=1,2……….N ……………………………………………………(13)
Pi
N
And PD   ( Pi )  0 …………………………………………………………. (14)
i 1

The above two conditions give n+1 number of equations which are to be
solved for the n+1 number of variables λ , P1, P2 ……. PN Equation (13) simply
says that at the minimum cost operating point, the incremental cost for all the
generating units must be equal. This condition is commonly known as equal
incremental cost rule. Equation (14) is known as power balance equation.
It is to be remembered that we have not yet considered the inequality constraints
given by
Pi min ≤ Pi ≤ Pimax i=1, 2 ….N

Fortunately, if the solution obtained without considering the inequality

constraints satisfies the inequality constraints also, then the obtained solution will
be optimum. If for one or more generator units, the inequality constraints are not
satisfied, the optimum strategy is obtained by keeping these generator units in their
nearest limits and making the other generator units to supply the remaining power
as per equal incremental cost rule.
4.5. TRANSMISSION LOSS:-
Generally, in a power system, several plants are situated at different places.
They are interconnected by long transmission lines. The entire system load along
with transmission loss shall be met by the power plants in the system.
Transmission loss depends on i) line parameters ii) bus voltages and iii) power
flow. Determination of transmission loss requires complex computations.
However, with reasonable approximations, for a power system with N number of
power plants, transmission loss can be represented as
 B11 B12 ...... B1N   P1 
B B22 ........ B2 N   P2 
 21  
 . . . .  . 
PL  P1 P2 PN  . . . .  .  (15)
  
 . . . .  . 
 . . . .  . 
 
 BN1 BN 2 ...... B NN   PN 

Where P1, P2, PN are the power supplied by the plants 1, 2… N respectively
From eqn.15
 N 
  B1n Pn 
 nN1 
 B P
 n 1
2n n

PL  P1 P2 PN  . 
 
 . 
 . 
N 
 B Nn Pn 
 n 1 

Thus PL can be written as

N N
PL   Pm Bmn Pn ……………………………………………………………... (16)
m 1 n 1

When the powers are in MW, the Bmn coefficients are of dimension 1/ MW.
If powers are in per-unit, then Bmn coefficients are also in per-unit. Loss coefficient
matrix of a power system shall be determined before hand and made available for
economic dispatch.
For a two plant system, the expression for the transmission loss is

B B12   P1 
PL  P1 P2  11
 B21 B22   P2 

Solving the above equation we get Since Bmn coefficient matrix is symmetric, for
two plant system
PL  P1  B11  2P1 P2 B12  P2 B22 ……………………………………………….. (17)
2 2

PL
In later calculations we need the Incremental Transmission Loss (ITL),
PI

For Two plant system

PL
 2 B11P1  2 B12 P2 …………………………………………………………… (18)
P1
PL
 2 B12 P1  2 B22 P2 …………………………………………………………. (19)
P2

This can be generalized as

PL N
  2 Bmn Pn m = 1, 2…N
Pm n 1

4.6. ECONOMIC DIVISION OF SYSTEM LOAD BETWEEN VARIOUS

PLANTS IN THE POWER SYSTEM
It is to be noted that different plants in a power system will have different cost
characteristics.
Consider a power system having N number of plants. Input-output characteristics
of the plants are denoted as C1 (P1), C2 (P2)…CN (PN).
Our problem is for a given system load demand PD, find the set of plant generation
P1, P2, ………………, PN which minimizes the cost function
CT = C1 (P1) + C2 (P2) +……. + CN (PN)…………………………………… (21)
Subject to the constraints
PD+PL-(P1+P2+…………. …+ PN) = 0 …………………………………….. .(22)
And

Pi min ≤ Pi ≤ Pimax i=1, 2 ….N……………………………………………. (23)

Inequality constraints are omitted for the time being. The problem to be solved
becomes
N
Minimize CT   Ci ( Pi )
i 1 ……………………………………………………… (24)

Subject to
N
PD  PL   PI  0 …………………………………………………………(25)
i 1

For this case the lagrangian function is

 N N
L (P1, P2… PN , λ) = CT   Ci ( Pi ) + λ ( PD  PL   ( Pi ) )........................... (26)
i 1 i 1

This Lagrangian function has to be minimized with no constraint on it. The

necessary conditions for a minimum are
L i=1, 2...……N and L
0 0
Pi 

From eqn. (26) for a system with two plants

N
L (P1, P2, λ) =  C (P )  C P + λ ( P
i 1
1 1 2 2 D  PL  P1  P2 )

L C1 P
   ( L  1)  0
P1 P1 P1

L C 2 P
   ( L  1)  0
P2 P2 P2

L
 PD  PL  P1  P2  0

Generalizing this, for system with N plants, the necessary conditions for a
minimum are
C1 P
  ( L  1)  0
P1 P1 i = 1, 2,....N.......................................................... (27)
And
N
PD  PL   PI  0 ........................................................................................ (28)
i 1

As discussed in earlier case

We can write
Ci Ci
 ................................................................................................................(29.)
Pi Pi
Using eqn (29) in eqn.(27) we have
C i PL
  
Pi Pi
N
PD  PL   PI  0..........................................................................................(31)
i 1

These equations can be solved for the plant generation P1,P2,......PN. as

shown in the next section the value of λ in eqn. (30) is the incremental cost of
received power. The eqns. described in eqn. (30) are commonly known as
C i
coordination equations as they link the incremental cost of plant incremental
Pi

PL
cost of received power λ and the incremental transmission loss equation (31)
Pi

is the power balance equation.

Coordination equations
Ci P
 L 
Pi Pi

Can also be written as

ICi+ λ ITLi = λ, i=1,2,3.......N .................................................................. (32)
The N number of coordinate equations together with the power balance
equation are to be solved for the plant loads P1,P2,......PN to obtain the economic
schedule.
4.7. UNIT COMMITMENT:
Economic dispatch gives the optimum schedule corresponding to one
particular load on the system. The total load in the power system varies throughout
the day and reaches different peak value from one day to another. Different
combination of generators, are to be connected in the system to meet the varying
load.
 When the load increases, the utility has to decide in advance the sequence in
which the generator units are to be brought in. Similarly, when the load decreases,
the operating engineer need to know in advance the sequence in which the
generating units are to be shut down. The problem of finding the order in which the
units are to be brought in and the order in which the units are to be shut down over
a period of time, say one day, so the total operating cost involved on that day is
minimum, is known as Unit Commitment (UC) problem. Thus UC problem is
economic dispatch over a day.
The period considered may a week, month or a year. But why is this
problem in the operation of electric power system? Why not just simply commit
enough units to cover the maximum system load and leave them running? Note
that to “commit” means a generating unit is to be “turned on”; that is, bring the unit
up to speed, synchronize it to the system and make it to deliver power to the
network. “Commit enough units and leave them on line” is one solution. However,
it is quite expensive to run too many generating units when the load is not large
enough. As seen in previous example, a great deal of money can be saved by
turning units off (decommiting them) when they are not needed
4.7.1. Constraints on UC Problem:
Some of the constraints that are to be met with while solving UC problem are listed
below.
1. Spinning reserve: There may be sudden increase in load, more than what
was predicted. Further there may be a situation that one generating unit may
have to be shut down because of fault in generator or any of its auxiliaries.
Some system capacity has to be kept as spinning reserve i) to meet an
unexpected increase in demand and ii) to ensure power supply in the event
of any generating unit suffering a forced outage.
2. Minimum up time: When a thermal unit is brought in, it cannot be turned
off immediately. Once it is committed, it has to be in the system for a
specified minimum up time.
3. Minimum down time: When a thermal unit is decommitted, it cannot be
turned on immediately. It has to remain decommitted for a specified
minimum down time.
4. Crew constraint: A plant always has two or more generating units. It may
not be possible to turn on more than one generating unit at the same time due
to non-availability of operating personnel.
5. Transition cost: Whenever the status of one unit is changed some transition
cost is involved and this has to be taken into account.
6. Hydro constraints: Most of the systems have hydroelectric units also. The
operation of hydro units, depend on the availability of water. Moreover,
hydro-projects are multipurpose projects. Irrigation requirements also
determine the operation of hydro plants.
7. Nuclear constraint: If a nuclear plant is part of the system, another
constraint is added. A nuclear plant has to be operated as a base load plant
only.
8. Must run unit: Sometime it is a must to run one or two units from the
consideration of voltage support and system stability.
9. Fuel supply constraint: Some plants cannot be operated due to deficient
fuel supply.
10. Transmission line limitation: Reserve must be spread around the power
system to avoid transmission system limitation, often called “bottling” of
reserves.
The most talked-about techniques for the solution of the unit commitment problem

are:

• Priority-list schemes,

• Dynamic programming (DP),

4.8 Priority-List Method

The simplest unit commitment solution method consists of creating a priority list of

units. As we saw in Example 5B, a simple shut-down rule or priority-list scheme

could be obtained after an exhaustive enumeration of all unit combinations at each

load level. The priority list of Example 5B could be obtained in a much simpler

manner by noting the full-load average production cost of each unit, where the full-

load average production cost is simply the net heat rate at full load multiplied by the

fuel cost.

4.8.1 Example

Construct a priority list for the units of Example SA. (Use the same fuel costs as in

Example First, the full-load average production cost will be calculated:

Unit
Full Load Average Production Cost (e/MWh)

1 9.79

2 9.48

3 11.188
A strict priority order for these units, based on the average production cost, would

order them as follows:

Unit R/MWh Min MW Max MW
2 9.48 100 400
1 9.79 150 600
3 11.188 50 200
and the commitment scheme would (ignoring min up/down time, start-up costs, etc.)

simply use only the following combinations.

Combination Min MW from Max MW from

Combination Combination
2+1+3 300 1200
2+1 250 1000
2 100 400
Note that such a scheme would not completely parallel the shut-down sequence

described in Example 5B, where unit 2 was shut down at 600 MW leaving unit 1.

With the priority-list scheme, both units would be held on until load reached 400

MW, then unit 1 would be dropped.

4.8.2 Algorithm

Most priority-list schemes are built around a simple shut-down algorithm that might

operate as follows.

• At each hour when load is dropping, determine whether dropping the next unit

on the priority list will leave sufficient generation to supply the load plus

spinning-reserve requirements. If not, continue operating as is; if yes, go on to

the next step.

• Determine the number of hours, H, before the unit will be needed again. That

is, assuming that the load is dropping and will then go back up some hours

later.
 • If H is less than the minimum shut-down time for the unit, keep commitment

as is and go to last step; if not, go to next step.

• Calculate two costs. The first is the sum of the hourly production costs for the

next H hours with the unit up. Then recalculate the same sum for the unit down

and add in the start-up cost for either cooling the unit or banking it, whichever

is less expensive. If there is sufficient savings from shutting down the unit, it

should be shut down, otherwise keep it on.

• Repeat this entire procedure for the next unit on the priority list. If it is also

dropped, go to the next and so forth. Various enhancements to the priority-list

scheme can be made by grouping of units to ensure that various constraints are

met. We will note later that dynamic-programming methods usually create the

same type of priority list for use in the DP search.

4.9 Dynamic-Programming Solution

Dynamic programming has many advantages over the enumeration scheme, the

chief advantage being a reduction in the dimensionality of the problem. Suppose we

have found units in a system and any combination of them could serve the (single)

load. There would be a maximum of 24 - 1 = 15 combinations to test. However, if a

strict priority order is imposed, there are only four combinations to try:

Priority 1 unit

Priority 1 unit + Priority 2 unit

Priority 1 unit + Priority 2 unit + Priority 3 unit

Priority 1 unit + Priority 2 unit + Priority 3 unit + Priority 4 unit

The imposition of a priority list arranged in order of the full-load averagecost rate

would result in a theoretically correct dispatch and commitment only if:

1. No load costs are zero.

2. Unit input-output characteristics are linear between zero output and full load.

3. There are no other restrictions.

4. Start-up costs are a fixed amount. In the dynamic-programming approach that

follows, we assume that:

1. A state consists of an array of units with specified units operating and

2. The start-up cost of a unit is independent of the time it has been off-line

3. There are no costs for shutting down a unit.

4. There is a strict priority order, and in each interval a specified minimum the rest

off-line. (i.e., it is a fixed amount). amount of capacity must be operating.

A feasible state is one in which the committed units can supply the required load and

that meets the minimum amount of capacity each period.

4.10 Forward DP Approach

One could set up a dynamic-programming algorithm to run backward in time

starting from the final hour to be studied, back to the initial hour. Conversely, one

could set up the algorithm to run forward in time from the initial hour to the final

hour.

The forward approach has distinct advantages in solving generator unit

commitment. For example, if the start-up cost of a unit is a function of the time it has

been off-line (i.e., its temperature), then a forward dynamic-program approach is

more suitable since the previous history of the unit can be computed at each stage.
There are other practical reasons for going forward.

The initial conditions are easily specified and the computations can go forward

in time as long as required. A forward dynamic-programming algorithm is shown by

the flowchart in Figure. The recursive algorithm to compute the minimum cost in

hour K with combination I is,

F cost ( K , I )=min[ P cost ( K , I )+S cost (K −1, L: K , I )+F cost (K −1, L)]

Fcost(K, I) = least total cost to arrive at state (K, I)

Pcost(K, I) = production cost for state (K, I)

Scost(K - 1, L: K, I)= transition cost from state (K - 1, L) to state (K, I)

State (K, 1) is the Zth combination in hour K.

For the forward dynamicprogramming approach, we define a strategy as the

transition, or path, from one state at a given hour to a state at the next hour. Note that

two new variables, X and N, have been introduced in Figure

X = number of states to search each period

Flow chart for Unit commitment via forward dynamic programming.
 Fig. Restricted search paths in DP algorithm with N = 3 and X = 5.

N = number of strategies, or paths, to save at each step.

These variables allow control of the computational effort For complete

enumeration, the maximum number of the value of X or N is 2” - 1. For example,

with a simple priority-list ordering, the upper bound on X is n, the number of units.

Reducing the number N means that we are discarding the highest cost schedules at

each time interval and saving only the lowest N paths or strategies.

There is no assurance that the theoretical optimal schedule will be found using

a reduced number of strategies and search range (the X value); only experimentation

with a particular program will indicate the potential error associated with limiting the

values of X and N below their upper bounds.

The following flowchart explains the priority list method for load contingency.