ECE666 Winter 2018

Assignment-1 Solutions

Q.1 The problem is of minimizing the total system cost, given as,

J = C1 (P1 ) + C2 (P2 ) + C3 (P3 )

The system demand is 975 MW.

We shall first obtain the optimal solution without considering the generation limits. The
Lagrangian F, can be written as,

F = C1 ( P1 ) + C2 ( P2 ) + C3 (P3 ) + λ (975 − P1 − P2 − P3 )

The Kuhn-Tucker’s conditions are as follows:

∂F ∂F
= 0.008P1 + 7.2 − λ = 0; = 0.005P2 + 7.3 − λ = 0
∂P1 ∂P2
∂F ∂F
= 0.006 P3 + 6.74 − λ = 0; = 975 − P1 − P2 − P3 = 0
∂P3 ∂λ
Solving the above equations, we have
P1* = 233.5MW ; P2* = 353.6 MW ; P3* = 388MW ; λ = 9.068 $ / MWh

Since the generation from unit-3 exceeds the upper limit of 300 MW, the generation
from this unit is fixed at its upper limit; i.e., P3* = 300MW . Now we solve a reduced order
ELD problem with the two generators unit-1 and unit-2.
Minimize, J = C1 (P1 ) + C2 (P2 )

The system demand is 975 – 300 = 675 MW. The new Lagrangian F, can be written as,

F = C1 ( P1 ) + C2 ( P2 ) + λ (675 − P1 − P2 )

The Kuhn-Tucker’s conditions are as follows:

∂F ∂F ∂F
= 0.008P1 + 7.2 − λ = 0; = 0.005P2 + 7.3 − λ = 0; = 975 − P1 − P2 − P3 = 0
∂P1 ∂P2 ∂λ
Solving the above equations, we have
P1* = 267.4 MW ; P2* = 407.8MW ; λ = 9.339 $ / MWh

Since generator limits are satisfied now, the solution may be acceptable if they satisfy
the optimality conditions. Let us verify if incremental cost of unit-3 is less than λ, since it
is operating at its upper limit.
dC3
IC3 = = 8.54 $ / MWh < λ which satisfies the condition.
dP3 P3 =300 MW
Hence the optimal dispatch is:
P1* = 267.4 MW ; P2* = 407.8MW ; P3* = 300 MW λ = 9.339 $ / MWh
Q.2 As per given conditions,
For Load = 550 MW we can write the ELD conditions as follows:
0.008P1 + 6 − λ = 0
2αP2 + β − λ = 0
P1 + P2 = 550
And given that λ= 8 $/MWh. We can then simply obtain from the above that,
P1* = 250 MW ; P2* = 300 MW
And therefore we have,
600α + β = 8 (1)

For Load = 1300 MW we can write the ELD conditions as follows:

0.008P1 + 6 − λ = 0
2αP2 + β − λ = 0
P1 + P2 = 1300
Also given that λ= 10 $/MWh. Then we can simply obtain from the above that,
P1* = 500MW ; P2* = 800MW
And therefore we have,
1600α + β = 10 (2)
Solving (1) and (2), we obtain α = 0.002; β = 6.8

Q.3 The cost functions are linear, their incremental costs are constants.
dC1 dC2 dC3
= 5.3 $ / MWh; = 5.5 $ / MWh; = 5.8 $ / MWh
dP1 dP2 dP3
Therefore the KKT conditions cannot be applied in this problem, and generators will be
dispatched in merit order of their incremental costs, while satisfying the limits. The
merit order is:

Generator-1  Generator-2  Generator-3

The dispatch will be as follows:

P1* = 450 MW ; P2* = 250 MW ; P3* = 100 MW
λ = 5.5 $ / MWh
The system marginal cost will be determined by the incremental cost of generator-2
which will take up the next 1 MW of load when it appears on the system.

Q.4 The coordination equations are given as:

( pf1 )(0.008PG1 + 8) = λ
( pf 2 )(0.012 PG 2 + 9) = λ
Since the incremental loss factor of Gen-2 = 0.2, therefore, pf2 = 1.25.
Also, given that, PG1 = PG2 = 500 MW.
Therefore, from the above equations, we have,
λ = 18.75
pf1 = 1.5625
Q.5 Pool Operation versus Independent Operation
a. Independent Operation
Utility Demand, MW Dispatch, MW Cost, $/hr Incremental Cost, $/MWh
A 425 425 469,841.5 2,187.55
B 320 320 1,923,596.6 11,944.03
C 400 400 878,385 4,337.54
Total Pool Cost 3,271,823.1

b. Pool Operation: The problem is of minimizing the total pool cost, given as,
J = C A (PA ) + C B (PB ) + CC (PC )
The pool demand is PD A + PDB + PDC = 1145MW .
This is same as an ELD problem, and will be solved first without considering the
generation limits. The Lagrangian F, can be written as,

F = C A ( PA ) + C B ( PB ) + CC (PC ) + λ (1145 − PA − PB − PC )

The KKT conditions are as follows:

∂F
= 5.092 PA + 23.45 − λ = 0
∂PA
∂F
= 37.08PB + 78.43 − λ = 0
∂PB
∂F
= 10.708PC + 54.34 − λ = 0
∂PC
∂F
= 1145 − PA − PB − PC = 0
∂λ

Solving the above we obtain the optimal solution as follows:

PA* = 712.5MW ; PB* = 96.36 MW ; PC* = 335.93MW λ = 3651.5 $ / MWh

Note that Utility-A and Utility-B generation violates the upper and lower generation
limits, respectively. Thus, we fix these at the respective limits, and we have,
PA* = 700 MW ; PB* = 100 MW ; PC* = 345MW λ = 3694.26 $ / MWh
The above value of λ = 3694.26 $/MWh is the incremental cost of Utility-C which
operates within the limits.

Let us verify if the incremental costs of the utilities operating at limits, satisfy the
conditions of optimality.
dC A
IC A = = 3587.85 $ / MWh < λ which satisfies the condition.
dPA PA =700 MW

dC B
IC B = = 3786.43 $ / MWh > λ which satisfies the condition.
dPB PB =100 MW
Therefore, the above solution for pool dispatch is optimal.

Summary of pool operation:

 Utility Demand, MW Dispatch, MW Cost, $/hr Incremental Cost, $/MWh
A 425 700 1,263,959 3587.85
B 320 100 193,246 3786.43
C 400 345 656,016 3694.26 (which is also λ)
Total Pool Cost 2,113,221
Pool savings = 1,158,602 $/hr
Optimal transactions:
Utility-A to Utility C = 55 MW
Utility-A to Utility B = 220 MW

Q.6: The increment cost of generation for each units are as follows:
dC1
= 0.00506 P1 + 3.19 $ / MWh
dP1
dC 2
= 0.00650 P2 + 5.11 $ / MWh
dP2
To minimize the cost of generation, the Lagrangian is given by:
F = (0.00253P12 + 3.19 P1 + 850) + (0.00325P22 + 5.11P2 + 1687) + λ (1200 − P1 − P2 )
The KKT conditions can be formulated as:
∂F
= 0 ⇒ 0.00506 P1 + 3.19 − λ = 0
∂P1
∂F
= 0 ⇒ 0.00650 P2 + 5.11 − λ = 0
∂P2
∂F
= 0 ⇒ 1200 − P1 − P2 = 0
∂λ
Solving above equations, we have the optimal solution as:
P1 = 840.83 MW P2 = 359.17 MW, λ =7.45 $/MWh

Including the transmissions losses, the demand-supply balance is now given as,
P1 + P2 = 1200 + PLoss ( P1 , P2 )
Where, PLoss ( P1, P2 ) = 0.00011P12 + 0.00006 P22 MW .

To minimize the cost of generation, the Lagrangian is given by:

F = (0.00253P12 + 3.19 P1 + 850) + (0.00325P22 + 5.11P2 + 1687)
+ λ (1200 + PLoss ( P1, P2 ) − P1 − P2 )
The KKT conditions are derived as follows:
∂F
= 0 ⇒ 0.00506 P1 + 3.19 + λ (0.00022 P1 − 1) = 0
∂P1
∂F
= 0 ⇒ 0.0065P2 + 5.11 + λ (0.00012 P2 − 1) = 0
∂P2
∂F
= 0 ⇒ 1200 + 0.00011P12 + 0.00006 P22 − P1 − P2 = 0
∂λ

The incremental loss factors for the two generating units are as follows:
 ∂PLoss ( P1 , P2 ) ∂PLoss ( P1 , P2 )
= 0.00022 P1 = 0.00012 P2
∂P1 ∂P2

The coordination equations can be written by re-arranging the above, as follows:

 
(0.00506 P1 + 3.19) = λ
1

 1 − 0.00022 P1 
 
(0.0065P2 + 5.11) = λ
1

 1 − 0.00012 P2 
1200 + 0.00011P12 + 0.00006 P22 − P1 − P2 = 0

( pf1 )(0.00506 P1 + 3.19) = λ

Or, ( pf 2 )(0.0065P2 + 5.11) = λ
1200 + 0.00011P12 + 0.00006 P22 − P1 − P2 = 0

In order to solve the above set of equations we use the iteration method, staring with
the ELD solution without losses, as the initial guess.

Iteration P1 P2 PLoss Pf1 Pf2 PD + λ P1 P2

MW MW MW PLoss $/MWh MW MW
MW
1 840.83 359.17 85.51 1.227 1.045 1285.51 8.7648 781.29 504.22
2 781.29 504.22 82.4 1.2076 1.0644 1282.4 8.7573 802.786 479.61
3 802.786 479.61 84.64 1.2145 1.0611 1284.64 8.7783 798.012 486.63
4 798.012 486.63 84.26 1.213 1.062 1284.26 8.775 799.27 484.99
5 799.27 484.99 84.39 1.2134 1.0618 1284.39 8.776 798.97 485.41
6 798.97 485.41 84.36

The solution is seen to have converged well, and the optimal solution is:
P1 = 798.97 MW, P2 = 485.41 MW, λ = 8.776 $/MWh

Q.7. The increment loss factors of the two generators are as follows:
∂PLoss
= 0.00022 P1 + 0.0045 P2
∂P1
∂PLoss
= 0.00012 P2 + 0.0045 P2
∂P2
From the loss coordination equations, we have,
dC1  ∂P 
+ λ  Loss − 1 = 0
dP1  ∂P1 
0.00643P1 + 3.37 + 30(0.00022 P1 + 0.0045 P2 − 1) = 0
0.01303P1 + 0.135 P2 = 26.63
 dC2  ∂P 
+ λ  Loss − 1 = 0
dP2  ∂P2 
0.00364 P2 + 7.19 + 30(0.00012 P1 + 0.0045P2 − 1) = 0
0.135P1 + 0.00724 P2 = 22.81

Solving for P1 and P2 we obtain, P1 = 159.21 MW, P2 = 181.89 MW.

Q.8: We know that when the generating units operate on economic dispatch, the
incremental cost of all generators should be the same, and be equal to the Lagrange
multiplier λ. We can also formulate a relation for the increment in total cost ∆Ct in terms
of the generator incremental cost functions as follows:
dC1 dC dC
∆Ct = ⋅ ∆P1 + 2 ⋅ ∆P2 + 3 ⋅ ∆P3
dP1 dP2 dP3

dC1 dC2 dC3

= = =λ
But since, dP1 dP2 dP3
⇒ ∆Ct = λ ⋅ (∆P1 + ∆P2 + ∆P3 )
Now, from the first reading, we see that:
0 = λ (1 + 1 − 2 ) = λ × 0 = 0
(i)
⇒0=0
This reading does not provide any clue about λ. From the second reading, we have:
30 = λ (1 + 1 + 1) = 3λ
(ii)
⇒ λ = 10$ / MWh
This value of λ satisfies (ii) and also (i), although any value of λ would satisfy (i), for that
matter. From the third reading we have:
− 20 = λ × (− 3 + 1 + 1)
(iii)
⇒ λ = 20$ / MWh
This value of λ will satisfy (i), as we already know, but (ii) is not satisfied. Therefore, we
can conclude that the system is NOT operating on economic dispatch. Jill was obviously
right!
Let us write (ii) and (iii) again, but now considering the fact that the generators operate
at their own individual incremental costs (which are not equal), since they are not on
economic dispatch. Thus we have:
dC1 dC dC
30 = ⋅ (1) + 2 ⋅ (1) + 3 ⋅ (1)
dP1 dP2 dP3
(iv)
dC dC dC
− 20 = 1 ⋅ (−3) + 2 ⋅ (1) + 3 ⋅ (1)
dP1 dP2 dP3
The set of simultaneous equations in (iv) can be solved to obtain dC1 dP = $12.5 / MWh .
1
However, no conclusions can be drawn about the incremental costs of the other two
generators except for the relation:
dC2 dC3
+ = 17.5
dP2 dP3
Q.9: For economic emission load dispatch we solve the following three equations:
0.055 P1 + 3.5 = λ
0.13P2 + 6.2 = λ
P1 + P2 = 900
Solving the above equations we obtain P1 = 647.05 MW, P2 = 252.95 MW which is the
economic emission dispatch solution. Since both generators operate within their limits,
their incremental emissions are the same, which is 39.088 kg/MWh, and this is also the
system marginal emission, λ. The significance of λ is that, it denotes the increase in
emission (in kg) for a 1 MW increase in demand.
Compare the total emissions in case of joint dispatch, with when utilities operate
independently. The results are tabulated below:

Economic Emission Independent

Load Dispatch Operation
P1, MW 647.05 MW 575 MW
P2, MW 252.95 325 MW
Emission from Utility-1 13,803.2 kg 11,129.7 kg
Emission from Utility-2 5,775.2 kg 8928.6 kg
Total emission 19,578.4 kg 20,058.3 Kg
Emission Reduction from System 479.9 kg/h

Note that, by shifting the generation from one unit to another, the total
emissions can be reduced by 479.9 kg for this load condition at an hour. Of course,
associated with such a shift is an increase in operating costs. The Emission Characteristic
for a generator is the pollution characteristic, usually available from utilities, and refer
to either their CO2 or SO2 emission characteristic, or a composite function of the two. It
depends on the type and quality of fuel (particularly for coal-fired units) and the
generator’s efficiency (heat rate).

Q.10
(a) Total energy demand: 200 MW x 168h = 33,600 MWh
Available hydro energy: EH = 28,000 MWh
Thermal energy required from steam generator: ES = EL – EH = 33,600 – 28,000 MWh
= 5,600 MWh
To minimize the operating cost, the thermal generator should operate at:
100
PS* = = 50MW
0.04
for a duration of
5,600
TS* = = 112h
50
(b) Total volume of water discharged over the week:
{ }
QTOTAL = 300 + 25 ⋅ (150 ) + 0.0075 ⋅ (150 )2 × 112
{ }
+ 300 + 25 ⋅ (200 ) + 0.0075 ⋅ (200 )2 ⋅ (168 − 112 ) acre − ft
= 786,100 acre − ft
(c) If the total water for drawdown is reduced by 5% of 786,100 acre-ft, i.e., by 39,305 acre-
ft, the thermal generator should operate for ∆TS additional hours. Accordingly, the
reduction in water drawdown is accounted for, as follows:
 39,305 acre-ft = [Savings in Water Drawdown for PH = 200 MW for ∆TS hours]
– [Increase in Water Drawdown for PH = 150 MW for ∆TS hours]

Thus, we have,
{ } {
39,305 acre − ft = 300 + 25(200) + 0.0075(200)2 × ∆TS − 300 + 25(150) + 0.0075(150)2 × ∆TS }
39,305 = 5600∆TS − 4218.75∆TS
∆TS = 28.46 hours
The thermal unit should operate for additional 28.46 hours; i.e., for a total 140.46 hours.

Q.11 The optimization problem is as follows,

2
Min Cost = J = ∑ n j C ( PSj )
j =1
Subject to the constraints:
PHj + PSj = PDj ∀j = 1,2
2
( )
∑ n j q j PHj = QTOT
j =1
The Lagrangian is written as,
F = 4C (PS1 ) + 4C (PS 2 ) + λ1 (725 − PH 1 − PS1 ) + λ2 (615 − PH 2 − PS 2 )
+ γ ((5.64 PH 1 + 180) + (5.64 PH 2 + 180) − 25,000)
The coordination equations can be formulated from the above as follows:
∂F dC (PS1 )
=4 − λ1 = 0.0216 PS1 + 36 − λ1 = 0
∂PS1 dPS1
∂F dC (PS 2 )
=4 − λ2 = 0.0216 PS 2 + 36 − λ2 = 0
∂PS 2 dPS 2
∂F dq(PH 1 )
= −λ1 + 4γ = −λ1 + 22.56γ = 0
∂PH 1 dPH 1
∂F dq(PH 2 )
= −λ2 + 4γ = −λ2 + 22.56γ = 0
∂PH 2 dPH 2
∂F
= PH 1 + PS1 = 725
∂λ1
∂F
= PH 2 + PS 2 = 615
∂λ2
∂F
= 4q(PH 1 ) + 4q(PH 2 ) − QTOTAL = 0
∂γ

The above set of equations can be solved iteratively as follows:

Iteration Interval λj PSj PHj ∈= 4q(PH 1 ) + 4q(PH 2 ) − QTOTAL
$/MWh MW MW = 22.56 PH 1 + 22.56 PH 2 − 23,560
(1) j =1 45.12 422.22 302.78 [22.56(302.78) + 22.56(192.78)] − 23,560 γ is too high, there is under-
γ = 2 $/acre-ft j =2 45.12 422.22 192.78 = −12,380 acre − ft utilization of hydro. In the
next iteration, γ is reduced.
(2) j =1 40.608 213.33 511.67 [22.56(511.67 ) + 22.56(401.67 )] − 23,560 γ is still high, there is under-
γ = 1.8 $/acre-ft j =2 40.608 213.33 401.67 = −2,955.2acre − ft utilization of hydro. In the
next iteration, γ is further
 reduced.
(3) j =1 39.48 161.11 563.89 [22.56(563.89) + 22.56(453.89)] − 23,560 γ is slightly reduced in the
γ = 1.75 $/acre-ft j =2 39.48 161.11 453.89 = −598.93acre − ft next iteration to bring the
margin of error within 300
acre-ft.
(4) j =1 39.2544 150.67 574.33 [22.56(574.33) + 22.56(464.33)] − 23,560
γ = 1.74 $/acre-ft j =2 39.2544 150.67 464.33 = −127.8acre − ft

After 4 iterations, the solution has converged well. The optimal solution is:
PS1 = PS2 = 150.67 MW
PH1 = 574.33 MW, PH2 = 464.33 MW
γ = 1.74 $/ acre-ft; λ1 = λ2 = 39.2544 $/MWh