ECE666: Power Systems Operation

Module-1: Power System Economic Operation

Part-1: Economic Load Dispatch, Multi-Area Dispatch and Power Pools

Prof. Kankar Bhattacharya

Department of Electrical & Computer Engineering
University of Waterloo, Waterloo, N2L 3G1, Canada
kankar@[Link]

1 Module-1, Part-1 ECE666: Winter 2024

 Coverage

• Power system operational timeframes

• Steam generator characteristics
• The ELD problem
• KKT conditions of optimum, concept of marginal cost
• Solution of the ELD problem
• ELD with transmission losses
• Loss formula, ELD model formulation, Concept of loss coefficients and
penalty factor, coordination equations and their solution
• Power Pools and Multi-Area Interchanges

2 Module-1, Part-1 ECE666: Winter 2024

 3

The Vertically Integrated Electricity Utility

Generation T&D
Customer

1 Company Utility-1 Utility-2

Generator Generator

Energy Flow Wholesaler /

Transmitter
Wholesaler /
Transmitter

Money Flow
Distributor Distributor
Information
Flow Customer Customer

3 Module-1, Part-1 ECE666: Winter 2024

 What is Electricity Sector Deregulation?

Generation, Transmission, Distribution- independent activities

Energy supply completely separated from transmission

Competition amongst generators for customers

Customer driven market

Transmission- still a monopoly

Basic transmission service provisions to all
eligible market participants

Ancillary services as separate products

4 Module-1, Part-1 ECE666: Winter 2024

 5

The Deregulated Electricity Disco PROSUMER

Sector

Customer
Transco Disco
Independent Power
Producer (IPP)

Disco Customer
Genco

Genco Independent System

Operator

Energy Flow
Information Flow Market
Money Flow Operator
Energy Flow

5 Module-1, Part-1 ECE666: Winter 2024

 Differing Objectives of Electric Utilities

Vertically Integrated Systems Deregulated Systems

 Minimize system cost  Generators
 Meet the demand  Maximize profit
 Ensure reliability  Meet demand contracts
 Get more customers
 Transmission
 Meet transaction contracts
 Maximize (regulated) profit
 System Operator
 Network security and
reliability

6 Module-1, Part-1 ECE666: Winter 2024

 Power System Operation and Planning
Time-frame Task Activity

5 – 10 years ahead Long-term planning Determine new generation

Power System

capacity installations and

transmission / distribution lines
Planning

commissioning.
1 – 2 years ahead Medium-term Fuel-supply scheduling,
production scheduling maintenance scheduling,
production scheduling.
1 – 7 days ahead Short-term generation Hydro-thermal coordination,
Power System Operation

scheduling reservoir drawdown schedules,

unit commitment decisions.

5 – 30 minutes Generation dispatch, Economic load dispatch, OPF and

ahead real-time operations power flow studies, capacitor
switching, load curtailment, load
following by generators.
1 – 300 seconds Dynamic operations Frequency regulation and AGC,
fault clearing, transients, low
frequency oscillations.

7 Module-1, Part-1 ECE666: Winter 2024

 Steam Generator Characteristics: I/O

• The Input / Output (I/O) characteristic of a steam generator

is an important parameter in understanding economic
operations
• Power output of a generator (MW output) depends on the
fuel input (in the boiler) and hence the heat input
• “Gross heat input” at boiler end and “net power output” are
used for I/O characteristics
• Net Power Output = Total power generated – Power consumed
by plant auxiliaries

8 Module-1, Part-1 ECE666: Winter 2024

 I/O Characteristics… contd.

• Actual operating data from a steam generator is used to

develop its I/O characteristic
• Hi(Pi) is the heat I/O characteristic of generator i which
denotes the heat required (in MBTU) to generate a certain
MW of power, Pi, per hour
• Hi(Pi) is represented in MBTU/h
• Typically, Hi(Pi) is a polynomial function of Pi

H i (Pi ) = f (Pi )

• For example,

H i (Pi ) = f (Pi ) = a / Pi2 + b / Pi + c /

• Parameters a/, b/ and c/ are determined from field tests

9 Module-1, Part-1 ECE666: Winter 2024

 I/O Characteristics… contd.

• The cost I/O characteristic of generator i, Ci(Pi), can

therefore be obtained using Hi(Pi) and the fuel cost, θFC

Ci (Pi )[in $/h ] = θ FC [in $/MBTU ]× H i (Pi )[in MBTU/h ]

• Typically, the cost I/O characteristic is thus,

( )
Ci (Pi ) = θ FC H i (Pi ) = θ FC a / Pi2 + b / Pi + c / = aPi2 + bPi + c

10 Module-1, Part-1 ECE666: Winter 2024

 I/O Characteristics… contd.

• A plot of generator input (heat or cost) versus net output

(power) is shown
• Typically the I/O characteristic representation is of second order
polynomial; which is an “ideal” form as a smooth convex curve
Ci (Pi ) = ai Pi 2 + bi Pi + ci
Cost ($/h) or Heat Input (MBTU/h)

PMin PMax MW

11 Module-1, Part-1 ECE666: Winter 2024

Minimum Loadability of Steam Generators

• Observe that the I/O characteristic is within the PMin and PMax
limits of the generator
• The PMin limit defines a Minimum Loadability of the steam
generator- the minimum load at which the unit can operate
• This is determined by the fuel combustion stability and other
steam generator design constraints
• Most super-critical units cannot operate below 30% of their
design capacity
• Super-critical units: advancement of conventional coal-fired steam
generation, involving use of supercritical steam pressure and steam
temperatures > 1100 F
• A minimum flow of 30% is required to adequately cool the tubes in the furnace of the
steam generator

12 Module-1, Part-1 ECE666: Winter 2024

 Incremental Cost Characteristic

• The incremental cost characteristic is obtained as the

derivative of the I/O Characteristic. For the given second-
order polynomial, we have,
dCi (Pi )
= 2ai Pi + bi
dPi

• The incremental cost characteristic denotes the additional

cost required to produce an extra MWh of energy
Incremental Heat Input (MBTU/
Incremental Cost ($/MWh) or

MWh)

PMin PMax MW
13 Module-1, Part-1 ECE666: Winter 2024
 Net Heat Rate Characteristic

• Heat Rate (HR): defined as the amount of heat required to

generate 1 unit (kWh) of electricity
• Denoted by BTU/kWh
• In effect, this is the reciprocal of the generator’s efficiency

• Example:
• Generator HR = 9000 BTU/kWh
• We know, 1 kWh = 3412 BTU thermal equivalent
• Thus Generator HR = 9000 BTU/3412 BTU
• Implying that 3412 BTU of equivalent heat is produced when the input is
9000 BTU. Thus the unit’s overall conversion efficiency is
• η = (3412/9000)x 100 = 37.91%

14 Module-1, Part-1 ECE666: Winter 2024

 Net Heat Rate Characteristic

• The heat rate characteristic is a plot of the unit HR versus

the power output. It is a function of the generator’s
design parameters.
• The lowest HR (i.e. highest efficiency) operation is
attained at P=PRated Unit heat rate, H/P (Btu/kWh)

PMin PRate PMax Output, P (MW)

d

15 Module-1, Part-1 ECE666: Winter 2024

 The Economic Load Dispatch Problem

Module-1,
16 ECE666: Winter 2024
Part-1
 Economic Load Dispatch (ELD) Problem

• The ELD problem primarily involves allocating the total load

between all available and synchronized generating units
• The main objective is that the total cost of operation, i.e.
the total generation cost, be kept at a minimum
• An ELD is executed in the dispatch stage (every 5 minutes)
• Hence the solution algorithm should be efficient
• However, the ELD should also represent the system in as much
detail as possible

17 Module-1, Part-1 ECE666: Winter 2024

 The ELD Problem… contd.

• Consider a system of N generating units (i ∈ 1, 2, …, N)

• Each unit feeds power, Pi, to the transmission system,
serving a total load PD
P1
G1

P2
G2 Transmission
system

PD

PN
GN

18 Module-1, Part-1 ECE666: Winter 2024

 The ELD Problem- Assumptions

• Vertically integrated environment

• The system operator has access to all generators’ cost
functions and is responsible for their dispatch
• System load is aggregated for the whole system at a single
node. Spatial distribution of loads is ignored.
• Transmission losses are usually neglected
• Or a simplistic representation is used

• Reactive power demand is neglected

• Consequently all bus voltages and reactive power relations are
ignored
• Generating units that are committed for the given time
interval are only considered for ELD
• A decision making has already taken place as to what units will
be on-line at that time interval

19 Module-1, Part-1 ECE666: Winter 2024

 ELD Mathematical Model

• Objective Function: Minimize the total cost of generation

from all generating units in the system
• Using the quadratic cost function, we have
N N
J = ∑ Ci (Pi ) = ∑ ai Pi 2 + bi Pi + ci
i =1 i =1

• Demand-Supply Balance: The operator dispatches the

generators to meet the total system demand
N

∑P
i =1
i = PD

• Upper and Lower Limit on Generation: Generation from a

unit is constrained by the limits
Pi Min ≤ Pi ≤ Pi Max ∀i ∈ N

20 Module-1, Part-1 ECE666: Winter 2024

 ELD Mathematical Model… contd.

• The overall model is thus,

N N
• Minimize J = ∑ Ci (Pi ) = ∑ ai Pi 2 + bi Pi + ci
i =1 i =1

• Subject to,
N
ϕ (Pi ) = 0 ⇒ PD − ∑ Pi = 0
i =1

g (Pi ) ≤ 0 ⇒ Pi Min − Pi ≤ 0 ∀i ∈ N

h(Pi ) ≤ 0 ⇒ Pi − Pi Max ≤ 0 ∀i ∈ N

21 Module-1, Part-1 ECE666: Winter 2024

 Structure of a General Optimization Problem

Minimize J
 ϕ i ( X ) = 0 ↔ λi ∀i∈N
Subject to 
 g j ( X ) ≤ 0 ↔ µ j ∀ j∈M

• The above is a constrained minimization problem which can

be converted to an unconstrained minimization problem by
defining a new function, the Lagrange Function or
Lagrangian, as follows:
N M
F ( X , λi , µ i ) = J + ∑ λiϕ ( X ) + ∑ µ i g i ( X )
i =1 j =1
• F(.) is formed by augmenting objective J with equality and inequality
constraints, using multipliers, λi and µj.
• These multipliers, λi and µj, known as Lagrange multipliers, are unknowns,
and are simultaneously determined from the optimization solution.

22 Module-1, Part-1 ECE666: Winter 2024

 The ELD Lagrangian

• The ELD Lagrangian can accordingly be written as:

N N
F ( Pi , λ , µi , γ i ) = J ( Pi ) + λϕ (Pi ) + ∑ µi g i (Pi ) + ∑ γ i hi (Pi )
i =1 i =1

N  N 
F (Pi , λ , µ i , γ i ) = ∑ Ci (Pi ) + λ  PD − ∑ Pi 
i =1  i =1 

[ ] [ ] [
+ µ1 P1Min − P1 + µ 2 P2Min − P2 + ... + µ N PNMin − PN ]
+ γ [P − P
1 1 ]+ γ [P − P ]+ ⋅ ⋅ ⋅ ⋅ + γ [P
1
Max
2 2 2
Max
N N − PN
Max
]

23 Module-1, Part-1 ECE666: Winter 2024

 Conditions for Optimum

• The conditions for optimum are based on the same principle

of finding the minima of a function
• Except that, the full derivative of the function is replaced by
partial derivative of Lagrangian
∂F
a. =0 ∀ i = 1,......, N
∂ Pi

∂F
b. = 0 ⇒ ϕ (Pi ) = 0
∂λ

• The above conditions yield N+1 equations in (N+1) variables,

Pi (i = 1, 2, …, N) and λ, and can be solved uniquely

24 Module-1, Part-1 ECE666: Winter 2024

 Conditions for Optimum… contd.

• Complementary Conditions
If g i(Pi ) = 0 (binding constraint), then µ i ≥ 0
c. µ i g i (Pi ) = 0 ∀ i = 1,...., N 
µi ≥ 0  If g i(Pi ) ≠ 0 (not binding), then µ i = 0

d . γ i hi (Pi ) = 0 ∀ i = 1,...., N If hi(Pi ) = 0 (binding constraint), then γ i ≥ 0


γi ≥ 0  If hi(Pi ) ≠ 0 (not binding), then γ i = 0

• The complementary conditions implies:

If Pi = PiMin (binding constraint) then µ i ≥ 0


 If Pi ≠ PiMin (not binding), then µ i = 0

If Pi = PiMax (binding constraint), then γ i ≥ 0


 If Pi ≠ PiMax (not binding), then γ i = 0

25 Module-1, Part-1 ECE666: Winter 2024

 Conditions of Optimum… contd.

• The optimization problem is first solved assuming

• µi = 0 and γi = 0, for all i
• Which means the inequality constraints are not “binding” and all
generation variables (Pi) are within their limits
• The conditions of optimum are the well known, Karush-
Kuhn-Tucker’s (KKT) conditions of optimality
• Applying KKT conditions on the ELD Lagrangian, we obtain

∂F dCi (Pi )
=0⇒ − λ − µi + γ i = 0 ∀i∈N
∂Pi dPi

∂F N
= 0 ⇒ PD − ∑ Pi = 0
∂λ i =1

26 Module-1, Part-1 ECE666: Winter 2024

 KKT Conditions Applied to ELD… contd.

• Applying the complementary conditions, we obtain:

[ ]
µ1 P1Min − P1 = 0, µ1 ≥ 0 [ ]
γ 1 P1 − P1Max = 0, γ 1 ≥ 0

[ ]
µ 2 P2Min − P2 = 0, µ 2 ≥ 0 [ ]
γ 2 P2 − P2Max = 0, γ 2 ≥ 0
....

[ ]
µ N PNMin − PN = 0, µ N ≥ 0 [ ]
γ N PN − PNMax = 0, γ N ≥ 0

27 Module-1, Part-1 ECE666: Winter 2024

 KKT Conditions on ELD... Contd.

• From the first KKT applied on ELD, we obtained

dCi (Pi )
− λ − µi + γ i = 0 ∀ i = 1,..., N
dPi

• Case-1: All Pi within limits, µi = 0, γi = 0

µi = γ i = 0 ∀i = 1,..., N
dCi (Pi )
⇒ =λ
dPi

• When all generators operate within limits, optimal dispatch is

at equal incremental cost

28 Module-1, Part-1 ECE666: Winter 2024

 KKT Conditions on ELD… contd.

• Case-2: Let P1=P1Max, • Case-3: Let P1=P1Min,

all other Pi within limits all other Pi within limits
γ1 ≥ 0
µ1 ≥ 0
γ i = 0 ∀i = 2,..., N
µ i = 0 ∀i = 2,..., N
µ i = 0 ∀i = 1,..., N
γ i = 0 ∀i = 1,..., N
dCi (Pi )
⇒ =λ ∀i = 2,..., N dCi (Pi )
dPi ⇒ =λ ∀i = 2,..., N
dPi

For i = 1,
For i = 1,
dC (P )
⇒ 1 1 = λ − γ1 dC1 (P1 )
dP1 ⇒ = λ + µ1
dP1
dC1 (P1 ) dC1 (P1 )
⇒ ≤λ ⇒ ≥λ
dP1 dP1

29 Module-1, Part-1 ECE666: Winter 2024

 To Summarize
• Conditions for optimum of a constrained ELD problem are:
dCi (Pi )
=λ PiMin < Pi < PiMax
dPi
dCi (Pi )
≤λ Pi = PiMax
dPi
dCi (Pi )
≥λ Pi = PiMin
dPi

• The ELD is first solved without the inequality constraints;

if generator limits are binding, the specific limiting
constraints are included in the Lagrangian, and KKT
conditions re-determined.
• In such cases, it is always necessary to verify that the
conditions for optimality are satisfied
• λ is known as the System Marginal Cost (SMC), the cost to
supply 1 additional MWh of energy in the system
30 Module-1, Part-1 ECE666: Winter 2024
 Economic Load Dispatch Considering
Transmission Losses

Module-1,
31 ECE666: Winter 2024
Part-1
 ELD Considering Transmission Losses

• So far we considered the ELD problem by ignoring

transmission losses
• However, transmission losses can alter dispatch solutions
and need to be taken into account for system economic
operation
• Representation of transmission losses in ELD problems is
usually approximate
• Use of Loss Formula Coefficients

32 Module-1, Part-1 ECE666: Winter 2024

 Kirchmayer’s Loss Formula
• System loss is expressed in terms of active power
generation
• Commonly referred to as the B-Matrix Loss Formula
k k k
PLoss (P1 , P2 ,..., PN ) = ∑∑ Pm Bmn Pn + ∑ Bmo Pm + K lo
m =1 n =1 m =1

• PLoss : Power loss

• Pm, Pn: Power generation from all sources
• Bmn: Loss coefficients in Watt/MW2
• A linear term ΣBm0Pm and a constant Kl0 are also included

33 Module-1, Part-1 ECE666: Winter 2024

 ELD Considering Losses

• The same ELD objective is considered, that of minimizing

the total cost of generation,
N N
J = ∑ Ci (Pi ) = ∑ ai Pi 2 + bi Pi + ci
i =1 i =1

• Demand-supply balance: the generators meet the total

system demand and losses
N

∑P = P i D + PLoss (P1 , P2 ,..., PN )

i =1

• Upper and lower limit on generation: generation from a unit

is constrained by the upper and lower limits
Pi Min ≤ Pi ≤ Pi Max ∀i ∈ N

• For the present discussions, generation limits are not

considered. But these can be introduced in the same manner,
as before

34 Module-1, Part-1 ECE666: Winter 2024

 ELD Mathematical Model With Losses
• The Lagrangian of the ELD problem with transmission
losses, can be formulated as follows
N
 
F = ∑ Ci ( Pi ) + λ  PD + PLoss (P1 , P2 ,..., PN ) − ∑ Pi 
i =1  i 

• The above can be written in elaborate form as:

F = C1 ( P1 ) + C2 ( P2 ) + ... + C N ( PN )
+ λ [PD + PLoss (P1 , P2 ,..., PN ) − P1 − P2 − ... − PN ]

35 Module-1, Part-1 ECE666: Winter 2024

 ELD Mathematical Model With Losses
• The KKT condition can be applied to the Lagrangian, and
we obtain therefore, a set of N- equations, as follows:

∂F dC1 ∂PLoss
= +λ −λ = 0
∂P1 dP1 ∂P1
∂F dC2 ∂P
= + λ Loss − λ = 0
∂P2 dP2 ∂P2
...
∂F dC N ∂P
= + λ Loss − λ = 0
∂PN dPN ∂PN

36 Module-1, Part-1 ECE666: Winter 2024

 ELD Mathematical Model With Losses

• In general form, these can be re-organized as,

∂F dCi  ∂P 
= + λ  Loss − 1 = 0 ∀ i = 1,..., N
∂Pi dPi  ∂Pi 
• We can further re-write as,
 
 
 1  ⋅ dCi = λ
 ∂PLoss  dPi
 1 − ∂P 
 i 

∂PLoss
∂Pi is Incremental Loss Factor at bus i
• Denotes sensitivity of system loss to power injection at
bus i
• Power injection at a bus i, Pi, is given by, Pi = PGi - PDi
• Pi increases when PGi increases or PDi decreases
• Pi decreases when PGi decreases or PDi increases

37 Module-1, Part-1 ECE666: Winter 2024

 Incremental Loss Factor… contd.

• If losses increase, for an increase in power injection at a

bus i
• Incremental Loss > 0
• A generator / customer at this bus contributes directly to
system losses when it injects power by increasing generation /
reducing load
• If losses decrease, for an increase in power injection at a
bus i
• Incremental Loss < 0
• A generator / customer at this bus contributes inversely to
system losses when it injects power by increasing generation /
reducing load

38 Module-1, Part-1 ECE666: Winter 2024

 Loss Penalty Factor
• The Loss Penalty Factor for bus i, pfi, is defined using the
Incremental Loss Factor as follows:
1
pf i =
∂PLoss
1−
∂Pi

• If incremental Loss > 0, Penalty Factor >1

• If injection at a bus contributes directly to losses,
the penalty factor acts as a “penalty”
• If incremental Loss < 0, Penalty Factor <1
• If injection at a bus contributes “inversely” to
losses, the penalty factor acts as a “credit”

39 Module-1, Part-1 ECE666: Winter 2024

 Coordination Equations
• The conditions for optimum can be written in terms of loss
penalty factors, as follows:
dCi
pf i ⋅ =λ ∀i = 1,...,N
dPi

• The above is termed as Coordination Equations, which

needs to be solved to obtain the economic dispatch.
dC1
pf1 ⋅ =λ
dP1
dC 2
pf 2 ⋅ =λ
dP2 Composite ELD Model
.... with Losses
dC N Considered
pf N ⋅ =λ
dPN

∑P = P i D + PLoss (P1 , P2 ,..., PN )

i =1

40 Module-1, Part-1 ECE666: Winter 2024

 Coordination Equations… contd.

dC1/dP1 With penalty factor

dC2/dP2 dC3/dP3

pf > 1 pf > 1 pf < 1

λ’
λ

P’1 P1 MW P’2 P2 P3 P’3

MW MW

No penalty factor

The criterion of equal incremental cost operation can be extended

from ELD without losses, to an ELD with losses, considering that for
a generator with pf > 1, its incremental cost curve is scaled up while
for one with pf < 1, the incremental cost curve is scaled down

41 Module-1, Part-1 ECE666: Winter 2024

 Iterative Solution of ELD with Losses

Choose starting value of Pi, i = 1,…,N

Calculate PLoss and hence penalty factor at bus i

Choose starting λ

Solve for each Pi

Adjust λ
dFi (Pi )
pf i ⋅ =λ for i = 1, ..., N
dPi

NO
∑P −P i D <ε?
YES

Compare Pi to Pi of previous iteration and save Max |Pik-1 - Pik|

NO
Pi k −1 − Pi k < ε o ? YES
STOP

42 Module-1, Part-1 ECE666: Winter 2024

 Power Pools and Multi-Area Interchanges

Module-1,
43 ECE666: Winter 2024
Part-1
 Power Pools

• Pools existed as a reaction to increasing price of electricity

• Reduce utilities’ operating cost by sharing cheaper generation

• Pooling was desirable because they

• Enhanced reliability and reduced costs
• Load change shared by all generators in the interconnection
• Pool ELD can be obtained if all cost information was available

• Structure of the Pool depended on

• Role of Pool Operator-
• Generation dispatch, unit commitment, setting up transactions,
maintenance scheduling and even long-term expansion planning
• How transactions were set up
• Joint dispatch of generation for all areas simultaneously
• Energy Brokerage System- transactions based on bid / offer system
• How cost savings were allocated to pool participants
• Split the savings equally, proportional, etc.

44 Module-1, Part-1 ECE666: Winter 2024

 Economic Exchange of Energy

• Simple form of arrangement involving exchange of energy

between two utilities with significant differences in their
marginal costs
• Utility with higher operating cost purchases power from
utility with lower operating costs
• Arrangement usually on hour to hour basis and conducted by
two system operators
• Benefit from such an arrangement is ensured if savings in
production cost from two cooperating utilities is more than the
cost incurred from payments towards charges for transmission
of energy

45 Module-1, Part-1 ECE666: Winter 2024

 Economic Exchange of Energy… contd.

• Let, utilities U1 and U2 be involved in economic interchange

of energy at a certain hour
• U1 transferring PT MW of power to U2
• C1*: Cost of U1 on Economic Dispatch, meeting its own demand PD1
• C2*: Cost of U2 on Economic Dispatch, meeting its own demand PD2
• C1: Cost of U1 to generate extra PT MW (generating PD1 + PT MW)
• C2: Cost of U2 after receiving PT MW from U1 (generating PD2 – PT MW)
• Increased cost to U1 from the transaction is
∆C1 = C1 − C1*

• Reduction in cost for U2 from the transaction is

∆C2 = C2* − C2
• The net savings (S) from a transaction is thus

S = ∆C2 − ∆C1

46 Module-1, Part-1 ECE666: Winter 2024

 Economic Exchange of Energy… contd.

• S>0 Feasible transaction [λ1 < λ2]

• S<0 Infeasible transaction [λ1 > λ2]
• λ1, λ2: incremental cost of exporting and importing utilities respectively
• U1 exporting PT to U2, a negative savings would occur if λ1 > λ2
• Seller’s advantage- when buyer’s incremental cost high
• Buyers advantage- when seller’s incremental cost low

• Savings allocation- simplest approach is to split the savings

equally
• U2 pays U1 for the PT MW import, an amount equal to U1’s
increased cost from the transaction and half of the savings
resulting therefrom

47 Module-1, Part-1 ECE666: Winter 2024

 Multi-Utility Coordinated ELD

• Independent Mode of Operation of Utilities:

• The set of M utilities obtain their own optimum schedules using
the classical ELD condition:
dC1,1 dC1,2 dC1, N 1
= = ... = = λ1
dP1,1 dP1,2 dP1, N 1
...
dC M ,1 dC M ,2 dC M , NM
= = ... = = λM
dPM ,1 dPM ,2 dPM , NM
• Each utility operate at “equal incremental cost” condition
• The respective marginal costs of generation are, λ1, … , λM
• Each utility meets its own demand, PD1, … PDM:
P1,1 + P1,2 + ... + P1, N 1 = PD1
...
PM ,1 + PM ,2 + ... + PM , NM = PDM

48 Module-1, Part-1 ECE666: Winter 2024

 Coordinated ELD… contd.

• All Generators Pool Dispatched to Minimize Pool Cost

• The condition for optimum can be written as:

dC ji
=λ ∀ j = 1,..., M
dPji
i = (1,..., N1 ); (1,..., N 2 );...(1,..., N M )

N1 + N 2 +...N M M
∑ Pi = ∑ PD j
i =1 j =1

49 Module-1, Part-1 ECE666: Winter 2024

 Multi-area Transmission Constrained Joint Dispatch

• Central dispatcher coordinates generation dispatch

• Minimize total system cost

J = ∑∑ C ( Pi ,m ) ∀ i = 1,..., NGm ; m = 1,2 ,. ...., NU

m i
• k, m is the index for utility
• NU is the no. of utilities participating in the coordinated dispatch
• i is the index for generator
• NG is the no. of generators within one utility
• Transmission constraint (transportation model)- algebraic
sum of power at a node is zero

∑ Pi ,m + ∑ Tk ,m = PDm
i k

Tk , m ≤ TkMax
,m

50 Module-1, Part-1 ECE666: Winter 2024

 Multi-area Joint Dispatch… contd.

• Elaborate representation of transmission system is possible by

using dc power flow model
1
Tk ,m = (δ k − δ m ) ⋅
x k ,m
• Ideally an ac load flow is most accurate representation

• High λm denotes large cost saving for system when PDm

reduces by 1 MW
• i.e. when utility m imports power from another utility

• γk,m (associated with transmission constraint) is high when a

transaction between k and m reduces the system cost

51 Module-1, Part-1 ECE666: Winter 2024

 Other Types of Utility-Interchanges

• Capacity interchange- agreement for peak reserve. No

energy transaction, except emergency
• Diversity interchange- energy transaction between utilities
with diversity in load curves
• Energy banking- transaction between utilities having
different generation types
• Emergency power interchange- to take care of forced
outages in one system
• Inadvertent interchange- due to error in controlling
interchange by automatic generation control

52 Module-1, Part-1 ECE666: Winter 2024

 Module-
53 ECE666: Winter 2024
1, Part-1