ENER4040 Lecture 1: Course Overview

and Basic Power System Component

Minghua Chen
http://www.ie.cuhk.edu.hk/~mhchen

Department of Information Engineering The Chinese University of

Hong Kong
 EEEN2030

Lecture 6

Locational Marginal Pricing (II)

Xu Yunjian
[email protected]

Department of Mechanical and Energy and Environmental

Automation Engineering Engineering (EEEN) Programme
 Today’s Agenda
• Review on DC-OPF
• Counter-intuitive example of LMP
 AC Power
Instantaneous Power :
p(t )  v (t ) i (t ),
v (t ) = Vmax cos(t  V ),
i (t) = I max cos(t   I ),
1
cos  cos   [cos(   )  cos(   )]
2

1
p(t )  Vmax I max [cos(V   I )  cos(2t  V   I )].
2

4
 AC Power, cont’d
Instantaneous Power is sum of average and varying terms :
1
p (t )  Vmax I max [cos(V   I )  cos(2t  V   I )],
2
T
1
P 
T0 p (t )dt , [average power]

1
 Vmax I max cos(V   I ),
2
 V I cos(V   I ),

Power Factor Angle =  =V   I .

Power Factor = cos( ) .
5
 Phasor Representation
The RMS, cosine-referenced voltage phasor is:
jV
V  Ve  V V ,
v (t )  2 V cos(t  V ),
V  V cosV  j V sin V ,
I  I cos I  j I sin  I .

• For detailed justifications, see Lecture 2.

• Also note that the convention in power engineering is that the
magnitude of the phasor is the RMS voltage of the waveform.

6
 Complex Power
S  V I*
  V cosV  j V sin V  I cos I  j I sin  I 
 V I cos(V   I )  j sin(V   I )  ,
: P  jQ ,
P = Real Power (W, kW, MW),
Q = Reactive Power (VAr, kVAr, MVAr),
S = Complex power (VA, kVA, MVA),

I * : the complex conjugate of the complex number I

7
 Complex Power, cont’d

Power Triangle
|S|
Q


P
1  Q 
P
2
S  P Q 2
  tan pf 
 
P P2  Q2
P P
S  P  jQ S  
cos( ) pf
8
 Impedance and Admittance
Z = Impedance  R  jX  Z  , (in Ohms)
R = Resistance; X = Reactance,

Z = R X ,
2 2  X
 =arctan   .
R
V  IZ .

Y = Admittance  1 / Z , (in siemens).


Y = G  jB   2
1  ( R  jX ),
2
R  X 
G = conductance; B = susceptance,
I  VY .
9
Advantages of Phasor Analysis
Device Time Analysis Phasor
Resistor v (t )  Ri (t ) V  RI
di (t )
Inductor v (t )  L V  j LI
dt
1 t 1
Capacitor v (t )   i (t )dt  v (0) V I
C 0 jC
Z = Impedance  R  jX  Z  ,
(Note: Z is a
R = Resistance, complex number but
X = Reactance, not a phasor).

Z = R X ,
2 2  X
 =arctan   .
R 10
 AC Power Flow in a Network
• A power network is a connected graph G=(N,E).
• For each edge (i, j)  E, let yij be its admittance (a
complex number).
• If (i, j)  E, then (j, i)  E, and yij  y ji .
• For each bus j N, let V j be its complex voltage.
• Let s j be the complex power generation at bus j.
• Let d j be the complex power demand at bus j.
• Let bus 1 be the slack bus (reference bus); its
voltage is fixed with a phase angle 0.
11
 Power Flow in a Network
• The current on edge (i, j) from i to j
I ij  yij (Vi  V j ). (Ohm's law)
• The total current out of node (bus) i:
 I ij . (Kirchhoff 's Current Law)
j: (i , j )E
• Conservation of power at node i:
si  di  Vi  yij* (Vi*  V j* ), i  N . (1)
j: ( i , j )E

12
 DC Optimal Power Flow (OPF)
Approximations made in the DC OPF:
• Consider only real power; ignore reactive power.
• There is no line loss, i.e., the impedance has only
the imaginary part Y=Bj.
• Recall that the loss of power on line (i,j) is
minimized at i   j  0.
• For each line (i,j), the phase difference is small,
and therefore sin(i   j )  i   j .
• Same voltage at all nodes, which is normalized to
be 1, i.e., Vi  1.
13
DC Approximation in A Two-bus System

• The real power injected to bus 1 is given by

Re( s1 )  gv1 ( v1  v2 cos  )  v1v2b sin 
• The real power injected to bus 2 is given by
Re( s2 )  gv2 ( v2  v1 cos  )  v1v2b sin 
• Since g=0, sin( )   , and v1  v2  1, the power flow from
node 1 to 2 is b .
• Since bj  1 / xj, b  1 / x. Power from 1 to 2 is  / x .
• In DC models, 1/x is usually referred to as the admittance.14
 DC OPF: Constraints
• AC power balance constraint
si  d i  Vi  yij* (Vi*  V j* ), i  N . (1)
j: ( i , j )E
• In the DC approximation, the power from i to j is simplified,
and
si  di   yij (i   j ), i  N . (5)
j: ( i , j )E
• The voltage limit constraint v i  Vi 2  v i is simply ignored.
• The constraints on power generation remain:
si  si  si , i  N .
• The power flow from node i to j cannot exceed the capacity
limit.
AC : Vi (Vi  V j ) yij  cij ,
* * *
(i, j )  E.
DC : yij (i   j )  cij , (i, j )  E. 15
 DC OPF: Formulation
• Let Ci(si) denote the convex generation cost at bus i.
• Here, si represents only real power.
min iN Ci ( si )
s.t. si  d i   yij (i   j ), i  N . (5)
j: ( i , j )E

si  si  s i , i  N . (6)
yij (i   j )  cij , (i, j )  E. (7)

• Decision variables are {si ,i }iN .

• It is a convex optimization problem, with a convex
objective function and linear constraints. 16
 Today’s Agenda
• Review on DC-OPF
• Counter-intuitive example of LMP

• Extra office hour today 2-4:30pm

 Locational Marginal Price
• In most deregulated markets, DC-OPF is used to compute
locational marginal prices.
min iN Ci ( si )
s.t. si  d i   yij (i   j ), i  N . (5)
j: ( i , j )E

si  si  s i , i  N . (6)
yij (i   j )  cij , (i, j )  E. (7)
• For simplicity, demand is assumed to be inelastic.
• The LMP at bus i reflects the incremental cost
associated with marginal increase in consumption at
bus i.
• Supply and demand at bus i face the same LMP.
18
 LMP and Lagrange Multiplier
• The LMP at bus i reflects the incremental cost associated
with marginal increase in consumption (at the same bus).
• Let F (i )  fi ( s* , * ) denote the optimal value (of
DC-OPF) with demand i at bus i.
• The locational marginal price at bus i is
F (i )
i   d
i i
• The envelope theorem shows that the above derivative
equals  i* , which is the negative of the Lagrange
multiplier associated with the equality constraint:
si  di   yij (i   j )
j: ( i , j )E
 Summary on LMP
• We formulate DC-OPF problem that approximates the
AC-OPF problem.
• The SO collects the bids submitted by generators, and
solves a DC-OPF to determine the dispatch and prices.
• We claim the locational marginal prices equal the
negative of Lagrange multipliers .
• We show locational marginal price equals a generator’s
marginal cost if its supply is not on the boundary.
• A transmission line’s payment is the product of the
power flow and the nodal price gap at its two ends.
• The total revenue of all lines equals the difference
between the total consumer payment and supplier total
revenue.
• Is it possible that a line receives a negative payment?
 Counter-intuitive Fact 1
• Usually, power flows from a node with lower
nodal price to a node with higher nodal price.
– Why Alaska ship oranges to California?
• As a result, transmission owners receive positive
revenue.
• Is it possible that power flows from an expensive
bus to a cheap bus?
– YES!
– This is counter-intuitive, but possible due to
Kirchhoff’s laws.
 Counter-intuitive Fact 1

• 3-bus system: the three lines have the same impedance.

• Consider the DC approximation of AC power flow:
1 1
pij  (i   j )  P1  P2  ( A  C )  P3
zij z
 Counter-intuitive Fact 1

• The marginal cost of generator at bus A is $50/MWh.

• The demand at bus A, B, C is 0, 50, and 100 MW.
• Line capacity limit: P1  25, P2  50, P3  100.
DC-OPF

min 50 s A  70 sB  100 sC
s.t. s A  P1  P3 ,
50  sB  P1  P2 ,
100  sC  P2  P3 ,
P1  P2  P3 ,
P1  25, P2  50, P3  100.
 DC-OPF and LMP

min 50 s A  70 sB  100 sC
Solution : s.t. s A  P1  P3 ,
s A  87.5, sB  62.5, sC  0. 50  sB  P1  P2 ,
P1  25, P2  37.5, P3  62.5. 100  sC  P2  P3 ,
LMPA  50, LMPB  70, P1  P2  P3 ,
LMPC  60. P1  25, P2  50, P3  100.
 LMPB is Higher than LMPC

Solution : • Say, the demand at bus C

s A  87.5, sB  62.5, sC  0. increases by 1MW.
• Generators A and B increase
P1  25, P2  37.5, P3  62.5.
output by 0.5MW.
LMPA  50, LMPB  70, • The incremental cost is $60.
LMPC  60. P1  25, P2  38, P3  63.
 Why LMPB is Higher than LMPC ?

• Because line (A,B) is congested!

• Incremental demand at bus B must be met by local generation.
• The incremental demand at bus C can be met by a mix of
generation at A and B.
• The total cost will decrease if line (B, C) is disconnected!
• This is why its owner gets negative payment.
 Counter-intuitive Fact 2
• An increase in the capacity of an existing line always
reduce the LMP?
• No.
• We know that adding a new line may increase the
total cost (L6).
• However, by increasing the capacity of an existing
line, we are guaranteed a solution (of DC-OPF) that is
at least as good as the original optimal solution.
• This is because with larger line capacity the set of
feasible solutions (to DC-OPF) is enlarged.
• This does not mean that LMP must decrease!
 The Original Network

P1  25, P2  25, P3  25.

s A  40.

Solution :
1 s A  25, sB  75, sC  50.
P1  P2  ( A  C )  P3
z P1  0, P2  25, P3  25
LMPA  50, LMPB  70,
LMPC  100.
Exercise: Line Capacity Increases…

P1  25, P2  40, P3  50.

s A  40.

Formulate the DC-OPF and solve it.

Calculate the nodal prices at each bus.
 Line Capacity Increases…

P1  25, P2  40, P3  50.

s A  40.
min 50 s A  70 sB  100 sC
Solution : s.t. s A  P1  P3  40,
s A  40, sB  90, sC  20. 50  sB  P1  P2 ,
P1  0, P2  40, P3  40 100  sC  P2  P3 ,
LMPA  85, LMPB  70, P1  P2  P3 ,
LMPC  100. P1  25, P2  40, P3  50.
 Why LMPA increases?

• With P1  25, P2  25, P3  25, it is impossible to send

more than 25MW of generation from A.
• Incremental demand at A can be met locally, i.e., LMPA=50.
• With P1  25, P2  40, P3  50, generator A is able to
send 40MW out, which is the generator’s capacity.
• Incremental demand at A has to be met by generator B and C.
 Why LMPA =85 ?

P1  25, P2  40, P3  50.

s A  40.
P1  0, P2  40, P3  40

• Say, the demand at bus A increases by 1MW.

P1  0.5, P2  40, P3  39.5.
• Generators B and C increase output by 0.5MW.
• The incremental cost is (100+70)*0.5=85.