Hydrothermal scheduling

problem
 NEED FOR HYDROTHERMAL SCHEDULING

Thermal plants:
• Has very high operating cost, low capital cost.
• Starting of thermal plants is slow and their speed of response is slow.

Hydro plants:
 Operating cost is low,capital cost is high.
 Can be started quickly ,has higher reliability and greater speed of response
, so can take fluctuating loads.

 So thermal plant is preferred as a base load plant and hydro electric plant
is run as a peak load plant.
 Hydrothermal scheduling problem

OBJECTIVE:
• To determine the water releases from each reservoir of the hydro system
at each stage such that operation cost is minimized along the planning
period meeting the forecasted demand.

• Operation cost includes fuel cost for the thermal units ,import costs from
neighboring systems and penalties for load shedding.
 Classification of Hydro plants

 Classification on the basis of type:

 Pumped storage plants
 Conventional plants
 Run-of-river plants
 Storage plants
 Classification on the basis of location:
 Hydro plants on different streams
 Hydro plants on the same stream
 Multi chain hydro plants
 Classification according to quantity of water available:
 Run off river plants without pondage
 Run off river plants with pondage
 Reservoir plants
 Classification of hydrothermal scheduling
Problem
• Long range Problem:
• Considers the yearly cyclic nature of reservoir water inflows and
seasonal demand
• Scheduling period of one year is used
• Considers the dynamics of head variations through the water flow
continuity equation.
• Classified into three groups:
– Multi storage hydroelectric systems
– Cascaded hydroelectric systems
– Multi-chain hydroelectric systems
 Classification of hydrothermal scheduling
Problem
Short term hydro thermal scheduling:
 Scheduling interval for short range -either a day or a week.
 As scheduling interval of short range is small, the head is assumed to
be constant.
 The short term hydrothermal scheduling problem is classified into two
groups:
 Fixed head hydro thermal scheduling:
Head of reservoir can be assumed fixed if the hydro plants have
reservoirs of large capacity
 Variable head hydro thermal scheduling
Head of reservoir is variable if the hydro plants have reservoirs of
small capacity.
 Short Range Fixed head hydrothermal
scheduling
 Output of each hydro unit varies with effective head and the rate of water
discharged through the turbines.

 For large capacity reservoirs , assume the effective head is constant over
the optimization interval.

 Consider an electric power system having N thermal generating plants and

M hydro plants , where M+N is the total number of generating plants.

 To find the active power generation of each plant in the system as a

function of time over a finite time period from 0 to T.
 Short Range Fixed head hydrothermal
scheduling
Thermal model : T N
J1    Fi ( Pi (t )) dt
0 i 1

Where Fi ( Pi (t )) is the fuel cost of the i th thermal unit ,given by

Fi ( Pi (t ))  ai Pi 2 (t )  bi Pi (t )  ci Rs/h
Hydro model:
The input-output characteristic of a hydro generator is expressed by the
variation of water discharge q(t) as a function of power output P(t) and
head h.
According to Glimn-Kirchmayer model, discharge is given by
q j (t )  K (h j ) ( Pj  N (t )) m3 / h
for Ф and  are two independent functions.
 Short Range Fixed head hydrothermal
scheduling
For a large capacity reservoir ,effective head is constant and for constant head
 (hj) becomes constant , therefore
q j (t )  K ' ( Pj  N (t )) (j  1,2,......M)

Each hydro plant is constrained by the amount of water available for the
optimization interval ie., T
 q j (t )dt  V j
0
(j  1,2, ......, M)
Where Vj-predefined volume of water available in cubic meters
qj-hydro performance given by

q j (t )  x j Pj2 N (t )  y j Pj  N (t )  z j m3 / h
 Short Range Fixed head hydrothermal
scheduling
Equality and Inequality constraints:
(i) Load demand equality constraint:
N M

 P (t )  P
i 1
i D (t )  PL (t )

(ii) Limits are imposed as

Pi min  Pi (t )  Pi max (i  1,2,......, N  M)
Transmission Losses:
Using Kron’s approximated loss formula
N M N M
PL (t )   P (t )B P (t )   B P (t )  B
i 1
i ij j
i 1
i0 i 00 MW
 Short Range Fixed head hydrothermal
scheduling –Problem definition
T N
Minimize J1    Fi ( Pi (t ))dt
0 i 1
N M
Subject to  P (t )  P (t )  P (t )
i 1
i D L

 q (t )dt  V
0
j j (j  1,2, ......, M)

Pi min  Pi (t )  Pi max (i  1,2,......, N  M)

Langrange function given by:
N T M
 N M

L( Pi (t ),  (t ), v)    Fi ( Pi (t ))   v j q j (t )   (t )  PD (t )  PL (t )   Pi (t ) dt
0  i 1 j 1  i 1 
M
-  v jV j
j1
where vj-water conversion factors
 Short Range Fixed head hydrothermal
scheduling –Problem definition
The optimality conditions w.r.t decision variables Pi(t) ,λ(t),vj

Fi ( Pi (t ))  P (t ) 
  (t )  L  1  0 (i  1,2 ....., N)
Pi (t ) 
 iP (t ) 
q j (t )  P (t ) 
vj   (t )  L  1  0 (j  1,2,....., M)
Pj  N (t )  Pj  N (t ) 
N M

 P (t )  P
i 1
i D (t )  PL (t )
T

q
0
j (t )dt  V j (j  1,2,...., M)
 Long Range hydrothermal Generation scheduling –
Problem definition
Fuel cost:

T
N 
F    ai ( Pi )  bi Pi  ci 
k 2 k
(1)
k 1  i 1 

Where,
ai , bi , ci - cost coefficients

Pi k - thermal power generation during kth sub-interval

 Long Range hydrothermal Generation scheduling –
Problem definition
Water Storage Equation:
Assuming all the hydro plants are on different water streams , the storage
at the end of kth sub-interval is given by,

Where, X kj 1  X kj  J kj  Qkj  S kj (j  1,2, ......, M) (2)

X kj is water storage for the jth turbine during the kth sub-interval
Q kj is water discharge through the jth turbine during the kth sub-
interval.
J kj is water inflow into the reservoir for the jth turbine during the kth
subinterval
S kj is spillage from the jth turbine during kth sub-interval.
 Long Range hydrothermal Generation scheduling –
Problem definition
Hydro Generation:
The average generation during the kth subinterval is given by,
 
Pjk N  h j 1  0.5g j (2 X kj  J kj  Qkj  S kj ) (Qkj  u j ) (j  1,2, ..., M) (3)

Where,
h j is the basic head of the jth hydro plant.
g j is the water head correction factor to account for variation in head
with storage of the jth hydro plant.
u j is the non-effective water discharge of the jth hydro plant.
 Long Range hydrothermal Generation scheduling –
Problem definition
Power Balance Equation :
The power balance equation for the kth sub-interval is ,
M N

 Pk  Pk  Pk  0
i 1
i D L
(4)

Where,
PDk is the load demand during the kth sub-interval.
PLk is the transmission losses during the kth sub-interval.
M N M N M N
P  B00 
L
k

i 1
Bi 0 Pi 
k
i 1
 i ij j
P k

j 1
B P k
(5)
 Long Range hydrothermal Generation scheduling –
Problem definition
Equality and Inequality Constraints:
(a)Fore-bay limit of reservoir
Xm
j
in
 X k
j  X max
j (j  1,2, ......, M) (6)
(b)Water discharge limit
Qm
j
in
 Q k
j  Q max
j (j  1,2 ....., M) (7)
(c) Output of thermal plants
Pi min  Pi k  Pi max (i  1,2, ....., N) (8)
(d)Total volume of water available at the end of kth sub-interval can be
obtained from the initial storage as follows.
T T T
X T 1
j X 
1
j 
k 1
J 
k
j  Q 
k 1
k
j 
Sk  0
k 1
j
(9)
 Long Range hydrothermal Generation scheduling –
Multiobjective optimization Problem
T
N 
Minimize F    ai ( Pi k ) 2  bi Pi k  ci 
k 1  i 1 
Subject to :
(i) Load demand constraint M  N
P
i 1
i
k
 PDk  PLk  0

(ii) Storage continuity constraint

X kj 1  X kj  J kj  Qkj  S kj (j  1,2, ......, M)
(iii) Total volume of water available constraint
T T T
X T 1
j  X   J   Q   S kj  0
1
j
k
j
k
j
k 1 k 1 k 1
(iv) Hydro generation equation
 
Pjk N  h j 1  0.5g j (2 X kj  J kj  Qkj  S kj ) (Qkj  u j ) (j  1,2, ......, M)
X min
j  X k
j  X max
j (j  1,2, ....., M)
Qminj  Q k
j  Q max
j (j  1,2 ....., M)
Pi min  Pi k  Pi max (i  1,2, ....., N)
 Long Range Hydrothermal Generation Scheduling –
Optimal Control Strategy
Lagrangian function is given by,

T
N k k
N M

L    (ai ( Pi )  bi Pi  ci )  1  PD  PL   Pi k 
k 2 k k

k 1  i 1  i 1 

 P 
M
 k
2j
k
jN  h j (1  0.5 g j (2 X kj  J kj  Q kj  S kj ))(Q kj  u j )
j 1

 3k j ( X kj 1  X 1j  J kj  Q kj  S kj )
M
 T 1 T T T

  4 j  X j  X j   J j   Q j  S kj 
1 k k

j1  k 1 k 1 k 1  (10)
 Long Range Hydrothermal Generation Scheduling –
Optimal Control Strategy
The dual variables1k ,k2j and3k j are obtained by equating the partial
derivatives of the Lagrangian function w.r.t the dependent variables to zero,
L k  PL 
k
 (2ai Pi  bi )  1  k  1  0
k
(i  1,2 ..., N)
Pi k  Pi  (11)
L  PLk 
 2 j  1  k  1  0
k k
(j  1,2, ...., M) (12)
Pj  N
k
 Pj  N 
L
k 1  3k j 1  3k j  k2 j h j g j (Q kj - u j ) (j  1,2, ..., M) (13)
X kj
L
Q j
k
 k
3j  4j  k
h
2j j 1  0
.5 g j ( 2 X k
j  J k
j  S k
j  2Q 
j  u j ) (j  1,2, ..., M)
k
(14)

L T T T
 X j  X j   J j  Q j   S kj
T 1 1 k k
(j  1,2 ..., M) (15)
4 j k 1 k 1 k 1
 Long Range Hydrothermal Generation Scheduling –
Algorithm
1.Assume a set of λ4j for (j=1,2, …,M).
2. Set k=1, 3k j =0 (j=1,2 …,M).
3. Assume/modify Q kj (j=1,2 …,M).
4. Calculate hydro generations Pl k (j=1, …, L; l=j+N) from Eqs. (3).
5. Calculate thermal generations Pi (i=1,2 …., N) and 1 by solving Eqs.(11)
k k

and (4).
6. 2 j and 3 j for (j=1,2 …,L) are solved using Eqs.(12) and (13),respectively.
k k

 PLk 
 k
  1 
k
0 (j  1,2 ....., M)

2j 1 k

 P j N 


3k j  3k j 1  k2 j  h j g j (Qkj  u j ) (j  1,2 ......, M)

 Long Range Hydrothermal Generation Scheduling –
Algorithm
7.Calculate the gradient vector from Eq.(14) .If the optimality conditions are
achieved within the prescribed accuracy then GOTO step 8,else adjust the
water discharges using the conjugate gradient method and Goto step4.
8.Calculate the values of storages from Eq.(2) for all the hydro generators.
9.If (k≥T) then GOTO step 10.
else set k=k+1 and GOTO step 3.
10.Check the convergence using the gradient vector given by Eq.(15).If
convergence is not achieved then adjust 4 j using steepest ascent method.
 L 
4 j  4 j  s 

(j  1,2, ......, M)
 4 j 

where s is step size. And GOTO step 2 else stop.