Sei sulla pagina 1di 22

INPEL Dam Duy Hung

Optimal Power Flow


Management for Grid
Connected PV Systems With
Batteries
Yann Riffonneau, Seddik Bacha, Member, IEEE, Franck Barruel, and Stephane Ploix, July 2011
University of Ulsan Seminar (25 February 2014)
Contents
Introduction
SystemArchitecture
System Modeling
What is Dynamic Programing (DP)?
System Modeling
DP and Power Flow Management
Simulations and Results
Conclusion
References
University of Ulsan Seminar (25 February 2014) 2 of 22
Introduction
An electrical storage element generates expensive investment and operation
costs with strong operating constraints.
The objective is to reduce operation costs by managing the power flows in the
system.
It is an optimization problem that consists of optimizing the use of the storage, the
use of the PV source, and to match the local production with the local
consumption.
This paper deals with power flows management for grid connected PV systems
with storage (GPVS) with a focus on optimal scheduling.
University of Ulsan Seminar (25 February 2014) 3 of 22
System Architecture
Presentation of the System
University of Ulsan Seminar (25 February 2014)
GRID PV BAT LOADS
( ) ( ) ( ) ( ) P t P t P t P t = + +
min max
min max
BAT BAT BAT
min
max
GRID GRID
SOC SOC( ) SOC
( )
SOH( ) SOH
( )
t
P P t P
t
P t P
s s
s s
>
s
4 of 22
System Architecture
Power Flow Supervisor
University of Ulsan Seminar (25 February 2014)
provides the necessary
predicted input data to the
optimization stage.
the power flow
management problem is
solved.
is in charge of assigning
the command to the
power electronic elements
in order to apply the
power flow schedule from
the optimization stage.
5 of 22
System Architecture
Power Management and Optimization Tools
The main purpose of this work is to find the power flow scheduling that minimizes
the energy bill of the owner of the system over the studied period.
Use Dynamic Programming (DP) to perform optimal power flow management.
University of Ulsan Seminar (25 February 2014) 6 of 22
What is Dynamic Programming (DP)?
Dynamic Programming is a method for solving complex problems by breaking
them down into simpler subproblems.
Dynamic programming algorithms are used for optimization (for example,
finding the shortest path between two points, or the fastest way to multiply many
matrices). A dynamic programming algorithm will examine all possible ways to
solve the problem and will pick the best solution.
University of Ulsan Seminar (25 February 2014)
Example: Fibonacci Number: 1, 1, 2, 3, 5, 8
Calculate n
th
Fibonacci Number with n = 1000008
Solve: Call F(n) is n
th
Fibonacci Number.
So we have:
( ) ( ) ( )
(2) 1;
1 2
(1) 1 F
n n F n
F
F F = +
= =

long Fibonacci(long n)
{
if (n==1) return 1;
if (n==2) return 1;
return Fibonacci(n-1)+Fibonacci(n-2);
}
7 of 22
What is Dynamic Programming (DP)?
Optimize step n by optimizing all the ways to step (n-1). And choice the way that
has lowest (or highest) total cost from step 1 to step (n-1) and from step (n-1) to
n.
University of Ulsan Seminar (25 February 2014)
Optimized
1 n-1 n
8 of 22
System Modeling
PV Generator
The PV generator has been modeled by a linear power source according to the
ambient temperature and the irradiance level
University of Ulsan Seminar (25 February 2014)
( )
( )
( )
( )
( )
PV PV,STC PVs P
2
Vp
0
0
,
W
/
1 25
1000

%/



T
j
PV STC
T
j
G
P P T N N
P
G The irradia
PV Power at MPP and STC
W m
C The Cell Tempera
nce leve
ture
l
T
C Temperature coefficient at MPP

(
(
=
(

( 20)
800
T
j amb
G
T T NOCT
NOCT nominal operating cell temperature
= +
9 of 22
System Modeling
Batteries
State of Charge (SOC):
University of Ulsan Seminar (25 February 2014)
0
( )
( )
( ) ( ) ( ) ( )
ref
c d
ref
0
c
d
C t
SOC
C t
C t Q t Q t Q t
C(t) capacity at each instant
C (t) capacity of reference
Q(t ) initial quantity of charge
Q (t) quantity of charge
Q (t) quantity i of d arge sch
=
= + +
ref
ref ,nom
( )
SOH( )
C t
t
C
=
State of health (SOH):
ref ref ref
ref ref ,nom
( ) ( ) ( )
( ) [SOC( ) SOC( )]
C t C t t C t
C t C Z t t t
= A A
A = A
linear ageing coefficients Z for different battery technologies have been carried out. The
coefficient Z has a value of 3.10
-4
for lead acid technology.
10 of 22
System Modeling
Batteries
Voltage (V
BAT
):
University of Ulsan Seminar (25 February 2014)
| |
( )
_
_
( ) 12.94 1.46 ( )
( ) 12.13 1.54 1 ( )
BAT BAT S
BAT BAT S
V t SOC t N
V t SOC t N
= +
= (

11 of 22
System Modeling
Converter Efficiency
Voltage (V
BAT
):
University of Ulsan Seminar (25 February 2014)
2
1
1 (0.0094 0.043 0.04 )
conv
In In
In
q = + +
12 of 22
DP and Power Flow Management
Problem Formulation
University of Ulsan Seminar (25 February 2014)
min max
min max
min
max
( ) ( ) ( ) ( )
( )
( )
( )
( )
GRID PV BAT LOADS
BAT BAT BAT
GRID GRID
P t P t P t P t
SOC SOC t SOC
P P t P
SOH t SOH
P t P
= + +
s s
s s
>
s
The objective function of supplying the user loads at the best cost is expressed by:
| |
0
( ) ( ) ( )
T
t t
Min CF Min CR t CP t
=
= +

Cash Flow Cash Received Cash Paid = +


13 of 22
DP and Power Flow Management
Dynamic Programming
The problem is formulated as a system evolution divided in a multistage decision
process.
The state of the system corresponds to the SOC of the batteries, discretized with
a step size of oS0C.
University of Ulsan Seminar (27 February 2014) 14 of 22
DP and Power Flow Management
Dynamic Programming
A transition between two states during one time step corresponds to a SOC
variation written S0C.
As each S0C corresponds to a battery's power value. This reduces the
computation time to check the satisfaction of the constraints:
University of Ulsan Seminar (27 February 2014)
min max
( , , ) SOC SOC xi xj t SOC A s A s A
15 of 22
DP and Power Flow Management
Dynamic Programming
For each S0C, first the P
BA1
and P
uRI
are calculated according to the a priori
knowledge of the loads consumption P
L0A
and the PV availability P
Pv
, and
second, the corresponding CF value is obtained:
University of Ulsan Seminar (27 February 2014)
FIT: is the feed-in tariff (renewable
energy payments)
EgP: is the electricity grid price
BrC: is the battery's replacement cost.
( ) 0 ( ) 0
( ) ( ) ( )
( ) ( ) ( ) ( ) ( )
GRID GRID
GRID GRID
P t P t
CF t CR t CP t
P t FIT t t P t EgP t t BrC t
A s A >
A = A + A
( (
= A A A + A A A + A
( (

( , , ) ( ) ( ) ( , , )
( , , ) ( ) ( )
xi xj
xi xj
SOH xi xj t SOH t t SOH t SOH xi xj t
SOH xi xj t Z SOC t t SOC t
A = A A
( A = A

min
min
1
( , , )
( , , )
1
( )
t such as SOH SOH
t such as SOH
SOH xi xj t
BrC xi xj t BiC
SOH
BrC t BiC
=
=
| | A
=
|

\ .
=

16 of 22
DP and Power Flow Management
University of Ulsan Seminar (27 February 2014) 17 of 22
Simulations and Results
University of Ulsan Seminar (25 February 2014)
E
C
is the energy measured with the meter
Ci.
3
2
1
( ) ( ) ( ) ( )
( ) ( ) ( )
( ) ( ) ( )
C
C2 C1
C
C
CP t E t EgP t BrC t
Since E <E
CR t E t FiT t
Else
CR t E t FiT t
= +
=
=
( )
m
a
ax max
x
If
( (

) )
GRID GRID GRID
m
GRID GRID
CP t EgP P P t P GpF
P
t
P
(
= + A

>
18 of 22
Simulations and Results
University of Ulsan Seminar (27 February 2014)
With DP Optimization
19 of 22
Simulations and Results
University of Ulsan Seminar (27 February 2014)
With simple ruled-based management.
20 of 22
Conclusion
A predictive control system based on a DP approach, that optimizes the power
flow management into a grid connected PV system with storage, has been
presented.
The study focuses on predictive optimization from a priori known forecasts.
The objective was to perform peak shaving with the lower cost for the owner of
the system.
Performances of the management in real conditions strongly depend of the
accuracy of the forecasts and of the mode of operation.
University of Ulsan Seminar (25 February 2014) 21 of 22
References
Optimal Power Flow Management for Grid Connected PV Systems With
Batteries, Yann Riffonneau, Seddik Bacha, Member, IEEE, Franck Barruel, and
Stephane Ploix, July 2011
R. Bellman, The theory of dynamic programming, RAND Corporation, Proc.
National Academy of Sciences, pp. 503715, 1952
University of Ulsan Seminar (25 February 2014) 22 of 22

Potrebbero piacerti anche