1
Intelligent Systems Subcommittee
Working Group on Modern Heuristic Optimization
Task Force Optimization Test Bed on OPF
Competition on Application of
Modern Heuristic Optimization Algorithms for
Solving Optimal Power Flow Problems
Sebastian Wildenhues
Fraunhofer IWES Kassel
Germany
2
Optimization Test Bed on OPF
Targets
• Common benchmarks
• Objective assessments and comparisons
• Black box implementation
• Standard / open source software
Test systems / optimization problems
• IEEE 57, 118, 300
Problem 1:
Problem 2:
Optimal Reactive Power Dispatch
Optimal Active and Reactive Power Dispatch
• Offshore Wind Power Plant
Problem 1 :
Modified Optimal Reactive Power Dispatch
96 times for 24hrs / 15mins scenarios
3
Optimization Test Bed on OPF
Problem overview
Test system
Problem #
Number of
control variables
WPP
IEEE 57 IEEE 118 IEEE 300
1
25
77
145
2
31
130
213
385
814
1
22
Number of constraints
123
165
Number of function evaluations /
independent trials
10000 /
31
50000 /
31
100000 / 300000 /
31
31
4
Problem Formulations
IEEE 57, 118, 300
Problem 1:
Optimal Reactive Power Dispatch
min f = Active power losses
s.t.
Power balance, Slack generator active power,


Load
bus
voltages,
Branch
loadings,




Generator reactive powers
x : [Generator voltages, OLTC taps, Shunts on/off ]
5
Problem Formulations
IEEE 57, 118, 300
Problem 2:
Optimal Active and Reactive Power Dispatch
min f = Dispatch costs using quadratic cost functions
s.t.
Power balance, Slack generator active power,


Load
bus
voltages,
Branch
loadings,




Generator reactive powers
Generator active powers,


x:

Generator
voltages,
OLTC
taps,
Shunts
on/off


6
Offshore Wind Power Plant
18x5MW DFIG Wind Turbines
C
C
C
C
C
C
0.73 km
0.74 km 0.59 km 0.58 km 0.57 km 0.57 km
PCC 220 /
110 kV
110 /
7.5 km land cable
33 kV
26.9 km submarine cable
C
C
C
C
C
C
0.60 km 0.60 km 0.60 km 0.55 km 0.55 km 0.55 km
Xsh,1
C1
Xsh,2
C
C
C
C
C
C
0.75 km 0.56 km 0.54 km 0.62 km 0.54 km 0.57 km
Grid
QPCC
Qset
CONTROLLER
-
OPF
(every 15 min)
POI/PCC set point
Loading of WGs
POI / PCC
Reactive
Power
Requirement
Active
Power
Generation
2
4
6
8
10
12
14
Time (hour)
16
18
20
22
24
7
Problem Formulations
Offshore Wind Power Plant
Problem 1:
Modified Optimal Reactive Power Dispatch
min f = Active power losses
s.t.
Power balance, Slack generator active power,


Bus
voltages,
Branch
loadings,


 POI / PCC reactive power requirement 
Wind turbine reactive powers,


x:

OLTC
taps,
Shunts
susceptanc
es
/
steps


8
Conditions
Test bed procedure
• Packages: Matlab, Matpower, Parallel Computing Toolbox
Encrypted black box with automated evaluation for results submission
• Initialization of individuals
x L + U (0,1)( xU − x L ) ∈ ℜ n
• Access only to
Constraint handling
Integer variable treatment
• Variable boundary checks
Continuous variables:
Integer variables:
no
yes
Algorithm tuning
• Allowed without restrictions
Evaluation criteria
• Statistics from internally calculated fitness
• Normalized variable boundary violations
f '= f + ρ
n
∑
c =1
max[0, g i ]2 , ρ := 10 7