Академический Документы
Профессиональный Документы
Культура Документы
I.
INTRODUCTION
2
system is tremendous. (2) The transmission and distribution
systems differ in voltage level, topology structure, parameter
of element and value level of power flow. Each of them needs
its own suitable power flow algorithm. (3) The models for
transmission and distribution networks are built and
maintained in geographically distributed EMS and DMS
respectively, which requires the algorithm supporting
geographically distributed computation. (4) Three-phase
unbalance is one of the main features of a distribution system
distinguished from the balanced transmission system, which
should be also considered in the GPF calculation.
Historically, several piecewise methods have ever been
proposed for calculating power flow of large-scale
interconnected power systems in parallel [10,11]. A series of
splitting methods have been proposed to support distributed
or parallel computation for solving some optimization
problems in large-scale interconnected power systems, such
as state estimation [12-16] and optimal power flow [17-20].
However, these methods were oriented to pure transmission
systems with equilibrium partitions (with same voltage levels),
and dont meet well the special requirements of the hybrid
GPF problem described above.
II.
METHOD
(1)
where,
S&
S&
and
S&
(2)
denote the complex vectors of power
Transmission
System
S
(Slave
System)
Distribution
System
k)
k + 1)
(k + 1)
Step3 Substitute S& BS
into (1) and solve it to obtain the
k + 1)
V&B(
k + 1)
3
transmission system is taken as an intermediate variable S& BS ,
which appropriately reflects the comparatively weak effect of
the distribution system on the transmission system. On the
other hand, V&B is specified and considered as a reference
voltage source for the distribution system, which reflects the
determinative effect of the state of the transmission system on
that of the distribution system. Such a special iterative method
reflects the different physical positions between the
transmission and distribution systems and ensures the
convergence.
(2) The MSS method for GPF calculation can be described
in a comprehensible way, i.e. in calculating the distribution
power flow, the voltage of root node of distribution system is
gotten from the transmission power flow solution and is fixed,
while in calculating the transmission power flow, the load data
of the transmission system are gotten from the distribution
power flow solution and are fixed, these two parts of power
flow calculations are alternated and repeated until
convergence is achieved. Obviously, the proposed method is
compatible with any existing power flow software, since no
specific technique is required for solving the transmission and
distribution power flow equations. As a result, we can take
suitable algorithm for each of them, which is regarded as an
important characteristic of the GPF problem presented above.
Furthermore, we can use engineering quantity instead of per
unit for data exchange to meet the needs for different bases
for transmission and distribution power flows to ensure the
good performance of the GPF calculation.
(3) Distribution power flow equation (2) is yet of a large
scale, and can be split further into numerous power flow subproblems of distribution feeders:
( )
( )
V&
B1
&
S&BS1 V B 2
S& BS 2 V&B 3
S& BS 3
DMS1
DMS2
DMS3
(DPF1)
(DPF2)
(DPF3)
WAN
4
transmission parts, while five radial distribution systems,
named as A, B, C, D and E, are connected into transmission
systems as the partial loads of the transmission systems. For
example, the test system 30E is the combination of the
transmission system IEEE 30 and the distribution system E, as
shown in Fig.3. The static characteristics of the loads in the
distribution systems are modeled as
PL = ( 0.3VL2 + 0.5VL + 0.2 ) PLN
(6)
2
QL = ( 0.4VL + 0.4VL + 0.2 ) QLN
where VL denotes the voltage magnitude of the load node
along distributed feeders, and (PLN ,QLN ) is the nominal
power of the distributed load. More detailed information
about these test systems can be found in [21].
IEEE 30
Transmission
System
Distribution
System
2
VB
(PU)
0.9731
1.0
0.9741
0.001
-0.0259
B
(o)
0.6847
0.0
1.2086
0.5239
1.2086
41
PBS
(MW)
18.40
18.19
17.77
-0.63
-0.42
QBS
(Mvar)
4.21
3.37
1.80
-2.41
-1.57
VB
(PU)
0.9682
1.0
0.9695
0.0013
-0.0305
B
(o)
-0.1186
0.0
0.4819
0.6005
0.4819
118
PBS
(MW)
32.28
31.30
29.94
-2.34
-1.36
QBS
(Mvar)
6.80
5.48
4.02
-2.78
-1.46
VB
(PU)
0.9530
1.0
0.9535
0.0005
-0.0465
B
(o)
-6.1436
0.0
-5.6851
0.4585
-5.6851
PBS
(MW)
28.81
27.71
26.67
-2.14
1.04
303
QBS
(Mvar)
5.04
4.52
3.91
-1.13
-0.61
1.2e6
5
3 and the following text, the FDLF and the forward/backward
sweep algorithms [8] are adopted by the MSS method to solve
the transmission and distribution power flows respectively.
Thus each sub-iteration for transmission system includes a P iteration and a Q-V iteration of FDLF algorithm, and a subiteration for distribution system is an iteration of
forward/backward sweep algorithm. As shown in Table 3, the
total numbers of sub-iterations for the transmission and
distribution systems are 6 and 11 respectively, which validate
the excellent performance of the MMS method.
TABLE 3
COMPARISON ON CONVERGENCE AMONG DIFFERENT METHODS FOR 30E
SYSTEM
N-R
method
46
FDLF
MSS method
method
NMSS
NT
ND
3
6
11
Iteration
Not
number
converged
>100
Notes: For the MSS method, NMSS denotes the number of MSS iteration, NT and
ND denote the numbers of sub-iterations for transmission and distribution
systems respectively.
NT
ND
TABLE 5
ITERATIVE NUMBER OF THE PROPOSED MSS METHOD
System
5A
14B
30E
118C
118D
NMSS
4
3
3
2
3
6
11
9
8
8
9
[2]
H.B. Sun, B.M. Zhang, Global state estimation for whole transmission and
distribution networks, Electric Power System Research, May 2005, V74, N2,
P187-195
[3]
[4]
W.
R.
Cassel,
Distribution
management
system:
functions and
SYSTEM
7
10
IV. CONCLUSIONS
COMPARISON ON ACCURACY BETWEEN THE N-R AND MSS METHODS FOR 30E
Mismatch at 1-th
Mismatch at 90-th
distribution node
distribution node
P
Q
P
Q
(kW)
(kvar)
(kW)
(kvar)
N-R method
6.3
-5.5
0.3
2.9
MSS method
<0.1
<0.1
<0.1
<0.1
Notes: the load data at 1-th and 90-th distribution nodes are 10.2kW+j3.1kvar
and 39.2kW+j10.5kvar, respectively.
14
12
[6]
[7]
[8]
reconfiguration
in
large
scale
unbalanced
distribution
[10]
[11]
6
Tearing, International Journal of Electrical Power and Energy Systems, V
11, N 4, pp.283-288, Oct. 1989.
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
BIOGRAPHIES
Hongbin Sun(M2000) received his double B. S. degrees from Tsinghua
University in 1992, the Ph.D from E&E Dept. of Tsinghua University in 1997.
He is now a professor in Dept. of E.E., Tsinghua Univ, and assistant director of
State Key Laboratory of Power Systems in China. His research interests include
energy management system, voltage optimization and control, applications of
information theory and data mining technology in power systems. He won the
second rank prize of china educational committee science and technology
progress in 1995 and the first rank prize of Beijing science and technology
progress in 2004 respectively.
Boming Zhang(M1994,SM1995) received his doctorate from Dept. of E.E.,
Tsinghua Univ. in 1985. He is now a chief professor in Dept. of E.E. of
Tsinghua Univ, and vice director of State Key Laboratory of Power Systems in
China. His research interests include of power system operation and control. He
won the second rank prize of Chinese national science and technology progress
in 1992.