Transcript Document

An Evaluation Procedure for Estimating Voltage Ripple
Caused by Cloud Shadows Moving Over High-Penetration
PV Distribution Networks
W. Mack Grady
Baylor University
Waco, Texas
[email protected]
Holly Thomas and Alvin Razon
U.S. Department of Energy
Golden, Colorado, and Washington, D.C.
[email protected],
[email protected]
IEEE-ICHQP
May 26, 2014
Bucharest, Romania
High-Pen PV - Should We Be Concerned About Days Like This One?
Plane of Array Solar Radiation
Panel Short Circuit Current
The answer is … it depends!
• Do panel max power curves have the same rapid changes shown above, or are
there response delays?
• How quickly do grid-tie inverters adjust to changes in the max power curves?
• To what extent will averaging due to geographical distribution of PV cancel the
rapid changes shown above?
Test Panels (total 130W) on the Front Side,
Monitoring System and Data Logger on the Back Side
View of Back Side
GH and plane of array sensors
GPS for time stamping
Summary of the “Curve Tracing”
Procedure
• The data logger controls a DC-DC
converter connected between the
test panels and a power resistor
load.
Analog converter
control board
By intelligently adjusting the DCDC converter, the system is able
to sweep the test panels’ IV and
power curves, from short circuit
to open circuit, twice per 5second time interval.
•
Other readings are also taken,
including short circuit current,
open circuit voltage, GH, and
plane of array.
•
Interpolation provides highquality integer 1-second spaced
estimates
Data Logger
Power dumping
resistor and meter
box
DC-DC
converter
Power supply
board
•
Data acquisition
board
Maximum power output (Watts) from a test panel during cumulus cloud activity
consistently shows
• power ranging from maximum power (no shadow) to about 30% of maximum,
• with ramp times of about 5-seconds.
PV MAX DC POWER Using I-V Curve Sweeper, 5-Second Spacing Between
Readings, for a 10 Minute Window, Dec. 20, 2010. Mack Grady, U.T. Austin.
160
160
Panel DC Watts
140
140
120
120
100
100
80
80
60
60
40
40
20
20
00
10-MinuteNear
Window
Near(dot
12 Noon
10-Minute Window
Noon
spacing is 5-sec.)
Concept of Moving Cloud Shadow
Grid Pattern for Planning Studies
Zoom-In of Grid Pattern
A = 40 sec.
5 sec.
transition
band
2C = 100 sec.
Square grid side
length = vel • C,
e.g., 10 m/s • 50
sec. = 500 m.
C = 50 sec.
Power
Distribution
Network

Vel. = 10 to 20
meters per sec.
Simulation Test System
A portion of the 500m grid pattern stacked wide enough
and deep enough to gradually encompass all buses
#20, Substation 138kV,
V = 1.02 pu
Isc = 34.4 pu,
10 MVA base,
X/R =5.0
10 m/sec.
1 km
#14, Wipe Out
(400 HP)
10/12.5/14 MVA Trans., Z = 6.54%,
10 MVA Base, X/R = 10,
Load side tap = 1.025,
Delta 138/Grounded Wye 12.47kV
#15, Big Boss
(1250 HP)
#13, Jupiter
(1200 HP + 0.5 MVA)
1370m, 1/0AL
1780m,1/0AL
1830m,1/0AL
1700m, 350CU
#2, Near Sub W.
860m, 350CU,
2 ckts
#10, Taylor
(400 HP + 1.0 MVA)
1610m,
1/0AL
1300m,
350CU
960m.
350CU
#4, PBW
(0.5 MVA)
1000m. 350CU
#9, Dorsey
(250 HP)
2700m,
350CU
Cable
Impedances
R+
(ohms/km)
X+
(ohms/km)
1000MCM
350 CU
1/0 AL
0.0929
0.1582
0.6460
0.1138
0.1228
0.2030
1300m,
1000MCM,
2 ckts
#5, PBE
(0.5 MVA)
1300m, 1000MCM,
2 ckts
1150m,1/0AL
#11, Longs
(200 HP + 0.5 MVA)
#3, Near Sub E.
#16, Shop
#7, Star
#8, Wilderness
(250 HP)
#12, Apollo
(1200 HP)
#1, Substation
12.47kV
600m,
1000MCM,
2 ckts
#6, Base
(3.0 MVA)
Load Summary
 5.1 MW linear, pf 0.88
 3.8 MW six-pulse DC
motor drives, pf 0.85
Voltage Magnitude Ripple, 50% PV, Wind = 10 m/s, Azimuth = 0 deg
0.000
-0.005
PU Voltage Drop
-0.010
Bus 12
Series2
-0.015
Series3
-0.020
Voltage if all PV is simultaneously shadowed
-0.025
-0.030
Voltage if all PV is simultaneously turned off
-0.035
0
5
10
15
20
Minutes
Fig. 3. Bus 12 Voltage Ripple with North Wind
25
Quasi-static loadflow voltage calculations performed in 1-second
intervals, coupled with the cloud shadow model presented in this paper,
show that for the worst-case bus and wind direction in this test system,
voltage ripple associated with 50% PV penetration level is
• In the range of 1.4%, which is consistent with 0.5-to-2% values in [2],
• Approximately 0.014 / 0.021 ≈ two-thirds of the value computed by
assuming that all PV generation is simultaneously shadowed, and
•
Approximately 0.014 / 0.031 ≈ one-half the value computed by
assuming that all PV generation is simultaneously turned off.
A reasonable method for estimating cloud shadow voltage ripple
for any feeder and any PV penetration level is to
• use a loadflow program to compute the worst-bus voltage drop
due to switching off-and-on distributed load of the same
magnitude as rated PV, and
• then multiply the above by one-half.