Title: DR on the GSA Williams Building Spot Network One year of Monitoring the Operation of the Interface S
1DR on the GSA Williams Building Spot NetworkOne
year of Monitoring the Operation of the Interface
System
2(No Transcript)
3A Quick Review of What and Why We Did a
Monitoring Study
- The System
- The Design
- The IEEE Standard
4Arrangement of Williams Building Spot Network
- 277Y/480 volt LV supply system
- two utility primary feeders supplying network
transformers - primary feeders may have other loads
- integrated transformer, relays, and LV air-break
switch (network protector or NWP) in a Vault in
the basement - Initial design had the protectors opening on
reverse power flow with no time delay
5DR on Spot Networks
- use inverter-based generation technology so
network protectors are not as likely to be opened
by fault current contribution from the local
generator or, - time-coordinate network protector relay with
feeder relaying to prevent NP opening from
generator fault current contribution
6Network Relay Modifications
- Electromechanical Master relay changed to a solid
state MPCV (also a very sensitive three-phase
reverse power relay but with the capability to
set a low reverse power level with time delay) - Time delay selected was 15 cycles
- The MPCV high reverse power setting for
instantaneous trip was set at 50 of the
transformers rated current
7Adjustable Reverse -Power Characteristic
time delay for low currents
time
adjustable delay time
adjustable instantaneous trip threshold
instantaneous trip for higher currents
100
current ( of xmfr rating)
8Also Added Network Interface Control Unit, The
CH Auxiliary Unit
9Problems with Local Generation on Network
- if local generation exceeds local load, even
momentarily, network protectors open and isolate
the network from the utility supply. - At very light load conditions, one of parallel
protectors may open due to a minor change in
feeder supply system which can lead to excessive
cycling - A preventative rule of thumb is to maintain
between 2 and 10 of transformer rated power flow
into the network
10DR on Spot Networks
- limit generation to less than local load, with an
adequate margin for sudden loss of load
conditions, to insure no undesired reverse power
conditions or, -
- control inverter power output with tie-line load
control so power flow from the utility to the
network never reverses or,
11DG with Network UnitsAlternatives
- trip or isolate local generation from network if
network protector relays sense low incoming power
flow or, - isolate local generation with critical loads by
sensing reverse power from critical load bus to
network (not an option for induction generators)
12DR on Spot Networks
- prevent islanding of the network trip generation
or isolate it from the network unit with a
circuit breaker whenever a network protectors open
13Over and Under-power Relays Required
14Reconnection of the DR must be blocked until
- 1. Both protectors are closed and
- 2. The minimum load is greater than (150 minimum
load DR rating) and - 3. Each protector is supplying one half of the
value under item 2, and - 4. After the minimum reconnection threshold is
reached, a selectable time delay must pass before
reconnection is permitted. This time delay must
be adjustable from 1 to 300 seconds. -
15Compliance with IEEE Standard 1547-2003
-
- In the clause applying directly to spot networks,
the Standard first requires that a protector
shall not be used to separate, switch, or act as
a breaker failure backup, or in any manner
isolate a network to which DR is connected unless
the protectors are rated for such an application.
It clear that the framers of 1547 intend that if
all the protectors were to open, the DR must have
already switched off the network bus.
16Continued--Compliance with IEEE Standard 1547-2003
- To meet the requirement that the protectors not
act as a breaker failure backup to the DR
breaker, the Tecogen engineers provided an engine
kill circuit which acts in parallel with the
tripping of the DR contactor.
17Continued--Compliance with IEEE Standard 1547-2003
- The second requirement in this clause of 1547 is
that the DR cannot cause operation of a network
protector or prevent re-closing of any of the
protectors. It further states that this
coordination must be accomplished without
requiring any change to the utilitys practice
for clearing times of the network protectors
18Continued--Compliance with IEEE Standard 1547-2003
- The interconnection package meets the first part
of this requirement by the under power relay
tripping the DR if the power level drops in
either protector to the point that a power
reversal might be experienced by a sudden loss of
a block of load. The approach that, if either
protector is open, the DR must be off the
network, meets the requirement of not preventing
re-closing
19Continued--Compliance with IEEE Standard 1547-2003
- The third requirement of the clause is that the
DR cannot be connected to the network unless more
than 50 of the spot network protectors are
closed. -
- The interconnection package meets this
requirement by the prevention of a DR connection
unless both protectors are closed.
20Continued--Compliance with IEEE Standard 1547-2003
- The fourth requirement is that the DR cannot
cause any protector cycling. -
- One of the principal missions of the under power
relays is to prevent this condition.
21Continued--Compliance with IEEE Standard 1547-2003
- The fifth requirement of the clause is that by
connecting the DR to the network bus, network
equipment loading and fault interrupting capacity
cannot be exceeded. -
- This requirement was addressed in the Phase I
engineering study where no issues were
discovered.
22What the Captured Data Shows
- Questions to be answered
- Salient Events
- Conclusions
23Questions to be Answered
- Time for the Tecogen unit to shut down after a
trip signal was issued by the CH unit? - How often did the CH units protective function
operate? - How effective was the CH unit?
- Did the PV or Tecogen units recognized all events
other than minimum power flow that required their
shutdown? For example, could they detect and
clear for faults on either of the feeders
supplying the network units, or determine when
one of the network units tripped, or when voltage
went below limits?
24Continued-Questions to be Answered
- Did any event occur during or near minimum power
conditions that required the present minimum
power settings to prevent reverse power flow
through either network protector? - Did protector cycling exist even with the
presently determined minimum load? - If any protector cycling occurs, was it
excessive?
25Continued-Questions to be Answered
- Would protector cycling be excessive if the
minimum power settings of the auxiliary
protection package were lowered? - Did the MPCV time delay of .25 seconds give an
adequate margin of error for all feeder events? - Was the generation fault response characteristics
under abnormal conditions such that time delay
was not required?
26Continued-Questions to be Answered
- What does the CH unit protection package demand
of the Tecogen (DR) unit? - What situations occur on the 480 volt system?
- How close to borderline between normal and
abnormal are the Tecogen protection settings
operating on the system? - How does the DR behave under abnormal conditions?
- Is the generator affecting voltage on the system?
27Continued-Questions to be Answered
- Confirm the CH units and Tecogen equipments
clearing times are consistent with those
demonstrated by the commissioning test. What
happens because of the ten-cycle limit allowed by
IEEE P1547 for the Tecogen or the PV unit in case
of a single-line-to-ground (SLG) fault on the 480
volt system, i.e., will there be any temporary
overvoltages on the GSA system as a result of
slow clearing of the DRs? - How many trips accurate, missed, unnecessary?
- What caused the event?
28Continued-Questions to be Answered
- Under abnormal situations out on the utility
system, how does the onsite generation respond?
If the Tecogen system were to be as benign as
motor loads, it would have to shut down well
before three cycles of abnormal behavior have
been experienced on the power system.
29Answers Determined from Data Captured
- The Cutler Hammer Auxiliary Protection Package
(CH unit) performs exactly as designed to trip
the Tecogen unit off line within 5 cycles when
either protector experiences momentary reverse
power. -
30Answers Determined from Data Captured
- The CH unit holds the Tecogen off line when
either protector is open. Since the CH unit
tripped the Tecogen unit when under power
occurred in either protector, initially the
Tecogen unit was being tripped nearly every
evening. By the time the monitoring system got
installed, the Tecogen operator had the unit
gracefully shut down on a time of day schedule.
Of the 37 transient events captured, there were
only six that resulted in the CH unit forcing the
Tecogen unit off line. -
31Answers Determined from Data Captured
- The CH unit underpower/overpower scheme
functioned as designed. It blocked Tecogen
operation when a protector was open and it
allowed Tecogen operation even when one
protectors net import dropped to very close to
the under power set point (down to 64.7 kW in one
of the example events). The trip point was set
to 60 kW.
32Answers Determined from Data Captured
- Very little data was collected to show how the
Tecogen or PV protection functions work under
fault conditions because the CH unit operates so
rapidly for necessary trip conditions. The only
data that reveals something of the Tecogen units
response is two apparent transmission fault
events. When the Tecogen unit was on line during
these voltage dips it did not shut down on a
voltage trip. However, these dips last less than
10 cycles.
33Answers Determined from Data Captured
- The data has not been completely screened for
minimum load network protector non-trip
conditions. To the extent that it has been
reviewed, only the day that building power was
deliberately reduced for maintenance did power
flows on a per protector basis drop below 60 kW. - No condition that suggests cycling has been
observed. - With no indication of cycling, there would
not be any excessive cycling. -
34Answers Determined from Data Captured
- To the extent that the data has been evaluated,
there are no data that would indicate cycling
might be increased if a lower reverse power
setting were applied to the CH unit at this
installation where the MPCV relays have a low
reverse power time delay of 15 cycles. Based on
the frequent observation that load current
dropped at a rate greater than voltage dips, in
any application that does not employ a small time
delay, a system voltage dip could result in
cycling or total network loss if it occurred near
the minimum loading conditions.
35Answers Determined from Data Captured
- The MPCV time delay of 15 cycles appears to be
appropriate for this application - There is limited data that demonstrates the
Tecogen unit increases its current output under
fault conditions. These data alone may not be
sufficient to draw conclusions on need for time
delay in protector relaying because of the quick
action of the CH unit.
36Answers Determined from Data Captured
- The 60 kW minimum import criteria, at least
during the Spring months, meant that the Tecogen
tripped off line every evening. Therefore the
Tecogen operators installed a timed shut down
each evening to avoid the non-graceful trip from
the CH unit.
37Answers Determined from Data Captured
- Many dips greater than 3 have been recorded.
However, they appeared to be related to
transmission faults. Because of the high
resistance grounding of the NSTAR network feeder
system, single-line-to-ground feeder faults do
not seem to cause much of a voltage dip at the
480 volt bus.
38Answers Determined from Data Captured
- From the limited data collected where the Tecogen
unit may have had time to operate before the CH
unit forced a trip, the Tecogen protective
settings were not sensitive enough to be of use
in detecting faults for which they should trip. -
39Answers Determined from Data Captured
- While we have some data on the dynamics of the
Tecogen unit during abnormal conditions, it has
not been tested against model data. The
three-phase fault that occurred on April 11, 2004
showed that the unit could feed 170 of rated
current for two cycles when its terminal voltage
was held at 36. The voltage dip event of
February 2, 2004 suggested that, during
unbalanced fault conditions, an induction
generator might supply current long enough to
cause network protector tripping
40Answers Determined from Data Captured
- The generation does not seem to be affecting the
480 voltage, but it does seem to have noticeable
harmonics in its current output as demonstrated
by the difference between the average rms value
and the average fundamental value. -
- The rapid and correct actions of the CH unit
appear to preclude gathering much useful data on
the trip times of the Tecogen or PV units from
their own control actions.
41Answers Determined from Data Captured
- All of the trip commands issued by the CH unit
were accurate and necessary. When the Tecogen
unit was on line to respond to these CH unit
required trips, the Tecogen unit appears to have
tripped in one cycle after receiving the trip
command as designed by Tecogen.
42Answers Determined from Data Captured
- All but two of the network protector trips were
clearly caused by faults. One was a maintenance
trip and the other was from an undetermined
cause. -
43Answers Determined from Data Captured
- The rapid and comprehensive action of the CH unit
may have precluded the gathering of meaningful
data on the nature of the Tecogens dynamic
response during faults and the effectiveness of
the Tecogens protective actions to integrate
with a network without requiring a time delay in
the protector.
44Conclusions
- The findings from monitoring at this site
verified - The functional performance of the CH unit
- The need for at least 15 cycles of time delay in
network protector with DR connected to its low
voltage bus
45Continued-Conclusions
- The voltage characteristics of loads must be
considered when determining if a low voltage
island might be formed - Network protectors can sense and clear a fault in
less than six cycles which, given the mechanical
nature of the switch lends credence to concern
that network protector relays can initiate
tripping in three cycles
46Continued-Conclusions
- Voltage sensing at the DR cannot be counted on
to trip the DR in a timely manner for
interconnections with network systems - Some interfacing device must be provided to trip
the DR when one of the protectors opens.