Reliable Power Distribution Design for Water

advertisement
Schneider Electric
February, 2009
Reliable Power Distribution Design for Water and Wastewater Facilities
By Van Wagner, P.E., Schneider Electric Water Wastewater Competency Center
Introduction
Reliable power distribution is critical to the safe operation of water and wastewater (WWW)
treatment facilities. This loss of power to critical loads can result in raw sewage being dumped in
streams, rivers and, lakes and forcing residents to boil water.
The U.S. Environmental Protection Agency guidelines suggest two independent power sources be
provided for wastewater facilities. This can either be two independent utility feeds, or one utility
feed and one on-site generation. For on-site power distribution systems, there is still debate on
whether to retain dual-source redundancy to the load. To determine the most failure-prone system
components and whether or not they can be bypassed, an analysis of several types of common
power system configurations needs to be completed. By analyzing configurations such as radial,
looped, and main-tie-main for reliability and installed cost, the components that have the lowest
reliabilities can be revealed. This provides not only insights on how to address the potential issues,
but guidance on the most cost effective designs for given criteria.
Background
The purpose of a distribution system is to reliably deliver power to the loads. The system may use
manual switches to select power sources, bypass malfunctioning equipment, or isolate equipment
for maintenance. Upon loss of power, a system that must be manually switched to be reconfigured
may be down for 20 to 30 minutes. In contrast, a system with automatic throw-over capability can
switch automatically within 3 to 5 seconds. Since solid state transfer switches operate in a quarter
cycle, each application needs to be analyzed so its requirements form the design specification for
that particular power distribution system.
Component Reliability
When analyzing reliability, the failure rate is defined as the average failures per year from the data
collected. Repair time is defined as the average time required to affect a repair of the failed
component. Although a component may have a low failure rate, a long repair time can be just as
disruptive as a high failure rate. To see this, it is helpful to compare the reliability of individual
power system components in the context of overall configuration.
Table 1 shows the data for several medium voltage power system components from one of the
most extensive industrial reliability databases, the Institute of Electrical and Electronic Engineers
(IEEE) Standard 493-19971.The components are ranked from least reliable to most reliable. The
table shows that power sources have the poorest reliability, while distribution equipment has good
reliability but can have very long repair times.
1
The utility circuit is assumed to consist of 15 kV to 35 kV single feed. Generally, higher voltage
utility service has greater reliability but service voltage is largely determined by load. Depending on
the total lengths, medium voltage conductors can be the next most likely component to fail. The
remaining components have similar reliability values but time to repair varies widely. For instance,
a rackable circuit breaker can be quickly replaced but a transformer would require days.
Components
Utility
Standby Gen
Switchgear Bus
MV Conductors
MV Switch
MV CB
Transformer
MV Termination
Failure Rate
(#/yr)
Repair Time
(Hours)
1.956
0.967
0.0102
0.00613
0.0061
0.0036
0.0030
0.000333
1.32
3.9
26.8
97
3.6
2.1
130
25
Comments
Single feed
per 1000 ft; Buried conduit
Replace CB (Drawout)
Replace
Table 1: Reliability data of medium voltage equipment
Utility Sources
In the following section, it is assumed that where there are two utility sources, the reliability of those
two sources is independent of one another. Realistically, though, there is always some coupling
between utility sources. In order to increase independence, the two sources should be from
separate utility substations or at least separate transformers within the same substation. For
instance, in a regional blackout the circuits to the facility should be in separate rights-of-way.
Whereas overhead conductors are subject to weather related disruptions, the most common cause
of underground circuit disruption is dig-ins by excavators. Better independence can be gained with
the second source as a standby generator.
However, additional utility sources can be expensive, , and the cost of a second utility circuit is
entirely born by the utility customer. Even in a dense metropolitan area where alternative circuits
are more available, a second distribution circuit could cost $500,000 to $1 million with a two-year
lead time. An alternative would be to use a standby generator for the second source. It’s about
twice as reliable as the utility, is fully independent and costs about $250,000 to install. However,
the standby generator has greater operating and maintenance costs. So for capacities over 5 MW
where multiple generators are required, the utility might actually be the least expensive.
Switching Components
The type of switching device employed in a design has significant impact on the down time and
cost of the system. The following is a short review of the characteristics of fused switches and
circuit breakers.
Load Interrupting Fused Switch: This is the most inexpensive means of power system switching.
At medium voltage this switch can average about $15,000, whereas a main circuit breaker section
will be about $50,000. At low voltage, a fuseable switch board might be 25 percent less expensive
2
than switchgear using circuit breakers. Although fuses provide short circuit protection at each
phase, they cannot open un-faulted phases. In addition, fuses cannot trip for sensitive ground fault,
but can have greater interrupting capability than some breakers. The switch provides a manual
means of reconfiguring the power system. It is appropriate where installation cost is a major factor
and reliability is not critical. At medium voltage, there are two major issues:, the need for available
trained personnel and the amount of time required to perform the switching. Repair time is not long
for any of the devices discussed, although the switch has a much more limited number of rated
operations.
Circuit Breaker: A major advantage of the circuit breaker is its ability to operate automatically.
This allows it to switch to a different source or reconfigure when necessary within a few seconds —
it could switch more quickly but must provide delay to allow residual motor voltages to decay.
When a circuit breaker trips, all three phases are opened, preventing single-phase operation.
Tripping is controlled by a separate protective relay that, in addition to simple over-current, can trip
for low or high voltage, ground fault, phase sequence, differential, or other functions.
Selective coordination can be more readily achieved with circuit breakers than fuses due to the
large variety of protective functions and setting adjustability. Selective coordination selects the
protective device setting that helps assure the nearest upstream device clears the fault. This
prevents a situation where a fault causes several protective devices to operate.
Circuit breakers are more durable than switches with a greater number of rated operations.
If and when they need to be replaced, medium voltage circuit breakers are draw-out and can be
quickly replaced with a spare. Low-voltage circuit breakers are available as draw-out or fixed while
lower-rated breakers are only available as fixed.
Approach
In this section, the reliability analysis method used is one described by Billington and Allen in
Reliability Evaluation of Power Systems2. The analyses are for single contingency only, which
means either a single failure or a planned shutdown, because the probabilities of independent
multiple contingences are so small that it has little impact on the type of comparison performed
here.
Note: The derived reliabilities should only be used to compare design configurations. They are not
absolute values that predict actual downtime in anything other than an approximate manner. Real
reliabilities depend on equipment age, operating history, maintenance level, environmental
conditions, etc.
The reliability analysis is applied at two parts. The first includes the electric utility and the facility
medium voltage (MV) distribution, while the second includes only the low voltage (LV) distribution
system to the load. This approach helps clarify the contribution of each part and allows a greater
number of configurations to be evaluated.
Reliability is evaluated at a single point for each configuration. In the MV case, it is the load side of
the feeder breaker or switch that supplies the load. For the LV case, it is the feed point of a load.
The LV case includes the MV feeder conductor and the step down transformer. Although this does
3
not cleanly divide the equipment by voltage level, it does reveal the reliability of the MV and LV
cases more distinctly.
Installed costs include equipment and labor for the circuit breakers or switches shown in the
figures. Engineered-to-order equipment costs are taken at current market prices. In addition,
conductors are included and assumed to be installed in buried PVC for MV and overhead
galvanized rigid conduit for LV. Wiring materials and labor costs are taken from RSMeans
Electrical Cost Data. 3 Conductor lengths are assumed to be 300 feet unless otherwise noted.
The reliability and cost determinations are sensitive to the number of feeders in the design. In the
reliability analysis, adding more feeders decreases reliability because more equipment is added
that can fail. To try to normalize the installed costs, they are determined on a per feeder basis.
Obviously, more feeders spread the fixed cost of the scheme and reduce the per feeder cost. With
the MV cases, three feeders are assumed fed off the bus while in the LV configurations, four
feeders are assumed.
Primary Distribution
Radial
The simplest power distribution system is a radial system, shown in Figure 1. A radial design
provides power distribution with the minimum initial equipment cost and is typically configured with
a load break switch. Loss of the utility, switch or conductors will make the system unavailable until
repairs can be completed, as there are no alternate paths or sources.
Figure 1: Example of single source radial power distribution system
The reliability and cost data are shown in Table 2. The radial system forms the basis of comparison
for all subsequent medium voltage designs. The failure rate is the probability of the loss of power to
the load side of the transformer switch in a given year. To yield the average outage hours per year,
the probability is multiplied by the mean time to repair. (For an example derivation of reliability
indices see the Appendix.) From an operating standpoint, the average hours down from a forced
outage (unavailability) is a more useful number than the failure rate.
The utility is by far the most unreliable element in the configuration and dominates the reliability of
the system. There is an estimated average of 3.30 hours of forced downtime per year due largely
to the utility restoration time as the configuration does not allow any scheduled maintenance to be
performed on the system without shutting the load down as well. The approximate cost for this
4
installation is $37,000 per feeder. This configuration could be appropriate where the loss of
capability can be tolerated or offset by other methods, such as a small remote pumping station.
Component
Radial
Primary Selective
Looped
Failure (#/yr)
1.990
1.990
1.992
Unavail
(Hours/yr)
3.30
0.657
0.932
MTTR
(Hours)
1.66
0.33
0.468
Cost (per
feeder)
$37k
$71k
$57k
Table 2: Reliability results and installation costs.
Primary Selective
Figure 2 shows a primary selective configuration. In this case a second utility feed is provided to a
radial system at the transformer primary. It requires a second primary conductor and disconnect
switch. A duplex switch can be provided at the transformer that utilizes a common fuse for a slightly
lower switch cost. In this configuration, one source feeds the transformer; these installations
normally do not parallel the sources. In this case and all subsequent configurations, it is assumed
the manual switching takes 20 minutes to reconfigure the system.
Figure 2: Example of primary selective feed
The reliability numbers are shown in Table 2 at one of the load sides of one of the switches. The
reliability is identical since the normal source is exactly the same as the radial case. There is
substantial improvement in the unavailability compared to the radial arrangement because the load
can be switched to the alternate source in 20 minutes. (It should be emphasized that the alternate
source is assumed to always be available in this comparative analysis).
The configuration allows work to be performed on the primary conductors to the transformer switch
while the transformer is fed from the other conductor. Work can also be performed on one of the
transformers while the others are energized.
Note: the cost has doubled due largely to the additional switches needed. For the per feeder cost, it
is assumed there are three feeders and the cost of the common components are shared.
5
Looped System
The looped network is a variation on the primary selective configuration and is shown in Figure 3.
In this case the loads are in series rather than parallel, as in the primary selective case. One of the
switches will be open so that the utility sources are not paralleled. The number of switches required
is equal to the equivalent primary selective scheme although duplex switches are shown here with
one fuse per transformer. The advantage of the looped system is that less conductor length is
required.
Figure 3: Example of looped primary system
A commercial version of the looped system can be applied as part of underground distribution at a
campus-type setting where several separated buildings are fed from individual pad mounted
outdoor transformers. Equipment cost can be further reduced by replacing the two disconnect
switches with dead front elbow terminators in the transformer to switch source conductors.
The overall failure rate of the looped system is slightly greater than the previous two schemes but
its unavailability falls somewhere between them. After a fault occurs, all the transformers on the
deenergized section will be down and all the equipment must be inspected to try to locate a visible
fault. When a conductor fails, it can take eight hours to locate the faulted segment and the
deenergized transformers cannot be restored until the faulted segment is isolated. If not for the
fault locating time, the unavailability would be close to the primary selective time.
The example looped system is 25 percent less expensive than the primary selective case and
offers a second source. In this case, conductors were $7,000 less expensive than the primary
selective case and the duplex switches were $7,000 less expensive than the full switches.
From a scheduled maintenance standpoint, this configuration is identical to primary selective. It
allows work to be performed on the primary conductors to the transformer switch while the
transformer is fed from the other conductor. A transformer can be maintained while the other
transformers are energized, although maintenance may require brief interruption of the other
transformers to reconfigure the system.
Main-Tie-Main
A very common two-source selective configuration is the primary main-tie-main (MTM) shown in
Figure 5. In this case, the primary source feeds an entire bus rather than a single load and a
normally open tie between the two buses provides source selection. Where there are many nearby
6
medium voltage loads, a bus arrangement is a more effective means to distribute power than the
previous configurations. With a MTM configuration, the buses must be rated for the load of both
buses with the tie closed. A MTM typically employs circuit breakers rather than switches and the
example shown uses circuit breakers as they can provide automatic switching within seconds of a
loss of one of the utility feeds.
Figure 5: Example of primary main-tie-main
This analysis assumes a three-second automatic throwover by the circuit breakers to the other bus
if the normal source is lost. The results are shown in Table 3. The three-second switching time
reduces the unavailability almost 100 times for this configuration. However, there is no alternative
source for the bus, feeder breakers or the feeder conductors.
Component
Radial
Primary Selective
Looped
MTM
Synch Bus
Ring Bus
Double Bus
Failure
(#/yr)r
1.990
1.990
1.992
1.996
2.000
1.967
2.004
Unavail
(Hours/yr)
3.30
0.657
0.932
4.67x10-2
5.42x10-2
1.29x10-2
1.77x10-2
MTTR
(Hours)
1.66
0.33
0.468
2.33x10-2
2.71x10-2
6.60x10-2
8.79x10-3
Cost (per
feeder)
$37k
$71k
$57k
$77k
$89k
$110k
$122k
Table 3: Reliability results and installation costs
This analysis shows the reliability at the load feeder breaker. If the point of evaluation is moved to
the transformer primary, the feeder conductor increases the unavailability to over 30 minutes per
year. This is because there is no alternative feed to the transformer and it’s assumed that it takes
97 hours to replace the conductor. In this case and the subsequent cases with MV buses, the
analysis is to the feeder breaker. Feeder conductor failures can obscure the characteristics of the
particular scheme. Medium voltage feeder conductor reliability will be included in the low voltage
analysis.
7
Scheduled maintenance can be performed on the main beaker or upstream of it without disruption
to the feeder loads. However, work on the bus, feeder breakers or tie would require shut down of
the feeder loads.
Synchronizing Bus
Additional buses and ties can be added to the MTM arrangement if there are more than two
sources. However, a more flexible scheme is to use the configuration shown in Figure 6. Here, a
separate conductor, bus duct, or switchgear bus links all the buses through circuit breakers. It does
use one more breaker than the MTM configuration, but it also provides the ability to tie any
combination of buses together. This arrangement is called a synchronizing bus (or, less commonly,
a star bus) even though none of the buses may be connected to a generator.
Figure 6: Example of primary synchronizing bus
Sometimes one of the buses will have a source but no load. For instance, a bus with several
paralleled standby generators (hence the synchronizing bus designation), or a bus that is fed by a
transformer, called a sparing bus.
The availability and the reliability of the synchronizing bus decrease a small amount compared to
the MTM arrangement. This decrease is due to the addition of the extra breaker. For the purposes
of this analysis, the reliability of the two configurations should be considered identical.
The cost of the synchronizing bus is about 15 percent greater than the MTM due to the extra
breaker and cabling but that cost buys additional switching flexibility.
As with the MTM scheme, scheduled maintenance can be performed on the main beaker or
upstream of it without disruption to the feeder loads. However, work on the bus or feeder breakers
would require shut down of the feeder loads.
Ring Bus
A configuration sometimes found in utilities but not often in industrials is the ring bus shown in
Figure 7. A utility would normally operate with the sources paralleled but the industrial version
8
typically would not. This arrangement has the advantage of using a minimum number of breakers
while having the ability to switch the loads between sources. The number of breakers required is
equal to the number of sources and loads. More load buses can be added as the loads or sources
are protected by a pair of breakers.
Figure 7: Example of primary ring bus
The availability of the ring bus in the figure is several times better than the best considered so far.
This is because the design uses minimal hardware and can accommodate automatic throwover.
The cost per feeder, however, is very high. Each feeder has two breakers sized as mains and the
total number of breakers per feeder is high. Another disadvantage is the ring bus still does not
address the loss of the bus or feeder conductor.
The good availability of the ring bus is offset by the high per feeder cost. Nevertheless, it is not
practical for industrial power distribution. An industrial version of the ring bus would need utility
breakers and more feeders with individual breakers on the load or source buses, increasing cost
even further.
Double Bus
The double bus arrangement can be considered a modification of the ring bus and is shown in
Figure 8. There is no tie breaker between buses but each feeder can be fed from either bus. The
switchgear lineups usually face each other across an isle and unlike the MTM, it allows throwover
to the other source if the bus is lost.
This configuration can be very flexible with multiple sources such as two utilities and a standby
generator. In that case one of the buses would be split into a MTM for the other source. The double
bus arrangement would be found in very large WWW facilities.
9
Figure 8: Example of a primary double bus
Like the previous configurations, the reliability and cost of the double bus are based on three
feeders.
The time unavailable is reduced by about half compared to the MTM. The feeder cannot be
restored after a bus or breaker fault with the MTM until the equipment is repaired, but with the
double bus scheme, the feeder is switched to the other bus in three seconds for a source bus fault.
However, a fault on the load bus or one of the load bus breakers does not provide an immediate
alternate source.
This configuration uses two breakers per feeder and increases per feeder cost 50 percent over that
of the MTM.
With the ability to feed loads from either bus, maintenance is possible at the MV bus while the
other bus serves the loads. However, maintenance is not possible without a shutdown from the
load bus to the load.
Low Voltage
Low voltage component reliability data from the Gold Book1 is shown in Table 4. Overall, the failure
rates are low but some of the repair times are quite long. Specifically, transformer and bus repair
times can take days, which will reduce availability.
Components
LV Switch
Transformer
LV CB
Switchgear Bus
Failure Rate
(#/yr)
Repair Time
(Hours)
0.0061
0.0030
0.0027
0.0017
3.6
130
4.0
24.0
Comments
Replace
Replace CB (Drawout)
10
LV Conductors
0.00141
10.5
per 1000 ft;
Table 4: Low voltage equipment reliability data
Single Ended Radial
This is the simplest low voltage radial configuration and is shown in Figure 9. It is the least
expensive arrangement and provides no alternative feed to the loads if normal power is lost. The
implementation would likely use fused switchboard rather than circuit breakers due to cost
considerations.
Figure 9: Example of single ended secondary substation
Each of these examples assumes a 2000 kVA transformer with a secondary main, four feeders,
and 300 feet conductor lengths. The single ended radial configuration assumes a fused
switchboard and the load evaluated for reliability is at the end of a 300 foot LV feeder conductor.
The results are shown in Table 5. As can be seen, the failure rate is low compared to the MV
examples primarily because no utility is included. However, the unavailability is similar, because of
the long repair time for the transformer and, to a lesser extent, the MV conductors. This provides
insight into what should be bypassed in subsequent configurations to improve reliability.
Component
Radial
MTM (LV)
Spot Network
Failure (#/yr)
3.59x10-2
2.29x10-2
1.25x10-1
Unavail
(Hours/y)
1.07
9.73x10-2
1.13
MTTR
(Hour)s
29.7
4.36
9.03
Cost (per
feeder)
$66k
$76k
$87k
Table 5: Secondary reliability results and costs
To determine the reliability of a single ended LV configuration together with one of the primary
distribution systems, add the reliabilities and the unavailabilities.
11
The radial design serves as a reference for comparison of the other designs.
Power must be shut down to the load for any maintenance that must be performed on the
equipment.
Secondary Selective
One of the more common LV configurations is the secondary selective or MTM, as shown in Figure
10. For maximum reliability, drawout power circuit breakers with automatic transfer should be used
in this configuration. The automatic transfer switches to the alternate source in three seconds
rather than the 20 minutes it may take to manually transfer.
Figure 10: Example of secondary Main-Tie-Main (MTM)
The unavailability is reduced more than 10 times compared to the single-ended radial value. The
transformer and MV conductor have long repair times and the MTM allows them to be
automatically bypassed. Still, loss of the bus, one of the bus breakers, or the feeder conductor to
the load, requires the load to be down until repairs are completed.
Cost per feeder is 15 percent more compared to the radial configuration due to the transition from
fused switches to breakers and the addition of the tie breaker.
Secondary Spot Network
Networks are implemented by utility companies to serve multiple loads in high density urban areas.
A network typically covers several city blocks with service to individual loads connected to the
network wherever convenient.
The secondary spot network is shown in Figure 11. In this case, the secondaries of two or more
transformers are paralleled through a special circuit breaker called a network protector. If there is a
transformer or MV feeder fault, the secondary bus will back feed it through the network protector.
12
The network protector has instantaneous tripping for reverse flowing current to isolate the faulted
equipment from the secondary bus. Typically, removable links are added to be able to isolate the
buses.
Figure 11: Example of secondary spot network.
A spot network refers to two or more parallel sources to serve a specific load such as the main
switchgear for a building. It is implemented in an industrial setting where loads are frequently
moved by running a bus duct between the substations and adding loads where needed. The
distribution is not likely used in WWW since loads tend to be permanent.
An interruption will only occur when all the sources are lost or one of the secondary buses fault.
However, the loads will see all the voltage sags on the sources or as a result of a fault within the
facility. With the voltage sag threshold of some electronic equipment at 85 percent of nominal
voltage, a voltage sag may be as disruptive as an interruption.
This evaluation assumes a power circuit breaker main and 300 feet of 3000 A aluminum bus duct
with eight 600 A bus duct plugs. In this case, the bus duct drops are assumed to be 150 feet since
the bus duct should be closer to the loads than a substation.
Interestingly, the reliability results don’t demonstrate a remarkable performance for this design.
While the spot network is immune to interruptions originating upstream of the network protector
(barring loss of all sources), it is vulnerable to loss of the bus duct, network protectors and the load
conductors. The bus duct failure rate and repair time are high and dominate the results, therefore
this design has an availability that is comparable to the single ended radial configuration.
The cost of the spot network is higher than any of the other configurations. The largest cost item is
the long length of high ampacity bus duct. The transformers and bus must be oversized and circuit
breakers must have a high interrupting rating due to the increased fault current available with the
paralleled sources.
13
Conclusions
This paper presents the relative reliability of several power distribution configurations with
approximate installation costs. These will help determine both the value of downtime and which
components have the greatest effect on reliability. Actual systems are a composite of those
evaluated and may have variations on the configurations presented. The key concept is that the
reliability of the design should be reviewed and balanced against the cost. For a complex system or
one where reliability is extremely important, a formal reliability analysis should be performed.
References
1IEEE
Standard 493-1997, IEEE Recommended Practice for Design of Reliable Industrial and
Commercial Power Systems (The Gold Book).
2Billington,
Roy & Allen, Ronald N, Reliability Evaluation of Power Systems, Plenum Press, 1984.
3RSMeans
Electrical Cost Data, 30th Edition, 2007.
Beeman, Donald, (editor), Industrial Power System Handbook, McGraw-Hill, 1955.
14
Appendix
Example Reliability Calculation for MV Radial Configuration
The reliability will be determined at the load side of the left transformer switch. The elements that
can cause loss of power to the left transformer when failed are:
•
•
•
•
Utility
Main Switch
Each cable segment (3@ 300 feet each)
Three transformer switches
The table below shows how the reliability indices are derived.
Component
Utility
Switch (X4)
Cable (X0.9)
Terminations (X12)
Result
Failure (#/yr)
1.956
2.44x10-2
5.52x10-3
4.00x10-3
1.990
Unavail
(Hours/y)
2.58
4.39x10-2
5.53 x10-1
0.100
3.30
MTTR
(Hour)s
1.32
3.6
97
25
1.66
The components are in series, which means failure of any one of them will result in loss of power at
the point of evaluation. For a series arrangement, the expected failure rate, λ, of the system is the
sum of the individual failure rates.
λ s = ∑ λi
The unavailability, U, of each component is the product of the failure rate and the mean time to
repair, r, of the component.
U i = ∑ λi ∗ ri
The unavailability of the system is the sum of the component unavailability’s.
U s = ∑U i
15
The mean time to repair of the system is the system unavailability divided by the system failure
rate.
rs =
Us
λs
In this analysis, it is assumed that an alternative source or path is always available. That is not the
case in actual systems, but including failure of an alternate source or path changes the indices very
little. This is a comparative analysis and the small change does not alter the relative effectiveness
of the configurations.
If the load can be switched to restore power, the mean repair time can be replaced with the mean
switching time for all the components upstream of the switching point. For a simple analysis, this
method suffices to represent the interruption time.
About the Author
Van Wagner is a staff power systems engineer for Schneider Electric. He is responsible for power
studies, design, investigations and training in the Midwest region and for strategic accounts. He
has 33 years of experience in power systems, 10 of which are with Schneider Electric. Wagner
received a Bachelor of Science degree (’74) and Master of Science degree in electrical engineering
(’93) from the University of Michigan and Michigan State University, respectively. He is a former
chair of IEEE 1346 and is the current chair of the new industrial chapter of the IEEE 1100 "Emerald
Book.” Wagner is a registered professional engineer in the state of Michigan.
16
Download