The inform ation contained w ithin this docum ent is for guidance only.

N o responsibility is taken
for errors or om issions. A ll inform ation contained herein is considered copyright.
2006 Eaton C orporation
Page 1 of 8
WHI TE PAPER
Proportioning Surge Protection and Fusing
- Fuse selection for surge applications

TVSS Engineering Group,
Eaton Power Quality,
2 Kent Rd, Mascot,
NSW Australia







Document: Fuse_selection_surge_applications.pdf
Prepared by: J on Point, Design Engineer TVSS
Date: 26 J uly 2006



The inform ation contained w ithin this docum ent is for guidance only. N o responsibility is taken
for errors or om issions. A ll inform ation contained herein is considered copyright.
2006 Eaton C orporation
Page 2 of 8

1 P U R P O S E
This docum ent provides inform ation about the selection of protective devices for surge
protection equipm ent. Issues covered include device type, current rating and connection
m ethods.
2 I N T R O D U C T I O N
The fusing of surge protective devices (S PD s) is a contentious issue and in m any cases has
been handled badly by S PD specifiers and installers because they dont necessarily
understand the special requirem ents of fusing S PD s. The fact that S PD s can and do fail is a
givenand w hilst w e all agree that some form of protection is required, there is little
agreem ent on the m ethods used. U sually, the m anufacturer only m akes a m inim al
recom m endation - usually a current rating and/or standards reference. A t best, the installer
follow s these instructions A N D actually consults a standard. U nfortunately, for ad-hoc
installations, little thought is usually given. This m eans that you end up either w ith no over-
current protector, an appropriate one or an inappropriate one.

W ith no over-current protector, substantial dam age m ay occur to the installation under
cases of overvoltage, extrem e surge current or the S PD m erely com ing to the end of
its useful life. This w ill alm ost certainly result in localised loss of pow er to the
installation. This also contravenes m ost, if not all electrical codes anyw ay, so it is
unlikely in practice. There are m any circum stances though, w hen a separate S PD
protector is not required read on.
A n appropriate over-current protector w ill w ithstand expected surge events w hilst
releasing under true fault conditions to lim it dam age to the S PD only.
A n inappropriate solution m ay protect the installation but cause a pow er outage. It m ay
allow dam age to occur by disconnecting the S PD under expected surge conditions. A t
w orst, it w ill never be able to protect the S PD because the lines rating is so low that
the m ain fuses w ill disconnect the supply first. In som e circum stances (low -current,
highly-exposed installations), this is unavoidable read on.

W hat is also im portant is to understand is w hy S PD s fail in the first place. The truth is, m ost
S PD s w ill last for m any years. The things that cause sudden failure (in descending order of
occurrence) are:
1. External supply faults such as overvoltage (faulty transform er, M V lines across LV
lines etc)
2. Local supply faults (broken or ungrounded neutral).
3. W rongly-selected S PD voltage.
4. A surge in excess of the S PD s rating.
5. A fault w ithin the S PD itself (usually tested at factory).

The first 2 (and m ain) causes are unavoidable, and w ill m ost likely cause further dam age to
other parts of the installation. N um ber 3 can be avoided by selecting S PD s w ith adequate
M C O V for the service. N um ber 4 is rare and generally only occurs w hen a direct strike occurs
to the supply infrastructure. A gain, as in 1 & 2, substantial dam age m ay also occur to other
parts of the installation. N um ber 5 is usually dealt w ith by factory testing and is unlikely to
occur in the field. Exposure to heat and m oisture can be a cause of early failure in S PD s that
havent been stored or installed correctly.


The inform ation contained w ithin this docum ent is for guidance only. N o responsibility is taken
for errors or om issions. A ll inform ation contained herein is considered copyright.
2006 Eaton C orporation
Page 3 of 8
U nderstanding these points, it can be seen that m ost faults that w ould dam age an S PD w ould
also cause other dam age to an installation. For this reason, it is usually accepted that it w ould
alw ays be safer to disconnect the site from the source of the fault than to sim ply disconnect
the S PD and leave the site pow ered. In som e cases (see below ), this is not acceptable, but for
m ost it is. Just bear in m ind that the client is expecting that their investm ent w ill be protected
under the m ajority of reasonable events that m ay occur.

A t the other end of the scale, it is pointless pretending that A N Y solution is 100% capable of
protecting a site against unforseen events, regardless of cost. To this end, A S 1768 and
various EN standards recom m end a risk-m anagem ent principle to be applied, w hich is
adm irable IF the system designer know s A LL the num bers. The truth is, m ost system s are
sim ply over-engineered because the designer sim ply doesnt have the required inform ation,
doesnt understand the problem , or fears the clients litigious nature!
3 T H E N O T -S O -S I M P L E P R O B L E M
S urely, an issue as sim ple as over-current protection requires little thought these days? W hen
it com es to protecting S PD s, som e different rules apply, com pared to other applications.

For exam ple, interrupt rating is very im portant w hen selecting any protective device but w ith
S PD protection, the ability to w ithstand com parable currents (15-100kA ) for short periods of
tim e (< 1m s) W ITH O U T releasing is m ore im portant.

In norm al operation, S PD s draw very little pow er few exceed 250m A , except devices that
incorporate filter com ponents. This m eans that the installation of an S PD generally does not
add to m axim um dem and, allow ing the fuses to be sized close to the supply rating.

A protective device for an S PD m ust:
H ave an interrupt rating suitable for the application. A s S PD s are norm ally connected
im m ediately after the m ain isolator, they m ay have to be able to interrupt 15-100kA
under fault conditions.
N ot release under expected surge conditions. This w ill effectively disconnect the S PD
from the system , rem oving protection from the site. S urge ratings range from 3kA for a
sub-board to 100kA + for a highly exposed or im portant site.
W hen required to release, they m ust do so quickly. Tim e delay m ethods increase the
likelihood of dam age occurring and of the fault current m itigating to a higher level, such
as a transform er fuse.

Then there is the inescapable fact that a system s ability to be protected w ithout loss of
supply is directly proportional to its supply rating. W hat this m eans is that the higher the
service rating, the easier it is to protect. In sizing S PD system s, m any designers forget this. It
is not a sim ply a m atter of site exposure either. For exam ple, a highly exposed 400A service
w ould be easier to protect than a m edium exposure 40A service. It all com es dow n to how
m uch protection the client requires so m uch it hurts, or som ew hat less?

It therefore m atters a great deal, w hether or not the client can accept a pow er outage in
preference to surge protection. If not, one m ust be careful not to m ake inappropriate decisions
regarding fusing.

The inform ation contained w ithin this docum ent is for guidance only. N o responsibility is taken
for errors or om issions. A ll inform ation contained herein is considered copyright.
2006 Eaton C orporation
Page 4 of 8
3 . 1 F u se s vs. C i rcu i t B re a k e rs
O ne recom m endation m ade by m any surge practicians is to avoid circuit breakers (C B s)
altogether. This m akes sense for a num ber of reasons. The m ain one being that there is no
standard for C B perform ance under surge conditions, except for transm ission line equipm ent.
This m eans that certain featuresin particular brands m ay m ake the device either m ore or less
suitable for surge applications. O ther reasons include:

If the trip coil has appropriate characteristics, it w ill operate under surge conditions,
releasing the C B . This w ill leave the site unprotected or, if a supply C B , unpow ered. It
also possesses a finite inductance, adding to the let-through voltage rating of the S PD .
M ost C B s include a tortuous pathto prevent dam age under high-current release
conditions by reducing the chance of destructive flashover inside the device.
U nfortunately, to a surge, the tortuous path represents a very short, high im pedance
part of the surge path. In certain cases, the surge energy sim ply jum ps across the
path. This m ay not alw ays dam age the C B im m ediately but w ill definitely shorten its
life and m ost likely cause it to release after the surge event.
If a C B has tripped, it is all too easy just to reset it. The result can be disastrous if the
fault hasnt been cleared beforehand. If the C B has tripped due to S PD m alfunction,
the S PD m ay ignite or the fault m ay m itigate to the prim ary disconnector, rem oving
pow er from the site. In extrem e cases, the operator m ay be injured if the C B is
adjacent to the S PD . A s fuses require replacem ent, there is m ore chance that the fault
w ill be investigated (because you just cant reset it again afterw ards, like a C B ) or a
service call to the S PD supplier w ill be m ade, avoiding the consequences of a quick
fix.
C ost-w ise, C B s have a lim ited interrupt capacity. S ince prim ary surge protection m ust
be fitted at the point-of-entry, an interrupt rating of 50kA m ay be required w hich is a
rather expensive C B , com pared to H R C fuses.

For these reasons, it is assum ed that S PD system designers and specifiers w ill use H R C
fuses for S PD protection. A s Eaton is a m anufacturer and supplier of both C B s and fuses,
w e could m ake m ore m oney selling C B s, but they are not the appropriate choice. In A LL
cases, gG /gL fuses are specified. D o not use delay or sem iconductor types.
3 . 2 F U S E S U R G E R A T I N G S
M any engineers assum e that fuses are not greatly affected by 8/20us surges. A s long as the
interrupt rating is adequate, they dont think too m uch about the fuse current rating and sim ply
apportion the fuse rating to be som ew hat less than the supply fuses. In m any cases, this
w orks w ell, but in others, it is courting disaster. The reason is that the service fuse rating
determ ines the m axim um surge current that m ay be conducted through the system .

W e have tested m any different values and styles of H R C fuses, and although different styles
of H R C fuse have slightly different characteristics, m ost seem to follow a sim ilar trend, at
least for fuses rated up to 690V . This m eans that for a given A C current rating, regardless of
the actual brand or style of H R C fuse used, there is a m axim um surge current that m ay be
carried by the fuse w ithout rupture. S ee Figure 1 and Table 1 follow ing.

The inform ation contained w ithin this docum ent is for guidance only. N o responsibility is taken
for errors or om issions. A ll inform ation contained herein is considered copyright.
2006 Eaton C orporation
Page 5 of 8


Figure 1.
A n 80A N H -style fuse after exposure to a 50kA 8/20us pulse. A s can be seen, the fuse
literally exploded, covering the inside of the test cham ber w ith filler m aterial. The fuse at rear
w as not connected at the tim e. The rating eventually attributed to this fuse w as 25kA .


S ervice fuse rating
(A )
S ervice fuse surge
rating
(kA )
M axim um TV S S fuse
(A )
(S ervice-33% )
C orresponding m ax.
TV S S fuse surge
rating (kA )
32 7.5 20 < 5
40 15 25 5
50 17.5 32 7.5
63 20 40 15
80 25 50 17.5
100 40 63 20
125 60 80 30
160 65 100 40
200 100 125 60
250 120 160 80
315 130 200 100
400 200 250 120
500 240 315 160
630 260 400 200

Table 1 Fuse ratings vs. surge ratings


The inform ation contained w ithin this docum ent is for guidance only. N o responsibility is taken
for errors or om issions. A ll inform ation contained herein is considered copyright.
2006 Eaton C orporation
Page 6 of 8
From table 1 it can be plainly seen that it is not possible to conduct C ategory C surges
through fuses rated below 40A . A s C ategory C protection (15kA 8/20us) is generally
considered the m inim um for system protection, a 63A service is the low est-rated service that
could be rated as fit for exposure. In this case, one has to question the purpose of specifying
highly rated S PD s on sm all or rem ote sites, beyond sim ply ensuring long life.

N ote that the surge current is not presented as either In or Im ax. The 15kA C ategory C level
given is assum ed to be In, requiring an S PD rated at 40kA Im ax. If the designer w anted a 40kA
In rating, the S PD fuse w ould be required to be rated at 100A . This im plies a service rating of
150A or m ore. If the service w as rated at only 63A , the supply fuses alone w ill only w ithstand
20kA ! Even w ith the S PD protected by a 40A S PD fuse, it is likely that any surge great
enough to rupture the 40A S PD fuse (> 15kA ) w ould also be likely to rupture the 63A supply
fuse (~ 20kA ) as w ell.

The TV S S fuse ratingis sim ply the supply fuse less 33% , to allow som e overhead. This level
should only be used in system s w ith low average dem and. For higher dem and levels, a low er
percentage of the supply rating should be apportioned to the S PD , although this becom es
problem atical for supplies below 200A .

TIP! From testing experience, the sm aller the fuse package, the better its ability to w ithstand
surges. B S 88-style or cylindrical fuses are preferred over N H -style or blockfuses. If you
have a choice, tag-style B S 88 fuses are m ore reliable and offer a low er voltage drop under
surge conditions. They are also less likely to explode or physically rupture, even under extrem e
surge conditions. The figures in table 1 are true for m ost types dow n to 63A . B elow 63A , the
data only applies to B S 88-style fuses but should be representative of m ost types. For N H
types, de-rate the surge rating by a further 25% below 63A .
3 . 3 Wh a t n o w?
It really com es dow n to how im portant it is for the site to rem ain active follow ing a m ajor surge
event and the likelihood of that event occurring. In som e cases, supply disconnection is not
acceptable:

The incom e generated by the system m ay far outw eigh the equipm ent repair costs and
running the system unprotected during peak tim es m ay be preferable to shutting it
dow n for repair.
Labour costs can be prohibitive. S ending a com pany technician to a rem ote site to
replace sm all pieces of equipm ent in a tim ely m anner is cheaper than having to send a
team of qualified people to the sam e site in an em ergency.
The site m ay process m aterials for m anufacture and the loss of pow er m ay cause
expensive w astage.
Public buildings, hospitals and hotels also require special treatm ent as law s generally
require evacuation follow ing loss of pow er.

S ites that w ill tolerate loss of pow er over protection are generally those that have large
am ounts of sensitive and expensive equipm ent and have been designed to w ithstand extended
periods of pow er loss. Exam ples are sites w ith com bined U PS and generator backup, such as
call centres, hospital operating theatres, telephone exchanges and large netw ork server
system s.


The inform ation contained w ithin this docum ent is for guidance only. N o responsibility is taken
for errors or om issions. A ll inform ation contained herein is considered copyright.
2006 Eaton C orporation
Page 7 of 8
S ites that are non-pow er loss tolerant m ust be rated in excess of 63A , just to m eet a C at C
rating (80A for a sm all m argin of safety). To m eet a 15kA In level (Im ax = 40kA ), the site m ust
be rated at 150A or m ore. The follow ing table lists som e points applicable to pow er loss
tolerant and non-pow er loss tolerant sites:

N o n -p o we r lo ss to le ra n t P o we r lo ss to le ra n t
N o backup or backup tim e less than repair
tim e.
Extended backup capability such as U PS w ith
extended runtim e or generator backup.
R aw m aterials subject to w astage Processes not left running or allow ed to stop.
C ostly re-start procedure S im ple or autom atic re-start procedure.
Pow er loss w ould involve evacuation of
personnel.
N on-populated areas or technical operators
trained in correct procedures.
B usiness incom e is directly related to
provision of pow er (m ovie theatre etc)
B usiness incom e not directly related to
provision of pow er (shop, outdoor etc)
Pow er loss m ay involve loss of property or life
(m ulti-storey building, outdoor lighting
system , parking station boom gates)
S ite is environm entally benign and poses no
threat under pow er loss conditions (dom estic
dw elling).
M edium to high pow er sites (> 80A ) Low pow er sites (< 80A )

A site doesnt have to posses all of these qualities it m ay just be one, depending upon the
clients needs. If the client is convinced that the site is non-pow er loss tolerant, rem ind them
that pow er outages do occur and ensure that the rest of the site is pow er loss tolerant
regardless of faults related to a faulty S PD . The last thing one needs is an irate client w ho
blam es his pow er outage on your recom m ended equipm ent!
4 S O L U T I O N S
O nce this decision is m ade, the rest of process is sim ple. The choice to the designer is
w hether to protect the S PD at or below the service rating.
4 . 1 N o n -p o we r lo ss to le ra n t si te s
For non-pow er loss tolerant sites, assum e that surge currents are below the level that w ill
rupture the supply fuses (see Table 1). Then, determ ine the appropriately-rated S PD fuse and
ensure that the S PD specified is rated for an In level at least equal to the expected surge
current.:
The m axim um S PD fuse rating w ill alw ays be at least 33% below the supply fuse rating
and m ay be m uch low er in value for higher-pow ered installations.
Low pow er sites and sites w ith high average load are harder to protect than those w ith
interm ittent loads. This is because the supply fuses have less headroom under norm al
operation. In som e cases, it m ay be necessary to connect the S PD fuses directly to the
supply, instead of the secondary side of the supply fuses. This preserves the service
fuses because fuses rated at 40 or 50A cannot conduct m ore than 10-15kA w ithout
rupture. B y connecting the S PD fuses directly to the supply, the service and S PD
fuses can be the sam e value. For exam ple, a 100A service using series fusing could
only support a surge rating of 20kA (63A S PD fuse) w hereas connecting the S PD
fuses directly to the supply gives a w ithstand rating of 40kA . This corresponds to a
device rated at In= 40kA , I m ax= 100kA , a relatively rugged S PD specification.
The greater the difference betw een the supply fuse rating and the S PD fuse rating, the
better coordination and the less chance that S PD failure w ill cause pow er loss. W ith
higher-pow ered sites (surge rating of supply fuse is 2-3 tim es the intended S PD rating

The inform ation contained w ithin this docum ent is for guidance only. N o responsibility is taken
for errors or om issions. A ll inform ation contained herein is considered copyright.
2006 Eaton C orporation
Page 8 of 8
or m ore), little consideration is required as the S PD and supply fuses w ill coordinate
correctly and S PD -induced pow er loss is unlikely. For exam ple, a 1200A service,
protected to a 100kA level (200A S PD fuse)

R eferring to Table 1, select the S PD fuse appropriate to the service and no less than w hat
w ill provide reliable operation at the desired surge current.
4 . 2 P o we r lo ss to le ra n t si te s
For pow er loss tolerant sites, it is im portant to m axim ise surge current tolerance, to provide
the m axim um level of protection. S uch sites usually have higher surge current w ithstand
requirem ents and therefore, m ore perform ance is required from the fuse.
In cases w here the fuse required for S PD protection is higher in value than the supply
fuses, the S PD fuses m ay be om itted altogether, utilizing the supply fuses for both
service protection and S PD protection. O N LY D O TH IS IF TH E S U PPLY FU S ES A R E
gG /gL H R C TYPES .
If the separate S PD fuse is om itted, the S PD m ay be w ired in K elvinconfiguration,
w hich prevents losses in cabling from increasing the let-through voltage under surge
conditions. In K elvinm ode, cables to the S PD m ay be relatively long w ithout
detrim ent to the protection level.
W ith low pow er sites, it m ay be necessary to connect the S PD fuses directly to the
supply, instead of the secondary side of the supply fuses. This preserves the service
fuses because fuses rated at 32 or 40A cannot conduct m ore than 10-15kA w ithout
rupture. It also m eans that the line fuses have m ore chance of surviving a surge as
they are not conducting surge currents. This m eans that the service and S PD fuses
can be the sam e value, as the pow er consum ption of an S PD in norm al operation is
negligible.

R eferring to Table 1, select the fuse according to the supply current rating or, for m edium
pow ered system s, use Table 2 below :

S PD per-m ode rating M inim um S PD fuse size (A )
20kA 63
40kA 100
80kA 200
120kA 250
200kA 400

Table 2 Fuse ratings vs. surge ratings
5 C O N C L U S I O N
The inform ation presented here should prove useful to engineering consultants, estim ators and
system designers. W hilst possibly appearing overly sim plified, there is no point providing an
intense level of descriptive detail w hen it is not required.

If installations are designed in accordance w ith the inform ation presented here, a cost-
effective, safe and reliable TV S S system should result.

Jon Point