On Thu, 11 Sep 2003 09:48:26 +0000 (UTC), Dr Ivan D. Reid wrote:

On 10 Sep 2003 15:31:14 -0700, robsignals
wrote in :
detection relays, connected via 1,200:1 transformer, are rated 1Amp or
5Amp with a wide ranging multiplier, 5A was specified but 1A installed

...

They were carried out but I guess those engineers assumed they had
been handed over correctly installed equipment and never thought to
check the relay rating, pity they aren't brightly colour coded.

Given that there appear to be only two different ratings, an obvious
question is, "Why not?"

The multipliers are usually small tapered plugs that bridge across a
tapered socket, and are hidden away inside the body of the relay, which
may itself be quite small. I don't think colour coding would have
helped, given that the reported procedure for checking the relay covered
both time and current settings, and had apparently been followed without
the wrong setting being noticed. The underlying reason could have been
as simple as a fitter bringing a mug of tea to the engineer at the time
he was checking the settings, causing him to lose concentration for
second, and.....

In message , Chris Game
writes
If the line went into overload
protection at a fraction of what it should have done, a simple test
of the line during commissioning (at 50% of rating) would have shown
this.

I wonder how you test 2500 amp circuits on load when a large section of
south London only draws half of that. What's the dummy-load going to
look like?
--
Roland Perry

Upon the miasma of midnight, a darkling spirit identified as Roland
Perry breathed:

I wonder how you test 2500 amp circuits on load when a large section of
south London only draws half of that. What's the dummy-load going to
look like?

A couple of thousand herbal Viagra spammers wired up in parallel? :-)

--
- Pyromancer, speaking for himself.
http://www.inkubus-sukkubus.co.uk -- Pagan Gothic Rock!
http://www.littlematchgirl.co.uk -- Electronic Metal!
http://www.revival.stormshadow.com -- The Gothic Revival.

On Thu, 11 Sep 2003 12:23:38 GMT, Roland Perry wrote:

In message , Chris Game
writes
If the line went into overload
protection at a fraction of what it should have done, a simple test
of the line during commissioning (at 50% of rating) would have shown
this.

I wonder how you test 2500 amp circuits on load when a large section of
south London only draws half of that. What's the dummy-load going to
look like?

You don't need 275,000v to inject a test current through the current
transformers. Protection testing is done at both primary and secondary
levels. A primary injection set may produce lots of amps, but at a
relatively low voltage. Testing on the secondary side only uses a few
amps.

On Wed, 10 Sep 2003 15:45:23 +0100 someone who may be Pyromancer
wrote this:-

http://news.bbc.co.uk/1/hi/england/london/3096098.stm

Seems to have changed now, but earlier versions of that piece had a
picture of the burnt out wreck of what looked like a very large
transformer with a couple of dodgy-looking air-con cooling units bolted
on to it.

Was that in any way connected with the fault, or just a random pic of
wrecked HV gear?

Yes and no.

Transformers used in high voltage systems are generally cooled with
oil. PCBs were tried, but didn't last long after the first explosion
spread PCBs around the surrounding buildings. Air cooled
transformers are expensive. The oil is cooled by convection in the
"radiators" attached to the transformer, which usually have a large
fan attached to them which is brought into play at times of heavy
demand, doubling the capacity of the transformer. On smaller
transformers the oil is cooled by the pipes that stick out from the
transformer.

Oil deteriorates over time for various reasons and as it does so it
gives off gas. The Buchholz alarm that caused the control room to
start the switching sequence is there to detect problems with the
oil. In some circumstances where there is an imminent danger of
explosion it will automatically disconnect the transformer, in less
serious circumstances it will issue a warning to the control room,
which can then disconnect the transformer in a controlled manner.

These alarms are not always able to detect problems with the oil
fast enough. In that case the transformer will probably explode,
with burning oil being thrown out. The picture in the report, which
I expect the BBC used, is of a transformer that has suffered this
fate. So the picture is not of what happened, but of what they were
taking steps to avoid happening.


--
David Hansen, Edinburgh | PGP email preferred-key number F566DA0E
I will always explain revoked keys, unless the UK government
prevents me using the RIP Act 2000.

Wanderer wrote in message . ..


It would have been nice to see an actual schematic of the network at
Hurst s/s. They have three supergrid t/frs on site. I'm guessing that
each would almost certainly be independently controlled, and
consequently capable of isolation, by circuit breakers on the hv and lv
sides of the t/fr.

Not necessarily so if this was configured as a mesh connected 4
breaker site or a three and a half breaker bay with the SGTs banked
with incoming feeders and disconnected by inter-tripping and power
operated dead break isolators.


Buchholz alarms are usually t/fr specific in the control room, so why
did the NG control engineer apparently disconect the incoming circuit
and not isolate the tranformer indicating the alarm? If this had
happened the overload situation would not have occured.

I wonder if the bucholz alarm was not for the main SGT, but for a
Voltage Transformer associated with the 275kV incoming circuit feeding
the bar. In which case the control engineer would have no option but
to de-energise the circuit. Operational procedures for certain VT's
call for immediate switchouts of certain 132kV, 275kV and 400kV VT's
to prevent explosions. The SGTs at this site step down to 132kV and if
they failed they would not have caused an impact upon the 275kV
circuit. My money is on the VT as this would cause the loss of a 275kV
circuit and some outgoing 132kV circuits. NR lost both Bromley grid
132kV circuits as a result of this.

It would also be interesting to compare areas of responsibility and
manning levels for this part of their network now and say 25 years ago.
What level of authority and/or delegation of that authority holds today
compared to 25 years ago?

NGC's main control centre is at Wokingham in Berkshire, which I have
visited. They have two backup control centres, locations of which I
cannot reveal in public. The control of the network is split into
three stations across two control rooms, in direct communication with
each other with CCTV and Sound. The SCADA system is extremely
comprehensive and list almost every alarm condition and control every
circuit breaker and powered isolator on the system. 45 persons are on
shift at any one time. Previously, NGC had 7 or 8 control centres with
more limited facilities. Alarms were often grouped and the control
engineer often had to send staff to site to determine which alarm was
actually operated.

With the demise of oil filled equipment and circuit breakers and the
extensive use of on line condition monitoring, the need for
maintenance staff is less than before, but more highly qualified on an
individual basis.

The installation of a 1A relay on a 5A seondary circuit is undoubtedly
an error, but the relays are stamped with the Seondary current on the
front panel. The relays concerned at Wimbledon are digital and thus
the multipliers and IDMT curves are set by software, not plug bridges
or bolted links. These are faster and much more relaible and are a
standard fitment on NR systems. At 275kV all protection systems are
duplicated with different maufacturers relays to ensure that a
malfunction with one set will not prevent a trip in the event of a
fault. A primary injection test would have revealed the discrepancy
with the 1A relay fitted to a 5A secondary circuit. I have equipment
which can circulate 2000A at 2V, and NGT have even larger equipment.
This seems to be a hole in their test regime.

I think that we shouldn't forget that the reliability of the National
Grid has actually improved over the last 15 years, with fewer
equipment failures, losses of supplies and interruptions despite
increasing loads and the deregulation of the energy market.

One other observation. If Mr Game is a quality Guru, perhaps he should
offer his services to NGT?, he obviously knows better than those
within NGT who have a professional knowledge of the system and its
requirements, a skill which Mr Game's posts show complete lack of
knowledge of, even though I and others have tried to educate him, to
the point where he accuses the more knowlegeable of us of knowing
nothing (see his earlier posts), despite the fact that some of us work
in the industry or hand in glove with it - for example I run the
electrification design department of NR.

Everyone is of course entitled to their view on usenet, I've expressed
mine as above, those who peddle half baked theories with no knowledge
and can't be bother to listen to those in the know can of course
broadcast their views till the cows come home, just the rest of us may
choose to cease to listen to their diatribe...........

Richard

On 12 Sep 2003 11:50:36 -0700, Richard Catlow wrote:

Wanderer wrote in message . ..


It would have been nice to see an actual schematic of the network at
Hurst s/s. They have three supergrid t/frs on site. I'm guessing that
each would almost certainly be independently controlled, and
consequently capable of isolation, by circuit breakers on the hv and lv
sides of the t/fr.

Not necessarily so if this was configured as a mesh connected 4
breaker site or a three and a half breaker bay with the SGTs banked
with incoming feeders and disconnected by inter-tripping and power
operated dead break isolators.

I think you've just proved my point. The report that has been published
is still woefully lacking in detail, leading to continued speculation.

Buchholz alarms are usually t/fr specific in the control room, so why
did the NG control engineer apparently disconect the incoming circuit
and not isolate the tranformer indicating the alarm? If this had
happened the overload situation would not have occured.

I wonder if the bucholz alarm was not for the main SGT, but for a
Voltage Transformer associated with the 275kV incoming circuit feeding
the bar. In which case the control engineer would have no option but
to de-energise the circuit. Operational procedures for certain VT's
call for immediate switchouts of certain 132kV, 275kV and 400kV VT's
to prevent explosions. The SGTs at this site step down to 132kV and if
they failed they would not have caused an impact upon the 275kV
circuit. My money is on the VT as this would cause the loss of a 275kV
circuit and some outgoing 132kV circuits. NR lost both Bromley grid
132kV circuits as a result of this.

The report seems to be quite specific about the alarm and it's
associated equipment.

It would also be interesting to compare areas of responsibility and
manning levels for this part of their network now and say 25 years ago.
What level of authority and/or delegation of that authority holds today
compared to 25 years ago?

NGC's main control centre is at Wokingham in Berkshire,

snip

I obviously didn't choose my words carefully enough. I was speculating
on work on site within the substations, associated with the testing and
commissioning of equipment, rather than NGT's control procedures.

The installation of a 1A relay on a 5A seondary circuit is undoubtedly
an error, but the relays are stamped with the Seondary current on the
front panel. The relays concerned at Wimbledon are digital and thus
the multipliers and IDMT curves are set by software, not plug bridges
or bolted links. These are faster and much more relaible and are a
standard fitment on NR systems. At 275kV all protection systems are
duplicated with different maufacturers relays to ensure that a
malfunction with one set will not prevent a trip in the event of a
fault. A primary injection test would have revealed the discrepancy
with the 1A relay fitted to a 5A secondary circuit. I have equipment
which can circulate 2000A at 2V, and NGT have even larger equipment.
This seems to be a hole in their test regime.

Thank you for the clarification.

I think that we shouldn't forget that the reliability of the National
Grid has actually improved over the last 15 years, with fewer
equipment failures, losses of supplies and interruptions despite
increasing loads and the deregulation of the energy market.

Do you have evidence to support that statement? I don't question your
claim, but it would be nice to see supporting evidence. I will, however,
admit to a very healthy dose of engineer's cynicism when it comes to the
validity and reliability of such evidence. I spent too long in the
industry.

In uk.railway, PJML writes:

"The units are colour-coded according to use and rating"
my father's company replied.

"The fitter who fitted it is colour-blind" admitted the
refinery operators after some days of the trial.

And I suppose that the disability anti-discrimination laws would have
prevented the company employing him on account of his colour
blindness. :-)

(Richard Catlow) wrote in message om...
Wanderer wrote in message . ..


It would have been nice to see an actual schematic of the network at
Hurst s/s. They have three supergrid t/frs on site. I'm guessing that
each would almost certainly be independently controlled, and
consequently capable of isolation, by circuit breakers on the hv and lv
sides of the t/fr.

Not necessarily so if this was configured as a mesh connected 4
breaker site or a three and a half breaker bay with the SGTs banked
with incoming feeders and disconnected by inter-tripping and power
operated dead break isolators.

Yes, a mesh site as is Wimbledon. I assume that only a few breakers
capable of interrupting live circuits are provided for reasons of
cost, if so this is surely a false economy that negates much of the
expensively provided diversity of transmission lines. The public and
political 'knee jerk' response is likely to call for more lines when
connecting all lines and SGTs via 'live breakers' would effectively
fire-wall most faults at a fraction of the cost.

Buchholz alarms are usually t/fr specific in the control room, so why
did the NG control engineer apparently disconect the incoming circuit
and not isolate the tranformer indicating the alarm? If this had
happened the overload situation would not have occured.

I wonder if the bucholz alarm was not for the main SGT, but for a
Voltage Transformer associated with the 275kV incoming circuit feeding
the bar.

The alarm was for a specific SGT and associated shunt reactor but
didn't show which (was later identified as the reactor which NG
isolated and re-energised the SGT at 23:00) but the only breakers to
isolate it also cover a transmission line which happened to be
carrying most of the S London load. The plan was to open the breakers
for a few mins while the SGT was isolated and then close them again,
the load displaced onto the line that tripped only took it to 33% of
rating - there was *no* overload anywhere.

The breakers were planned to be open for no more than 10 mins,
probably much less, during which it was accepted that the normal
standard of being immune to a double circuit (pylon) failure was not
met and it puzzles me why NG control didn't urgently complete the
planned switching thus reconnecting the unsupplied section within a
couple of minutes, the report doesn't explain this. Presumably they
re-energised in many small steps as a standard procedure to guard
against potential faults being worsened, because of possible damage to
the supplied system (EDF when offered supplies took up to ten minutes
to accept) and because it may be damaging to reconnect ~800MW in one
go. In the event I wonder if they had 'bitten the bullet' and
reconnected in 2 mins it wouldn't have been a much better outcome.

It would also be interesting to compare areas of responsibility and
manning levels for this part of their network now and say 25 years ago.
What level of authority and/or delegation of that authority holds today
compared to 25 years ago?

NGC's main control centre is at Wokingham in Berkshire, which I have
visited. They have two backup control centres, locations of which I
cannot reveal in public. The control of the network is split into
three stations across two control rooms, in direct communication with
each other with CCTV and Sound. The SCADA system is extremely
comprehensive and list almost every alarm condition and control every
circuit breaker and powered isolator on the system. 45 persons are on
shift at any one time. Previously, NGC had 7 or 8 control centres with
more limited facilities. Alarms were often grouped and the control
engineer often had to send staff to site to determine which alarm was
actually operated.

The report does say that alarm grouping and presntation requires
further work, the engineers had to deduce that the alarm came from the
shunt rather than the SGT.

The installation of a 1A relay on a 5A seondary circuit is undoubtedly
an error, but the relays are stamped with the Seondary current on the
front panel. The relays concerned at Wimbledon are digital and thus
the multipliers and IDMT curves are set by software, not plug bridges
or bolted links.

The report concedes that the relay rating could've been more clearly
stated on the documentation so the engineers probably did see it was a
1A relay but didn't spot that 5A had been specified. The multiplier
was correctly set for 5A, infuriatingly it could have been
recalculated to work correctly with a 1A relay!
[...]
fault. A primary injection test would have revealed the discrepancy
with the 1A relay fitted to a 5A secondary circuit. I have equipment
which can circulate 2000A at 2V, and NGT have even larger equipment.
This seems to be a hole in their test regime.

I find it surprising that in 2 years the loading can't have exceeded
23% of rating, would've thought they'd want to give it a good run now
and then to guard against possible 'rusty points' syndrome and to
confirm it does work. October would seem to be a good time when the
duplicate line can be switched back in instantly when this line
would've false tripped. Is there any way of fine tuning the load on
alternative circuits? Before the increased load the line was carrying
only 72MW of New Cross's 359MW load the rest coming what seems to be
'the long way round'. If it had been possible to gradually transfer
load the trip would only have lost part of Wimbledon including LT.

I think that we shouldn't forget that the reliability of the National
Grid has actually improved over the last 15 years, with fewer
equipment failures, losses of supplies and interruptions despite
increasing loads and the deregulation of the energy market.

They quote independent comparisons showing their record is much better
than many other countries. This does seem to be a case of not spending
a few thousand £ costing millions to their customers and the public. I
think NG Control were shown to be unprepared for dealing with a large
emergency, I don't really blame them but there's clearly a need for
modelling major and improbable faults to feed into simulator training.

One other observation. If Mr Game is a quality Guru, perhaps he should
offer his services to NGT?, he obviously knows better than those
within NGT who have a professional knowledge of the system and its
requirements, a skill which Mr Game's posts show complete lack of
knowledge of, even though I and others have tried to educate him, to
the point where he accuses the more knowlegeable of us of knowing
nothing (see his earlier posts), despite the fact that some of us work
in the industry or hand in glove with it - for example I run the
electrification design department of NR.

Everyone is of course entitled to their view on usenet, I've expressed
mine as above, those who peddle half baked theories with no knowledge
and can't be bother to listen to those in the know can of course
broadcast their views till the cows come home, just the rest of us may
choose to cease to listen to their diatribe...........

I wonder how Mr Game will respond if, or probably when, he's involved
in an embarrasing cock-up...

Wanderer wrote in message . ..
On 10 Sep 2003 15:31:14 -0700, robsignals wrote:

Wanderer wrote in message . ..

snip

On any interconnected system, the National Grid being a prime example,
events are by definition related.

Bad choice of words on your part, I think.How can a lightning strike in
the west country and a 3rd party cable damage that happened to occur at
more or less the same time in the west midland be related? No, that may
never have happened, but I give it as an example of the flaw in your
argument.

If you're talking about the 400kV Supergrid you'd be surprised how
small it is in terms of connection points especially away from big
cities, may only be 2 or 3 nodes between 'the west country' and 'the
west midlands'. As power is always flowing south from the the midlands
a trip in the SW will certainly affect them. As a general point we're
both right to a degree, some effects may be very minor but they do
exist.

This incident was caused by a
classic very simple installation 'cock-up' made 2 years ago and not
discovered or exposed since; system design, configuration and incident
response were correct and supplies would have been securely maintained
otherwise.

Not true. Somewhere between design, configuration and commisioning a
current multiplier was wrongly set, and was not picked up by testing
procedures that should have picked it up.

The multiplier was correct but the relay was wrongly rated though the
multiplier could have been altered to put it right. The system was
designed correctly but was wrongly set-up, it's very unfourtunate it
didn't show in a way that caused little effect.

Paragraph 177 in the report is quite explicit about the procedures used
to commission the protection. I would not fault them as written.
Certainly the responsibility for returning a circuit to commision in
good and proper order rests *exclusively* with the guys carrying out the
work.

Or in other words management have covered themselves, theory is one
thing and practice another. Competent engineers fell into a trap that
management could and should have prevented, as far as the public is
concerned NG fouled-up and the buck stops at the Chief Executive...

Lessons have been, painfully, learnt and the relay error is not likely
to happen again. All 45,000 are being checked (none found after
9,000). I would say the system is now secure

I don't share your confidence. The report paints a picture of a calm and
ordered working environment that almost certainly did exist two or three
decades ago. I suspect those working for NGT might have some difficulty
in recognising this environment today.

I don't think the authors would win the Booker prize, they can only be
describing the current (ouch!) situation.

[EDF] should have reconnected LT in a couple of mins at most. I get the
impression there was only one engineer in their Control who didn't
know what hit him...

Unlikely if they operate a centralised control system. But.....

Outside maintenance hours as this was I doubt there's any switching of
their system...