CT Meter Ratio Error

[Lectrician]

I have just looked at a CT meter on a site which is metering a sub-unit.

The CT's are 150/5 on a 100amp sub-main.

The meter has it's dip switches configured for 200/5 CT's and not 150/5 CT's. It has been like this for years I would guess.

Trying to work out a quick way to correct the readings, and the maths is confusing me...

If you had 7000 units, would deviding it by 200 and multiplying by 150 correct the ratio?

So a reading of 7000 units would really be 5250 units?

 

[Porque223]

I think yes.

The three cases are:
1. 100A (for example) reduced by 30:1 with the CT (giving 3.33A) and then the dipswitch giving a 3.33:1 ratio so you end up with 1A which indicates 100A so the overall ratio is 100:1.

2. 100A reduced by 40:1 with the CT (giving 2.5A) and then the dipswitch giving a 2.5:1 ratio so you end up with 1A which indicates 100A so the overall ratio is 100:1.

3., your case, 100A reduced by 30:1 with the CT (giving 3.33A) and then the dipswitch giving a 2.5:1 ratio so you end up with 1.33A which reads high by a factor of 1.33.

1.33 x 5250 = 6983.

 

[Wontdothatagain]

That is correct, CTs are as cheap as chips now, i would probably change them so you would not have to keep getting a calculator out every time they want to take meter readings,

[url]http://uk.rs-online.com/web/p/current-transformers/0331452/?origin=PSF_421300|alt [/url]

 

[Lectrician]

The meter is now programmed correctly for the CT's, but I want to go back over the last years readings.

So, devide by 200 and multiply by 150 should give me the expected readings?

 

[Wontdothatagain]

So, devide by 200 and multiply by 150 should give me the expected readings?

Correct

 

[Porque223]

If you have enough historical data you can tell to some degree of certainty when the wrong setting was made by using Analysis of variance, ANOVA. It takes up about as much paper as my previous post but uses somewhat more messy formulas and calcs.
Depending on how far you go back, a single energy use number for summer and one for winter may be enough, two data points per year.

Then you can back up your claim (for overbilling?) with numbers so it is more nearly a matter of fact rather than a matter of opinion.

Plus I can use your real world data for practice!

 

[JohnW2]

If you have enough historical data you can tell to some degree of certainty when the wrong setting was made by using Analysis of variance, ANOVA.

If adequate historical data does exist, then I really don't think that inferential statistical methods would be necessary to establish when the change happened. A simple graph of annual usage ought to show an sudden 25% change pretty clearly, and that would be much more easily understood, and probably treated with less suspicion than any statistical analysis, by the electricity company. From what we've been told, it seems likely that the incorrect settings have probably existed since the meter was changed/installed, the date of which must be recorded somewhere. If the step in the graph corresponds to that point in time, and has a magnitude roughly similar to that which would result from the known error in meter settings, it would really be a matter of 'QED'.

If (genuine) changes or variabilities in usage over time had been so great as to obscure the 'step' in the graph, then it's very unlikely that any type of statistical analysis would be able to ascertain a 'point of change' i n a manner that would convince anyone - human eyeballs are remarkably good at 'analysis'!

Kind Regards, John

 

[Porque223]

Humor me, John. . .

 

[JohnW2]

Humor me, John. . .

Nothing to do with humour - I'm just telling you how it is, and I can but presume that you have not got much experience of presenting evidence about this sort of issue to organisations such as we are talking about. If there were a change in recorded usage between 'before' and 'after' a point in time (when meter settings changed) which could be 'demonstrated' by inferential statsitical methods but was not apparent ('obvious') to the naked eye on a graph, then I think your chances of convincing them would be very low. Believe me, I've 'been there' more times than I care to recall

Kind Regards, John

 

[TheVictorian]

If you can provide me with the data then I am happy to do the analysis as this is part of my day job..

 

[JohnW2]

If you can provide me with the data then I am happy to do the analysis as this is part of my day job..

Everyone seems to be assuming that there is 'before' and 'after' data available (i.e. before and after the incorrect meter settings). However, from what Lectrician has said, it sounds as if the recent situation (incorrect settings) has existed for many years, probably since the meter was installed, and it is thus quite likely that metered usage data from before this 'incorrect settings' era may well not be available.

Kind Regards, John

 

[Lectrician]

That's correct John. There is no 'before' data.

The meter has been installed for a couple of years, and has been used to privately invoice the unit which is leased to someone else (large unit split into two). Readings are taken monthly, and the tenant invoiced.

The only reason this came up was that when the main unit was empty for a period of time, the units being consumed by the main meter were less than that of the sub meter

It is not actually the tenant making a fuss, it is the owner of the unit that noticed, and wants to correct things for the past year, which is when the current tenant same in.

Looking at the meter and CT's, the DIP switches where set to ON|OFF|OFF|OFF and not OFF|ON|ON|ON, so set to 200:5 and not 150:5. Luckily the instructions for the meter were still with it, saved having to download them!

 

[JohnW2]

That's correct John. There is no 'before' data. .... It is not actually the tenant making a fuss, it is the owner of the unit that noticed, and wants to correct things for the past year, which is when the current tenant same in.

That's what I suspected from what you'd said. So, as discussed at this start of this thread, all you have to do is to multiply the metered figures by 0.75 to get the correct answer, without any need to resort to any statistical methods. (which probably wouldn't have helped, anyway)

Kind Regards, John

 

[Porque223]

your chances of convincing them would be very low. Believe me, I've 'been there' more times than I care to recall

Kind Regards, John

American judges also don't like probability and statistics to determine guilt in spite of the methods and tools offered by the Reverend Thomas Bayes and one UK gentleman named D.V. Lindley who wrote on decision theory. One judge heard the odds and then ordered the court proceedings sealed.

It's sad.

 

[JohnW2]

American judges also don't like probability and statistics to determine guilt in spite of the methods and tools offered by the Reverend Thomas Bayes and one UK gentleman named D.V. Lindley who wrote on decision theory. One judge heard the odds and then ordered the court proceedings sealed.

Exactly my point, and the same is certainly true in the UK. One of the problems is there is usually so much wiggle-room in statistical opinions and approaches that 'the other side' will usually put up an expert witness who will challenge the primary statistically-based evidence, fuelling the judges (and/or jury's) belief that it must be 'mumbo jumbo', since 'even statsitical experts cannot agree'.

I don't know what the situation is in the US, but in the UK we have a big difference in the 'burden of proof' in criminal and civil courts. In criminal courts, conviction requires certainty 'beyond a reasonable doubt' (which is variously interpreted by people as meaning anything foirm a 90% to a 99.99% degree of confidence), whereas one 'wins' in a civil court on the basis of 'the balance of probabilities' - so a degree of 'confidence' of 50.001% theoretically can result in a 'win'). In that latter situation, there is much more scope for statistically-based arguments to be of importance.

It's sad.

In some senses, but in others it's quite refreshing to see courts adopting common sense. As I said earlier, if a step in an electricity usage curve is not apparent to the eye, then a 'subtle step', only apparent as the result of statistical analysis, will cause many judges to argue that "a reasonable man-in-the-street would not be convinced". The other big problem is that, unfortunately, increasingly sophististic statistical methods (hence increasingly incomprehensible to non-statisticians) are often used in attempts to 'twist' evidence in one direction or the other (often both) in a court, thereby rightly increasing suspicion about it all!

I've had judges telling me that if I can't make my point without using statistical terminology and without refering to statistical methodology, I should "sit down and be quiet"!

Kind Regards, John