Alarm on temperatur deviation

somlioy · Jun 17, 2017

Hello

I am measuring the temperatur from 8 temperature sensors. They are all supposed to be within range of eachother but if one of them gets a deviation from the others I want to set an alarm.

How would one solve this problem?

The sensors are measuring exhaust temperature on 8 cylinders in a diesel engine.

willxfmr · Jun 17, 2017

Average the 8 temps together, then compare the individuals to the average.

Bubba.

daba · Jun 17, 2017

willxfmr said:
Average the 8 temps together, then compare the individuals to the average.

Bubba.

I would sort the 8 readings and only average the middle 6, that way the one deviating will not affect the average.

somlioy · Jun 17, 2017

That was the issue I had in mind aswell, the deviating temperatur affecting the average.
Dyu think it'll be reasonable just averaging the middle 6?

Peter Nachtwey · Jun 17, 2017

If just one deviates from the others?

somlioy said:
Hello

I am measuring the temperatur from 8 temperature sensors. They are all supposed to be within range of eachother but if one of them gets a deviation from the others I want to set an alarm.

To do exactly what you stated I would simply find the min and max and subtract the two to find the difference then compare that to your specification.

g.mccormick · Jun 17, 2017

Peter Nachtwey said:
To do exactly what you stated I would simply find the min and max and subtract the two to find the difference then compare that to your specification.

I agree, if you need a general temp deviation or range alarm. You probably should check for open signal or burnout element. You may want to not include open/burnout in deviation alarm, but have a separate alarm for them.

daba · Jun 17, 2017

Peter Nachtwey said:
To do exactly what you stated I would simply find the min and max and subtract the two to find the difference then compare that to your specification.

But the Min and Max would include the one that was deviant ....

And in this instance I doubt there's a "specification" to work to, just actual readings. I suspect there'd be a Max (and a Min after warm-up), and I guess there'd be separate alarms for those, but the OP wants to check if one or more temperature readings are not in correlation to the others.

It's a "voting" system, and somlioy asked "Dyu think it'll be reasonable just averaging the middle 6? ", yes I do, unless you expect more than one deviation, in which case I'd average the middle 4 ....

willxfmr · Jun 17, 2017

daba said:
I would sort the 8 readings and only average the middle 6, that way the one deviating will not affect the average.

+1
That makes good sense.

somlioy · Jun 18, 2017

daba said:
But the Min and Max would include the one that was deviant ....

And in this instance I doubt there's a "specification" to work to, just actual readings. I suspect there'd be a Max (and a Min after warm-up), and I guess there'd be separate alarms for those, but the OP wants to check if one or more temperature readings are not in correlation to the others.

It's a "voting" system, and somlioy asked "Dyu think it'll be reasonable just averaging the middle 6? ", yes I do, unless you expect more than one deviation, in which case I'd average the middle 4 ....

Yes you are correct. Ideally all temperatures should be in range, but if theres something wrong with a cylinder (missing fuel etc) the temperature will change, this is what I want to detect with this system.

shooter · Jun 21, 2017

keep it simple, calc the average and keep a low deviation and a high dev, and apart from this also a high alarm.
keep in mind that for every rack setting the temperatures will be different.

daba · Jun 21, 2017

shooter said:
keep it simple, calc the average and keep a low deviation and a high dev, and apart from this also a high alarm.

I seem to think that idea has already been covered, but the question was raised whether the deviating temperature should, or should not, be included in the average calculation. That statement offers nothing new to the discussions.

shooter said:
keep in mind that for every rack setting the temperatures will be different.

huh ??? what's this mean "rack setting" ?

The temperatures can't be different, they are what they are, you just need to measure them, represent them internally on whatever temperature scale you want to use, and make whatever comparisons are needed.

shooter · Jun 23, 2017

on a diesel the speed is controlled by the fuel rack setting: it is a bar for the fuel pumps.
if one pump gives more it will also heatup that cylinder.

daba · Jun 23, 2017

shooter said:
on a diesel the speed is controlled by the fuel rack setting: it is a bar for the fuel pumps.
if one pump gives more it will also heatup that cylinder.

got it - i thought you were talking I/O "racks"

NetNathan · Jun 23, 2017

Which PLC are you using and which TC module?? Maybe a function block that would do most of the work?
Would calculating the "Mean Deviation" work?

Alarm on temperatur deviation

somlioy

Member

willxfmr

Lifetime Supporting Member

daba

Lifetime Supporting Member

somlioy

Member

Peter Nachtwey

Member

g.mccormick

Lifetime Supporting Member

daba

Lifetime Supporting Member

willxfmr

Lifetime Supporting Member

somlioy

Member

shooter

Member

daba

Lifetime Supporting Member

shooter

Member

daba

Lifetime Supporting Member

NetNathan

Lifetime Supporting Member

Similar Topics