Temperature loop: partial calibration

[Brenda]

Hi all,

To perform the temperature calibration in a furnace, we always take out the sensors and put them in a liquid bath
and calibrate the whole loop as one instrument.

This equals downtime for the furnace

For the reduction of downtime we are requested if we can perform a partial calibration on parts of the entire loop.
As we will be adding extra uncertainty to the calibratin (by the use of a multifuntion dmm) it will only be applied to furnaces with low criticality.

So we evaluate to be using a second set of temperature sensors and 4-20 milliamps transmitters
Have those calibrated in our liquid bath while a multifunction dmm reads out the milliamps.
and afterwards use the multifunction dmm to simulate the milliamps to the dcs card and check the temperature on the controlpanel.

Afterwards replace the first set by the second set and then start to calibrate the first set in the liquid bath, again readout by the dmm.

But: how to set tolerances?

Do we split the tolerances equal for both the sensor&transmittor and the dcs?
Or do we take the actual milliamps and only check the readout on the controlpanel is in specification

More detailed:
The furnace is a type that cycles thru heating and active cooling
Calibration of the entire loop is performed in the range of 0 - 100 °C with 6 data points (0 ; 20 ; 40 ; 60 ; 80 ; 100)
Specification for the entire loop, seen as one instrument, is plus minus 1 degree celcius

0 - 100 °C equals 4 - 20 mA, so the spec in milliamps can be said to be plus minus 0,16 mA

Option 1:
Measure sensor in liquid bath with dmm: use a tolerance of 0,08 milliamps compared to theoretical milliamps
Simulate theoretical milliamps with dmm to plc card: use a tolerance on the control read out of 0,5 °C

Option 2:
Measure sensor in liquid bath with dmm and note down the milliamps
Simulated the real values recorded with the dmm in to the plc card
Tolerance on the read out: 1°C

Any thoughts on both options?

I prefer option 2 as it does not stipulate what individual part of the loop is allowed to have drift
and best case drift on one of the parts are cancelled out by another part of the loop

But it means that after we have calibrated the first set (when it was replace with the second set) we will have to do one more simulation of those recorded values to the plc card.

And what about if we need to adjust settings to the plc card during the change from sensor set one to sensor set two?
Perform an extra theoretical milliamps simulation to be able to investigate the impact when we have the results of sensor set one?

 

[Marc]

Thanks in advance to anyone who can help with this.

 

[dwperron]

A couple of starter questions:

What type of temperature sensors are you using?
What do you mean by "calibrated in our liquid bath "

 

[Ninja]

Is it cheaper to buy another sensor and controller than to have the downtime?

If so, Why not have a "spare" system, and put it in the oven in addition to the governing sensor.
No changing the furnace, no pulling things apart...just an outside reference to see if the two (governing and checking) sensors agree.

You can then do your water bath check on the "spare" at your leisure without downtime.

Another (similar) approach...calibrate one system (sensor and plc together as a unit), and swap for the pair in the furnace. You can then calibrate the removed set without additional downtime...and it becomes "spare, known good".

Directly answering your question: I would NOT take amperage and simulate...I would measure temperature of a known while keeping the sensor and controller together. You never know if they were tuned together during manufacture by tweaking the ROM lookup tables...they aren't all interchangeable...

 

[Brenda]

A couple of starter questions:

What type of temperature sensors are you using?

Pt100 3 wire, connected to a temperature transmitter.
Transmitter connector to input of the plc

A couple of starter questions:

What do you mean by "calibrated in our liquid bath "

We take the sensor and put it in our liquid bath together with a reference
Low temperatures we have a methanol bath
for mid range water
high temperatures silicone oil.

Is it cheaper to buy another sensor and controller than to have the downtime?

Yes

If so, Why not have a "spare" system, and put it in the oven in addition to the governing sensor.
No changing the furnace, no pulling things apart...just an outside reference to see if the two (governing and checking) sensors agree.

You can then do your water bath check on the "spare" at your leisure without downtime.

The furnace room is not safe to enter when in process, due to toxic fumes

Directly answering your question: I would NOT take amperage and simulate...I would measure temperature of a known while keeping the sensor and controller together.

This is far from answering my question.
I agree that this is not the most optimal setting, but your answer does not give any advice on the tolerance setting

You never know if they were tuned together during manufacture by tweaking the ROM lookup tables...they aren't all interchangeable...

We validate our furnace controls
so yes: we know that there is no tuning or tweaking.

 

[Ninja]

The furnace room is not safe to enter when in process, due to toxic fumes

Hello Brenda,
This need not be a hardship...a data logger such as Veriteq means you do not have to be present during the validation.

but your answer does not give any advice on the tolerance setting

The tolerance setting should be based on your process needs, nothing else.
How tight a tolerance is needed for the process to yield good results? Set the tolerance there.

We validate our furnace controls
so yes: we know that there is no tuning or tweaking.

I suggest that if you keep the loop and plc together, you don't know if there has been adjustment made.
This is good, since if you keep them together moving forward, you won't need to care if adjustment is made.

 

[Brenda]

The tolerance setting should be based on your process needs, nothing else.
How tight a tolerance is needed for the process to yield good results? Set the tolerance there.

Did you even bother to read the text in the initial post? Instead of answering beside the question?

Specification for the entire loop, seen as one instrument, is plus minus 1 degree celcius

0 - 100 °C equals 4 - 20 mA, so the spec in milliamps can be said to be plus minus 0,16 mA

Do we split the tolerances equal for both the sensor&transmittor and the dcs?
Or do we take the actual milliamps and only check the readout on the controlpanel is in specification

Option 1:
Measure sensor in liquid bath with dmm: use a tolerance of 0,08 milliamps compared to theoretical milliamps
Simulate theoretical milliamps with dmm to plc card: use a tolerance on the control read out of 0,5 °C

Option 2:
Measure sensor in liquid bath with dmm and note down the milliamps
Simulated the real values recorded with the dmm in to the plc card
Tolerance on the read out: 1°C

 

[dwperron]

"Did you even bother to read the text in the initial post? Instead of answering beside the question?"

A bit harsh for someone looking for advice, I think.......

If you really want to know the tolerance for your total system based on the components, you need to look up the tolerance for each piece (since you did not supply a manufacturer / model or anything so we can check these out ourselves).
There are a couple of ways to combine the tolerances. First, get them all into the same units, the you just algebraically add the for the most conservative result, one that accounts for the possibility that each component is operating at the same extreme of its tolerance.
You can also combine the terms using RSS, assuming a normal distribution of the deviations from nominal. This will give you a tighter system tolerance, but can lead to possible false fails.

 

[Ninja]

Hi Brenda,
Took a couple of days to let others answer...before responding to your post...
All bold and colored are my emphasis added...

Some perspective in case you missed it (and you obviously missed it)...

For the reduction of downtime we are requested if we can perform a partial calibration on parts of the entire loop.

Option 1:
Measure sensor in liquid bath with dmm:

Option 2:
Measure sensor in liquid bath with dmm

I would NOT take amperage and simulate...I would measure temperature of a known while keeping the sensor and controller together.

Recap:
You asked "How should I set the parameters for doing this wrong?"
I answered "Don't do it wrong, do it right"

Is it cheaper to buy another sensor and controller than to have the downtime?

Yes

Recap:
I asked "is the equipment for doing it right cheaper than not doing it right?"
You answered "Yes"

This is far from answering my question.
I agree that this is not the most optimal setting, but your answer does not give any advice on the tolerance setting

Did you even bother to read the text in the initial post? Instead of answering beside the question?

Please be assured that I read your posts before answering, and answered beside your question intentionally because you are asking the wrong thing.
I have given some options for doing it a low risk, reliable way and cautioned you from pursuing your proposed solutions which are higher risk and take more time.
I have also offered you solutions that are cheaper and easier than your proposals...and are lower risk for known, common fails.
You may do whatever you wish...
Please feel free to take more time, spend more money and have more headache to gain an unreliable result.
I wish you the best in your efforts, and hope that you read my offerings and take the time to understand.
When someone asks me the best way to drink poison, my typical response is "Don't drink that".

I've worked with temperature sensors for decades, and if you think the ROM in the plc is not altered per sensor, you are sorely mistaken...

 

[BradM]

Brenda,
In future posts, please be more patient and cordial to the helpers. Everyone here is trying to assist the best they can.