Understanding temperature readings

I recently had a long consulting session on the phone, guiding a customer through the operation of a reflux still.  A number of isssues came up regarding temperature readings that I thought I should discuss here.  These issues are quite tricky so if you don’t understand what I’m talking about, it’s ok, take out of the discussion the fact that all is not as simple as it seems.

If you are making vodka (neutral alcohol) you will be watching the temperature at the top of the column to get it as close to 78°C as possible, which is about the boiling point of pure alcohol.  Imagine you are looking at a bimetal gauge and there is a gap between the needle and the graduation marks; if you are looking at an angle you will be getting an incorrect reading by up to 3°C I would say.  Let’s say it was just 2°C, this is the difference between alcohol at 94% ABV and 83% ABV which is huge!  This is the degree of error from the error of parallax. Have a look at the Vapour Liquid Equilobrium chart to see if you can work it out for yourself (CLICK THE LINK)…Equilibrian chart for alcohol 101.35kPa.

Imagine you have made a change to the power input and you want to observe the change in temperature at the top of the column with the same bimetal gauge, you look up and nothing has happened, so you wait a minute and you make a decision on what to do next based on 60 seconds.  I know these days we all want instant results, but any temperature instrument needs time to respond.  Bimetal will respond slowly as the two bonded metals need to heat up or cool down fully and become stable for the temperature reading to be correct.  This can take quite a few minutes.  Add to this the fact that in a thermowell, between the vapour and the bimetal, there is a thin piece of metal then an air gap then another thin piece of metal, then another air gap then the bimetal. The temperature change passes very slowly through the air gaps and it can take up to 10 minutes to see a change in temperature of about 10°C.

If that wasn’t enough, consider also that a bimetal gauge is not very accurate, say +/- 2°C and you certainly won’t get any decimal points.

Ok, so you can get a digital thermometer and everything will be hunky dorey.  Not so.  The same issues apply regarding response time and accuracy.  I have a $500 calibrated digital thermometer that measures to 0.01°C but I have to wait about 3 minutes for the temperature to stabilise.

Different digital thermometers have different accuracies and if you buy one for less than $50 you will probably only get +/- 1°C accuracy out of it and you won’t know where you sit on that range unless you test it against a calibrated thermometer.  Generally this type of accuracy is not a problem in distilling except for two prominent cases:

• the first case is for licenced distillers who have to measure the bottling alcohol for payment of excise duty.  The measurement of % ABV needs to be temperature corrected and for 40% ABV a 1°C temperature change can mean 0.4% alcohol which is quite a lot when multiplied by hundreds of bottles.
• the second case is when measuring the temperature at the top of a reflux column to assess the % alcohol in the vapour.  There are some stills out there that are ‘automated’ and give a digital reading of % ABV using the temperature reading.  Let say this very firmly – I do not recommend this at all.  Firstly, you need to understand that between 90% and 95% ABV the vapour temperature changes from 78.06°C to 78.11°C and there is not a commercial thermometer that can do this with any level of confidence. Secondly, you also need to understand that changes in pressure will affect the VLE curve and those temperatures I gave above are only for 101.35kPa (1 atmosphere).  As the weather changes, so does the VLE curve and this throw your calcualtions out the window. It is better to use the temperature as a guide and use the alcometer in the parrot to make decisions (with the necessary temperature correction).

I spoke of response time in a thermowell above and there are 2 ways to improve this:

• don’t use a thermowell, have the temperature probe directly in the medium you are measureing.  This will work if the probe is permanently installed and maintenance is easy during a shutdown.
• add some thermal paste to the thermowell as it will transfer the heat much quicker.  Look at the specifications of the paste for the thermal conductivity which is measured in W/K.m (watts per Kelvin (°C) per metre).  The higher the number the quicker the response time, but you have to balance this with price as the higher the thermal conductivity the high the price on an exponential curve.

In summary, take care to understand the limits of your temperature instruments in the contect of their application.  The sensitivy of the application will determine how you use the data to make good decisions.