Temperature measurement of mixed stream 2

[Suddo]

I am currently working on a process where two streams of the same liquid at different temperatures are mixed. There is a temperature indicator at the outlet of the mixing position, however, it shows erranous readings. It is suspected that this is because the two streams have not mixed properly. we have tried to change the location of the TT by several metres, still the result is not satisfactory. Is there a method to measure this mixed stream temperature correctly?
I can provide more details if needed.
Difference in temp > 30 degree celsius
Density difference (due to temperature change) is about 30 kg/m3
The flow is quite turbulent, 650 m3/hr in an 18" pipe

 

[LittleInch]

How are they being "mixed" and where is the temperature probe in relation to that?
Is this a tee into a pipe or what?
Relative flow rates of hot and cold fluids?
Relative temperatures?

What sort of temperature probe? surface mount, thermowell, direct immersion?

How much deviation are you getting?
How much accuracy do you want?

It might take upwards of 80 D for mixing to occur without some sort of induced swirl.

Can you install an inline mixer?
E.g.

https://www.koflo.com/

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Suddo]

Thanks a lot for responding @littleInch
Yes, its a simple Tee in a pipe.
The pipe diameters are all same, hence I assume flowrates would also be similar. Its a system where a pipe line splits into two, goes into two different sections of the heater (radiant and convective), and then come out and join as one again.
Temp outlet of radiant=340 and convective is 290 to 300.
Temperature probe = thermowell
Deviation is inexplicable, there are two TTs at a distance of about 40 cms ( 0.4 m), and the temperature difference between the two reads upto 6-7 degrees.

We want to use this data to calculate heater efficiency, hence as close to accurate as possible, because even 1 degree variation gives, 3 percent error in efficiency.

Inline mixers would call for a plant shutdown as it is a continuous process. We tried installing a TT at 20-25 metres from the point of mixing, however the problem still kind of persists. deviation is still of the order of 10 degrees.
Is there any other method to measure this temperature?
Thanks for the help in advance.

 

[LittleInch]

Depending on your actual piping geometry, flows could be quite different. Even a small difference in losses from one to the other could impact flows by ~20-30%.

without seeing the piping set up it's a little difficult to double guess at a difference, but you clearly have some significant separation of the two fluids.

You could try a strap on transmitter and move it around the pipe to see if you can plot a difference around the pipe, but a mixer is the only real way to get a reading that close to the joining flows. Even 20m is only 40D from the mixing point.



Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[rotw]

You did mention the liquids are same, but are they in the same phase before mixing.
Are you not mixing vapor with liquid given the density and temperature difference?

 

[georgeverghese]

Talk to the plant instrumentation engineer to see if he is happy with the calibration records for these 2 TTs'.

 

[Suddo]

@LittleInch, thanks a lot for these suggestions. I just want to ask something for clarification. What is the basis of the number 80D (the distance of TT from the point of mixing) to be used. Is there some empirical correlation for a given flowrate, reynold's number etc. to be calculated?
Or is it just a practice to keep it as 80D based on prior experience to keep the result accurate.(Is there a recognised source or guidelines in that case?)
I am asking this question because I'll have to justify this number to the management in my report, and the need for other instruments that you suggested.

 

[Suddo]

@rotw, yes they are in the same phase, that's for sure. (Liquid phase) The density difference I'm talking about is the property of the oil at different temperatures (supplied by the manufacturer's lab). And the temperature difference is due to the difference in heating provided by the radiant and convective section.


@georgeverghese, the TTs have been calibrated and recalibrated again and again.

Thanks a lot for bringing these things to my knowledge. Do you guys have more suggestions?

 

[LittleInch]

80D was based on experience when we looked at mixing of a condensate stream into a Crude Oil line without a mixer.

However this subject has been extensively studied, though some are on mixing of one liquid following another in a multi-product pipeline. Relative branch and main pipe sizes, flowrates, velocities, densities etc all can make a big difference.

Typical searches throw up things like:

https://pubs.acs.org/doi/pdf/10.1021/ie0209935

Interestingly this paper seems to show about 8-10D for complete mixing where they use different temperatures. The temperature does peak and vary in that junction just after the tee connection which is where your TT's are?

https://eprints.kfupm.edu.sa/1562/1/P171.pdf

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[georgeverghese]

If you interchange the 2 TT's, does the one with the higher reading (when it was at the old location) still read high? If it is, then I would guess it would be some thing to do with the instrumentation loops. If it now reads low, then it may be a poor mixing issue.

 

[Suddo]

@georgeverghese, on interchanging the two TTs, the high one begins showing low reading, while the low one shows high. So I guess you are suggesting a poor mixing issue. In that case, please let me know how to calculate the optimum distance of the TT from the mixing point so I do not lose on the accuracy, but the the point is near enough so that losses are insignificant.

@LittleInch, the paper you cited talks about Uj/Um ratio of 17,9 and 6. But in my case it would be about 1. So what length should I take?

 

[hacksaw]

Without proper mixing, your temperature measurements are meaningless

Sensor placement cannot solve your problem

 

[georgeverghese]

Okay, so this is not an instrumentation problem.
Other than poor mixing, it may be that
(a) the low reading location thermowell has a thicker fouling deposit layer on it compared to the one with the high reading. There is no way you can clean the outside of the the thermowell ( facing the process fluid) since it is welded to the pipe. If the 2 locations are close by, this seems improbable.
(b) poor contact of the TT with the thermowell - ask the instrumentation engineer to apply some conducting paste on both thermowells.
I wouldnt know what mixing lengths to use - @LittleInch's recommendations / links would be useful.

 

[hacksaw]

Without proper process and piping/mechanical design...measurements are a waste of time

 

[rotw]

At what point in your process do you need the temperatures fully established / homogeneous?
If it is say 100 meters (arbitrary) downstream the branching, why do do you care about the temperatures at the branching?
If it is 5 meters ahead, you do not have instrumentation problem, but something might be wrong with the process.
my two cents...

 

[georgeverghese]

In relation to item (b), check if the thermowell lengths are the same - perhaps the location producing the lower reading has a longer insertion length?

 

[Suddo]

@ hacksaw, I know that, but we cannot change the process without a plant shutdown and the analysis has to be done now.

@rotw, as I stated above, we want to calculate the heat load across the heater to calculate the efficiency. 100 metres down the line would be stable but then it would include all the losses occuring in that distance. Hence I was thinking to take the nearest accurate point.

@georgeverghese, I presume you are saying this considering the fact that the colder fluid being denser settles at the bottom. However, I would like to bring to your notice that the reynolds number is in range of 100,000+, hence the flow would be turbulent enough. The TT length are also same.

Thank you guys for helping in whatever capacity you can.

 

[rotw]

@rotw, as I stated above, we want to calculate the heat load across the heater to calculate the efficiency. 100 metres down the line would be stable but then it would include all the losses occuring in that distance. Hence I was thinking to take the nearest accurate point.

I would question how pertinent is a local temperature measurement for heat balance purpose when it is applied to a mixture that has a temperature profile not established.

A thought:

Could you pick a stable point further down the pipe distance (temperature T_B) and then estimate the heat losses (Qloss)?

Maybe an idea is to take N samples (=1000?) of temperature measurement at the point nearest to the branching (temperature T_A) ; then make a statistical average over the N samples. Then calculate Qloss based on measured difference T_A and T_B.

Knowing Qloss you can use it to improve efficiency accuracy calculation for heater?

 

[rotw]

While trying to learn on the subject, I found this article (seems quite a complex subject)...maybe it is interesting

https://pdfs.semanticscholar.org/959e/14be2e59d50f4ea2d39841caacfd7062009f.pdf

 

[LittleInch]

Why not use two surface mounted temperature transmitters on the outlet of each heater. However you will then need to know how much flow is going through each. You can't just assume 50/50. The possible variance in flow will dwarf your temperature measurement variance.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.