Temperature control through flowmeters

  • Thread starter Pierre Desrochers
  • Start date
P

Thread Starter

Pierre Desrochers

Hi all -

I convince that you can help me on this one.

In a cement factory, we want to control the final temperature of a mix which consist of rocks, sand, cement and water. We know the weight and
temperature of the solids and we have to add water. Its comes from to pipes (one for cold water and one for hot water). We detect the temperature of both water lines and we have flowmeters on each ones.

We must control the total addition of water (recipe) and we must control the final expected temperature (22C). Baer in mind that the water temperature could vary while its being poured.

Can you suggest a formula for the water ratio addition, the cold and hot water are never at the same temperature...

Thancks.

Pierre Desrochers
Integral Instrumentation Inc.
[email protected]
 
P
We use a similar approach for batch temperature control.
Use a Mix Temperature Controller,
Cascaded to a Heat Addition Controller,
Cascaded to the two Flow Controllers.

The Temp Controller SP (Tsp) is your desired mix temp (22 C).
The Temp Controller PV (Tpv) is the measured mix temp.

The Heat Addition Controller SP is TC's output.
The Heat Addition Controller PV is a calculated
value, the sum of:
Cp coldwater * (T coldwater - Tsp) * Flow coldwater
+ Cp hotwater * (T hotwater - Tsp) * Flow hotwater.

The Heat Addition Controller's output provides flow setpoints to the flow controllers, you will have to come up with some 'splitting' scheme such as keeping cold flow stable while adjusting hot, or keeping hot flow stable while adjust cold,
depending on whether you need to raise the mix temp or lower.

This scheme combines a kind of feedforward with feedback control. It works.

best success,

Paul R Nelson, PE
Dow Corning Auburn Site
Engineering and Process Controls
Email: [email protected]
Phone: (517) 496-7013
Fax: (517) 496-7100
 
T

Tanweer Ahmed

Hi Desrochers,

I assume that you actually don't need to calculate the volume but the over all temperature of the water to be added, i.e. temperature of the mixture of the two (hot and cold) waters. Because your process recipe allows you to add a specific ratio of water. This can be calculated by using the specific heat of both, the minerals and water.

For example you calculate that the overall temp. of water to be added is 40 deg C and you have to add 1000 gal. of water. Suppose your hot water temp. is 80C and cold water is 15C.

Total heat energy of 1000 gal
=> Vhot*80*C + Vcold*15*C = 1000*40*C
=> 80 Vhot + 15 Vcold = 40,000 -----------(i)
Vhot + Vcold = 1,000 -------------------(ii)
Now if you solve the two linear equations you get
Vhot = 384.6 gal
Vcold = 615.4 gal

You have one drawback that the temperature may vary while pouring in. I assume your heat exchanger is insufficient and you need to increase its capacity to ensure that the temperature does not change while pouring in.

Hope this helps. If you find some other better method I'll love to share.


Tanweer Ahmed
Phone: +9221-636 5519
Fax: +9221-568 2972
E-mail: [email protected]
 
L

Leon Winitsky

Hi Pierre,
I recommend you look at a fuzzy logic controll FLC for this application. It is relatively easy to set up and configure. Modern PLC's such as MODICON have built in functions to fuzzify and
defuzzify control variables to provide crisp results. Conventional PID, PD fail to meet high precision and robustness requirements, mainly because of the presence of large disturbances. Fuzzy controllers have been proven to be relatively immune to noise. Also a fuzzy controller does not require a formal Temperature model of this problem, and simple linguistic terms are used to set up the FLC.

Leon Winitsky

QUAD Engineering
75 Scarsdale RD
Toronto, On
tel 416 391-3755 ext 427
fax 416 391-3645
 
J

Johan Bengtsson

If you would add water with the same desired temperature all the time it would be faily easy to determine the desired temperature. That is - if no chemical reaction occur to change the temerature (ie rise it), do it do that?

endTemp=(HeatCap1*weight1*Temp1+...+HeatCapn*weightn*Tempn)/
(HeatCap1*weight1+...+HeatCapn*weightn)

breaking that out would mean:
DesWaterTemp=((HeatCapStone*weightStone+HeatCapCement*weightCement+
HeatCapWater*weightWater)*endTemp-
(HeatCapStone*weightStone+HeatCapCement*weightCement)*
startTemp)/(HeatCapWater*weightWater)

Assuming a solid start temperature. (otherwise you can easily expand it to cover that too).

Would it be possible to assume that the cold water always will be cooler than the desired water temperature and the hot water always hotter, in that case you can do like this:

percetHot=100*(DesWaterTemp-ColdTemp)/(HotTemp-ColdTemp)
percentCold=100-percentHot

or simply put in a controller opening one valve while closing the other for temperature control and one opening both valves for flow control
ie:
hotWaterValve=tempControlOutput*FlowControlOutput
coldWaterValve=(1-tempControlOutput)*FlowControlOutput
(assuming your control signals are scaled 0-1)


/Johan Bengtsson

----------------------------------------
P&L, the Academy of Automation
Box 252, S-281 23 H{ssleholm SWEDEN
Tel: +46 451 49 460, Fax: +46 451 89 833
E-mail: [email protected]
Internet: http://www.pol.se/
----------------------------------------
 
J

Johan Bengtsson

fuzzy control might be good, in some cases. This is not a case I would have considered for fuzzy (it does not mean it wouldn't work) but it would have taken me longer time to implement a suitable control using fuzzy than PI(D) (see my previous post), or if control is actually not needed, pure calculations (also in my previous post).

Good luck with whatever you choose!


/Johan Bengtsson

----------------------------------------
P&L, the Academy of Automation
Box 252, S-281 23 H{ssleholm SWEDEN
Tel: +46 451 49 460, Fax: +46 451 89 833
E-mail: [email protected]
Internet: http://www.pol.se/
----------------------------------------
 
E
Let me add to my prior note. In the case of two transmitters into one RTD. you would have 2 ma going into one RTD instead of 1 ma, and the voltage drop due to ohm's law would result in an incorrect temperature measurement.

Regards
Erich Mertz
 
Check the Pyromation or Omega websites for info on the probe or transmitter that you want. There are others that offer more than one probe in a single sheath or a transmitter with more than one output. I remember an application I worked on that used three thermocouples in one sheath.
Cheers
Mark Ray
 
B

Bruce Durdle

You can obtain a temperature probe containing 2 temperature elements in a single body - this is certainly the case with thermocouples and I believe also for RTDs. If you need to run 2 totally independent systems off a single process tapping, try this.

Bruce.
 
E
In order to make the measurement, the transmitter supplies a fixed small current to the RTD. In the case of a 100 ohm platinum RTD, the usual
practice is to put 1 ma into the RTD. The voltage drop across the RTD via ohm's law is then used to obtain the temperature. If you were to put two
RTD's into the input, the transmitter would measure the voltage across 50 ohms. i.e., two 100 ohm resistors in parrallel. This, of course, would
result in a gross error.


Regards
Erich Mertz
 
L

Leon Winitsky

Hi Pierre,
I recommend you look at a fuzzy logic controll FLC for this application. It is relatively easy to set up and configure. Modern PLC's such as MODICON have built in functions to fuzzify and
defuzzify control variables to provide crisp results. Conventional PID, PD fail to meet high precision and robustness requirements, mainly because of the presence of large disturbances. Fuzzy controllers have been proven to be relatively immune to noise. Also a fuzzy controller does not require a formal Temperature model of this problem, and simple linguistic terms are used to set up the FLC.

Leon Winitsky

QUAD Engineering
75 Scarsdale RD
Toronto, On
tel 416 391-3755 ext 427
fax 416 391-3645
 
Top