Heat Exchanger

[font=verdana, arial, helvetica]I arrive at a site where I am starting up a used piece of equipment of which a component was a steam to water to product heat exchanger that was controlled by a dedicated PID controller. The site does not have steam, so they installed two on-demand hot water heaters as a heat source, removed the steam control valve and added a mixing valve. Thermocouples from the second water circuit and the product were connected to analog inputs of the PLC-5. The mixing valve accepts discrete open or close outputs and is actually plumbed as a diverting valve where heated water always enters the AB port and A port supplies the heat exchanger and B port recurculates back throught the on-demand heaters. The mixing valve takes approximately 40 seconds to switch from fully open to fully closed or vice-versa.

The customer asks whether I can have it operating by noon!

I devise a cascade PID control system where the product loop is the master and its output is the setpoint of the slave water loop. The heated water source temperature is not controlled by the PLC. The output of this PID system is then used in a Time Proportional fashion where if the PID output is 60%, the mixing valve will be opened for 6 seconds out of ten to allow heated water into the heat exchanger water loop. The time base has not been established, so the ten seconds is an example. I suggest that an analog valve be used so that finer control of the mixing valve can be achieved.

To get things up and running I suggest setting the heated water source close to the product setpoint and fully opening the mixing valve to see what happens. It turns out the mixing valve or plumbing is no good as the on-demand heaters would not operate because of no flow with the mixing valve in any position. The mixing valve is bypassed and the product heats up.

The customer now wants to discuss the control system and my suggestion of the analog mixing valve. They are convinced by the supplier of the valve that it will work perfectly in this application and don't recognize the advantage of an analog valve. They further think that simply turning the pump on and opening the valve when the product is cold and then tuning off the pump and closing the valve when the produt is hot will result in what they want. The product must stay within 4 degrees F of setpoint or bad things happen. After much discussion, I have swayed their opinion to the point that they recognize the need for closed-loop control, but they want me to "role-my-own" with timers that make a variable pulse based upon error and fire either open or close output dependant upon direction of error.

Now, what do you guys think? I think I approached this unexpected situation properly and proceeded in a standard fashion that woud result in the best control possible with what I had to work with.

I know I don't like working with plumbers! [/font]

Hmmmmmmmmm

Derek,

Can you elaborate more on the Steam>Water>Product exchanger? If what I am reading is correct then this process is possibly exothermic?

As for controlling the steam valve this can be easily done by making a "roll your own" time proportional control. I use this in situations where I have boilers that operate at too high an output even at low fire. I modulate the firing rate with time proportional control.

Quick and dirty it works like this........
Set up a timer to work as your total firing rate..On time + Off time. Allow this timer to run and recycle contiuously. Scale the output of a PID algortithm so that 0% = 0 Time and 100% = Total firing rate as established in the timer above.

Use a <= (less than equal to) statement to compare the Scaled PID output to the firing rate timer acc value.

Fire the valve any time the statement is true.

Say the PID output scales to 20 seconds and your firing rate timer is 30 seconds. This will operate as valve on 20 seconds valve off 10 seconds.

This water mix presents another problem. So if you could give more details on the plumbing that would help.

Hope this helps,

Dave

Thanks for reply Dave,

Originally the steam heated a water loop in the heat exchanger. Now hot water is introduced into the heat exchanger water loop as steam is not available. Basically there are three loops; heated water, heat exchange water and heat exchanger product. The mixing valve meters heated water into the heat exchanger water when product heat is called for. The steam valve is long gone.

I already have logic like what you described to contol the open/close state of the mixing valve. With the valve taking 40 seconds to switch this may be problematic, hence my suggestion of an analog valve. The analog valve would follow the output of the cascaded PIDs and provide finer metering of heated water into the heat exchanger than a time proportioned setup. With TP, the valve will always be opening or closing because of the switching nature of TP.

Where did exothermic come from? It is not summer up here yet, so cooling has not been considered.

What I was asking is if the approach I took was appropriate. Rather than "rolling-my-own" quasy closed loop logic, I decided to use standard PID techniques on which standard tuning methods can be used to achieve tight control of the process. I encountered quite a bit of opposition to this approach and am wondering if the this opposition was justified.

The PLC analog valve will likely be installed when I return to site Monday.

Your approach is acceptable based upon the information that you supplied.

The installation of an analog valve will help to stabile the process. A TP on a 40 second valve "might" be too slow depending of the size of the load and the capacity of the line. After that point, whether the application needs a standard closed loop control, cascade control, feed forward, etc, depends upon the response of the process.

I am confident that a stand alone PID controller or a PLC can both accomplish the 4 degree F requirement, once you work out the details.

Exothermic

The water loop in the equation threw me.........thought it might also need cooling during process.

The 40 seconds to cycle can be factored in....it only becomes a problem at close to 100% output as any amount of off time becomes 40 seconds +. I have the same problem with hot oil systems as the purge time in the firebox is 60 seconds before relight. The only saving grace is that the thermal mass of the oil system, reactor, and process material prevents significant cooling during this time.

I will now take this opportunity to inject some humor:

The following is supposedly an actual question given on a University of Washington chemistry mid-term.

The answer by one student was so "profound" that the professor shared it with colleagues, via the Internet,which is, of course, why we now have the pleasure of enjoying it as well.

Bonus Question: Is Hell exothermic (gives off heat) or endothermic (absorbs heat)?

Most of the students wrote proofs of their beliefs using Boyle's Law (gas cools when it expands and heats when it is compressed) or some variant.

One student, however, wrote the following:

First, we need to know how the mass of Hell is changing in time. So we need to know the rate at which souls are moving into Hell and the rate at which they are leaving. I think that we can safely assume that once a soul gets to Hell, it will not leave. Therefore, no souls are leaving.

As for how many souls are entering Hell, let's look at the different Religions that exist in the world today.
Most of these religions state that if you are not a member of their religion, you will go to Hell. Since there is more than one of these religions and since people do not belong to more than one religion, we can project that all souls go to Hell.

With birth and death rates as they are, we can expect the number of souls in Hell to increase exponentially. Now, we look at the rate of change of the volume in Hell because Boyle's Law states that in order for the temperature and pressure in Hell to stay the same, the volume of Hell has to expand proportionately as souls are added.

This gives two possibilities:

1. If Hell is expanding at a slower rate than the rate at which souls enter Hell, then the temperature and pressure in Hell will increase until all Hell breaks loose.

2. If Hell is expanding at a rate faster than the increase of souls in Hell, then the temperature and pressure will drop until Hell freezes over.

So which is it?

If we accept the postulate given to me by Teresa during my Freshman year that, "it will be a cold day in Hell before I sleep with you, and take into account the fact that I slept with her last night, then number 2 must be true, and thus I am sure that Hell is exothermic and has already frozen over.

The corollary of this theory is that since Hell has frozen over, it follows that it is not accepting any more souls and is therefore, extinct...leaving only Heaven thereby proving the existence of a divine being which explains why, last night, Teresa kept shouting
"Oh my God."

THIS STUDENT RECEIVED THE ONLY "A

Enjoy,

Dave

The valve is already 'analog', what you're suggesting is you might need an actuator that responds to an analog signal.

It must be an electric motor actuator. Contact closure driven and 40 seconds stop-to-stop infers that it is. Most pneumatic actuators would have to be slugged down to respond that slowly.

If it is electric, is it spring loaded so the actuator drives either closed or open is neither contact is made (closed)? If it isn not spring loaded and the actuator maintains position when no contact is made, then perhaps you can run the algorithm known as 3 position step control (3PSC)?

3PSC is an open loop algorithm that uses motor-on, CW/CCW timing to establish in memory, where the motor position is, given that it has to initialize or start at a zero position (either full closed or full open). For instance, 10 seconds on CW from closed = 10/40 = 25% open. 5 seconds CCW from the prior point is 5/40 or 12.5% subtracted from the prior position = 12.5% open.

This works fairly well as long as you have some reset/integral factor in your PID to adjust for motor deviations from reality, over time. The Europeans came up with the idea some years ago and some function block based controllers offer a 3PSC function block, but the logic behind it is pretty clear.

Dan

Hello Dan,

You got it right about the electric valve and my suggestion is to change it to one with an actuator that accepts a 4-20mA signal. The valve is not spring loaded and stays where you last commanded it. In the TPO I used, when it is not commanded to open, it is commanded to close.

What are the criteria used to select 3PSC with contact closure over an analog actuated valve? Are there advantages one has over the other? The first valve (3PSC) was a spare onsite and costs 4.5 times a new analog actuated replacement. I still prefer the analog actuated valve as it is quite simple to scale the %OUT of the PIDs to a 4 - 20 mA output.

Hello Dave,

Good Story.

The valve revisited

You can still use this valve to achieve the same effect.

Split the PID output signal at 50% and use it as a duplex heat/cool controller.

Set up a timer that will "nudge" the valve open or closed in increments.

When the PID output is above 50% pulse it open, below 50% pulse it closed. According to the process you may can put a little deadband in the PID output split so that its not constantly adjusting the valve.

Dave

Dave,

Your assessment on 4-20m is dead nuts on. I would STRONGLY agree with obtaining an actuator that can take 4-20mA directly. Then your cascade loop output connects directly to the actuator; it's a clean design and understandable.

The strongest argument for getting a 4-20mA input motor actuator is that your PID block with time proportioning output will NOT work. That configuration provides only one on-off output, and the existing actuator requires TWO on-off outputs; one to drive the valve closed, the other to drive the valve open. Single output PID will either continue to drive the motor to its stop, or oscillate continually between the on and off modes, giving sawtooth results with the diverting valve setup you described.

Actuators that use 4-20mA use internal feedback from actuator to determine whether the actuator is at the position called for by the input. That feedback makes 4-20 inherently more "accurate" than 3PSC, which is open loop (no feedback). That's why 3PSC needs some reset/integral factor in order to compensate for accumulated misposition error over time.

3PSC was invented by the Europeans (in the late '80's?)to eliminate the wiring and maintainence associated with "slidewire" feedback control, which was used before 4-20mA became the defacto standard.

I've used 3PSC sucessfully in installations to eliminate slidewire problems, but then again, I had preprogrammed, drag & drop 3PSC function blocks to use.

The typical practice today for new installations is to use 4-20mA. The feedback element used by the input circuit in new electric actuators is rarely a mechanical slideware anymore, it's a Hall effect transducer that doesn't have the wear problems of slidewires.

But if you have to invent 3PSC from raw logic (because your PLC-5 doesn't have it, I don't know whether it does or not) you'll eat up hours on programming/testing versus the relatively small cost of a replacement actuator.

Dan

Whoops, I got people mixed up on that last post, went to correct and hit reply by mistake.

Derek, good idea on 4-20mA.

Dave, heat /cool is good idea for using a single output, but it still requires the logic to establish the timing from current position to next position, as you suggest.

Dan

No Timing Required

If it actually is a motorized valve that has an open signal and a seperate close signal to operate then you don't have to worry about position.

If more heat is required fire the open pulse bumping the valve open. If less bump it closed.

You dont really care where the valve is but rather it is delivering the heat quantity you desire which is detected in the product temp.

Dave

Thanks Dan and Dave,

It appears that Dan agrees with me while Dave agrees with the plumbers. I am thinking both scenarios can be made to work, but the 3PSC method adds complexity. The time base of the PIDs and the time base of the 3PSC would need to be determined.

There is lots of older Finnish equipment at this site, so maybe the site simply has not seen an alternate method of control and that is why they opposed it. I had a peek at the PLC logic that controls a similar system while onsite and it was quite extensive. There were about 75 rungs to produce the CV and about 30 to convert the CV into the "nudging" control described by Dave.

Yeah it will work.......

I've used that scenario in refrigeration systems to regulate ammonia flow thru the evaporators to regulate temperature in an area.

Made the temp controls alot cheaper by not having to supply 24vdc and using cheaper 120vac motorized ball valves instead of control valves.

Dave