A little PID Guidance If You Would

I have a environmental storage cabinet that my employer bought off the used market. It is controlled for temperature and relative humidity. Two West 6100 controllers are used. One for temp and one for RH. The temp is single output to a heater, and I have that looking pretty good. The RH controller uses dual output. One on a small electric steam generator, and the other on a compressor solenoid feeding an evaporator. The steam genny is basically a tea kettle, no control valve, just cycles the heater bar, so the response is a bit slow, and by its nature tends to overshoot. I not much on PID control, but have educated myself quite a bit in the last couple weeks. I have the RH control pretty close, but cant quite get it consistently within the 10% RH limits. I'm not quite using the WAG method, but a true PID expert would probably say I'm not far from it. I've used a mix of the controller manual and the Good Gain Method. As I say I'm pretty close, but not quite. My main problem is overshoot on the high side of the setpoint. I started at 0% P band overlap, and saw a regular overshoot period. This AM I put in a 10% OL. Still waiting on whether that helps or not. Anyway short story long, I'm no PID expert, so any guidance folks can spare the time for is appreciated. I have attached a jpeg of a typical multiple hour output. You can see the periodic overshoot I am referring to.

Hi
I would use the auto tune function ... I used these West instruments a few yrs ago and found the auto tune was pretty good

What was your setpoint, 60 or 61% RH?

Either way, it appears that your max (66) and min (57.5) are both within about 10% of your setpoint (0.10 x 60 = +/- 6% error band). Basically you have an On/Off controller, which means that your humidity control outputs are either 100% ON, or 0% OFF. A PID function does not help much with on/off controllers.

You might try slowly reducing the P or gain, and slightly increasing the I (integral) function parameter.

The auto tune seems to do a pretty good job on the single output controller. I ran pre tune and auto tune on the temp controller and it is running dead on, but it can't seem to dial in the dual output unit. Proportional control is accomplished by on/off percent of cycle time. I have the steam genny CT set short and the evaporatorator CT set long in an attempt to reduce overshoot. I misstated my limitations. It's plus minus 5 % from SP of 60%.

Are you using a true PID controller with proportional, integral, and deriviative settable parameters? It sounds like it might have only the "P" part.

There are 6-7 oscillating cycles (of approx 2 minutes duration) within each of the 15 minute major time interval on the graph. Most cycles look to be about 50% output for each 2 minute cycle time, one minute on, one minute off.

My question is, what consumes the periodic (each cycle) increase in humidity to cause the RH to drop so suddenly?

The steam generator comes on, puts moisture into the air, the RH measurement senses the increase, the steam generator turns off, the humidity decreases. Why does the humidity decrease? Where do the water molecules go?

Heat, being energy, transfers through the product and walls of the chamber.

Water molecules are mass. Where do they disappear to? They can't escape a sealed chamber . . . .

The load must be like a sponge, sucking up moisture from the air, which is now my operating assumption; the load is absorbing the moisture.

I also assume that the ~2 minute cycle duration reflects the time duration that the steam release valve is open. The steam valve opens for about 1 minute as the humidity rises, then the valve closes and the humidity drops for a minute, until the next cycle starts.

Since the rising slope (adding moisture) and falling slope (losing moisture) are about the same, I would conclude that if the cycle-time interval is shortened, say halved, then the RH level will not rise to the same (Y) amplitude/height, because the mass release of moisture over half the time (30 seconds instead of 60 seconds) is only half as much. The moisture gets consumed/absorbed for 30 seconds and another small shot of moisture is introduced 30 seconds later.

I'd halve the cycle time and see if the overall magnitude of oscillation doesn't drop down.

I am using all three. I got P and I from observing PV in manual mode. I roughed out D as I/6. This gave me a pretty good control except I got periodic overshoot. Reducing the P value by a factor of .8 reduced the overshoot some and a second reduction at a .8 factor improved it more. Adding some P band overlap has helped as well but I just can't completely eliminate the occasional overshoot.

Dual control. Output 2 drives a solenoid that cycles refrigerant through an evaporator. Dries the air and condensation drains out of the unit. As I say the response is slow. Off the top of my head I think the heater CT is less than a second and the evap valve is around 8 seconds.

>a compressor solenoid feeding an evaporator.

As in a dehumidifier/air conditioner evaporator?

Is the steam generator in a constant battle with a dehumidifier to regulate the humidity?

You got it. Not to mention the second process controller independently controlling temperature.

In normal heat/cool control applications, there's a call for heat below setpoint, a gap (not an overlap) between a call for heat and a call for cooling, and then a call for cooling above the setpoint.

One never heats and cools at the same time.

I've never actually implemented humidity control, but I would work on the same assumption as heat/cool control - one either humdifies, or one de-humidifies, one doesn't do both at the same time.

When you mention overlap, does that mean that you can have steam and evaporator cooling (dehumidifying) action at the same time?

If so, go in the other direction. Get a negative overlap or, in effect, a gap.

Your RH sensor is probably capacitive technology RH capacitive needs some time to equilibrate, that is, to sense the actual humidity.

If it were me, I'd go into manual, give it shot of humidity for some period, the watch and see what the response is. If you're not dehumidifying at the same time, the RH is likely to increase over a time constant and then hold, maybe decay gradually if the load absorbs humidity. That should give you some idea of how much humidification is likely needed.

It sounds like your temperature control is now pretty stable. It has to be stable, because RH is temperature dependent. If the temp changes, the RH will inversely follow because that's the nature of RH, the ability of air to hold a quantity of moisture at a certain temperature.

I thought about that. I did try 0% overlap, but I haven't tried a deadband. Maybe a smaller P band oron output 1 compared to output 2, so it shuts off sooner along with deadband.

As an fyi, during early shake down the compressor shut down for a while and that is exactly what happened. RH went up quickly, leveled off and slowly decayed.

If you're testing at elevated humidity, then one would not expect to use the dehumidifier because the job is to get moisture up to an elevated level and then hold it there.

Once the RH is at setpoint, the need for additional moisture is the rate which moisture is lost when the door opens, or lost as the load absorbs it.

The 'hold SP control' task is to add small amounts of moisture on a periodic basis in order to hold the elevated level.

I assume you've got the controller in primary and secondary (heat/cool) mode:


It isn't clear whether overlap or deadband is positive or negative, but you'll figure it out. Just get some deadband to turn the dehumidifier off.

Duh! I can't believe this didn't occur to me. I got so hung up on the dual controls moving the PV back and forth over small range I lost sight of what I was actually dealing with. Once the RH is at SP the natural tendency would be for a slow drop off. Dehumidifcation would only be needed to damp the overshoot. After that it should be rarely needed. Dead band is a negative value fyi. I can't wait until Monday morning to try it. I might have to go in Sunday and change the setting. Thank you for having the wisdom to see the forest despite the trees.