More experimentation with mixing valve

[velvetfoot]

I'm thinking of doing another experiment with my electric 3 way mixing valve that controls flow through/around my buffer tank.

I currently control the return temp to the boiler with the 3 way mixing valve. It works pretty okay, but sometimes it oscillates, maybe due to the Cv of the valve, cold water of the return at times, and not superfast response of the actuator. Integral valve is currently 60 seconds which seems to limit the oscillation but allows a 5 degree overshoot.

What I'm thinking of is trying a sensor at boiler/tank supply instead of return. There might a slower response to changing return temps so it might (or might not) be good.

I can see a bonus for those who heat with baseboard convectors. Use full temp when boiler is on, and use an outdoor reset controlled supply temp for tank heating.

Any opinions?

 

[Nofossil]

I'm not sure I understand how sensing at the supply would allow the valve to manage inlet temp. On my system I can have really hot supply temps and a really cold return temp (when I first start heating storage, for instance). I can also have a relatively low supply temp and a relatively high return at the end of a fire when the boiler is burning out and storage is really hot.

Most active control loops have the possibility of either too slow a response or oscillation (which is pretty much an indication of too fast a response), or both. There's a whole arcane body of knowledge about the math behind this phenomenon. I highly recommend it if you have difficulty falling asleep.

I use a PID loop and I've tuned is as best I can. It will overshoot and oscillate a little bit, but not enough to be an actual problem.

Here's my last fire. I use a circulator (Grundfos 15-58) as inlet protection. The speed signal is labeled 'Wood Boiler Recirc'. I'm aiming for 145 degrees at the inlet. You can see that most of the time it's either running at full speed or it's off.

 

[TCaldwell]

I'd try adjusting the pid parameters first, is the p a gain or a band.

 

[velvetfoot]

Thanks for the link. I may already have tried to read that. Only things I can change on my Honeywell are integral and derivative times. I've kept derivative at zero.

I'm not sure I understand how sensing at the supply would allow the valve to manage inlet temp.

My concept is to change things up and control the supply temp via changing the return temp. Since the boiler provides some hysterisis , if that's the right word, or delay, perhaps the not-fast-operating motor operated mix valve could control the supply temp better.

Just an idea, and easily tried. I'll try it on the next heat up (oil).

My further idea is this. Some people are devotees of running low temperature in their baseboard emitters ( hi JohnDolz ), and mix the water from a big storage tank (mine is only 120 gal, so it's probably moot for me anyway). I thought I could try this by placing the supply temperature where in the supply header where it would see both boiler and tank hot flows. If the boiler was off, the supply temperature (from tank) could be made lower (reset would be ideal) and the zones would run for longer times. I've never seen the benefit with that since there's the same amount of heat in the tank, but I guess it would creak less at a minimum. As I said, there do seem to be devotees to this idea.

Who knows how it will work when the modulating pellet boiler. If it works with the oil boiler, maybe I'll find out.

If the boiler is on, I'd like to remove as much heat from that as opposed to sending it to the tank, so I'd set a high supply temperature while it's running. Again, what I'm looking at now is an oil boiler that charges the buffer tank to 180 or so. Still, with a wood boiler running baseboards, I'd think you'd want to do the same.

I think the thing is, with one type of emitters in the structure, and with the right controls, only one mix valve, serving double duty, would be needed.

Again, only an experiment. Easily reversible.

 

[TCaldwell]

Try adding 25pct time to the integral

 

[brack86svo]

View attachment 187507

Will the graphing you are doing work with the Controlbyweb X300?

 

[Nofossil]

Experiments are good.

In my thinking, low-temp baseboard operation is better because you can return colder water to storage.

My boiler has a LOT of lag time from a change on the input to a change at the output. I guess, knowing the capacity of the boiler in gallons, you could figure out your flow rate that way. You'd have to dial back your integral gain quite a bit in that scenario.

 

[Nofossil]

Will the graphing you are doing work with the Controlbyweb X300?

Dunno. We tried talking with them a few years back and had no luck. We decided to go open source with our controller and make our data available with an open API, but most people aren't doing that. If they have a published API to get data out of their unit then we could graph it.

 

[velvetfoot]

Try adding 25pct time to the integral

My controller is in seconds. It was considerably lower than the 60 seconds I have now-the larger valve does seem to help with the oscillations.

 

[TCaldwell]

Typically there is a proportional band or gain, integral and derivative that would comprise a standard 3 term controller. If your I term is in seconds , 25 pct of 60sec would be 15. 15 plus 60=75, that's a start to see what happens

 

[velvetfoot]

The experiment failed. Restored to original.
Trying Integral =80 seconds. Seems okay too, but no zones turning on, so steady state with warming of tank. Now reading 62% recirc, at the target return temp of 155 , tank top =166, bottom=111. Recirc percentage goes down as tank return gets warmer.

 

[TCaldwell]

Will need a disturbance such as zones on and off to test but try reducing the integral from your 60 sec?

 

[velvetfoot]

Will keep it at 60 for a while and watch behavior. Thanks.