StillDragon® Community Forum


Be part of our community & join our international next generation forum now!

In this Discussion

Recirculating Coolant Flow Design - Feedback requested



  • edited February 2015

    @CothermanDistilling - Your last video had about 137/138F coming off both condensers, what's your average GPM from each at that temp?

  • Just scanning back, looks like about 0.5-2gpm

    If you can do 2gpm at 140F - That's good enough for a nice sized hydrionic blower - Something like Modine or similar - prepackaged, in an enclosure, with a fan. Just plumb it up and enjoy the free heat.

  • I think the last thing he needs in Florida is heat ;)

  • I got a big smile reading that!

    I am thinking a radiator outside with a fan...... I wonder if a gutted residential AC outdoor unit could provide the flow I want if I ran water through the refrigerant lines... aesthetic, quiet, easy...

  • Nah, pump will choke to death on that.

    Easier to keep an eye on the surplus market to see if you can find a good deal on a dry cooler. Some of the smaller dry coolers look just like central AC units.

  • dry cooler.... let me google..... ahh... I have a friend that takes a lot of these down for scrap, small to really big, like ten thousand ton ones... will see if he has anything my size, I need to look at my cooling water spreadsheet to figure out BTUs to get a couple gpm down 80 degreesF...

    again... google:


    Incoming water 150F outgoing water 80F 150 - 80 = DELTA T 70F
    BTU's for 1 gallon of water is 70 (delta T) * 8.34 (pounds per gallon) = 583.8 BTU's
    Flow Rate 2 Gallons per Minute
    2 Gallons * 60 minutes = 120 Gallons per hour
    120 Gallons * 583.8 BTU's per Gallon = 70,056 BTU's per Hour

    so this is 6 tons, I am running about 1/4th of that normally, so if I get it cheap, something around a 2-5 ton unit should give a lot of research potential

    maybe get a scrap mini-split housing and replace the coil with a automotive radiator?

  • edited February 2015

    You can find hot water heating units cheap, perhaps you have an eave on the side or back of the building you could put it under?

    There are Modine 150kbtu units that are under $500 on ebay, but they don't have exterior enclosures.

    Or something like this?

    141k BTU Air Handler Outdoor Furnace Heat Exchanger @ eBay

  • interesting, never seen anything like that before!

    Of course, I saw this at the bottom of the page... put it in conjuction with an exhaust fan and have a way to dump a lot of heat before returning to the tank or the chiller...

    18x18 Water to Air Heat Exchanger Hot Water Coil Outdoor Wood Furnace @ eBay


    333 x 400 - 27K
  • Adding 20% headroom puts you closer to 7 ton. In my case probably double that, 15 ton. Those are big chillers, but even those numbers seem somewhat light.

    On second thought, those small heat exchangers will help, but there is no way they'll be making a big dent, unless you can get the ambient air temperature down significantly.

  • edited August 2015

    Cleaning up the plumbing. All 1.25" sch 80 cpvc.


    800 x 600 - 55K
  • So the chilled line pops through the wall and feed the two loops through your three ways?
    What do the ball valve do? (I'm assuming the three ways divert?)
    The red and green thing are little recirc pumps?
    The line that wraps around the tank goes to the chiller? What comes out of the tank?
    What's sitting on the Deph? (cat?) Do you think there'll be much preferential flow coming from one side? Looks like a nice tidy layout. Congrats. :-bd

  • edited August 2015

    Chiller line not plumbed in yet. There is a second tank (cistern) buried underground that serves as a feed to this tank.

    Ball Valves near the 3-way mixing valves are balancing valves. Mixing valves are Johnson (blue) and Belimo (orange) - both are 1.25".

    Red thing is a Grundfos Alpha and Green is a Taco 0015 Circulator. Need to puck up another Alpha, the flowrate (GPM) display is great.

    This is plumbed how I originally proposed, in a constant flowrate/volume arrangement. The pumps always circulate coolant at full speed through the condensers, the 3 ways control the temp of the loops. My theory is that this the higher flow rates should increase velocity/turbulence and increase the capacity of the heat exchangers.

    The pipe off of the right of the tank return line is to bypass the tank and dump the coolant in a single-pass arrangement. In this arrangement the top tank is kept full by pumping water up from the cistern. If the cistern is too warm, I'll run the top tank with chiller, when the cistern is cool, the tanks will run piggyback.

    On top of the dephl. is the "cat". The cat is sitting up on legs above the dephl. output pipes, with the outer edge somewhat sealed, vapor will run straight up and through or the most part. It's 1/2" tubing, there is little restriction.

    It's hell to work up there, that is on top of the boiler room, that floor is 10 feet off the deck. I'm on a ladder to snap the photo.

    Should be interesting to see how the higher tank works, there isn't nearly any head, maybe 6 feet or so.


    800 x 600 - 65K
  • @grim said: Should be interesting to see how the higher tank works, there isn't nearly any head, maybe 6 feet or so.

    ?? Aren't the high points the returns? You must have a supply pump in there somewhere?

    With the high flow rates your loops temps are going to up near the vapor temp too. At the end of a run it could potentially be +95°C. Is the CPVC cool with that at low pressure?

  • Ok, so...

    Recirculating design/mixed flow/constant flow rate works great for the product condenser, ever time we run we are shocked at the knockdown power of the 4"x1m.

    ...doesn't work so great for the reflux condenser though. Can't seem to get the gremlins out of the PID setup. Can't get rid of the oscillation, and once or twice an hour the loop seems to go just wacky and swing +-10F with no apparent reason. I have a feeling I'm getting pulled around by the fact that this is a very fast responding system, but the control valve is slow (30 seconds full swing), and the lag time between valve change and temp sensor change is pretty long. The RC volume is pretty damn big, nearly 5 gallons, and we have a good amount of 1.25" piping between the valve and the RC (see above). Depending on the power input, it can take nearly 2 minutes to register a change if I manually step the valve. I'm basing most of this on the fact that if I slow the recirc pump down to about 5-8gpm, we get larger oscillation, pushing it to 15gpm seems to help.

    So, we're moving the valve closer to the RC, and moving the pump outside the loop, which would be the traditional 3-way mixing plumbing. I know there are some other strategies to manage loop dead time, but after pulling my hair out trying to retune a few times, it just seemed easier to try to minimize the dead time.

    Not sure why the PC seems to work well, other than the fact that the valve position is generally more open on the PC (less recirculation). The RC sticks near the full recirc/bypass, so perhaps I'm also dealing with some additional non-linearity issues.

  • you have the PID probe inside the RC, right?

    yes, I agree, put the valve closer, as in actually on the RC if possible (although at 5gpm, you have completely replaced the water in the RC in a minute, slightly more with piping...

    Random thoughts include slowing down the PID by reducing the effect of P and I..

  • edited November 2015

    We're on the same page, for sure.

    Probe is inside the RC, we moved the valve as close to the RC as possible yesterday - basically sitting on the lateral copper coolant and return lines from the wall to the dephleg, about 15" away from the input/output.

    Old controls guy told me my problem is partially due to integral wind-up caused by the lag time of the plumbing (transport time) and valve (response time). Initially, holding it holds full reflux is livable, it's when we start dropping the RC temp do we start to see the problem get worse. I didn't get a chance to run with the auto-reset/anti-windup turned on, I'm going to do that on Tuesday, along with trying the new valve location.

    Reducing the proportional gain (increasing the proportional band on my controller), actually increases the amplitude of the oscillation, as the RC temp actually increases faster than the valve can chase it, this causes HUGE oscillations, like +-30F. I would have thought the sweet spot is between the two, but I end up with less overshoot and shorter oscillation period, neither go away (or I haven't sat around long enough to see if it hones in on the setpoint).

    Also, the probe in the RC makes sense in the traditional plumbing arrangement, however in this setup, where the flow rate in the loop is fixed, the temp probe could have easily just been immediately after the valve. By nature of the loop, while the temp reading would have been lower than in the RC, it would have provided the same result with less lag time. I didn't have an easy way to test this, so I didn't bother trying.

    I'm not new to PID, I've used these all over the place for years, but this one is a bug-bear. I've dealt with large dead time processes before (temperature control in a large aquarium), however, what's different here is the slope of the process change (if that makes any sense) is wickedly steep. Seems that everything here is "nothing, nothing, nothing, JUMP!".

    Also, maybe one of our resident process engineers could weigh in - could very well my valve Cv is too large, that that's what's causing the instability.

  • Crossing fingers that valve relocation works...

    Just verifying, you integration variable is the typical and not the inverse? I have seen that before... change it an the thing goes the wrong way... whatever way slows it down should result in less overshoot, but you have experience, so you know the three way balancing act...

    PID controller - Manual tuning @ Wikipedia

  • edited November 2015


    800 x 600 - 85K
  • edited November 2015


    800 x 600 - 72K
  • edited November 2015

    Yeah used compression unions so we can swap that valve out if necessary.

    If there was some way to introduce the deadband delay, I could probably nail it - which would mean keeping the gain more aggressive than it is now.

    That animation is pretty sharp, wondering if I should have tried pushing the derivative higher (along with the anti-windup). But, the valve is moved now.

    We've got a batch fermenting now, planning to run on Wednesday before T-day.

  • edited November 2015

    +-0.5F in the new position, with just a bit of tuning, how can you not be happy with that? I can probably get it tighter.

    The crazy waver not due to valve position after all (or valve cV), looks like we have a problem with the pump intermittently stopping and starting again after the PID starts opening the valve to compensate for the temp rise. This is sending a slug of cold water in, collapsing the temp, and throwing the PID into fits and oscillation. So, not a PID problem after all.

    What a long day to figure that out... May mean that moving the valve wasn't necessary (although its MUCH easier to tune in the new position).

    At least I'm not losing my mind, that's always nice to know.

  • Btw - it was the fancy low wattage circulator with the problem.

  • Don't you love chasing the bug through a whole system only to find at the end something silly?!

    Seems to happen to me all the time...

  • Yep.....check the fuel level first :)

    StillDragon North America - Your StillDragon® Distributor for North America

  • While we were at it, we split the loops.

    The previous arrangement had a shared header on the pump inlets, and merged the loop outputs to a single set of piping before either recirculating or dumping. We're strongly suspecting that the loops were being coupled by the inlet friction or maybe some kind of venturi or suction on the outputs.

Sign In or Register to comment.