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 in Configuration

So here is the coolant plumbing design we've settled on for the big dragon, this is what we'll implement when we plumb her up in the next two weeks or so. Finally got the construction permits finalized, so now we're going to rock and roll. Here is what we've got:


What I wanted to run by the crew here is the plumbing design for the condensers. This is not a typical plumbing design that you see here, but it's one that we think may have some benefits. This is a constant flow rate setup, it's very common in process plumbing. No matter what the set point, the flow rate through the condensers will be the same, and will likely be very high compared to designs that slow the flow rate at set point.

My theory is that this design will essentially eliminate the temperature gradient in the condensers, or at least significantly reduce it. Being able to push a higher flow rate, without cooling lower than the set point, will make the condensers perform more efficiently (you can keep a larger percentage of the condenser surface area cooler, greater overall delta t). The other benefit here is that for the RC, you lower the risk of sub cooling the distillate, since the input coolant temperature will be warmer than the reservoir temperature (unless of course you are hitting the limits of the design).

Both the PC and the RC will have the same plumbing design, but they will be completely independent, and each will have their own pump, mixing valve, and controller.

The chiller will also have it's own pump and controller, separate from the still condenser controllers, to pre-cool the tank prior to a run, and to try to keep the coolant temp stable through the run.

We're going to try to get by with a coolant tank of 500 gallons, since we've got one. I have a line on a monster tank, 2000 gallons, that's free, but the thing is just too damn big.

Chiller is 3hp, we think it should be sufficient.

We may add a liquid-to-air heat exchanger on the return lines to the coolant reservoir. This would be more useful in the winter than the summer, but we should be able to bleed off a some heat prior to returning the coolant. Might consider just putting it on the roof, and just not needing to run the chiller for 5 months out of the year.

Thoughts? Comments? I did a small trial run of the condenser plumbing design, and it did work just fine. I ran it on a PC and swore I could push it harder than usual, but I generally don't record coolant temps, so I have no idea if it was just because of a lower coolant temp.



  • A two way on leg A would do the same and be cheaper. A check valve might be a good idea too.

    If your running multiple items off the tank or its far away, then a low flow, high presure pump on a separate feed feed line would be a good idea. If you were to do that then you could do away with the purple orifice too which would increases your recirc pump efficiency a bit.

    That's my 2 cent anywho...

  • Problem I ran into when testing the two way valves is depending on the cooling demand, and the coolant tank temperature, you could run into a situation where the pump was nearly dead headed against the nearly closed valve restriction. With no bypass, the pump wasn't too happy. When I tried to rig up a passive bypass (tee and a valve), I ran into a situation where the flow was no longer linear and the control range suffered (at 25% valve opening, all flow was through the bypass). You would need to adjust the bypass restriction depending on where you were in the run, and then the pid would overshoot as you changed the game on it. What's the point of a controller if you need to fiddle with the bypass valve? I could have run it without bypass, but there is always the question of how much additional wear and tear am I putting on the pump. Don't get me wrong, it worked, just didn't seem ideal.

    Running the two-way setup on city water was no problem at all, with high supply pressures and no ill effects due to the restriction it was great.

    I agree on the checks - my pumps are Taco 0015 3-speed circulation pumps, they've got integral check valves.

    Now, I've only plumbed this on the floor with shorter lengths of plumbing, so I'm not 100% sure the pumps will have enough flow at higher head pressures. The fact that I'm running two pumps gives me a little bit more headroom I think. I went with the Taco because I had a spare from plumbing the radiant heat floor in my house, and I thought the 3-speed would give me a little bit more flexibility to dial in system, it's not a huge range, but it's a little bit of insurance. Plumbing is all 1" PEX, trying to keep the flow restriction to a minimum. I suspect that pressure drop across the shell side of the condensers is close to nil.

  • Not sure your picking up what I'm putting down. The recirc pump is on the loop. The control valve and check are in between the loop and the cooling system. A closed control valve wont effect the recirc loop flow rate at all. It just wont be bleeding off any of the warm water in the loop or allowing fresh cool water in. I can't seem to google to the term but I've heard them referred to as an attemperation loops or something like that anyway. If you want to minimize delta T then they are the go. I've got some old P&IDs somewhere that I might be able to dig up. Commonly used in pasteurizers and such. Someone will know what I'm on about. You could probable do away with that tank with this system too (or reclaim it).

  • Pump in the loop is the point, not an error.

    With the pump in the loop, the flow rate in the loop stays the same, regardless of valve position. Utilizing the full capacity of the pump, flowrate stays high all the time. My hope is that this results in high turbulence in the condenser and minimizing temp gradient.

    When controller calls for maximal cooling, B closes, A is full open - we have full flow to and from the reservoir. When the controller calls for no cooling, A closes and B is fully open - the pump will only recirculate the condenser loop, no exchange with the coolant tank.

    The balancing valve on the B input line is to ensure a linear response in the range from full A to full B.

  • edited February 2015

    +1 @grim

    Dustin ended up installing a pumpIin the loop for this very reason.......after not having a pump in the loop initially.

    StillDragon North America - Your StillDragon® Distributor for North America

  • @grim, what are you using for the three way mixing valve, and how are you controlling it? Not sure if you mentioned it elsewhere, but can a standard PID handle controlling temps with the variable of a three way valve?

  • edited February 2015

    Omega CNi16D53-EIT Controller

    Johnson M9206-GGA-2S with 3 way valve

    Analog Control 0-10v/0-20ma

    You don't need a controller as fancy as the CNi16, it's got a lot of bells and whistles (ethernet connectivity and web interface). As long as it has an analog output, you just need to make sure it can do 0-10v or 0-20ma proportional control, depending on your valve.

  • edited February 2015

    M9206 is the big brother to the M9106 discussed here in other threads. Not necessary unless you you need to run valves larger than 1". It's also a bit slower than the M9106.

  • edited February 2015

    I still don't think you get what I'm saying. I'm too busy to draw a pic right now, maybe later or someone else can explain better. Keep everything how you've got it drawn, just get rid of the three way and make it a tee. Now put a two way on leg A and it should achieve the same thing. It's a bit hard to visualise the flow but it'll work. You don't need to 'divert' the flow with a three way, the pressure drop over the balancing valve will send water back to the tank when the control valve opens.

    In addition, if you get control over the feed line pressure (A) you can get rid of the balancing valve too.

    Your concerns in the first response just come down to design and getting the pressures right. They can be solved by good engineering.

  • edited February 2015

    @jacksonbrown This is what you are looking for, its called mixed flow. I posted the various types of piping, pump placement and control valve strategies on another thread ages ago. Seems to have fallen off the radar.


    Visio-Dephlegmator Valve Piping Mixed flow.png
    798 x 331 - 16K
  • edited February 2015

    @Mickiboi - yeah that's the layout, only real difference is the balancing valve on leg b. Forgot about that thread.

  • edited February 2015

    err, no. That looks just like a mirror of the original without the balancing valve to make it work properly (a simple orifice or reduction in pipe size on leg B would do it). Depending how it all physically sits, you may be asking a bit much from that pump too, suction wise. NPSH and all that. MechWarrior seems to talk the talk maybe he has some input.

  • I like the idea... and the last drawing above, will be watching... thoughts about bleeding air?

  • edited February 2015

    Why not something super simple like this:


    Use a simple on/off solenoid valve to send water back to the tank when it's too hot. This set up will prevent the pump from NPSH issues

    Single Pump Temp Control.jpg
    800 x 305 - 17K
  • with that setup, you can never send 100% cool water to the dephleg.

    Also, if the valve is closed, you still get way more reflux than with the pump off, and with the pump off, the sensor will not get warm...

  • edited February 2015

    There is only one other way to plumb a 3-way mixing valve in this application, which is:


    In this approach, the flow rate through the condenser will be variable based on cooling demand.

    587 x 406 - 12K
  • edited February 2015

    In my particular application, the pumps and valves will be located approximately 15 feet from the still. Because they are not explosion proof, they can not be in the classified area. So, in both my drawings, simply extend the plumbing that is directly attached to the condenser to get a more realistic view.

    Because of the distance from the bypass to the condenser, and corresponding time delay, the approach that keeps the flow rate through the condenser loop higher should reduce the lag time from a control perspective.

  • @CothermanDistilling said: with that setup, you can never send 100% cool water to the dephleg.

    Also, if the valve is closed, you still get way more reflux than with the pump off, and with the pump off, the sensor will not get warm...

    Why would you want to? (send 100%) and if the pump fails you've got bigger problem than a cool sensor!! With any critical system like this, a flow switch backed up with a check valve should always be included.

    Is this an extreme budget, commercial operation or an over engineered garage one?

    Not sure why you would put out a call for feedback/opinions when you've already made your mind up that yours is the correct way forward. :-?

    So the loop is extended to bring the valving and pump past the tank (and away from the still)?

  • @jacksonbrown said: So the loop is extended to bring the valving and pump past the tank (and away from the still)?

    Grim was the OP who was looking for opinions. I think Cotherman already has his system setup.

  • I'm just trying to find the best approach, not afraid to try something a little bit different, but don't want to waste my time on an approach that's got downsides that I'm not seeing.

    Commercial op, but the approach should work either way.

  • Just looked through five or six P&IDs of cooling loops and they were similar to how I suggested, usually with the control valve on the return line. Most with a check on the return too. I assume to prevent back-flow from a return line pressure spike. Some also had check valves in the loop in between the supply and return lines but that might not be necessary. A few had an auto stop valve in this position as well to allow a full purge as Coth requested but they were combined with a VSD on the recirc pump so the flow is still being controlled. Non had threeways. They would all have had proper supply pressure which you don't seem too.

    With no dedicated supply pump and a convoluted return path, the threeway might be a necessity here. Not sure. Relying on pump suction may be the downfall. You need to walk the path and work out your pressures, otherwise it could just be a blind (and expensive) stab in the dark. When things aren't sized correctly like pumps you just end up throwing good money after bad. Do it right, do it once.

    If you're running multiple loads off this tank you should have one pressure controlled supply line. You might even be able to take off between the red pump and the chiller to achieve this.

  • I don't think NPSH will be an issue here, we are talking 300 watt circulation pumps, not 120kW 250 litre a second condenser water pumps. Can't see a little Grundfos circ pump cavitating. :D

  • @jacksonbrown said: Why would you want to? (send 100%) and if the pump fails you've got bigger problem than a cool sensor!! With any critical system like this, a flow switch backed up with a check valve should always be included.

    I use 100% full flow to purge air from the Dephleg and PC

    regarding pump failure, even better reason to have the sensor in the Dephleg!!!

    like @floridacracker said, I am not the OP, but I have quite a few videos of my cooling system for my 380L w/ 8" CD on this forum...

    also, my personal thought on check valves is that they restrict flow more than anything in a system that is not under pressure... if you had your cooling tank and everything else under pressure but that is just my two cents. (I get 3.9GPM measured using a 809/815 march pump or my other ignition proof condensate pump, but I usually see the flow at .5gpm during a spirits run. This is an inexpensive and fairly quiet pump.)

  • I would use a manual ball to purge. Adjusting the balancing valve to achieve this would screw up your pid settings of you didn't get it back to the exact same setting every time. You are right that under sized pumps are very limiting on what you can do.

  • @grim Have you thought about a Primary/Secondary system both with pumps? It's commonly used in cooling systems in HVAC.

  • Yeah I replumbed my boiler from zoned to primary/secondary because added a number of radiant and hot air loops to an existing zoned system. Learned more about "closely spaced tees" and hydraulic separation than I'd ever cared to.

    I think the coolant tank negates the need for a primary/secondary approach. It would work, though - I could move the coolant tank and chiller even further away if necessary, and put the condenser loops closer to the still (but still out of the classified area). So might be some benefit there.

    Can't really use the chiller loop for this purpose. The chiller pump is sized to provide the optimal flow rate for the chiller, I don't want to push it too high or too low and risk losing efficiency, since the chiller likely isn't big enough.

  • edited February 2015

    My radiant floor loops use the same plumbing design as posted at the top. Because I have hot water baseboards that run at 180F, and the radiant loops that run at about 130F, I needed temperature control on the radiant loops. That along with your post Micki was kind of the inspiration to go this route.

  • edited February 2015

    What about this?


    Dephlegmator Primary Secondary cooling system.jpg
    754 x 532 - 30K
  • edited February 2015

    Or this with the chiller


    Dephlegmator Primary Secondary cooling system with chiller.jpg
    800 x 550 - 29K
  • edited February 2015

    Another local guy uses two large air-water heat exchangers (radiators) with a large fan to cool his return coolant prior to returning to the holding tank. He's not using a chiller at all. He says it works great, and in the winter provides supplemental heat. I believe his setup has the coolant water going to the product condenser first, then tees to the dephlegmator. Suspect his coolant output temps are hot hot hot. Pretty sure he splits the return line and runs the radiators in parallel, not series.

    There are a number of options for integrating the radiator. Recirculating with the tank, on the dephlegmator return line, on the condenser return line, or on the mixed return line.

    I wonder if putting it on the dephlegmator return line is the best option from an efficiency perspective, lower flow rate compared to the product condenser (except during stacking), and a higher output temperature (higher delta T in the radiator).

    Might be less effective on the product condenser, if you are pushing high flow rates.

    Blended return line might be a good middle ground, more cost effective.

    We'd considered putting the radiators on a loop with the tank, only because in the winter, the ambient temperatures in the building are stable around 60F with the heat off. As long as the coolant tank temp is a couple degrees above ambient, we can let it run. This doesn't quite work as well in the summer though when ambient is in the 80s.

Sign In or Register to comment.