Do turbulators actually work?

[Condensifier]

The next project is a condenser that is as light and efficient as I can possibly make it. It will be an offset Nixon and the plan is to engineer it rather than just guessing.

Hi engunear.
A Nixon-Stone still like a boka cooling system is pretty straight forward. Put a decent coil in it and it will act as a reflux and a product condenser at the same time and it's a pretty small condenser. So if your goal is to make a smaller efficient condenser maybe you should focus on one for a pot still If you look around there are some really long liebigs some longer than 5'. We sure could use some more efficiency in that area.

Just a suggestion. Keep up the good work.

[engunear]

It would be really interesting to know how long is the region where condensation is happening is. Do you have a thermocouple setup by any chance?

The math says (and maths can be wrong if the underlying model is wrong which is really what this is about) that a spiral like that makes no difference, that the key parameter is the area of contact between the centre pipe and the water, rather than the path length. So it would be great if you could measure it and see how closely it fits the calculator.

To do that just crank the flow up high, poke the thermocouple up its arse till you hit the hot area, and measure its length. Then measure inlet and outlet water temp. Also need inner pipe diameter. I'll happily run the numbers, though its pretty simple with the website calculator. Is the spiral copper?

One thing that possibly makes this confusing is that since we enter to water to our condensers through pretty skinny pipes we have somewhat turbulent flow from the start, and it doesn't get a chance to settle before it leaves the system. Its turbulent enough to stop layers from forming. So the standard youtube clops of flow with and without a turbulator are a bit irrelevant. A measurement in a worm and tub condenser would be interesting but that seems like a hard thing to set up.

[Condensifier]

It would be really interesting to know how long is the region where condensation is happening is. Do you have a thermocouple setup by any chance?

No sorry and I no longer have the boka as I chopped up to use the copper for other things. It was a 2" and it would have been ideal to check the temp on because you could just drop the temp probe down from the open top. There was no turbulator on it though because I used a coil made from 1/4" tube like most people do and I didn't use a separate product condenser.

[engunear]

OK. I understood your first post to mean that you have a pipe-in-pipe condenser with a coil of metal that forced the water into a spiral. But if its a Boka then its a variant on the coils in the Nixon, Dimroth etc. Yeah, it would have been interesting. But if its recycled then its all moot anyway. Anyone else out there with a Boka and a thermocouple?

The 3X sounds right. Thinking about the geometry, the exact ratio is pi (3.14159 ...) if the gap between the coils is equal to their diameter. 3X is near enough for Government work!

I've made some measurements with a Nixon/Stone and the condensation zone is surprisingly short. I'm gonna repeat before I post. You can check that with your hand by seeing where the jacket is hot. The 3m of coil recommendation that Nixon makes seems way off. Though Nixon condensers tend to be small to start with so it doesn't matter as much.

The other reason it would have been interesting is that it is 1/4 inch pipe. The coils I have are 3/16 pipe - automotive brake line from the UK. I'm getting HTC numbers that are really high. For this diameter pipe the Reynolds number is really high and the dimensions quite small (2.7mm ID), so maybe turbulence is relevant here. A measurement with 1/4 would be another data point. I'm thinking the high HTC numbers, the 3X benefit from length and the fact that a coil condenser has very little water in it at any time leads to very small and light unit.

[engunear]

Error in last post - its 3X if the pipes are pressed together, with no gap, which is what NZChris said.

[NZChris]

Error in last post - its 3X if the pipes are pressed together, with no gap, which is what NZChris said.

I did??

My 'coil' is coaxial cable wrapped around the vapor tube and soldered to it at each end. The outer jacket is forced on over it.

The circumference at the center of the water path is about 2X the length of the condensing section, plus it still has to travel the length of the section, hence my 3X estimate.

[shadylane]

Just realized the turbulator your talking about is used in the water jacket and not inside the vapor part of the condenser.
Guess I need to read and completely understand the post before commenting.

[engunear]

Hi NZChis, I'm just confused. I interpreted what you meant as per the left hand drawing, with water going down the coiled pipe, but if its coax then it is like the right hand drawing where the blue circles are the coax, and the water runs in the white gaps? The LH one has a surface area per unit length equivalent to 3X what you'd get with straight pipe.

If I've still misunderstood maybe you could post a pic? 1000 words and all.

Wishing to save people time rather than being critical, its not clear to me that the RH arrangement is any better than not having the coax because the area in contact with vapour is the same. It may be worse because it cuts area in contact with water. I'm happy to be wrong in this, its understanding I'm after.

[NZChris]

RH drawing.

The vapor contact area is the same.

The velocity of the water is higher because the cross sectional area it travels through is smaller and the same volume has to travel 3X the distance in the same time.

[engunear]

At first I was thinking that the velocity would make no difference, but I have seen hight HTC in 2.7mm ID pipe which is presumably because of velocity. Need data on velocity vs HTC. That is a core question.

Not this weekend, but maybe next I'm planning to make a trial setup as shown in drawing. (I'll buy the parts today.) The vertical pipe will be plastic and the horizontal ... a variety of copper pipes. The boiler will run at 2KW and push steam completely around the short horizontal section. By measuring the input and output water temperatures and flow rate it will be possible to work out how much power the short section is absorbing. By knowing its diameter we can find its surface area and thus the HTC. Flow can be varied for a given diameter and we will have different diameters as well.

[engunear]

Measuring HTC

This summarises the measurements of how HTC changes with flow rate and with pipe diameter.

These experiments were done with 2.7mm ID and 6.5mm ID pipe. (I think its 6.5. Its 8mm OD and I've misplaced my vernier so this is close enough.)

Theory predicts that HTC should be higher in thin pipe compared to thick as the heat has less distance to travel, and the water velocity is higher and so the flow is more turbulent. One also expects HTC to increase with water velocity for the same diameter pipe as the turbulence increases.

See the photo. A piece of copper pipe was passed through a plastic pipe exposed to steam at temperature 100C. The length of copper that is exposed to the steam is 29mm. The steam was generated with a 2kW boiler (wrapped in a towel and stuff). There is too much power to be absorbed over the length, so the length over which power is absorbed is fixed.

The temperature of the water at the inlet (Ti) and outlet (To) were measured with a thermometer (see pic). Inlet temperature was measured with the boiler turned off. The water flow rate was measured by timing the filling of a 750ml bottle with a stopwatch on a phone. The flow was varied by adjusting the tap. The energy absorbed can be calculated as (To-Ti) x flow x specific heat water.

By knowing the area of the surface of the pipe, the HTC can be measured, as the power absorbed is: [100-(Ti+To)/2] x HTC x area.

The first surprising thing in the data is how much power the thin pipe can absorb. In just 29mm it was absorbing between 200 and 400W, depending on flow. So a 2kW condenser only needs about 30cm of pipe ! ... plus safety margin of course. The thicker pipe absorbed more, but not as much as one might expect because of its lower HTC.

The graph shows the data points. The dots are measured values and the lines are best fit. The minimum flow is the flow below which the power cannot be carried away for any condenser design at all.

The HTC values are quite large, but consistent with the 2000 to 3000 W/deg/sq m measured previously. The red points were measured over two days. It took some practice to get into a groove of what had to be done. The blue points were measured on the second day and are much more consistent.

The Reynolds number for the 2.7mm pipe is around 7000 at the minimum flow point which puts it into the region of turbulent flow.

Conclusion ? Coil based condensers (Nixon Stone, Dimroth, Boka etc) have very high HTC values and so take very little material to do their job. They rely on turbulence to do this, but this is obtained naturally from the narrow diameter pipe and high fluid velocity it creates. This makes NZChris probably right that increasing length for the water to travel improves the performance of a Leibig (Eyeballing the red line I get about a 30% improvement from that). So would making the outer pipe to inner pipe gap as small as possible, allowing for water flow of course.

I'd love it if someone else repeats these measurements and calculations. It is not science till its done in two places getting the same answer. My work is not checked for dumbf errors and that makes me uneasy about believing it too much. I'm just about to start my new head, which will be a Boka, plan changed after a helpful comment.

[still_stirrin]

Good stuff engunear. Confirms hypothesis learned 35 years ago in school.

An old adage, "figures lie and liars figure" seems off basis here. Your tests and analyses have proven the suppositions.

Thanks for the revelation.
ss

[Maritimer]

Hi engunear,

If you were to take the slopes of the lines in the graph you've presented, you could have a single number that would describe a condenser. The units of the slope would be HTC/Flow, or (W/m2 C)/(liters/sec).

There are various ways to express the units. The flow rate could be eliminated by turning the watts into joules/sec, giving joules/(m2 C liter), or if you elimate the liters and turn it into m3/1000, you get kJ/(m5 C). When you get such unwieldy units, it's time to name it.

So, your graphs give
(18000-9000)/(.035-.01) = 360000 unnamed units for the 2.7 mm condenser
(7000 - 4500)/(.035 - 0.01) = 100000 for the 6.5 mm condenser.

M

[engunear]

Hi Martimer

Not quite sure what you are thinking as the lines don't pass through the origin so two points are required to describe them.

To save your calculator, the lines are 3483+flow*128,482 and 5322+flow*416,318 (from the trusty NumPy polyfit function).

BTW, this makes me wonder about those stills that seem to run the cooling water through the column for fractionation "vapour management" and mechanical support. The bottom pipe has the cold water, so it must return a lot of distillate without much fractionation benefit, as the column below it is so short. A bit off topic I know.

[Maritimer]

Hi engunear,

Hmmm, yes, I hadn't considered that the lines don't pass through the origin. Would it be possible to do the 2.7 mm experiment over? It would be interesting to see how close the lines would come together at zero flow with a more accurate experiment. From your equations, they are 3483 and 5322 W/m2/C.

What does zero coolant flow mean? The still coolant would heat up at a certain rate.

engunear's graph.jpeg (13.24 KiB) Viewed 2708 times

Given the broad distribution of the data set for the top line, it might be possible for the lines to be much closer at zero flow.

M

[shadylane]

From a practical perspective. Since many of the hobby distillers use liebigs with 1/2" inner tubes and 3/4 or 1" outer.
How much advantage is a turbulator for each ?

[engunear]

From my calculator, for US sizes (bring on that Royale with Cheese) 1/2 inch down 3/4 type L has an annular area of 97 sq mm. 1/2 down 1 inch has a annular area of 332 sq mm i.e 3.3X as much. The smaller diameter will increase the water velocity by about 3X at the flow. This will make the condenser lighter (less water volume) which is always good. Reading from the graph, looks like a 3X increase in velocity corresponds to a 30% reduction in the length of the condensation region (thats an eyeball measure, will do better later).

If using US pipe, I'd think seriously about 5/8 for the outer pipe. Any reason it is not popular?

I plan to test the leibig flow velocity theory by taking my leibig and replacing the outer pipe (currently 30mm ID) on with pipe with 23mm ID.

If the length of the condensation region is short you can turn the coolant flow down, which then makes it get longer while still maintaining a distance between it and the end of the leibig. If the condensation region starts close to the end of the condenser then a fluctuation in water pressure leads to it extending beyond the end i.e. vapour loss. Running a longer condenser at lower flow rates only buys you so much, because as the coolant heats it becomes less effective.

That graph might not be applicable because the thin pipe has turbulent flow, but these leibigs do not. So it may be it makes no difference other than making the condenser lighter which is a good thing. So the test will be interesting.

The coiled condenser stuff is off-topic for this, so I've started a second thread just for it: "Engineered coiled condenser" http://homedistiller.org/forum/viewtopi ... 87&t=57922

[engunear]

I'm still chugging at this, and have learnt that since my measurements were with water, they are not really representative. The HTC is reduced by a layer of alcohol that forms on the inside of the condenser pipe. This means HTC will vary along the condenser.

This would mean that things that improve the HTC of the water jacket (i.e. turbulators) are very much secondary. I was at the point of adding NZChris's spiral to my Leibig, too, and I've cut my Leibig down. Damn.

On the distillate side of the wall, something that gets the alcohol off the walls of the condenser might help e.g. scrubbies. A tough proposition to test, but anyway. Also, horizontal mounting or a shotgun seem like better ideas than they did before. All speculation at this point.

[engunear]

Having done a lot of expts on turbulators and reading about fluid dynamics and heat transfer, and having built a Leibig with a turbulator, I'm gonna summarize the gospel according to St Engunear. I want to do a final dump to empty the brain and move on to other things to think about.

Do turbulators actually work?

There are (at least) four kinds of turbulators:

1) Coolant stream ones that create turbulence by disrupting flow, usually with a spiral. See youtube under "laminar flow" https://www.youtube.com/watch?v=WG-YCpAGgQQ" onclick="window.open(this.href);return false;" rel="nofollow.

My measurements showed that they don't make any difference. The hypothesis is the scale of the turbulence is too large compared to what is needed. The layer of water that provides the thermal barrier is around 0.1mm thick. The turbluence a spiral turbulator produces is of order 1-3mm.

2) Velocity modifiers. These increase fluid velocity by either creating a race for the coolant (NZChris suggestion), or simply by narrowing the gap bewteen the outer wall of the condenser and the inner wall of the coolant (outer) pipe. This creates turbulence via the Reynolds number. These work well. You can get a 10X increase in velocity with a spiral, but it has to fit snugly betweent the two pipes. A combination of these two things will give a 2X reduction in the length of the condenser. See pic. There is a 2mm gap between the inner and outer walls for this example. Even just choosing a smaller diameter outer pipe will give benefit, and be cheaper. Having a thinner outer pipe also helps because it means the condenser holds less water so is lighter, so there are lots of benefits here.

3) Vapour path turbulators at the hot end (usually scrubbies). I don't have data to say one way or another. I suspect they don't help as they cause distillate to pile up and this reduces the HTC. Thats a hypothesis. They would also increase back pressure that would encourage leaks, and for that reason I've not gone down this path.

4) Vapour path turbulators at the cold end e.g. scrubbies. I have not measured the effect of these. But it seems likely that they will slow the distillate flow and so give it more time to cool. They also catch the mist that is created in a Leibig, which is better than losing it. (This mist is cool and does not burn, as opposed to true vapour that is hot and very aggressive.) Lots of people swear by vapour path scrubbies, so this is probably right.

I think the book numbers for HTC on the calculator are wrong because they are for industrial size condensers. Our little toys are just outside their worldview. That being said, if you build off the website calculator this error puts a safety margin in there, which is a good thing. Just don't add a second one or your unit will be big and unwieldly. But HTC varies with lots of factors (alcohol purity, coolant velocity, jacket dimensions) so measurements in context are the most accurate source.

Heat transfer coefficient is lower for alcohol than water. Pure alcohol seems to have the lowest HTC. (Thanks DeepSouth)

As evidence of the above, see the plot that shows flow vs length data for a variety of Leibigs. The red line is the website calculator with HTC=850. The two green dots are mesaured from a Leibig with a 3mm gap bewteen the inner and outer pipes, and straight flow. This is measured with 85% alcohol. The blue dots are tighther fit (2mm) and a 10X spiral. See pic. The black line is the website calculator with HTC=5000. The red dots is 50% water 50% alcohol with the spiral. This is a little odd as I had previously figured that 50/50 would be worst. But this is data that is hard to argue with.

It is also probably true that a nearly horizontal Leibig works better than a vertical one (limited data on this, but the distillate slops to one side rather than coating the sides evenly and adding to HTC.). If you do a saxophone design you can get the balance to work for you as well, so the unit is easier to support. See pic.

So the answer is ... "it depends".

Time to make some whiskey, methinks.

[skow69]

Could you summarize your conclusions, please? Are you saying that a spiral doesn't help, or it only helps if it is a press fit, or it doesn't help as much as reducing the gap to 0.1mm, or what?

[engunear]

Sorry if I wasn't clear.

Anything that increases the velocity helps to increase HTC. This can be having a smaller gap between inner and outer pipe. This helps because the velocity is the volume per unit time divided by the area ; smaller area, larger velocity.

Adding a spiral that fits against the outer pipe also increases the velocity so it helps as well. This increases the path length. It is easier to get an increase in velocity this way than with a narrow gap. HTC change with velocity is weak effect but because you can get 10X or 20X increase in path length, you get a significant (2X) shortening in the overall condenser.

A spiral that does not fit against the outer pipe does not help. The youtube clip is instructive but misleading for us. A spiral that fits against the outer wall but only gives a small increase in path length will only help a little bit. The effect is weak so a 2X increase in path length will not give a dramatic reduction in required length.

It would be possible to make the spiral fit snugly against the outer wall and solder it to the outer wall as well as the inner. This increases the area of copper in contact with the coolant so gives an extra benefit there. Need to be careful you don't get a blob of solder blocking the path. This would take a lot of thought and care to do right, but should work well.

Just turning the flow up increases velocity. You get more benefit from the extra flow from the extra velocity, but they do both contribute.

Although it is off topic for this thread, the coolant velocity in a coiled condenser (offset Nixon, Boka etc) is much larger than in a Leibig so these can be very small for their cooling ability because the coolant pipe area is comparatively small. There is a related thread "Engineered coiled condenser", which talks about this and drifts into Leibig's as well and has a lot more discussion of HTC and flow velocity.

I found no other things that you can do with coolant flow that help.

These benefits allow us to build shorter condensers. Once you have built one, and its working to your satisfaction: not venting at reasonable flow, and not too unwieldily, then it works. And if you build one and its too short then its a prick. If you have one that is unwieldy and you suspect is too long, before you cut it down, you can stick a thermocouple up it and measure where it is condensing and cut it down based on data. If you do that, remember to do it with the highest concentration of alcohol you normally make.

[skow69]

Thank you. That is more clear. So basically your saying, "Forget turbulation, do whatever you can to make the water run faster." And then you equate faster running with 1) smaller cross section, and 2) longer path.

The first is true up to a point. The second is just rubbish.

1. Yes the river runs faster through the rapids, the fluid flows faster through a venturi, until the restriction passes a threshold where the river backs up into a lake and the fluid builds pressure. Beyond that point the restriction is counterproductive. I'm still not clear whether you are advocating a 0.1mm gap, or a 1-3mm gap, but I suspect those may be in the counterproductive zone.

2. Making the river longer doesn't make the water hurry to get to the end on schedule, it just makes it arrive later.

[engunear]

If it arrives later then the velocity is lower.

If the flow (litres/sec) is the same, and the path length is longer, then the flow velocity is higher.

The outer diameter of the inner pipe is 19mm. With no coil, the cross sectional area of the path is pi * 19 * height (2mm) i.e. 80sq mm.

The gap between the spiral wires is 4mm. Cross sectional area = 4mm x height (2mm) = 8 sq mm.

Decrease in cross sectional area 10X.

Flow is the same. Therefore increase in velocity is 10X.

[skow69]

No. Arriving later just means it spent more time in the condenser. Liters per second is flow volume. Velocity is inches per second.

[skow69]

The outer diameter of the inner pipe is 19mm. With no coil, the cross sectional area of the path is pi * 19 * height (2mm) i.e. 80sq mm.

The gap between the spiral wires is 4mm. Cross sectional area = 4mm x height (2mm) = 8 sq mm.

Decrease in cross sectional area 10X.

Flow is the same. Therefore increase in velocity is 10X.

Pi*19*2=119. Your new cross section is 119mm2 minus the area of the ellipse described by the intersection of your spiral wire with a plane perpendicular to the axis of the condenser. I think it is evident from the photo that wrapping the wire didn't eliminate 90% of the cross section. To calculate from the gap between the spirals, you would have to account for the increased length of the spiral path.

[engunear]

Yes you are right, I did make an error. The area before was 120 sq mm. Apologies if I missed that the 2X in your calc which is the height.

The area after is 8 sq mm less the wire thickness. So the difference is > 15X. But that does not change the basic premise, since we don't have the exact function relating velocity and HTC. You can measure the change in path length by comparing a length of a piece of the wire before and after coiling. And its about 10X. Its a bit sloppy as I don't have the means to make a neat spiral.

The problem with the river analogy is that if the path length increases due to uneven bottom, but the depth stays the same, then the volume has increased. In this case the volume has stayed the same, or reduced even.

But if you want an argument, please pick someone else to have it with, preferably on another website. I really don't have the energy. I'd also suggest you try out a few of these ideas and see for yourself. The data on performance speaks for itself. If you take a measurement and find differently I'd be interested to discuss. Maybe I'm missing your tone, if so I apologise.

[skow69]

Oakey doakey, Smokey.