meerkat

It would seem that your blending targeted at 95 proof was actually spot on (and not 95.6) because if you calculate 190 lb of 95 proof plus 1.65 lb of water you get 94.25 proof. I would guess that everything was correct, but your gauging after the first blend gave a slightly higher reading than it should have. Others have also mentioned that immediately after blending the measured proof can be slightly high and the most plausible theory I have come across for this is that micro bubbles of air are mixed into the spirit during blending lowering the density and making the proof appear high. It could of course also be caused by the exothermic blending raising the temperature and decreasing the density, but I presume that you would have corrected for the temperature. Whatever the cause, leaving the blend to settle for 12 hours or so will give the correct proof. What is strange in this case is that the second blending immediately gave the correct/expected proof. Was there any difference in the way you did the first and second blends? How soon after the first blend was completed did you do the gauging, and how long after the gauging did you add the extra water?

@Still_Holler I am a bit confused about what exactly you are asking because in an earlier post you asked whether temperature corrections need to be applied if the quantity of spirit is measured by weight, but then in your example the quantity is given in volume. Let's look at both alternatives, as applied to your example. If you measure 1000 gallons at 76°F then you would have to use Table 7 to correct the volume to 60°F. As you showed before, the factor is 0.991 and there would be 991 wine gallons at 60°F. For reporting purposes you need the proof gallons and this is simply 991 x 190 / 100 = 1882.9 proof gallons. If you had weighed the same tank you would have got a weight of 6733 lbs. From Table 4 we get that 190 proof spirit has 0.27964 PG/lb so we can calculate the total proof gallons as 6733 x 0.27964 = 1882.8 PG. So in summary, if you have the weight and the proof you can go directly to the proof gallons using Table 4. If you have the volume and the proof then you also need the temperature at which the volume was measured so that you can obtain the volume at 60°F, and then obtain the proof gallons by correcting this volume by the ratio of the actual proof to 100 proof (190/100 in the example).

Thanks to @John Bassett and @HedgeBird for the kind words. @PeteB - please will you make a screen shot of where the problem occurred. I guess it wasn't on the Hydrometer Correction calculator? If it was on one of the blending calculators then maybe I could lock the choice between Proof, ABV and Mass the way that the Standard Temperature can be locked. This would allow all the choices to be fixed for normal work, but leave the option open for the one weird day when you need to blend some European (or Aussie!) ABV spirit with some American Proof spirit.

@Hudson bay distillers If there is such a thing as a universal optimum temperature for the feed to a continuous column then I would say it is the boiling point corresponding to the composition of the feed. If it is colder than this it causes internal reflux in the bottom part of the column which does not do much towards providing a high strength top product. On the other hand if it is hotter than the boiling point you will find that you need to provide a higher reflux ratio from the condenser which increases the water requirements on the condenser but saves a bit of heat on the boiler. But I have seen feed temperatures from substantially below the boiling point all the way through to feeding the column with 100% vapour. If the feed is not a liquid at its boiling point then a change in diameter of the column at the feed point may be required.

From one physics nerd to another.... I believe you are on the right track with your interpretation that it is the boiling that stops and starts. I have seen a similar phenomenon in continuous columns that have significantly oversized external reboilers. These get into a similar cycle of pulsing and the accepted interpretation was always that boiling was stopping and starting. With the very oversized reboiler, heat can be passed through the tubes into the liquid so fast that a vapor blanket is formed on the inside of the tubes. The heat transfer through vapor is much slower than through liquid and suddenly the heat transfer stops. The bouyancy of the vapor gradually causes the vapor to rise out of the reboiler and colder liquid flows in from the bottom of the column. Heat transfer picks up because of the higher thermal conductivity of the liquid and the cycle repeats.

To eliminate the liquid U-seal formed by the tapered section below the condenser and the parrot itself you can either lower the parrot or raise the entire condenser. Once you have the levels changed so that the section of pipe coming out of the condenser stand no longer stays flooded with liquid you can install a vent as suggested by Dehner. It probably sounds like a lot of hassle to do, but if you don't this setup will always be awkward to operate.

It appears that the stainless vessel to the right of the parrot in the photo is the condenser. If that is so, then you will have a problem installing a vent. In order for the condensed liquid to develop sufficient head to flow over the top edge of the parrot the level of liquid in the condenser will have to build up to slightly higher than the top edge of the parrot. This will create a liquid seal at the bottom of the condenser and gas will not be able to get to the outlet. If you were only concerned with flushing trapped air at startup then the suggestion by PeteB to open the bypass would be adequate. But there will also be dissolved air and CO2 in the fermented beer and this could continue to come out during operation. There will also be light ends (aldehydes etc) that some operators allow to escape from the vent by setting the cooling water flow to get a fairly high temperature for the condensate. Colder isn't always better. By creating a seal at the bottom of the condenser you will force the pressure in the still and condenser to gradually build up (due to incondensible gases) to the point where the seal is blown and you get the surging you are seeing now. In my experience the outlet from the condenser either flows to a small vented pot and then drains from the bottom of the pot to the parrot, or else the vent is mounted on the body of the condenser itself. This second option is common when the condenser is mounted at floor level. We cannot see the other side of the condenser in the photo - perhaps there is already a vent mounted there?

When I have seen this in the past it has always been because the product line from the condenser down to the parrot (or flow meter) has been too small and the liquid flowing down entrains air sucked in via the condenser vent. The solution is just to use a larger diameter drain pipe from the condenser. But it seems this is not the reason in your case. If you genuinely do not have a vent after the condenser there is nowhere for the air that is in the system at startup to escape - except via the parrot. As the water is turned to steam and displaces the air, the air will be forced out via the parrot. Seeing that it stops after 10-20 minutes I suspect this is the cause rather than the small drain pipe I mentioned above.

PeteB - The calculation for the quantity of water required to dilute to a target proof is illustrated in the article I referenced before. The example is based on volumes, but it could easily be converted to weights using a similar method to what I showed above when using Tom's suggestion to use Table 4. whiskeytango - The questions you are asking are exactly those AlcoDens was designed to answer. The formulas are way too complex to use in manual calculations. They are freely available from the OIML site if you want to see just how involved they are. The AlcoDens calculator for doing this can convert back and forth between Mass (Weight) %, Density, Volume % (ABV), Proof and Molar % As an example assume there is an unknown quantity of 150 proof spirit to which we add 100 lbs of water. After mixing the proof is 100 proof. What was the unknown volume of 150 proof spirit and what volume of 100 proof spirit has been produced? In the example shown below Source 2 is the pure water (Proof = 0). I have set the Source 2 quantity to mass and given the weight as 100 lbs, but you could equally well give the quantity as volume. In the bottom right panel you tell the program that the quantity of Source 2 is the known quantity. Set the Source 1 and Final Blend quantities to Volume and it will calculate how much 150 proof spirit was used and how much 100 proof has been produced. THis could be solved using Tom's Table 4 method provided all quantities are in weight terms by starting off with the classic "Let X be the mass of 150 proof spirit" and noting that the number of proof gallons does not change when you add pure water. Once the answers have been calculated in weight terms they can easily be converted to volumes.

Thanks Tom, using Table 4 is a bit easier because it provides data for every 0.1 proof and it is probably easier to understand. For comparison with my earlier calculation here is the Table 4 method. 160 pf spirit has 0.13903 WG/lb. Table 4 only starts at 1 proof and from standard engineering tables the specific volume of water at 60 degrees F is 0.12008 WG/lb. The weight of the spirit is therefore 100/0.13903 = 719.27 lbs and the weight of the water is 100/0.12008 = 832.78 lbs The total weight is 719.27+832.78 = 1552.05 lbs. This contains the same 160 PGs that were in the original spirit so we have 160/1552.05 = 0.10309 PG/lb Table 4 shows (without interpolation) that this is 81.6 proof and the specific volume is 0.12633 WG/lb Total volume is 1552.05 x 0.12633 = 196.07 gallons This is definitely shorter and more direct than my earlier calculation. But AlcoDens still wins, especially if the temperature is not 60 F and the calculated PG/lb requires interpolation.

Disclaimer: I am the author of the AlcoDens blending app mentioned below, so I am probably biased. This question of how to do blending calculations comes up fairly often, so I wrote this article to compare 3 of the methods. I have not seen any really comprehensive guide to performing calculations using the TTB Tables, but I am aware of two distillers who are in the process of preparing books that will cover this. Hopefully early next year (2018) things will be better. The theoretical blend proposed by whiskeytango is unfortunately not in the form for which the TTB Tables were structured. The tables are designed to easily answer the question of how much water to add to dilute the 100 gallons of 160 proof down to 80 proof. But it can be done. I will show my solution but perhaps someone who actually uses the TTB tables can show an easier way. From Table 6, 160 proof spirit contains 80 parts of alcohol and 22.87 parts of water. BTW, the fact that these add up to 102.87 shows how the tables do include the effect of contraction. If we now add 100 gallons of water to 100 gallons of 160 proof spirit we still have 80 parts (gallons) of alcohol but we now have 122.87 parts of water. The ratio of parts of alcohol to parts of water is 80/122.87 = 0.6511. Now we have to scan through Table 6 to find what proof corresponds to this ratio. As I said, the tables are not structured to make this easy. At 81 proof the ratio is 40.5/62.95 = 0.6434 At 82 proof the ratio is 41.0/62.47 = 0.6563 By interpolation we see that a ratio of 0.6511 corresponds to a strength of 81.6 proof. This diluted spirit obviously contains 81.6/2 = 40.8 parts of alcohol. This came from the original 80 gallons of alcohol in the 160 proof spirit. If 80 gallons is 40.8 parts then the total parts (i.e. 100) must be 80x(100/40.8)= 196.08 gallons. Naturally, I checked this with AlcoDens to make sure I had the correct answers. Because I do not do these calculations using the TTB Tables very often I did make a few mistakes along the way, but the values above compare well with the AlcoDens answer shown below.

Yes, the interpolation in Table 6 is very similar to that for Table 1. In fact it is a bit easier because you only have two parameters to deal with (proof vs parts of water) while in Table 1 you have three parameters (apparent proof, true proof and temperature). I have not seen the TTB data in the form of an equation - only as the Tables. There must be an equation because the Tables could not have been generated from experiments at every published point, and must have been generated from an equation. I just have not seen it. The data that is openly available as a formula is the European OIML alcohol-water density data. This is the data that forms the basis of my AlcoDens program. The OIML and TTB data are based on much of the same original experimental data but they are presented in different terms. The TTB data is (mostly) presented as densities and masses measured in air but the OIML data is presented as absolute or "in vacuum" values. Fortunately it is possible to convert between "in air" and "in vacuum" data so it is possible to select in AlcoDens whether you want to match the TTB or OIML tables.

Distillers have been gauging their products successfully with hydrometers for over a 100 years and you shouldn't stress over that process. The TTB have posted some excellent videos that show and describe the process in detail. See https://www.ttb.gov/spirits/proofing.shtml If you know that you have dissolved solids in your product then the first step is to measure them using the petri dish and scale method you mentioned. If the total dissolved solids (TDS) content is less than 400 mg per 100 ml (or 4 gram per litre) then you are allowed to ignore the effect of the solids and use the standard TTB Tables. If the TDS is between 400 and 600 mg per 100 ml (4 to 6 gram per litre) then you must use the TTB formula to adjust the apparent proof to get the true proof. This is also described in a TTB video at the page I linked to above. This formula applies to a fairly narrow range of proofs - 80 to 100 proof from memory but subject to correction. If the TDS is above 6 gram per litre then you must use the lab scale distillation procedure also shown in one of the TTB videos linked above to get to the true proof. There are several useful discussions on this subject in previous threads and using "obscuration" or "lab distillation" in the search box at the top of the page will find them for you. Once you know what your proof is, adjusting it to your target proof is a whole different kettle of fish. As RobertS said above "be ready to repeat this a few times as you come down to bottling proof". The more accurately you can measure the true proof and perform your proofing calculations the fewer times you will have to repeat the process. Joe Dehner offered a manual procedure for these calculations in the thread You can see the computer program that I finally developed as a result of all these discussions at http://www.katmarsoftware.com/alcodenslq.htm and download a trial version if you want to test it.

See Part 3 of the TTB video series https://www.ttb.gov/spirits/proofing.shtml

The Australian Distillers Association group on Facebook seems to be where most of the professional Aussie distillers hang out. There is also the Aussiedistiller forum at http://aussiedistiller.com.au which is very active but seems to be mainly home distillers. And of course there are some very active and knowledgeable Australian distillers right here in the ADI forum.

One part of your question which I did not address is whether it is detrimental to have a column that has a larger diameter than necessary. Unfortunately columns that are larger in diameter than necessary do not work as well as correctly designed columns. Depending on the type of column, there will be a narrower or wider range of flows over which it works well. With packed columns if they are too large the liquid will not wet all the packing and there will be zones where the vapor can rise up through the packing without contacting the liquid at all. WIth sieve (perforated plate) columns a minimum vapor velocity is required to prevent the liquid from weeping through the holes. A small amount of weeping does not affect the efficiency, but too much will cause the plates to run dry and in larger columns can lead to mechanical problems like vibration. Correctly designed bubble cap trays will not weep and these columns have the widest range of capacity. But even these trays will fall off in efficiency if they are grossly oversized because the intensity of the bubbling will decrease. This reduces the mixing and the vapor-liquid contact - reducing the separating efficiency of the column. Of course, for all types of columns a detrimental impact of over sizing is the cost of the column. The quantity of liquid held up on each tray also gets to be significant with very large columns and this can affect the recovery of alcohol in batch systems, but is not really a problem for continuous columns.

This subject was discussed in Unfortunately that thread got a bit messy with some irrelevant side-issues causing a bit of bickering and hair-splitting. Basically the situation is that the diameter of the column and the size of the pot are determined by different factors, so there is no fixed ratio between them. The column diameter is mainly determined by the vapor velocity up the column. In a small R&D column of around 2" diameter the vapor velocity will be in the region of 6 to 10 inches per second, but on a large vodka column of say 10 ft diameter you can get vapor velocities of 6 to 10 feet per second. So the column diameter is determined by the rate at which you want to run. In a pot still the pot size is determined by the heating method and the size of the batch you are working with. Let us imagine your fermenters produce 100 gallons per batch. You will probably want a pot of around 150 gallons to be able to boil this safely. If you want to process this in 12 hours you will need a column roughly double the diameter (4 x the area) than if you want to process it in 48 hours. And of course you need to put heat into the pot at 4x the rate for the 12 hour scenario. Distillers generally want to be able to process a batch in 8 to 12 hours (one shift) so it turns out that in practice there is an approximately consistent ratio between the pot size and the column diameter (or more correctly the column cross sectional area) but this is a coincidence - as explained above there are different drivers in determining the pot and column sizes.

Your 1674.8 lbs of 80 proof will give you 804.1 liters at 73.54°F. AlcoDens and the TTB Tables agree on this. But because 750 ml at 60°F grows to 754 ml at 73.54°F you should expect to get 804.1/0.754 = 1066 bottles. This agrees with PeteB’s mass based calculation. The underfill is 4 ml per bottle so you could expect to have 1066 x 4 / 750 = 5.7 extra bottles. The fact that you had 14 too many means that we still need to find where the extra 8 bottles came from. I agree that your scale is unlikely to be the source of error, but keep it in mind to check once you have eliminated all other possible reasons. If your proofing was out and the 1674.8 lbs actually gave you 810.0 liters ( = 1080 x 0.750) instead of the calculated 804.1 then your density at 73.54°F was 7.8262 lb/WG and this would correspond to a proof of 88.2. It is unlikely that you could be this far out. Another possibility for error would be if your bottling temperature was not the 73.54°F in your storage tank. But the temperature would need to be in the region of 90°F to explain the difference. This should be easy to check. As PeteB has said, it would be better to do your quantity checks during the run based on mass rather than volume. If you have an accurate lab scale you could also use it to calibrate your measuring cylinder. Use AlcoDens to calculate the expected weight of 0 proof (i.e. water) when your cylinder is full, and fill it with RO or well filtered water. If you have a bottling machine that uses a fixed head (pressure) and adjustable timer to control the fill quantity then you can set it to give a target weight rather than volume. The weight filled will vary with the temperature of the spirit and if you make a note of the time required and the temperature each time you adjust it you will soon be able to draw up a calibration curve to speed up the job.

No, the original AlcoDens is still available for those who only work with alcohol/water mixtures and do not need the capability of calculating liqueurs. The new product is AlcoDens LQ and it will do everything that the regular AlcoDens does, plus a few calculators that deal with mixtures containing alcohol, water plus sugar.

Thanks for the kudos PeteB. It is a complicated calculator to set up the first few times, and if anyone else is having similar problems they are welcome to send me some numbers and I will set up the calculation for them with an explanation of how/why I made the selections. Other users have also reported that after a few runs it becomes much easier, but the learning curve is a bit steep.

The lab still will allow you to determine the proof and sugar loading of your spirit, but if they are not on target then you still need the math to determine what to add to get to the correct levels.

If you know the proof and weight (or volume) of the rum and the target sugar loading then you can calculate the quantities of sugar and dilution water required to achieve 80 proof using the calculator downloadable from www.katmarsoftware.com/alcodenslq.htm But the final proof must be verified as described by bluefish_dist

The reason for the difference between adding water and sugar is that proof is based on volume and a given mass of sugar has a different volume from the same mass of water. For example. if you have 10 gallons of 100 proof spirit and you add 10 lbs of water the volume will increase to 11.169 gallons. But if you added 10 lbs of sugar to 10 gallons of 100 proof spirit the volume would increase to only 10.749 gallons. In both examples no additional alcohol is added, so both contain the same amount of alcohol after dilution. Adding the water would lower the proof from 100 to 89.54, but adding the sugar would lower it to only 93.03 because the volume has increased by a smaller amount. If you used a standard proof hydrometer to determine the proof of the sample with the sugar added it would read about 30.4 proof because the sugar would have significantly increased the density of the sample and the hydrometer is not calibrated for that.

The volume will decrease, and part of it is due to the evaporation of the ethanol, but the main reason for the volume decrease is the loss of CO2. If we take the hypothetical example of 1000 kg of water with 200 kg of sugar added we would have a volume of about 1120 litres, since 1 kg of sugar increases the volume by about 0.6 litre. Roughly 100 kg of the sugar is converted to CO2 and is evolved. The other 100 kg will be converted to alcohol with an SG of about 0.8 and ignoring shrinkage we would have a volume of 1080 litres, or a loss of 40 litres. In large commercial distilleries the CO2 that is evolved is taken via a scrubber to recover the alcohol that is carried off. A very rough calculation indicates around 3 or 4% of the ethanol is carried off with the CO2 and in large distilleries this is worth recovering. If you lost 3% of your 80 litres that is another 2.4 litres. So the evaporation losses are real (2.4L) but small compared with the CO2 loss (40L).

Towards the end of last year (2016) I thought that I had virtually completed the liqueur blending calculator. But then some of the liqueur makers who have kindly been giving me advice pointed out that sometimes in the proofing process the sugar level is "close enough" to not need any further correction while the alcohol must be adjusted to within the TTB rules. At that stage my calculator required that both the alcohol and sugar be corrected at the same time. I have now added the ability to correct only the alcohol proof while allowing the sugar level to find its own level. Strangely, the math for correcting only alcohol is more complicated than that for correcting both together. A few examples of liqueur blending calculations can be seen at http://www.katmarsoftware.com/alcodenslq.htm I am now drawing a line in the sand and will not make any further changes to the program unless I discover a bug. Hopefully the program will be ready fot public testing by the end of February. Thanks to all of you who have continued giving me advice and encouragement - and for putting up with my broken promises regarding the end date.