We are getting more birectifiers in the wild so I thought I should write a guide to running it for the first few times to develop a heating routine. The system requires very precise management of energy input so that fractions are faithfully collected at 25ml / 15 minute increments. The more faithfully collected, the better we are able to make meaningful comparisons across distilling runs. To maintain a constant rate of flow, the energy required is dynamic and ramps upwards because of how alcohol is exhausted over the course of the run.
For starters, I advocate using a 500 ml Glas-Col brand mantle (model 100B TM106) over a Chinese version. They do have long lead times, but they are worth it. You can also pay a small premium to buy them elsewhere with no lead time. Do not use a Glas-col heating controller because you’ll need something far more precise with a digital read out that can eventually also integrate into our automation kit.
For heating controllers, I own a lot of options and have systematically tried everything, but now I wholeheartedly endorse Auber controllers and have a custom specified version for the birectifier I can direct anyone to ($400 shipped with a precision thermometer probe). Auber sells what are basically affordable Chinese clones of American Watlow controllers. Auber’s customer service is awesome and Watlow prices will make you cry. (Instruction manual for the Auber birectifier controller [instruction for entering manual mode start on page 5])
We will be using the Auber controller manually, but you can also use it as a PID for other lab applications. For the birectifier, we’ve carefully selected an RTD style thermometer probe to measure the vapor temp which implies distillation proof (automation will be able to log data from it).
Make sure your birectifier column is setup plumb (not tilted!) and that the condenser is roughly at a 45° angle, inline, and not twisted. The condenser uses a ball/socket valve to limit potential for breakage, but if the angle is too extreme, the nature of the mechanism will be to reduce the aperture of the bore and create reflux you do not intend or want. If you arrive at one set of heating numbers and they are very different another time your system is setup, it could be because your condenser is not at the right angle and the slight difference varies energy needed by a percentage point or more.
It is safe to call the birectifier initial heat up frustrating (because you want to go fast!). You cannot apply a lot of energy to heat quickly then turn it down. The heating mantle simply has too much inertia. If your first fraction is optimally collected at 39% power then that is how much energy initial heating should require and it simply takes as long as it takes. Botching the first fraction is a pain in the ass. Sometimes initial heat up for a 250 ml sample takes 20 minutes.
For fractions 1-8 my current heating profile looks like this:
Fraction 1: 39% <–Near azeotrope
Fraction 2: 39% <–Near azeotrope
Fraction 3: 40% <–Near azeotrope
Fraction 4: 42% <–Near azeotrope
Fraction 5: 42-75% <–Alcohol exhausted in the beginning
Fraction 6: 76% <–Aqueous
Fraction 7: 76% <–Aqueous
Fraction 8: 76% <–Aqueous
The first four fractions are very similar in energy required and so are the last three. The fraction that changes dramatically is fraction 5 because near all alcohol is exhausted in that fraction. On your very first running, you have multiple tries to arrive at the correct number for the first four fractions and then again for the last three. On your next running, because you’ve established your beginning energy needs and ending, the focus will be about perfectly executing fraction 5 which may need the voltage changed multiple times. I often change it at the very end of fraction 4 as the vapor temperature starts to rise. Changing it again towards the end of fraction 5 ends up being sufficient. To get good results, it is definitely not necessary to sit there and increase the energy by one point every minute.
These numbers have stayed fairly consistent in my lab as spring weather rolled into summer and humidity increased. After a while, I have been able to look at the drips almost like a metronome to know things are proceeding correctly. As a rule of thumb, it is better to be too slow than too fast. As another rule of thumb, for the first four alcoholic fractions, you should be able to count off nearly one drip per second so you can set your stop watch and count out thirty drops in roughly thirty seconds. For the last four fractions, where alcohol is exhausted, the drips change in size and even become less consistent in interval.
Some thing to keep in mind is that the number percentage is a fraction of the voltage which we only assume is 115V, but isn’t always. A change of a single percentage point can add or subtract an entire minute off a fraction. Heating is that sensitive!
Eventually our automation kit will have a PID take care of everything. A high end scale will measure the flow rate and we will use a combination of voltage adjustment and on/off to hopefully track perfect flow rates.
For every fraction I collect, I note the time and update my ramped voltage charts. The more systematic you become, the less time you’ll burn creating your routine. Being very patient with the very first start up will save you a lot of time down the road. It may also be useful to collect your very first run in a 25 ml graduated volumetric cylinder. That way at 5 minute intervals you will know if you are on track (roughly 8ml / five minutes).
Initial heating can take almost 15 minutes before a rapid boil even begins. In the next approx 5 minutes before output is collected you can watch vapor slowly start to creep up the walls of the birectifier like a lurchy phantom (its very cool). The slow pace of this process may also help the vapour come to equilibrium.
If it looks like I am coming up slow, two or three minutes before a fraction is due, I increase the energy by a point. Fifteen minutes is an ideal, but do not panic if you are collecting a few fractions at 16 or 17 minutes.
Arroyo likely had an army of interns fidgeting with Bunsen burners and counting off on pocket watches while smoking cigarettes (The first four fraction by pocket watch, the last four by cigarette intervals).
To first prepare my sample, I find the ABV. If the sample is 40% then 250 ml are required. The formula to use is 10000 / ABV so if the sample is 45%ABV then it is 10000/45=222.22ml. This sample would then need to be watered to a minimum of 250ml (I use a 250ml volumetric flask). Notice the formula simply has the numerator and denominator multiplied by 100 to avoid decimals. You could also simply use 100/.40=250ml.
The math adds up to there being roughly 50 ml of aqueous stillage (which it is beneficial to analyze just like a fraction), but what would happen if we wanted to analyze something like Chartreuse or Cointreau that have sugar? As a rough estimate, the sugar in a 250ml volume of Cointreau occupies 32ml. If all eight fractions were collected, the stillage would be as thick as molasses! The solution to analyzing liqueurs may simply be to not collect the last fraction or possibly the last two. There is likely enough information in fraction 6 that the last two are redundant.
This brings me to another point. When you understand your fractions and how they correlate to your production, you may not need to collect all the fractions. With gin, we have found the last two fractions contain no usable information that isn’t already in fraction 6. Not collecting the last two can save a half hour. The same can be said for extremely routine work with distillates you know well. If your objective is only to assess decisions related to your heads cut, you may not benefit from collecting more fractions than you need to make a decision. However, do not run with this concept too far, and until you are very experienced. The last three fractions contain any potential fermentation faults so doing a check up is always a good idea if you have the time. You may also discover variance in fraction 5.
I would appreciate any feedback that could make this clearer or more concise.