The Physics of Double Retort Pot Stills and Thumpers

Pot stills are the oldest form of distillation and continue to be used across the world, but one of their main limitations is that the maximum ABV that can be attained in a single distillation is ~45% from standard 8-10% ABV wash. The small amount of high proof spirit from the first distillation will also be highly contaminated with low boiling compounds that range from unpleasant to unsafe, so it has traditionally required at least two pot still distillations to produce flavorful, drinkable spirit.

But double distillation is slow and expensive. Each run requires charging the still, consuming lots of fuel to heat it up, and cleaning out the remains in the pot after a run is complete. It also requires complex logistics to balance the flow of raw material through mashing, fermentation, and distillation so that equipment is being efficiently utilized. Distilling has always been a volume-driven business, so the more time it took to produce marketable spirit the less money a distiller was making on the vast amount of capital they had sunk into their plant, inputs, and labor. Somewhere in the 17th or 18th century distillers had the clever idea of hooking multiple pot stills together to perform multiple distillations simultaneously in series.

These types of stills are now uncommon, but can still be found in a number of rum distilleries across the Caribbean such as the double retort systems at DDL in Guyana (both the Port Mourant wooden 'double' pot still and John Dore high ester still), Appleton, Hampden, and Worthy Park in Jamaica, and Foursquare and Mount Gay on Barbados. They can also be found in many bourbon distilleries coupled to column stills under the title of 'doubler' or 'thumper'. All perform a secondary or tertiary distillation to boost the ABV of the output without having to manually perform a second or third distillation.

I've written before about the physics of pot stills and that background will be important for understanding what happens when they are connected to a retort. In essence all of this comes down to a bit of plumbing - while the lyne arm of a traditional pot still is connected directly with a condenser, a retort pot still passes the lyne arm into a additional pot still. This can either direct the hot vapor into liquid where it bubbles through and heats the contents through residual heat or the vapor can first be condensed then passed into the next pot where it is heated again and undergoes another distillation. In either case some portion of the liquid has to be passed back to the previous pot to maintain the liquid level as water and feints are left behind from the increasingly enriched vapor. Importantly, when this is a batch process being fed by a pot still all that is being changed is how many times the vapor is being redistilled. The distiller still makes heads, hearts, and tails cuts just like with a simple pot still.

Double retort pot still with rectifying column at the Worthy Park distillery from The Floating Rum Shack

One of the most important parts of this process is what goes into the retort. If you put pure water in the retort the ABV of the output will not be significantly boosted, but some of the more water-soluble compounds may be scrubbed out, kind of like a hookah or bong. At many distilleries that use these systems, the retorts are loaded with what are called 'low wines' and 'high wines' (see labels on retorts in photo of Hampden Estate below), which are respectively the tails and heads from previous distillations diluted to differing degrees depending on the desired output. Others, such as DDL, combine the heads and tails together before loading them into the retort. This replicates the practice in many distilleries with simple pot stills of recycling feints back into the wash still for redistillation. A visual description of that process can be found here. For more flavorful spirits, stillage or dunder (what remains in the pot after a previous run) can also be charged into the retorts to boost the ester content in the Cousin's process (this is a sufficiently complex topic that it will get its own post at a later date).

To cite one example of how a retort pot still operates, this report claims that Appleton's double retort pot still starts with 8% ABV wash that is converted into roughly 30% ABV output, which goes through the first retort charged with 30% ABV low wines and is converted into roughly 60% ABV output, which goes through the second retort charged with 75% ABV high wines to give a final product at 80-90% ABV.

Double retort pot still at Hampden Estate from Leonardo Pinto

While the dynamics of retorts fed with the condensed output from the previous still (doublers in bourbon parlance) are basically the same as any other pot still, a vapor feed creates far more complex dynamics. What happens to the vapor bubbling through the liquid in the retort is dependent on a large number of influences that will shape the output. Thanks go out to user The Black Tot from the Rum Project forums, who did a pretty thorough job of thinking through what's happening in a retort.

Vapor from the pot still emerges into the liquid in the retort, initially at a much higher temperature than the liquid. The height of the liquid in the retort creates pressure that compresses the bubble. These forces will make the bubble partially or completely collapse as the temperature drops and the pressure rises, driving the vapor within the bubble below its condensation point. The heat from the vapor, both from its initial temperature and the gas to liquid phase change, will be added to the liquid. That process will be more or less complete depending on the temperature of the liquid, the pressure in the liquid where the bubbles emerge, and the size of those bubbles. Low temperature liquid with a lot of depth and small bubbles will encourage complete collapse, while higher temperature liquid without much depth and larger bubbles will be more likely to reach the surface of the liquid and burst. The first case will give better separation as the liquid is gently heated, while the second case will give less separation as the liquid is quickly heated and boils turbulently, mixing up heavier and lower boiling components.

The interplay between the size of the retort and the volume of the charge in it play an important role in determining how much heat will be lost from the system through radiant cooling and influence how much reflux is generated in the retort. A larger retort with a smaller charge will result in more cooling and more reflux, while a smaller retort with a larger charge will result in less cooling and less reflux. The charge will be influenced by how the stills are set up to handle the mass balance of the system - vapor enters the retort, gives up its heat, and the alcohol is preferentially vaporized again. The enriched vapor stream leaves water behind, which will tend to increase the amount of liquid in the retort. This is usually dealt with by passing some of the liquid back to the previous pot, but that can be plumbed in different ways. An outlet with a vapor lock part way up the wall of the retort can help to maintain a constant liquid level, while one leaving at the bottom will have a flow dependent on relative pressures in each vessel, though this can also be controlled with a valve if the distiller wants to vary the conditions over the course of a run. In some ways this is also analogous to a purifier pipe in the lyne arm of a pot still, passing material back to be redistilled and giving a greater amount of total reflux through the system.

All of these parameters give a distiller multiple ways to control the process and output, resulting in full-bodied 'pot still' spirits in a single run that would take a standard pot still two to three distillations to match. In my next article in this series I will describe how this concept was transformed into the batch column stills that have become so common in the craft distilling industry.

New Cocktails: Turkey Shoot

While rye and Ramazzotti go together in any number of drinks ranging from the Brooklyn to the Black Manhattan, I've never seen them together in the mold of a Boulevardier, with the close exception of the Canon Cocktail by Jamie Boudreau, which still fancies up that basic combination. But sometimes simplicity is just as good as complexity.

Turkey Shoot

1.25 oz Wild Turkey 101 Rye
0.5 oz Ramazzotti
0.5 oz sweet vermouth

Combine all ingredients, stir with ice for fifteen seconds, then strain into a chill cocktail glass and garnish with a strip of orange peel.

The nose is balanced between rye grain, orange peel, and herbs/mint. The sip opens with sweet grape notes, then it counterbalanced by bitter notes from the rye and amaro/vermouth, with a grain and herbal fade at the back, and a rounded orange note throughout.

The herbal character of Ramazzotti goes really well with the spiciness of rye, giving a different emphasis than with Campari in the classic Boulevardier. With that said, it does take a fair amount of rye to hold up against the very strong flavors of the amaro, so if you're making this with a lower proof rye you might need to up it to 1.5 oz to keep everything in balance. Overall I'm really happy with how this turned out.

Whisky Review: A.D. Rattray Macallan 15 Year 1995/2011

Independently bottled Macallan has grown even thinner on the ground lately as the blending casks the distillery was willing to sell before the 2000s were bottled. This particular cask was a double rarity - teenage Macallan from an ex-bourbon cask, very different from the sherry-driven style they're known for.

This whisky was distilled on October 23rd 1995, filled into an ex-bourbon cask, then bottled on April 27th 2011 at 46% without coloring or chill filtration in an outturn of 334 bottles.

Thanks to MAO for this sample.

A.D. Rattray Macallan 15 Year 1995/2011 Cask #11251

Nose: rich bourbon cask influence - caramel, vanilla, graham crackers, and mild oak - apple/pear/apricot notes, powdered lemonade, gently floral, some ethyl acetate, a little dusty/musty (but in a good way). After adding a few drops of water the vanilla and floral notes expand, the orchard fruit are joined by berries, the oak becomes honied, and the ethyl acetate is better integrated.

Taste: sweet up front balanced by some alcohol heat, bourbon cask influence of caramel and mildly tannic oak around the middle, with a vague orchard fruitiness and citric tang throughout. After dilution the alcohol heat largely disappears, leaving big sweetness up front and less tannic oak in the middle, but the slide into the finish falls a little flat.

Finish: a little savory/tannic oak, clean malt, vanillin, a little fruit residue

While far more drinkable than the Whisky Galore Macallan, the clear ethyl acetate note throughout establishes a clear lineage. I think this one is rescued from disaster by more time and a more active cask, but the flaws keep it from being an unqualified winner. I can also see how this spirit works better when augmented with sherry casks. While it doesn't quite click for me, dilution helped in a way that makes me wonder if it would have been better off at 43% to begin with.

Whisky Review: Whisky Galore Macallan 10 Year 1992/2003

If Macallan is known for one thing it is its sherry cask matured spirit. While the "Oh crap, we can't keep up with demand" Fine Oak series blended bourbon casks (probably originally intended for blends) with a smaller proportion of sherry casks, the only way to get pure bourbon cask matured Macallan has been independent bottlers. This one is from one of Duncan Taylor's lines that was eventually dropped in favor of their NC^2 line. Both offered largely younger whisky at 46% without coloring or chill filtration at fairly reasonable prices.

This whisky was distilled in 1992, filled into (probably Nth refill) ex-bourbon casks, then bottled in 2003 at 46% without coloring or chill filtration.

Whisky Galore Macallan 10 Year 1992/2003

Nose: so much ethyl acetate, clean malt, a little green/grassy, unripe apples/pears/peaches, some vanilla in the background, very little oak, orange peel. After adding a few drops of water the ethyl acetate settles down a bit and it gets maltier, the vanilla is amplified but the fruit notes fade a bit into the background.

Taste: malt sweetness with an undercurrent of ethyl acetate throughout, a pleasant thickness around the middle, and a bit of savory oak/malt going into the finish. After dilution it becomes much sweeter, the fruitiness around the middle is amplified, and the savory notes at the back are joined by some orange peel, but there's also more grassy bitterness in the middle.

Finish: surprisingly long recapitulation of the aromas - bittersweet malt, grassy, unripe apples and pears, a little vanilla, distant oak, then straight ethyl acetate lingers for minutes afterwards

This was a somewhat disappointing whisky. While there's clearly some good, fruity spirit in there, it's the strong ethyl acetate notes throughout make it hard to enjoy what it has to offer. With slightly more active cask with more porous wood I can envision this being much better, but as is I feel like it would need many more years of evaporation to settle into something good. It also seems rather similar to the Whisky Galore Glenlossie I tried a while back, which also had some prominent ethyl acetate notes. I'm more than a little surprised that these casks were chosen for bottling, even with such a famous name attached, but I've questioned the bottling decisions made at Duncan Taylor during the 2000s and this seems of a piece with them.