Continuous Mash Engine (CME)

[stevea]

This note describes the ideas behind a working lab-scale mash system I built, with a partner. I think this system probably has little direct application to small-scale distilling or brewing, but perhaps someone can prove me wrong. I hope.

The lab system devised was able to produce 125ml of mash per minute, or 180 liter [47.5gal] day. If you want 47.5 gallons of wash once in a while, it’s far easier to get a drum, a big burner and do a batch mash. If instead you want to make 47.5 gallons of mash every day, then an automated continuous system has advantages.

The goal was to develop this system, and controls, at lab scale then upsize for a full production whiskey distillery at several liters per minute capacity. I’ll describe an example implementation at 1 liter/minute capacity in subsequent posts. Keep in mind that 1 lpm is 1440 l/day, [380gal/day] and enough wash to recover 1 to 1.15 barrels of barrel-strength whiskey per day. The lab system had some basic automation, such as temperature PIDs for each tank, software controlled enzyme pumps, and stirrers, and some alarm software. We performed lab-runs up to about 13 hours. A little added automation & debugging should make the system capable of 24x7 operation, with periodic attendance to manage grist, enzymes and other inputs & fermenter output.

Rather than launch into theory, I think it’s best to describe the system, then add some background later.

(continued)

[stevea]

Processing grains for the production of mash involves a lot of intricate steps.. The CME implements a viscosity rest, a gelatinization rest, and a saccharification rest; however it performs these in separate tanks connected by pumps. The tanks are kept at distinct temperatures using PID control, stirred continuously, a constant flow rate is maintained with pumps between the tanks. Each tank is sized to account for the reaction time of the rest and the flow rate. Enzymes are added to each tank using metering pumps. The pH of the tanks must be adjusted (specific to the enzyme selection) using manually adjusted metering pumps that add acids or bases.

For the lab-scale prototype system Dow Danisco enzymes were used, specifically.
LAMINEX® C2K - for viscosity reduction (TANK1)
AMYLEX® 5T - a high temperature alpha-amylase for gelatinization (TANK2)
DIAZYME® SSF2 - for saccharification (TANK3)


Viscosity Rest (TANK1):
Viscosity reduction is not needed for corn, but is definitely needed for rye and wheat. The Danisco enzyme chosen works reasonably well, however it was ‘finicky’ wrt pH, and a pH < 4.8 was needed for a 100% rye mash. The temperature of 58C was chosen based on Danisco data sheets. This mash step was acidified using phosphoric acid in the lab.

Gelatinization Rest (TANK2):
Gelatinization involves two reactions. First, there is a simple physical chemistry process as macro-molecular starch granules, in the presence of water and heat, expand and unfurl into massive tree-like molecules while trapping large amounts of water molecules between the branches. A second reaction, where heat-tolerant alpha-amylase prunes the tree into modest size branches and frees many water molecules which prevents turning the mash into “mashed potatoes'” (iow very thick). A temperature of 90C was found to be entirely sufficient for excellent gelatinization. Unfortunately the Amylex T5 prefers a higher pH than the C2K, so a lot of pH juggling was necessary. This step was made more alkaline, using a calcium hydroxide solution. At scale an alkaline yeast nutrient, like diammonium phosphate might be used to adjust pH upward.

Saccharification Rest (TANK3):
The Saccharification rest uses an amylase that converts the starch and dextrins into almost-pure glucose. The SSF2 enzyme prefers a pH around 4.5, so acidification with phosphoric acid was used at this step. Because this enzyme continues to operate in the fermenter, we found we could reduce this step from ~99% conversion to ~75% conversion in a smaller TANK3. The saccharification completed in the fermenter and the final ABV levels were not affected!

Enzyme Note:
I believe that Danisco and others now produce an alternative to Amylex T5 which has a lower pH optima. This should eliminate at least one of the pH “wickets”, and maybe two. It *might* be possible to set the pH in TANK1 and do no more into the fermenter.
Also b/c the T5 is stable at TANK1 pH & temperature, we decided to add both C2K & T5 enzymes into TANK1, thus eliminating a metering pump.

(continued)

[stevea]

Lab Prototype:
We used micro-pumps (60ul per jot, periodic pulses) controlled by a raspberry pi, sending dilute enzymes. Pumping of the in-process mash used peristaltic pumps. The tanks were conventional restaurant scale aluminum pots, each with a motorized stirrer and temperature senders attached. The stirrers were controlled by relays from the raspberry pi. The pots, each sat in a separate water filled “food service” type steamer table bath, and the pot temperatures were controlled by a PID & TRIAC circuit to the water bath.

Peristaltic pumps are capable of transferring some amount of solids, however it is necessary to keep those solids in a sort of “suspension” with high enough flow velocity. Also you need to select the tube material carefully to avoid excessive wear. These are barely OK for the lab, but probably not production.

Production Scale (1 lpm and greater) considerations:
Here is the big headline. The tanks needed for a 1 lpm system are remarkably small!
TANK1 holds 34 liters [9 gal]
TANK2 holds 64 liters [17 gal]
TANK3 holds 114 liters [30.5 gal]
Of course you need to add 25% extra for a practical tank size, so ~{42, 80, 145liter} or [11, 21, 38 gallon].

If you need a 2 lpm CME, just double the tank volumes, and flow rate. We’ve considered systems as large as 15 lpm.

It seems that a lot of components required for a production sstem (>1lpm) are too large for basic lab equipment, and too small for small brewery equipment - an awkward size.

Grist & Water. 1 liter of mash contains ~240 grams of grist (~a cup), and ~810ml of water (similar to 2lb/gal distillers mash). Metering process water with a pump is easy, but how do you feed ~half a pound of dry grist per minute from your grist-case holding a ~750lb day supply ? You need a very small mini-rotary valve & control GEAR motors controller to a few RPM.. The sort used for powder & chem processing. BTW each addition (water, grist, enzymes, acids) can be delivered in 1-a-minute pulses or faster. This only causes a small perturbation of the continuous process.

Mash pumps: There are a lot of positive displacement pumps able to transfer solids, and each has pros and cons. Peristaltics - I’d choose something else, but they may operate better at a larger scale. Diagram pumps are (oddly) mostly air-operated = noisy low efficiency and a bit ‘pulse-y’. The diaphragms will wear. Moyno/Netzsch pumps have fantastic pressure characteristics. Some are used to pump damp, spent grist out of brewery mash tuns. I think that lobe pumps are the best choice. *** Special note: B/c the tanks at 1lpm or even 2lpm are so small, it may be possible to arrange them vertically and use a solenoid valve a a down-pipe, and avoid or reduce the considerable cost of the pumps.


Enzyme Pumps/Acid Pumps: These enzymes are applied at rates between 180 to 700 grams(almost = ml) per metric tonne of grist. So ~50-200 microiter/minutefor a 1lpm CME. There are micropumps made in the 50-200uL per ‘squirt’ range. We used some on the lab-scale system by diluting the enzymes 10x. It’s easy to adjust the strength of acid and base solutions, by dilution, and there are loads of metering pumps that will control in the 0-50ml/min range.

Heating&Cooling Energy: An interesting problem. Aside from TANK1 dealing with ambient temperature water & grist, power (heating and cooling) are constant. For 1lpm at …,
TANK1: heat 2.4 - 3.1 kW (varies with ambient temperature).
TANK2: heat 2.2 kW
TANK3: cool -1.9kW
Fermenter: cool -2.5 kW

If you assume electric (there are cheaper choices at larger scale), and 90% efficient heating, 65% efficiency cooling you are using ~300kW-hr/day.

*** NOTE: Most of that energy could be saved by use of heat exchangers between the obvious points. Unfortunately passive heat exchangers capable of managing grist-in mash (non-plate HX) have limited capabilities. Cheap “homebrew” type counter-flow exchangers might save ~20% of energy. Active (heat-pump) type exchangers could recoup a great majority of energy, but most are made for large industrial apps. At small scale (1-2 lpm) adding a couple inexpensive counterflows is likely worthwhile and woud pay-off in a couple months of operation (but grist sedimentation becomes an issue).

(continued)

[stevea]

How to apply the heat or cooling ? The obvious approach uses jacketed tanks. We have to stir the tanks anyway, so … but you can only transfer about 300W/meter^2/degC of heat to a load, and you can only cool ~150W/meter^2/degC. Heating with 115C water at 1 atm above atmospheric, there is no problem heating and no problem cooling to TANK3 with 20C water. The heat-exchanger to the fermenter will require active cooling ~10C & an excellent 1-pass heat exchanger. Inserting in-tank HX coils can be very effective at larger scale (>3lpm). Resource:
https://thermopedia.com/content/547/


As you increase the flow rate and size of the CME the jacket heating problems become more difficult. This is because the power required increases linearly to the flow rate(F), but the jacket area increases at F^⅔. [example, 8lpm requires 8 times the power, but the tanks have only 4 times the area 4=8^⅔ ]. At ~5 lpm you can’t apply all the power needed to TANK2 by jacket. You’ll need an input heat exchanger or internal coil or … At ~7 lpm TANK2 can’t get enough jacket cooling.

Time for some theory.
(continued)

[stevea]

Theory:

For a first order reaction, including approximately these enzyme catalyzed reactions, the reaction slows as the concentration of reactants C is used up. See the reference for details, but we can say for a closed batch reactor:

C(t)/C(0) = e^-kt
where.
t is time
C(t) is the concentration of reactants at time t
k is a rate constant with units of 1/time.

At t=0, e^0 = 1, So 100% of the reactants remain, and 0% of product.
At t = 1/k, then C(1/k)/C(0) = e^-1 = 37%, so only 37% of reactant remains,
63% of final product.
At t = 2/k, then C(2/k)/C(0) = e^-2 = 13.5%, only 13.5% reactant remains,
86.5% of product.
…
Now consider a mixed-flow reactor, where a tank is filled to volume V with an in-flow = out-flow = F (liter/sec for example) and the reaction still takes place. The mixing means that the entire tank has an equal concentration of reactant. Eventually the system reaches a steady-state where the rate of product produced (from reactants) equals the amount of product in the outflow.

Let’s call the reactant concentration in the tank Ctank, the inflow concentration Cin. The product concentration Ptank compared to the final product possible Pfinal. We know the fraction of product formed so-far is directly related to the amount of reactant remaining by:
Ptank/Pfinal = 1 - (Ctank/Cin)
IOW of 70% of the product has formed, then only 30% of the reactant remains.

So at steady state, and for a tiny time interval dt, we can compare this to the batch case and (I’ll spare you the algebra and differential) …
V/F = ((Cin/Ctank)-1)/k ;or
k*V/F + 1 = Cin/Ctank
and
Ptank/Pfinal = 1 - (1 / ((kV/F)+1) )

(continued)

[stevea]

EXAMPLE:

Let’s try an example calculation, to clarify the confusion. Imagine a BATCH saccharification tank, similar to TANK3 but no flow. Say we determine after 50 minutes that 77% of the glucose-glucose bonds are broken. That means 23% remain.
C(50min/k)/C(0) = 0.23 = e^-k*50min
1.47 = k*50min
k = 1/(34 minutes).

Now imagine that we want to create a mixed-flow reactor that gets the same 77% completion. How big must the tank be ?

Ptank/Pfinal = 0.77 = 1 - (1/ ((kV/F)+1))
0.23 = 1/((kV/F)+1)

kV/F = 3.35
V/F = 3.35 * 34min = 114 minutes

So if F = 1 liter/minute, the V = 114 liters (the tank volume); V/F = 114 minutes.

(continued)

[stevea]

More Theory: CSTR
A “Continuous Stirred Tank Reactor” consists of one or more sequential tanks performing the same reaction. *Sometimes* splitting a reaction this way results in dramatically smaller tank volumes. Just as an example, the 1 lpm CME requires a single TANK3 with 114 liter liquid volume. If we split this among two sequential reactors, each would only be 37 liters! Three sequential 21l reactors would to the same job. Of source given the small size (114 l) and the cost of pumps and stirrers and controls, this isn’t justified here.



Resources:
"Chemical Reaction Engineering", by Octave Levelspiel

“Brewing Yeast and Fermentation”, Boulton & Quain



How much product (or reactant conversion) do we really need ?

It’s a complex question, but let’s consider the 3 ‘rests’.

The viscosity rest uses enzymes to break apart complex and viscous pentosans, gums, and arabinose. It appears that we don’t need to do much removal of these polymeric bonds to reduce viscosity a lot. OTOH the enzyme reaction rate seems pretty slow. We estimate that removing 20% of bonds is sufficient and that in a batch reactor this takes 30 minutes. Which give us a k value of (1/(135 minutes)). V/F = 34 minutes.

The gelatinization rest has two components. The unfurling and swelling of starch granules is very rapid and not first order, it is complete in a couple minutes. The alpha-amylase reduction of those complex carbs is rapid, and only a very incomplete reduction is needed to prevent “mashed potatoes”. We estimate 10% completion of G-G bond hydrolysis is needed and this is accomplished in a batch system in under 5 minutes. This gives a k value of ~ 1/(43 min)). The enzymatic action is the drive for the tank size calculation. V/F = 64 minutes (some extra margin was allowed here).

We originally targeted a 99% G-G bond reduction in the saccharification rest tank, however this requires a rather large TANK3 volume, It soon became apparent that the SSF2 enzyme could do a lot of work in the fermenter. We chose and tested a 77% conversion ratio, (~50 minutes batch, k = (1/(34 minutes)). V/F = 114 minutes. This works well in practice. If time and budget had allowed, we would have tested lower conversion rates (smaller TANK3 arrangements). Reducing TANK3 to the same size as TANK2 (untested) would result in a ~40% conversion in TANK3. Whether SSF2 can complete the other ~60% of the task in the fermenter remains untested, but it seems likely.

The k values above are conservative, with the saccharification tank numbers being the most well studied and accurate.

(continued)

[stevea]

UNEXPLORED FERMENTER OPTION:
Adding a TANK4, for yeast propagation and starting the fermentation strongly, without lag was considered, but not implemented or tested. Levelspiel text covers the dynamics of generic ‘fermentations’. The yeast dynamics are not ‘first order’ tho’ they are quite close during the “log” growth phase. Boulton and Quain document that yeast ‘doubling time’ in “log” phase is under 167 minutes (and often lower). If we do the calculations, then we could add a TANK4 with a V/F = 241 minutes [a 241 liter volume (64 gal) for a 1lpm CME]. Once mash made it through the system, into TANK4, aeration and yeast sufficient for 64 gallons could be added. A delay sufficient to account for yeast start lag period, might be added. Then aerated mash is added, and the yeast growth & concentration should keep up with the inlet mash.

Because SSF type enzymes can saccharify right in the fermenter, it's possible to combine TANK3 & TANK4 into one batch-oriented cooling tank, and add SSF enzymes when the temperature is @ 63C ,then add yeast when the temperature is ~28C. This allows for slower passive cooling therefore a lower cooling power requirement. This approach (and elimination of TANK1) is how the corn-fuel ethanol operations work.

(continued)

[stevea]

Operational Considerations:

How do we get the system to ‘steady state’(SS), at startup ?

First, let’s consider basics. At 1lpm flow the tanks (optimally) fill at times:
TANK1: T=34min
TANK2: 34+64 = T=98min
TANK3: 34+64+114 = T=212 min (3.6 hr)
So ~3.6 hours (212min) after startup the first drops of fermentable wash appear in the fermenter.

But if we start filling TANK2 at the moment TANK1 fills - are we at “steady-state” ? No! The content of TANK1 (in this example) is “under-reacted” for several reasons. The tanks start’ “cold” and the concentrations of reactants only increase at full-rate once the tank is filled. We should delay for an amount of time equal to ~10 minutes per tank (specific nums available), Yes, but let’s consider the costs.

STUTTER-START REGIME: imagine you start the system, fill TANK1 @34min later, then *wait* (stirrers, but no pumps nor grist or water&-in). Then start the pump. to reach SS ? Once TANK2 fills, we again delay 10min till reactants reach SS. …

What does STUTTER-START do ? We estimate (high estimate) that STUTTER-START prevents loss of ~6% of volume (212l) initial fill of mash. The delays after each fill allow the tank to reach steady state. If you start filling TANK(n) immediately after TANK(n-1) fills then you are pushing a little under-reacted product through the system. With the stutters, you save~12.7 liters of mash,but you lose half an hour of production time (30l).

How do we shutdown ?

Stop inflow to TANK1. (T=0).
At T=34 minutes, when TANK1 becomes empty, shutoff the first pump.
At T=98 minutes, when TANK2 beconmes empty, shutoff the second pump.
Also at T=98 min, the contents of TANK3 is fully reacted, and TANK3 can be emptied into the fermenter as fast as the heat-exchanger can handle the load.

--end--

Happy to answer question, or add, correct details.

[stevea]

TL;DR ?
I suppose I should have begun with a thumbnail sketch.

The system presented allows continuous mash production using some remarkably small tanks for <viscosity, gelatinization, saccharification>. A 1 liter/minute flow rate system requires tanks liquid volumes of <60 (228l) gallons total, and produces 380 gal (1440 l) of wash from ~760lb (345kg)of grist per day!. The system can scale up or down, although applying & removing heat becomes more challenging above ~4lpm (~1 gal/minute) flow rate.

A 1lpm system, operated 24hr * 20 days per month (240 days per year) would generate enough recoverable ethanol (~33000 LA) using conventional continuous still to fill ~265 barrels at cask strength.

The heat exchange issues for very large scale (>4lpm or >1000 bbl/yr) are addressable, but only by uncommon heat exchangers (see 'spiral heat exchangers'), evaporative cooling or unusual tank design.

The system also presents an opportunity to recoup a large fraction of the heating & cooling via heat exchangers, however heat-exchange of a grist-in wash is technically challenging, particularly in a continuous system.

Although the small tanks are relatively inexpensive, the required grist conveyor, metering pumps, stirrers and especially mash pumps add to the capital cost. The main saving is therefore due to very low labor costs of production. Also in some cases to lower power (not energy) costs.

This schema was tested at lab scale (~125ml/min) and worked well, producing good quality wash with excellent fermentability. Yeast nutrients were added.

-S

[Demy]

For the preparation of beer it uses a small mash-tun with motorized blades, the engine is a 12V powered car wiper, it works beautifully.

[greggn]

The goal was to develop this system, and controls, at lab scale then upsize for a full production whiskey distillery at several liters per minute capacity.

What's the advantage of producing a continuous output of wort given that fermentation is batch process ?

[LWTCS]

Why no jet cooker?

[NormandieStill]

The goal was to develop this system, and controls, at lab scale then upsize for a full production whiskey distillery at several liters per minute capacity.

What's the advantage of producing a continuous output of wort given that fermentation is batch process ?

There's a possibility of continuous fermentation as well I presume, but I can easily imagine that in terms of work load and scheduling, having a system that generates wort continuously is easier to manage. You can easily ensure that you have an store of raw materials and the number of people involved will be fewer meaning that when Bob calls in sick with <ILLNESS> you don't have to send admin staff into the washbacks!

Obviously there'd be a cost analysis to be carried out, but I could imagine it being interesting even if your fermentation continues to be a batch process.

[stevea]

For the preparation of beer it uses a small mash-tun with motorized blades, the engine is a 12V powered car wiper, it works beautifully.

Will car wiper motors operate continuously for days under load ?

[stevea]

The goal was to develop this system, and controls, at lab scale then upsize for a full production whiskey distillery at several liters per minute capacity.

What's the advantage of producing a continuous output of wort given that fermentation is batch process ?

Cost. This system would allow a medium size distillery to approach (not reach) the product cost structure of big producers.
The capital equipment is very small, and somewhat less expensive for the capacity.
The amount of labor involved is dramatically less (a large cost item).
Power use is constant; there is no peak energy demand significantly above average.
As noted, there are great opportunities for power saving by heat recovery since you are heating one side of the process while at the same time cooling the other.
You can make make small incremental changes for improvement dynamically rather than seat-of-the-pants batch-by-batch tweaking.

Most small distilleries are very labor intensiveness, have poor or non-extant quality-control, and generally must sell their products at ultra-premium prices (>$50/btl) regardless of actual quality or age.

==
There actually is work, at a defunct Tasmanian brewery IIRC, and recently in Germany, toward continuous beer fermentation. These methods are not optimal for grist-in mash fermentation. Also note that a set of fermenters can be filled, cycled, and monitored, temperature controlled with some simple automation. But ultimately it's necessary to wait the 60-72 hours for fermentation to complete, and this is somewhat variable.

The system is "batched" at the still too. You can't realistically or safely operate a still without humans present, and manning a small still 24x7 is expensive and problematic. Most commercial continuous stills have a feed capacity several times larger than this CME. So logically you'd run the still periodically; liked 3 shifts per week.

[stevea]

Why no jet cooker?

Excellent question. For those unaware, a jet cooker is basically an eductor that pulls wet mash through an oriface gap using steam pressure and thus instantly gelatinizes the mash grist starch. Corn-fuel ethanol pants use these, however their corn has been wetted for several days and degermed. We discussed this with a vendor and got a work-up from their engineers. There is a relationship between the rate of mash flow through the jet cooker, and the orifice gap size. But the oriface gap determines the largest grist particle size that the system can handle w/o blockage.

At our flow rate (I'd have to look it up, but it was probably ~5 l/minute) we'd need to guarantee that the largest grain particles would be under 300 microns IIRC. Our milling specs was close, but guaranteeing no blockage wasn't likely w/ a hammer milling operation.

Also some Scottish distilleries have tested, then rejected even steam injection systems, claiming it causes excessive Maillard sections impacting flavor.

It was seriously considered, but risk and cost were factors.

[Demy]

For the preparation of beer it uses a small mash-tun with motorized blades, the engine is a 12V powered car wiper, it works beautifully.

Will car wiper motors operate continuously for days under load ?

The maximum time I have used is 2 days for my malt drying system, you could consider wipers for trucks, bigger. Of course they are not designed for that job but I have used them without problems also to mix the mash, the idea was born from some homebrewers here in Italy (forum) but it is probable that the idea has also been adopted elsewhere.

[LWTCS]

Excessive Maillard? Hmm?
Were you able to sample for yourself how excessive?

Not everybody likes foie grass evidently.

[zapata]

Lol, bit late, but I need to say this is a fantastic thread, kudos!

And I figure I can add a bit of tangential info on fermentation. Grey Goose is continuously fermented, as is the New Zealand brewery whose name escapes me but is pretty well documented (and patents long ago expired). A number of other breweries have decided continuous fermentation is not conducive to the variety demands of modern breweries at all but the largest scales. But many have found continuous secondary fermentation is much faster at maturing beer eliminating days and weeks of conditioning time.

As mentioned grain in would pose different challenges than a brewery has, but continuous breweries have to deal with excess yeast flocculant too so it may not be too different. I imagine that separating yeast from grain would be the logistical problem, and the easiest solution would be to just not recycle yeast. Of course separating yeast and grain solids would be possible it would just require additional settling vessels. And it's just the cost of yeast vs. vessels that would determine the outcome.

Most continuous fermentation schemes use multiple vessels in a cascade arrangement. One proven but abandoned method differed in using one tank tall enough that it naturally stratified into a sort of "plug flow" with fresh wort and yeast fermenting quickly at the bottom, settling and finishing in the middle, and clear finished beer exiting the top. This arrangement may work for grain in ferment depending on how grain particles differ from yeast. It is entirely possible that grain will settle and can be removed from the bottom below the active yeast.

Ultimately the largest hindrance to whiskey distilleries is probably the required sterility. Continuous breweries are even more attentive to sterile technique than batch breweries. An infection can take the plant down for days or even weeks. But infections are endemic in distilleries and as many of us brewer / distillers know infection free spirits can be excessively clean and bland. But if maintaining one organism in a continuous reactor is difficult, I can't imagine balancing several! Of course you may accept sterile booze, or attempt to compensate in other ways.

[stevea]

Excessive Maillard? Hmm?
Were you able to sample for yourself how excessive?

Not everybody likes foie grass evidently.

Luv the pun : Maillard, mallard, goose => foie gras.

Definitely not. I base this comment on reports on "Whisky Science ..' by Piggott .... They report what some Scottish distillery (Grant's IIRC) tried steam injection and didn't like the results in terms of flavor. I *suspect* that may not be true for US style whiskey where grist-in distillation is destined to create some Maillard products.

There is zero doubt that a jet cooker can wonderfully gelatinize starch .... but no one [well very few] care what fuel ethanol tastes like.

I wouldn't base a biz-plan on a hunch that it may not taste bad. The cost of an experiment is around $10k for the jet cooker then ~$40k the rest of the hardware and funds for a dozen tests. That *seems* like a good test for an established distillery.

[stevea]

Lol, bit late, but I need to say this is a fantastic thread, kudos!

And I figure I can add a bit of tangential info on fermentation. Grey Goose is continuously fermented, as is the New Zealand brewery whose name escapes me but is pretty well documented (and patents long ago expired). A number of other breweries have decided continuous fermentation is not conducive to the variety demands of modern breweries at all but the largest scales. But many have found continuous secondary fermentation is much faster at maturing beer eliminating days and weeks of conditioning time.

Thanks for the great comments!

Morton Coutts 'Dominion Breweries' of NZ. IIRC the NZ brewery was bought up by an (OZ?) brewery then they killed the ~5+-decade continuous fermentation process. From several reports they always had problems keeping their c-fermentation "stable". There are also many studies that continuous fermentation (any stirred/flow fermentation by yeast) increase fusels - not a big issue in beer, maybe not vodka, but a potential disaster in whisk[e]y.

As mentioned grain in would pose different challenges than a brewery has, but continuous breweries have to deal with excess yeast flocculant too so it may not be too different. I imagine that separating yeast from grain would be the logistical problem, and the easiest solution would be to just not recycle yeast. Of course separating yeast and grain solids would be possible it would just require additional settling vessels. And it's just the cost of yeast vs. vessels that would determine the outcome.

There are some interesting studies that show yeast flocculence can be manipulated on in ~10 ferment selections (~40 generations).

You correctly assess one major, of several problems. Many of the papers on continuous fermentation (CF) use "immobilized yeast". These are yeast cells made to adhere to a (typically plastic) matrix. Then there is some German work to create continuous fermentation across a series of tanks with SOME separation, but these all involve taking yeast from stage N+1 and pumping it back to stage N. That's great for a beer that has been cleanly separated after a hops-boil and lauter. It's very hard for a grist-in ferment where lacto-bacilli are an important flavor-factor and secondary ferment.

Can it(continuous distillation) be done - heck-yeah. Is the flavor of distillation product good ... uhhh - dunno. No one knows.

Most continuous fermentation schemes use multiple vessels in a cascade arrangement. One proven but abandoned method differed in using one tank tall enough that it naturally stratified into a sort of "plug flow" with fresh wort and yeast fermenting quickly at the bottom, settling and finishing in the middle, and clear finished beer exiting the top. This arrangement may work for grain in ferment depending on how grain particles differ from yeast. It is entirely possible that grain will settle and can be removed from the bottom below the active yeast.

My recollection is that NZ Dominion brewing system was close to a plug flow case; The Weihenstephaner tech work was ~ 4 semi-batch stages. A MAJOR problem wrt any plug-flow schema is that the product ethanol dilates much better than even water. ContinuosF is interesting - but requires a LOT of work ti justify/rationalize.

Ultimately the largest hindrance to whiskey distilleries is probably the required sterility. Continuous breweries are even more attentive to sterile technique than batch breweries. An infection can take the plant down for days or even weeks. But infections are endemic in distilleries and as many of us brewer / distillers know infection free spirits can be excessively clean and bland. But if maintaining one organism in a continuous reactor is difficult, I can't imagine balancing several! Of course you may accept sterile booze, or attempt to compensate in other ways.

Distilleries aren't sterile, nor even close. Unlike beer-brewing, distilling is closer to wine making where we NUDGE fermentation in our favor. There are many papers on the topic, but essentially all take the stance that LIMITED lacto fermentation, esp late in the process provide a positive flavor impact. We might agree - that some *balance* of a majority yeast product and a minority lacto product *might* be ideal. HOW to achieve that in a continuous fermentation remains a mystery. The answer MAY involve inoculating a wash in late ferment. What we know is that mash gelatinization temps won't eliminate all lacto spores. Lactobacillus grow relatively slowly. It also makes timing of fermentation completion mor problematic.

Nice to chat with someone who has considered the matter deeply.

[LWTCS]

You should get grim's thoughts on this to round out your thought process.

He put together a low budget, tri clamped solution for steam injection with an eductor head. All tri clamp.
Not 40k. Not 10k. Puh,,,prolly not more than 1k.

How many barrels do you plan on laying down a year? More than 600?

He is Silk City Spirits over on ADI.

[stevea]

You should get grim's thoughts on this to round out your thought process.

He put together a low budget, tri clamped solution for steam injection with an eductor head. All tri clamp.
Not 40k. Not 10k. Puh,,,prolly not more than 1k.

How many barrels do you plan on laying down a year? More than 600?

He is Silk City Spirits over on ADI.

Excellent. Yeah I know Silk City from ADI, and respect his comments there greatly. I appreciate his ingenuity as well.
Did he post on that somewhere ? OTOH, when you roll your own production hardware, then YOU are the maintenance,
support & design team, and ... that can make things very slow.

What does a jet-cooker buy you in this CME ? Well, fast gelatinization, but TANK2 size is determined by the
high-temp alpha-amylase reaction. The open question about Maillard reactions remains. You need culinary grade steam,
to inject and that's a different story than a closed steam system. It means you either buy a separate small boiler for the jet cooker,
or you drive the entire mash system from an ro-water, OR* use the rather expensive FDA approved water additives to avoid scaling.
jet cooker will req about 8 gallons of RO water steam every hour, 24x7, on a full size 7.5lpm system (~6-7% dilution).

The biz is defunct. The second iteration biz-plan showed we need to be at least at high-500s bbl/yr to maintain a good margin.
(IOW pay ALL costs, return on capital, plus a 10-14% margin). 700 bbl/yr was the target with expansion capability to 1100 bbl/yr.
With barrel production costs slightly above $600/bbl, then all the capital, overhead and marketing costs - you'll have
a good sum invested by year 5 when you have a 4yo product to sell. (it takes almost a year to get a still fab'ed and delivered).

==

FWIW - these links are about the German continuous fermentation efforts out of he Technical University of Munich (TUM) and seems to be related or perhaps funded by Weihenstephaner.

https://phys.org/news/2013-05-sustainab ... ation.html (fluff news piece)
https://www.asbcnet.org/events/archives ... ermann.pdf
https://www.brewer-world.com/dynamic-fe ... -poseidon/

They went from a lab model ~50l(13gal = 0.4ber-bbl) per day, to a test plant at 5000l(1300gal = 46.2 beer-bbl)/day. And in 2019 they were going to scale up to 200kliter (53k gal = 1700 beer-bbl) per day. That last is abt half the production capacity of Sierra Nevada. So it seems they expect market acceptance of continuously fermented beer.

[LWTCS]

@stevea,
Yeah, he posted a bit about it on the ADI forum.
He hangs out pretty regularly on the SD forum as well.

I think his goal was to eliminate man hours associated with cooking.

[Evil Wizard]

I built a jet cooker recently out of copper fittings. Steam was supplied by a bathroom/spa steam generator. Grist was malt from CMG that I hammer milled fine, stirred up with enzymes and citric in cool water.

I had a valve to control slurry flow, the 11,000 watts of clean steam was full blast.

The internal steam nozzle is able to slide so I can adjust the slurry gap. I had to ride the controls too frequently so I think I'll put the steam injector well back from the slurry input.

The cooker screwed into triclamp fittings so I could use 2" sight glasses containing helical mixers elements to increase dwell time. (Thanks Grim for that one photo you posted years ago)

This setup dumped gelatinized malt into my fermenter with the consistency of soft serve ice cream or pudding. This was about 65C. Upon stirring my fermenter, it was all nice and thin liquid.

I ran 7 full bags of 55lb in a shift, made 600L at 25brix. That's much higher conversion than I could get with a mash tun, not that I have space for a mash tun.

This is fermenting madly and I'll run it through my continuous still which I just outfitted with plates instead of packing. It is set up for 20,000 watts of steam heat injected into the column and 4500watts in the reboiler.

Its a big advantage being only limited by fermenter capacity and not cooker (mash tun or pot) capacity or still kettle capacity.

[LWTCS]

I built a jet cooker recently out of copper fittings. Steam was supplied by a bathroom/spa steam generator. Grist was malt from CMG that I hammer milled fine, stirred up with enzymes and citric in cool water.

I had a valve to control slurry flow, the 11,000 watts of clean steam was full blast.

The internal steam nozzle is able to slide so I can adjust the slurry gap. I had to ride the controls too frequently so I think I'll put the steam injector well back from the slurry input.

The cooker screwed into triclamp fittings so I could use 2" sight glasses containing helical mixers elements to increase dwell time. (Thanks Grim for that one photo you posted years ago)

This setup dumped gelatinized malt into my fermenter with the consistency of soft serve ice cream or pudding. This was about 65C. Upon stirring my fermenter, it was all nice and thin liquid.

I ran 7 full bags of 55lb in a shift, made 600L at 25brix. That's much higher conversion than I could get with a mash tun, not that I have space for a mash tun.

This is fermenting madly and I'll run it through my continuous still which I just outfitted with plates instead of packing. It is set up for 20,000 watts of steam heat injected into the column and 4500watts in the reboiler.

Its a big advantage being only limited by fermenter capacity and not cooker (mash tun or pot) capacity or still kettle capacity.

Nice.
What have been your challenges with sanitation?
Are you using this for production or just a proof of concept experiment?
Any evidence/feedback on too much Maillard reaction?
I'm thinking bourbon and rye here so I don't see Maillard reaction being an undesirable issue? Especially with the material being on the move,,,,continously.
Also, I don't see "clean steam" being a major issue as long as good boiler maintenance with companies like Nalco's water treatment/maintenance is in place. We are not making beer for consumption. Ultimately we are distilling no?
And ,,about how much $$$ have you invested into the process?

[LWTCS]

I envision a slurry mixing tank pumping over to a vessel with the jet cooker.
Then over into a series of jacketed tubes/pipes that are heated with conventional steam. Perhaps static mixers installed in the tubes?
The array of tubes purposes is to ensure adequate dwell / cook time.
Temps of 190 may not be as critical since the jet cooker will expedite gelatinization?

Just visualizing from a much larger system that cereal companies ( like General Mills) use to make breakfast cereal.

[LWTCS]

Curious to know those of you using your thumper outfits to cook grains with direct steam how much faster you are able to achieve adequate gelatinization?