• Advertisement

Column design considerations for information.

These usually have the means for returning the spirits to the column and usually allow metering of the take off to help in the reflux.

Column design considerations for information.

Postby snuffy » Sun Jan 31, 2010 2:58 pm

Packed Columns

Introduction

If you are a reader in a hurry, just skip to the conclusions at the end. Everything between here and there is mostly long-winded explanation leading to the conclusions.

The packed column still was introduced through Mike Nixon and Mike McCaw's book, The Compleat Distiller. Through their work with the Amphora Society, the use of structured copper mesh packing was refined and made available to the distilling community. Structured copper mesh is a knitted tube of fine, flattened copper wire. It has two main uses in commerce: as a mild abrasive for cleaning cooking vessels and plastic molding equipment; and as a material for blocking rodents from entering buildings. The mesh was originally introduced as copper scrubbing pads for kitchen use, but later the industrial suppliers of bulk mesh became a more affordable source.

A basic distilling column is typically, 2" (51mm) or 3" (76mm) in diameter. The 2" column is the most popular, but 3" is becoming more widely used for the greater production capacity. Typical boiler sizes range from 7.5 gal to 15 gal with stainless steel beer kegs or large stock pots being the most common forms from which they are made. The boilers can be heated with natural gas, propane (LPG) or electricity. 4500W electric water heater elements are frequently used in conjunction with some sort of power controller. The so-called "low density" elements are especially esteemed because their lower surface temperature is less likely to cause scorching of suspended solids in the wash or mash.

The dynamics of a packed column consist of rising vapor from the boiler at the bottom contacting liquid falling from the reflux condenser at the top of the column. The repeated condensation and evaporation of the liquid and vapor streams creates a dynamic thermal equilibrium with the column containing a lower boiling (and cooler) liquid / saturated vapor mixture at the top and a higher boiling (and warmer) mixture at the bottom. The cooler mixture at the top of the column is richer in alcohol and the warmer mixture at the bottom is richer in water.

1. Surface tension and reflux - centering collars

Alcohol has a lower surface tension than water. This is responsible for the lower boiling point of alcohol compared to water and also affects how the liquid interacts with the mesh at various heights in the column. Lower down, the higher surface tension of the water predominates and the liquid reflux is more prone to form large droplets, pools and streams. Where streams form, called channeling, the liquid flows rapidly downwards without much opportunity to contact the rising vapor, exchange heat, and separate the alcohol. In the upper section of the column, the lower surface tension allows the reflux to spread more evenly over the mesh, thus exposing more surface area to the rising vapor and more efficiently exchanging heat between the liquid and vapor phases.

This leads to an interesting -- and troublesome -- situation where the column operates at different thermal efficiencies at different heights. Less efficient lower down, and more efficient higher up. The best remedy to date is the introduction of "reflux centering collars" as first described by Riku in his book, Designing and Building Automatic Stills. The collars are funnel-shaped and collect the reflux into the center of the column. The size of the center hole is best kept to more than 20% of the cross-sectional area of the column. When the restriction is greater than this, the increased velocity of the rising vapor is sufficient to entrain the falling reflux and blow it back up through the central hole. When this occurs, a layer of liquid will form in the column above the centering collar and be unable to descend. A rough rule of thumb for finding the minimum hole radius for a centering collar is to divide the radius of the column by the square root of 5 (approx 2.24). Thus the minimum center hole for a 2" column would be greater than 1 / 2.24 = 0.45" radius for a diameter of approximately 0.9" ( or 15/16" or 24mm). For a 3" column, the center hole in a collar should be greater than 1.34" (1 3/8" or 35mm). If the hole in the collar is smaller than this, it will cause the column to "choke" on a layer of liquid above the collar. More about choking can be found below.

There is dispute about the proper location for centering collars. This is due to confusion over the effect of a centering collar of reducing the effective diameter of the column. This can have beneficial effects when a collar is placed above the tee of the vapor management (VM) takeoff, but this is the result of restricting the vapor, not centering the liquid reflux. This will be covered in more detail in the later section on the vapor management head. For the time being, it should suffice that the best locations for reflux centering collars is in the lower half of the distillation column. This is the area where the increased surface tension (due to the higher water content of the reflux) is most prone to cause channeling and poor contact between vapor and liquid. Riku has reported substantial increases in the rate of output with the use of two reflux centering collars in the lower half of the column.

2. Packing - copper and stainlees steel - copper annealing

Column packing remains somewhat of a mystery. I've experimented with several different weights of copper mesh and also one weight of stainless steel mesh. To date, I have not used scrubbers (stainless scrubbers have a wider profile to the metal strands and this has been reported to be more effective than thinner strands.) I have not been able to measure any difference between the different weights of copper mesh. I have found that tighter mesh is better than looser, but have not tried extremely tight packing. In all cases, it is easy to blow through a packed 48" (120 cm) long column made of stainless 2" (51 mm) OD tubing.

Measurements with a very sensitive manometer have shown that moderately loose packing in a 1 3/4" (44 mm) ID column has about the same back pressure as an open tube of a little less than 3/4" ID (19mm). Tighter packing is preferred and it probably causes no more restriction than an open tube of 1/2" (13 mm) ID. This also suggests that free vapor tubes need not be any larger than 3/4" ID.

I have experimented in a glass column with rolling the mesh so that it forms points for recentering the reflux. Mesh rolled to form a downward point does reduce the channeling against the walls, but does not eliminate it as effectively as a collar does.

The copper mesh also plays a role in binding up sulfur compounds in the wash. It has been reported that the copper also plays a catalytic role in breaking down some of the higher alcohols (fusel oils) but I haven't been able to verify this experimentally (and probably won't be able to without access to gas chromatography testing. This removal through chemical binding is only effective during the stripping run of distillation. A second pass over copper during the spirit run does not appear to have much benefit. The sulfur compounds will turn the mesh black.

I have found it useful to remove the mesh, unroll it and wash it in the upper rack of an automatic dishwasher. The blackening of the mesh can be removed by dipping the mesh in a weak acid solution of vinegar and salt, dilute hydrochloric acid or either of these acids activated with a small amount of hydrogen peroxide.

One thing I have observed with the copper mesh is that it becomes softer over time. I believe this is due to annealing of the metal at relatively low temperatures. In one instance I wanted to dry some mesh by baking it in a 350F oven. The mesh came out noticeably softer after baking. This is surprising because I thought that annealing did not take place at temperatures much below red heat. The softening of the mesh causes it to fit the column more loosely. After a period of time I have had to add more mesh to a roll to keep a reasonably firm packing in the column. In the glass column I observed an increase in channeling as the mesh softened and pulled away from the sides of the column. Stainless mesh or scrubbers are not subject to this annealing/softening effect.

The function of the mesh is to spread the liquid reflux out over a larger surface area. It does not play any role in heat transfer, despite the high thermal conductivity of the copper. The strands of the mesh are too thin to be able to move much heat over any appreciable distance. This became quite manifest during experiments with packing copper mesh around cold finger reflux condensers. The tighter packed mesh was less effective and the outer shell of the condenser ran hotter than more loosely packed mesh. I'll have more to say about this when I cover reflux condenser designs.

The relative unimportance of the thermal conductivity of the mesh suggests that stainless mesh may be as effective as copper; other than the removal of sulfur compounds and perhaps the catalytic action on fusel oils. As noted above, stainless scrubbers may present more liquid surface area for vapor contact.

3. Heat transfer by direct contact - insulation

The heat transfer between reflux and vapor is entirely due to direct contact between the vapor and liquid. There is no action at a distance with heat transfer. Both liquid and vapor are very poor conductors of heat, so the heat transfer takes place over a very short distance of the liquid/vapor interface. This also means that channeling of both liquid and vapor can form hot pathways in the column where the heat efficiency is very low. The solution to this is even packing density; thin films of liquid over the surface of the packing; and turbulence in the vapor to promote mixing.

Because the enrichment of alcohol in the column is due entirely to liquid/vapor heat exchange, insulating the column is an important means of increasing thermal efficiency. Heat loss through the walls of the column to the environment removes energy that would otherwise be available for increasing the output of alcohol. This energy can be replace by increasing the heat to the boiler, but more work can be done with less heat if the column (and boiler if heated electrically) are well insulated. The two most popular forms of insulation are closed cell foam and aluminized "bubble wrap" known under the brand name of Reflextix in the United States.

4. equalization - heat pipe effect - low power equalization

The thermal efficiency of the column is most important during the equalization of the column. This is the initial heating of the boiler and column without removing any product. During equalization, the low-boiling components of the mixture -- heads -- will migrate to the top of the column. Typically, this equalization takes about an hour, though many people will let the column equalize for much longer periods.

Once the wash starts to boil, the vapor will start moving up the column. At around 1100W the vapor will move up the column at a couple of inches per second. Some of the heat from the vapor will be absorbed by the column and packing. Once the vapor has filled the entire column, there will still be a small amount of air entrained in the packing. Once all the air has been displaced and the column contains nothing but vapor, the heat transfer efficiency rises dramatically.

This can be seen in a glass column by varying the power going to the boiler. Once the air is purged from the column, an increase in boiler heat is almost instantly reflected by an increase in the reflux at the condenser. This is because pure vapor will transmit the heat energy through the column in a domino-like effect of cascading boiling and condensing throughout the entire length of the column. What this means in practical terms is that the equalization process will require very little energy once all the air is gone and there is nothing but vapor in the column.

A well insulated 2" column that is delivering distillate through the takeoff will usually require between 750W and 1200W at the boiler. The equalization process can operate a much lower power levels. I have had good equalization at as little as 150W. Observations of the reflux condenser in a glass column have shown that reflux begins to form at as little as 75W power to the boiler. This, in turn, suggests that the heat losses from the insulated boiler and column were less than 75W. Losses from uninsulated boilers and columns can measure in the hundreds of watts.

The fact that equalization can take place at very low power densities has some implications. If higher power is used during equalization, there will be more mass transfer in the column as larger quantities of vapor and reflux move up and down. This can impede the equalization process because the amount of heads is very small and mixing will increase the dilution of heads contaminants throughout a larger volume of alcohol. Heads occupy such a large volume of product because it can be smelled and tasted in tiny dilutions; in pure form, heads contaminants would measure only a few milliliters.

5. Heads compression and sharp cuts

What this means for distillers is that anything that can prevent the dilution of heads components will increase the yield by holding the losses due to heads to a minimum. The general term for these techniques is "heads compression"; the volume of heads is compressed as small as possible. Any agitation or mixing inside the column will tend to dilute and spread out the heads, so reducing the mass transport inside the column while allowing the temperature gradient to move the heads contaminants into a small region will give the greatest amount of compression. This can be aided by specific design, to be discussed later, of the upper column to increase this "trapping" of heads contaminants in the smallest possible volume of alcohol. The way to do this is to use as little heat as possible during equalization and running off the heads volume. This will give the smallest amount of heads and the sharpest "cut" between heads and hearts. This principle of low-energy heads separation also carries over to pot stills as well: the slow and gentle distilliation of heads will give the least volume of alcohol with off-flavor contaminants and the sharpest cuts.

Once the heads have been stripped from the column, the power level can be increased to run off the hearts at a faster rate. As the tails approach, a reduction in power and slowing the takeoff rate can compress the tails contaminants in the same way. A thermometer in the upper 2/3 of the column will be of considerable help in monitoring this process, since it will show a wider change in temperature and show it earlier than the thermometer in the takeoff arm of the head. On my current column, 80 deg C appears to be the upper limit for keeping the column stable during hearts. If the temperature rises above this, the column is running too fast and both purity and %abv suffer. Below this temperature, the purity is good, but the rate of production is lower. Each still will have a threshold temperature in the upper portion of the column that will indicate the min/max point where purity and speed are maximized. This simple addition has not been widely adopted yet, but as more people gain experience with it, the benefits will become more widely accepted.

6. Liquid loading, puking, choking, power levels

The interaction of vapor and reflux in a column is complex. The usual measure of activity is vapor speed in an open tube with the same inside diameter as the column. This is a conceptual measurement because in a packed column, the velocity will be higher due to the restricted volume due to packing and reflux. The packing will increase the actual velocity of the vapor because the packing occupies space and thus decreases the effective cross-sectional area of the column and thus gives it a smaller effective diameter. In addition, the fluid drag of the packing further increases the restriction of the vapor flow because the twists and turns of the vapor path actually make the distance traversed by the vapor longer. Also, because the vapor is being diverted from a straight path, there is an inertial resistance that must be overcome. The constricted cross sectional area due to the space occupied by the packing is a constant effect. The turbulence and fluid drag effects are dynamic and the resistance increases with the vapor velocity, thus higher levels of boiler power will increase the resistance to flow in a non-linear manner. Since this resistance is accompanied by slightly higher temperatures and pressures, the overall response to higher rates of mass transport are non-linear. This means that the packing in the column offers increasing resistance and also increasing local vapor velocities that rise faster than just the increase in power levels to the boiler would suggest.

In addition, as the power from the boiler increases, the volume of reflux also increases. The higher volume of reflux further restricts the free cross-sectional area available for vapor to travel in. The net effect of the sum of the various factors that simultaneously restrict the flow of vapor and increase the velocity of the vapor causes the thermal efficiency of the column to fall dramatically once a certain level of input power from the boiler is exceeded.

Two other effects will drastically effect the performance of a column. The most severe of these is "puking": when the boiling of the wash is so violent that liquid is forced into the column. Puking usually happens when the boil is so violent that the liquid foams up and the foam is forced into the column. Once a foaming boil enters the base of the column, the nonlinear effects described earlier create a positive feedback that increased the foaming so that things go out of control very rapidly. The foam then shoots up the column and exits from the takeoff and in really extreme cases will come out the venting above the reflux condenser. The absolute worst case is puking when the coolant water to the reflux condenser is not flowing. Once experienced, this disaster is not soon forgotten.

A less disastrous form of excessive liquid in the column is flooding or choking. The cause is different, but the effect is the same. Flooding/choking occurs when the volume of reflux is so high that sections of the column are completely filled with liquid and the vapor starts to push the liquid up the column. Both flooding/choking and puking are caused by too much heat at the boiler. At least with choking, you don't have to unpack the column and clean out the packing.

Choking can be detected by listening to the vent at the top of the column. It sounds like gurgling or a coffee percolator. The onset usually gives you time to turn down the heat before things get totally out of control: if you are listening and hear it when it first starts. Puking happens quickly and is usually undetectable until you are in full disaster mode.

The rule of thumb for tuning a column for maximum output is 70% - 80% of the power necessary to cause choking. You can run at a higher power, but your column will be less efficient and the output will be lower. This condition of running at excessively high power (yet short of choking) has been referred to as "over-refluxing." If you are using a mid-column temperature probe, this point will be a degree or two higher than the mid-column temperature at the end of stripping the heads and the beginning of taking off high %abv product.

7. Column height

The rule of thumb for the height of industrial packed columns is twelve to 20 times the column internal diameter. This is a little short for home distillers. Home distilling column heights of 24 to 30 diameters are typical. As will be explained below, faster takeoff rates will demand taller columns, but the 30 diameter rule is about the upper limit of what makes a noticeable difference. The reason: as you get closer and closer to azeotrope, the difference made by each successive distillation becomes less and less. Also, because most home distillers are limited by the ceiling height, practical limits on column length usually determine the longest column possible.

In experiments with insulated 8 and 9-foot columns of 2" OD stainless tubing, there did not appear to be much benefit once the column height was over 6 feet. If you don't have height restrictions, let 30 diameters be your guide. More than that is just succumbing to column envy. As will be explained below, column envy is an unrewarding vanity.

8. Packed columns, theoretical plates, reflux rates and HETP

The alcohol-concentrating mechanism inside a packed column worked by repeated distillation and condensation of vapor. For any particular mix of substances, there will be a fixed band of temperature between the vapor and liquid state. It takes a small amount of heat energy to tear the molecules loose from the liquid state into the vapor state. Where liquid, vapor and heat are all present, the hot vapor will evaporate the liquid and the cooling liquid will condense the vapor. Distillation of alcohol is frequently represented by a McCabe-Theile diagrams which show how many repeated jumps from liquid to vapor it takes to reach some level of %abv. Each of these steps is referred to as a "Theoretical Plate." The imaginary plate is based on the ideal notion of a single plate in a Coffey distilling column. Each plate is a device that allows intimate contact between a thin layer of descending liquid and vapor rising through it.

Real plates don't work at 100% efficiency because of thermal losses to the environment and insufficient contact between the vapor and liquid. In a plate-type distilling column, the more plates, the higher the intensification of alcohol. A pot still has the equivalent of one plate. A pot and thumper has two. To obtain azeotrope, it typically takes between twenty-five and thirty physical plates.

The McCabe-Theile diagrams can be misleading when trying understand how a packed column still operates. First of all, the diagrams are drawn for pure binary mixtures: just alcohol and water. That's the end product of distilling, not the starting point. The congeners in heads and tails are present in sufficient quantity to alter the boiling point of the wash. So there is a continuously changing mixture in the still, not just a changing alcohol level. This is particularly important when it comes to heads because the acetyl compounds that make up the majority of heads congeners have azeotropes with ethanol with significantly lower boiling points. This will be discussed in greater detail in the later section on heads compression and upper columns.

The second thing that is not readily apparent from the McCabe-Theile diagrams is that the size of a step between theoretical plates is highly dependent on the reflux rate. The effect of changing reflux rates is not as pronounced in a plate still -- because there are a fixed number of plates -- but in a packed column at a high reflux rate, the number of theoretical plates can be very large.

At this point, it would be wise to clarify what I mean by reflux and takeoff rates. The reflux rate is the percentage of liquid reflux that stays in the column and the takeoff rate is the percentage of the liquid reflux that is removed as product. The two sum up to 100%, so 100% reflux means no product is being removed; 90% reflux means 10% is being removed, etc. The terms are sometimes used differently, even occasionally reversing the relationship between product and reflux. Needless to say, this can be confusing.

The relationship between the amount of packing and a theoretical plate is called the Height Equivalent to a Theoretical Plate (HETP.) In some writings about home distilling, the figure of roughly 6" HETP is given for typical packed columns. This is misleading because the HETP varies enormously with the reflux ratio. In fact, at 100% reflux, as is used during equalization, the HETP shrinks to nearly zero! Two things affect the HETP in a packed column under normal conditions: the thermal efficiency and and the takeoff rate (which is the inverse of the reflux rate.)

At high reflux ratios and good thermal efficiency, the column will perform as if it had a high number of theoretical plates. However, when the vapor velocity in a 2" ID column exceeds 20 in/sec, the thermal efficiency drops because of insufficient contact time between rising vapor and falling reflux, increased channeling, etc. In a well-insulated column and boiler, this is around 1000W - 1300W. This can lead to the counter-intuitive situation of having the %abv (and sometimes the takeoff rate) decrease as power in increased.

The reason why the takeoff rate can decrease with higher power is that as the thermal efficiency of the column decreases, the %abv drops as well. Since the pressure that drives the vapor out the takeoff arm in a VM head comes from the fact that alcohol vapor is heavier than air and low %abv vapor is lighter than high %abv vapor, the vapor in an overdriven and thermally inefficient column will exert less pressure at the takeoff port and (assuming the valve setting is left the same) the takeoff rate will decline. This is more pronounced in stills where the reflux condenser is close to the takeoff arm of the tee in the VM head and less pronounced in stills that have an upper column between the tee and the reflux condenser. These effects will be discussed in more detail later, when the head and upper column are covered in more detail.

The important point here is that the efficiency of the column to concentrate high %abv vapor is varies with the amount of power driving the column. An over-driven column will lose efficiency and this can be corrected by decreasing the power and increasing the reflux rate until both the %abv and takeoff rate are maximized. If higher takeoff rates are needed the solution lies in geometry and thermal efficiency of the still as a complete system, not just putting more power into the boiler.

Summary and conclusions

The operation of a packed column is a complex and dynamic system. There are counter-intuitive and non-linear effects involved. In getting top performance, it is useful to get a feel for how the column responds over a wide range of conditions.

The following guidelines are not hard and fast rules, but they will provide help in designing and operating a packed column:

* 2" and 3" columns are widely used. These are good sizes for home distillers.

* Columns should be 20 to 30 internal diameters tall. As columns get above this height, there are diminishing returns to making them taller.

* The thermal efficiency can be maximized by insulating the column and boiler.

* Channeling and poor contact between vapor and liquid is most pronounced the lower portion of the column and this is the best area to place reflux centering collars. Two collars in the lower half of the column will suffice. The center hole in the collar should beat least
15/16" (24mm) for a 2" column and at least 1 3/8" (35mm) for a 3" column.

* Copper or stainless mesh or scrubbers work well as packing material. If the column is to be used for stripping, copper mesh had the advantage of binding and removing sulfur compounds.

* Equalization is most efficient a very low power levels; typically 200W to 700W in a well insulated 2" column or 15% to 40% of the maximum power during high %abv hearts separation.

* Lowered power is also useful during the final leg of the hearts run to reduce the amount of tails, though the compression effect is greater for heads than it is for tails.

* Maximum power levels can be approximated by slowly raising the power level until choking begins and then reducing power to 70% - 80% from where choking begins.

* The use of a mid-column thermometer to monitor the temperature in the column is very useful.

For ongoing discussion, contributions and questions, refer this thread...

viewtopic.php?f=3&t=2326
User avatar
snuffy
 
Posts: 655
Joined: Mon Dec 22, 2008 10:05 pm
Location: Upper corner of the left coast

Return to Packed Columns

Who is online

Users browsing this forum: No registered users and 3 guests