All machines have drivers. A distillation column is also a machine, driven by a reboiler. It is the heat duty of the reboiler, supplemented by the heat content (enthalpy) of the feed, that provides the energy to make a split between light and heavy components. A useful example of the importance of the reboiler in distillation comes from the venerable use of sugar cane in my home state of Louisiana.
If the cut cane is left in the fields for a few months, its sugar content ferments to alcohol. Squeezing the cane then produces a rather lowproof alcoholic drink. Of course, one would naturally wish to concentrate the alcohol content by distillation, in the still shown in Fig. 6.1.
The alcohol is called the light component, because it boils at a lower temperature than water; the water is called the heavy component, because it boils at a higher temperature than alcohol. Raising the top reflux rate will lower the tower-top temperature and reduce the amount of the heavier component, water, in the overhead alcohol product. But what happens to the weight of vapor flowing up through the trays? Does the flow go up, go down, or remain the same?
There are two ways to answer this question. Let’s first look at the reboiler. As the tower-top temperature shown in Fig. 6.1 goes down, more of the lighter, lower-boiling-point alcohol is refluxed down the tower. The tower-bottom temperature begins to drop, and the steam flow to the reboiler is automatically increased by the action of the temperature recorder controller (TRC). As the steam flow to the reboiler increases, so does the reboiler duty (or energy injected into the tower in the form of heat). Almost all the reboiler heat or duty is converted to vaporization. We will prove this statement mathematically later in this chapter. The increased vapor leaving the reboiler then bubbles up through the trays, and hence the flow of vapor is seen to increase as the reflux rate is raised.
Now let’s look at the reflux drum. The incremental reflux flow comes from this drum. But the liquid in this drum comes from the condenser. The feed to the condenser is vapor from the top of the tower. Hence, as we increase the reflux flow, the vapor rate from the top of the tower must increase. One way of summarizing these results is to say that the reflux comes from the reboiler.
The statement that the mass, or weight flow of vapor through the trays, increases as the reflux rate is raised is based on the reboiler being on automatic temperature control. If the reboiler were on manual control, then the flow of steam and the reboiler heat duty would remain constant as the reflux rate was increased, and the weight flow of vapor up the tower would remain constant as the top reflux rate was increased. But the liquid level in the reflux drum would begin to drop. The reflux drum level recorder controller (LRC) would close off to catch to falling level, and the overhead product rate would drop in proportion to the increase in reflux rate. We can now draw some conclusions from the foregoing discussion:
• The flow of vapor leaving the top tray of the tower is equal to the flow of reflux, plus the flow of the alcohol overhead
• The overhead condenser heat-removal duty is proportional to the reboiler heat duty.
• The weight flow of vapor in a tower is controlled by one factor and one factor only: heat.
An increase in reflux rate, assuming that the reboiler is on automatic temperature control, increases both the tray weir loading and the vapor velocity through the tray deck. This increases both the total tray pressure drop and the height of liquid in the tray’s downcomer. Increasing reflux rates, with the reboiler on automatic temperature control, will always push the tray closer to or even beyond the point of incipient flood.
This is perhaps an idea you remember from high school, but never quite understood. The phase rule corresponds to determining how many independent variables we can fix in a process before all the other variables become dependent variables. In a reflux drum, we can fix the temperature and composition of the liquid in the drum. The temperature and composition are called independent variables. The pressure in the drum could now be calculated from the chart in Fig. 5.4. The pressure is a dependent variable. The phase rule for the reflux drum system states that we can select any two variables arbitrarily (temperature, pressure, or composition), but then the remaining variable is fixed.
A simple distillation tower, like that shown in Fig. 5.2, also must obey its own phase rule. Here, because the distillation tower is a more complex system than the reflux drum, there are three independent variables that must be specified. The operator can choose from a large number of variables, but must select no more than three from the following list:
• Tower pressure
• Reflux rate, or reflux ratio
• Reboiler duty
• Tower-top temperature
• Tower-bottom temperature
• Overhead product rate
• Bottoms product rate
• Overhead product composition
• Bottoms product composition
As an operator reduces the tower pressure, three effects occur simultaneously:
• Relative volatility increases.
• Tray deck leakage decreases.
• Entrainment, or spray height, increases.
The first two factors help make fractionation better, the last factor makes fractionation worse. How can an operator select the optimum tower pressure to maximize the benefits of enhanced relative volatility, and reduced tray deck dumping without unduly promoting jet flooding due to entrainment?
To answer this fundamental question, we should realize that reducing the tower pressure will also reduce both the tower-top temperature and the tower-bottom temperature. So the change in these temperatures, by themselves, is not particularly informative. But if we look at the difference between the bottom and top temperatures, this difference is an excellent indication of fractionation efficiency. The bigger this temperature difference, the better the split. For instance, if the tower-top and tower-bottom temperatures are the same for a 25-tray tower, what is the average tray efficiency? (Answer: 100 percent / 25 = 4 percent.)
Figure 5.5 illustrates this relationship. Point A is the incipient flood point. In this case, the incipient flood point is defined as the operating pressure that maximizes the temperature difference across the tower at a particular reflux rate. How, then, do we select the optimum tower pressure to obtain the best efficiency point for the trays? Answer: Look at the temperature profile across the column.
The process design engineer selects the tower design operating pressure as follows:
1. Determines the maximum cooling water or ambient air temperature that is typically expected on a hot summer day
in the locale where the plant is to be built.
2. Calculates the condenser outlet, or reflux drum temperature, that would result from the above water or air temperature.
3. Referring to Fig. 5.2, the designer calculates the pressure in the reflux drum, assuming that the condensed liquid is at its bubble point. Adding 5 or 10 psig to this pressure, for pressure loss in the overhead condenser and associated piping, the designer then determines the tower-top pressure.
Of course, the unit operator can physically deviate from this design pressure, but to what purpose?
Why are distillation towers designed with controls that fix the tower pressure?
Naturally, we do not want to overpressure the tower and pop open the safety relief valve. Alternatively, if the tower pressure gets too low, we could not condense the reflux. Then the liquid level in the reflux drum would fall and the reflux pump would lose suction and cavitate. But assuming that we have plenty of condensing capacity and are operating well below the relief valve set pressure, why do we attempt to fix the tower pressure? Further, how do we know what pressure target to select?
I well remember one pentane-hexane splitter in Toronto. The tower simply could not make a decent split, regardless of the feed or reflux rate selected. The tower-top pressure was swinging between 12 and 20 psig. The flooded condenser pressure control valve, shown in Fig. 5.1, was operating between 5 and 15 percent open, and hence it was responding in a nonlinear fashion (most control valves work properly only at 20 to 75 percent open). The problem may be explained as follows.
The liquid on the tray deck was at its bubble, or boiling, point. A sudden decrease in the tower pressure caused the liquid to boil violently. The resulting surge in vapor flow promoted jet entrainment, or flooding.
Alternately, the vapor flowing between trays was at its dew point. A sudden increase in tower pressure caused a rapid condensation of this vapor and a loss in vapor velocity through the tray deck holes. The resulting loss in vapor flow caused the tray decks to dump.
Either way, erratic tower pressure results in alternating flooding and dumping, and therefore reduced tray efficiency. While gradual swings in pressure are quite acceptable, no tower can be expected to make a decent split with a rapidly fluctuating pressure.
All vendors now market a high capacity tray. These trays have a 5 to 15 percent capacity advantage over conventional trays. Basically, the idea behind these high capacity trays is the same. The area underneath the downcomer is converted to bubble area. This increase in area devoted to vapor flow reduces the percent of jet flood.
But what keeps vapor from blowing up the downcomer? What prevents loss of the downcomer seal? If the downcomer seal is lost, surely the downcomer will back up and flood the upper trays of the column.
The design I’m most familiar with is the NorPro high capacity tray shown in Fig. 4.6. The head loss through the orifice holes in the downcomer seal plate shown is sufficiently high to prevent loss of the downcomer seal. These trays flood rather easily when their design downcomer liquid rates are exceeded. However, when operated at design downcomer liquid rates they perform very well indeed, and have shown quite a high vapor-handling capacity as compared to conventional trays.
The downcomer seal plate shown in Fig. 4.6 is an example of a dynamic downcomer seal. The Koch-Glitsch “Nye” tray also uses a dynamic downcomer seal to increase vapor-handling capacity. All trays with a dynamic downcomer seal suffer from two disadvantages:
• Loss of flexibility in that the liquid rates cannot be varied over too great a range without either flooding or unsealing the downcomers.
• Tray installation complexity is always increased, sometimes with terrible consequences.
For these reasons, high capacity trays using dynamic downcomer seals are best avoided on new columns. They should be reserved for use on retrofit tower expansion projects.
The problem we have been discussing—loss of tray efficiency due to low vapor velocity—is commonly called turndown. It is the opposite of flooding, which is indicated by loss of tray efficiency at high vapor velocity. To discriminate between flooding and weeping trays, we measure the tower pressure drop. If the pressure drop per tray, expressed in inches of liquid, is more than three times the weir height, then the poor fractionation is due to flooding. If the pressure drop per tray is less than the height of the weir, then poor fractionation is due to weeping or dumping.
One way to stop trays from leaking or weeping is to increase the reflux rate. Assuming that the reboiler is on automatic temperature control, increasing the reflux flow must result in increased reboiler duty. This will increase the vapor flow through the trays and the dry tray pressure drop. The higher dry tray pressure drop may then stop tray deck leakage. The net effect is that the higher reflux rate restores the tray efficiency.
However, the largest operating cost for many process units is the energy supplied to the reboilers. We should therefore avoid high reflux rates, and try to achieve the best efficiency point for distillation tower trays at a minimum vapor flow. This is best done by designing and installing the tray decks and outlet weirs as level as possible. Damaged tray decks should not be reused unless they can be restored to their proper state of levelness, which is difficult, if not impossible.
The first continuous distillation tower built was the “patent still” used in Britain to produce Scotch whiskey, in 1835. The patent still is to this day employed to make apple brandy in southern England. The original still, and the one I saw in England in 1992, had ordinary bubble-cap trays (except downpipes instead of downcomers were used). The major advantage of a bubble-cap tray is that the tray deck is leakproof. As shown in Fig. 4.5, the riser inside the cap is above the top of the outlet weir. This creates a mechanical seal on the tray deck, which prevents liquid weeping, regardless of the vapor flow.
Bubble-cap trays may be operated over a far wider range of vapor flows, without loss of tray efficiency. It is the author’s experience that bubble-cap trays fractionate better in commercial service than do perforated (valve or sieve) trays. Why, then, are bubble-cap trays rarely used in a modern distillation?
There really is no proper answer to this question. It is quite likely that the archaic, massively thick, bolted-up, cast-iron bubble-cap or tunnel-cap tray was the best tray ever built. However, compared to a modern valve tray, bubble-cap trays
• Were difficult to install, because of their weight.
• Have about 15 percent less capacity because when vapor escapes from the slots on the bubble cap, it is moving in a horizontal direction. The vapor flow must turn 90°. This change of direction promotes entrainment and, hence, jet flooding.
• Are more expensive to purchase.
But in the natural-gas fields, where modern design techniques have been slow to penetrate, bubble-cap trays are still widely employed, to dehydrate and sweeten natural gas in remote locations.
We stated that the top edge of the outlet weir is maintained about 0.5 in above the bottom edge of the inlet downcomer to prevent vapor from flowing up the downcomer. This is called a 0.5-in positive downcomer seal. But for this seal to be effective, the liquid must overflow the weir. If all the liquid is weeping through the tray deck, there will be no flow over the weir, and the height of the weir will become irrelevant. Figure 4.4 shows the result of severe tray deck leakage:
1. The downcomer seal is lost on tray deck 1.
2. Vapor flows up the downcomer between tray decks 1 and 2.
3. Liquid flow is backed up onto the tray above, i.e., onto tray deck 2.
4. The dry tray pressure drop through tray 2 decreases due to low vapor flow through the tray deck.
5. The hydraulic tray pressure drop on tray 2 increases due to increased liquid level.
6. Tray 2 will now start to weep, with the weeping concentrated on the low area of the tray.
7. Tray 2 now has most of its vapor feed flowing up through its outlet downcomer, rather than the tray deck, and most of its liquid flow is leaking through its tray deck.
The net result of this unpleasant scenario is loss of both vapor-liquid contacting and tray efficiency. Note how the mechanical problems (i.e., levelness) of tray 1 ruins the tray efficiency of both trays 1 and 2.
When trays weep, efficiency may not be significantly reduced. After all, the dripping liquid will still come into good contact with the upflowing vapor. But this statement would be valid only if the tray decks were absolutely level. And in the real world, especially in large (>6-ft)-diameter columns, there is no such thing as a “level” tray. Figure 4.3 shows the edge view of a tray that is 2 in out-of-level.
As illustrated, liquid accumulates on the low side of this tray. Vapor, taking the path of least resistance, preferentially bubbles up through the high side of the tray deck. To prevent liquid from leaking through the low side of the tray, the dry tray pressure drop must equal or exceed the sum of the weight of the aerated liquid retained on the tray by the weir plus the crest height of liquid over the weir plus the 2-in out-of-levelness of the tray deck.
Once the weight of liquid on one portion—the lowest area—of a tray deck exceeds the dry tray pressure drop, the hydraulic balance of the entire tray is ruined. Vapor flow through the low area of the tray deck ceases. The aeration of the liquid retained by the weir on the low area of the tray deck stops, and hence the hydraulic tray pressure drop increases even more. As shown in Fig. 4.3, the liquid now drains largely through the low area of the tray. The vapor flow bubbles mainly through the higher area of the tray deck. This phenomenon is termed vapor-liquid channeling. Channeling is the primary reason for reduced distillation tray efficiency, because the vapor and liquid no longer come into good, intimate contact.
The common reason for out-of-levelness of trays is sagging of the tray decks. Sags are caused by pressure surges and sloppy installation. Sometimes the tray support rings might not be installed level, or the tower itself might be out of plumb (meaning the tower itself may not be truly vertical).