Geoffrey Widdison
1656 Sunnyside Avenue
Salt Lake City, Utah, 84105
Dr. Terry Ring
Department of Chemical Engineering
University of Utah
Salt Lake City, Utah, 84102
Dear Dr. Ring,
On October 18, 2005, you assigned the members of project group E the task of operating the distillation column located in the MEB Senior Lab. We were asked to use the column to separate a water-isopropanol mixture under conditions of total reflux, as well as under continuous feed with a reflux ratio of twice the minimum. We were asked to determine the efficiency of each tray, and to make recommendations for the operation of the column.
We determined that measured efficiency rates within the column varied widely within the column and from one set of operating conditions to another. The minimum reflux ratio was found to be 1.11. The average efficiency of the column was found to be approximately 31.1% under total reflux conditions and 23.84% with a reflux ratio twice the minimum. The values of individual trays, however, ranged from near 100% to approximately 1%. While this may be partly due to defects and variation within the column, it seems important to reexamine the methods used to measure composition at each tray.
The lack of a vapor sampling port on each tray inhibited the data range we were able to gather, installation of additional vapor ports could increase the accuracy of models developed for the separation process. Syringes that form better connections with the liquid sampling ports could also facilitate the testing process.
Examination of the separation data leads us to conclude that the ninth tray is at or near the optimal feed location for most operating conditions, though additional testing may be desired to confirm this. The range of conditions we used are near to those used to separate a number of mixtures. The information on tray and column efficiency, and the related information concerning column capacity and flow rates make it possible for a wide variety of separations to be modeled, the reaction of this specific column to be predicted.
Sincerely,
Geoffrey Widdison
OPERATION OF A BUBBLE-CAP CONTINUOUS DISTILLATION COLUMN
By
Geoffrey Widdison
Project No. 2M
Projects Laboratory I
Assigned: October 18, 2005
Due: November 15, 2005
Submitted: November 16, 2005
Project Team Members for Group E:
Brandon Grimm
Stanley Holbrook
Geoffrey Widdison
Geoffrey Widdison
INTRODUCTION
According to Seader (1998), distillation processes are a major component of chemical processing with a great deal of scientific and economic significance. A large amount of industrial distillation takes place in staged columns. Each stage consists of a tray designed to contact rising vapor with descending liquid, which increases the vapor composition of the more volatile component. This process allows the separation of a wide variety of mixtures into separate species.
The relationship between the configuration of a distillation column and the degree of separation achieved can be predicted using a combination of theoretical and empirically derived models. Since complete equilibrium is rarely achieved in any stage in distillation processes, experimental data must be gathered in order to determine the efficiency of the column.
By collecting and analyzing samples from different stages, the efficiency of the distillation operation can be calculated. These calculations can be used to confirm the mathematical model for the distillation operation, and modify it for the specific equipment being used.
THEORY
The mathematical basis for staged distillation columns is well established and presented in a number of texts. The following treatment is adapted from Seaders (1998).
The separation dynamics are most heavily dependent on the vapor pressures of the species in the column. The more volatile component is also known as the light key, the less volatile component is also known as the heavy key. For a two-component separation:
α=KL/KH. (1)
KL= PL/P. (2)
KH= PH/P. (3)
All symbols are defined in the nomenclature section at the end of the text. By Raoult’s law, this means that relative volatility can be described in terms of mole fractions in the liquid and vapor phases.
(4)
This can be rearranged to give
(5)
The traditional approach to modeling distillation processes is known as the McCabe-Thiele approach. This model requires making the following assumptions:
¨ That the relative volatility is constant over the temperature range in the column.
¨ That the components have equal and constant molar enthalpy.
¨ That enthalpy changes and heat of mixing are negligible.
¨ That the pressure is uniform within the column.
These assumptions allow equations to be derived relating flow rates to molar compositions in each stage of the column. The column is divided into a stripping section, in which the more volatile component is selectively driven into the vapor phase, and the rectification section, in which the less volatile component is selectively condensed. In each section, the liquid and vapor compositions can be related to flow rates.
Rectification section:
Vn+1yn+1=Lnxn+DxD. (6)
yn+1=(Ln/ Vn+1)+(D/ Vn+1) xD. (7)
Where the stages are numbered consecutively, beginning with the top stage.
L/V=R/(R+1). (8)
. (9)
Stripping section:
yn=(Ln/Vn)xn-(Bn/Vn)xB. (10)
VB=V/B (11)
(12)
. (13)
Given equilibrium data for the components of the distillation, these equations can be used to model the stages graphically, as shown in Figure 1.
Figure 1. Sample McCabe-Thiele diagram. Each step indicates a theoretical stage
going to complete equilibrium. The calculations are shown for the rectification
section only. Reprinted from Seader (1998).
Because true vapor-liquid equilibrium is unlikely to be achieved on each stage, this method must be adapted in order to properly model the distillation system. A widely used method for this is known as Murphree Efficiency.
, (14)
where y* used in this equation is derived from experimentally measured liquid-phase composition values. The Murphree Efficiency can be used to alter the equilibrium line in the Mcabe-Thiele graph. This adapted graph can be used to more accurately predict the separation under different conditions of feed composition and reflux ratios.
APPARATUS AND PROCEDURE
The primary piece of equipment used was the distillation column located in the Merrill Engineering Building on the University of Utah campus. The top of the column, control system and condenser were located in room 3290 on the third floor. The column extended directly downward to the second floor, where the reboiler and bottoms storage tank were located. The feed tank and pump, as well as the distillate storage tank were located on the first floor, directly below the column.
The distillation column contained 12 bubble cap trays spaced at 11” intervals. The trays were numbered in descending order, with the top tray designated tray 1. Each tray had a diameter of approximately 14” and contained 18 bubble-caps. The bubble-caps were 3” in diameter with square openings. The total column height was 145”. A full schematic can be found in Appendix C.
Figure 2. Illustration of a bubble cap. Reprinted from Seader (1998)
The column design allowed for variation in the feed point, but for the purposes of our experiments it remained stationary above tray 9. T-type thermocouples were positioned inside each tray with digital readouts to the control computer. Each tray was equipped with a sample port for the liquid phase. Sample ports for the vapor phase were located on even-number trays only. Additional sample ports were available to sample liquid from the feed, distillate and bottoms. Orifice plate meters equipped with digital pressure gauges were used to measure flow rate of the feed, reflux, distillate and bottoms.
A computer equipped with Opto 22 control software was used to alter valve positions to control feed, distillate and reflux flow rates, as well as the flow of steam to the reboiler, and the flow of cooling water to the condenser. The software also received and reported data from the thermocouples and flow meters.
Composition analysis was done with two pieces of equipment. A Mettler/Par DMA46 densitometer (SN: 451038) was used to calculate sample densities. A Reichert-Jung ABBE Mark II digital refractometer (SN: 10952-9) was used to measure refractive indices of samples. Assuming binary mixtures at every point, these measurements allow the composition to be calculated. The correlations between densitometer and refractometer data and composition were developed by creating and testing standards with known volumetric compositions. These compositions were then converted to mole fractions. For details, see Appendix A.
The column was first operated at total reflux. The pump was used to mix the feed, then a sample of the feed was taken prior to distillation. Approximately 17.5 gallons of feedstock was introduced into the column, and then the valve was closed and the pump shut down. The steam flow valve was opened until the steam flow rate was 0.456 kg/min, which required manual adjustment of the valves. The flow rate of the cooling water was set at 50% of capacity using the computer control system, which corresponded to a flow-meter reading of 136.8 kg/min. The temperature profile within the column was monitored until the temperatures stabilized. When the temperatures no longer changed significantly, the column was assumed to be at steady state.
A syringe with a plastic connection hose was used to extract a liquid sample from each the top six trays. A glass syringe with a stainless steel needle was used to sample the vapor from trays 2, 4 and 6. Samples of the distillate and bottoms were also taken. All of these samples were saved and tested for refractive index. The samples with sufficient volume to allow a densitometer reading were tested for density as well. Once the samples were taken, the steam was shut down, and valves were opened to drain the bottoms, the distillate, and any remaining fluid in the column into the feed tank. Once the column had cooled below 70°C, the cooling fluid was shut off, and the valves were closed.
Due to time constraints, only the top six trays were tested in this fashion on the first distillation run. The column was subsequently started and brought to steady state using the same procedure previously described, with a new sample of the feed taken before running the column. Once the column was at steady state, liquid samples were taken from trays 6 through 12, as well as the distillate and bottoms. Vapor samples were taken from trays 8, 10 and 12. As before, all of these samples were tested for refractive index, and as many as had sufficient volume were tested for density. The column was shut down as previously described.
The feedstock analysis allowed the calculation of inlet composition. This composition was used to graphically determine the minimum reflux ratio for a distillate product of 60 mol% isopropanol. Given this data, it was calculated that 69% of the distillate needed to be recycled for the column to run at twice the minimum reflux (see Appendix D). The column was once again brought to steady state using the previously described process. Once steady state was reached, the feed pump was turned on and the valve was opened until the flow-meter reported a flow rate of 4.4 kg/min. The bottoms valve on the column was immediately opened to release fluid at a similar flow rate, and the reflux settings were changed to recycle 69% of the distillate.
Rotameters which reported the flow rates of the feed and the bottoms were monitored and valves were adjusted to maintain a similar flow rate from each. Additionally, the fluid level in the reboiler was observed, and any significant changes were compensated for by adjustment of the flow rates of the feed and the bottoms.
As in previous experiments, the temperature profile of the column was watched. When the profile no longer changed significantly, the column was assumed to be at steady state. Liquid samples were taken from each tray, and vapor samples were taken from the even numbered trays, using the methods previously mentioned. The samples were tested for refractive index and density, and the compositions of the vapor and liquid at each tested point was calculated.
Using known vapor-liquid equilibrium data for water-isopropanol systems and the experimental data, the efficiency of each tray was estimated. These efficiencies, which reflect the specific characteristics of the column, could then be used to predict the behavior of other distillation systems run on the same equipment.
RESULTS AND DISCUSSION
The calculated mole fraction in each tray was as follows:
Table 1: Liquid and vapor compositions from the first run of the distillation column under conditions of total reflux. Given in mole fractions of isopropanol in water. Calculated from density and refractive index data.
Tray / Liquid Composition / Vapor Composition1 / 0.55±0.015
2 / 0.42±0.015 / 0.45±0.015
3 / 0.33±0.015
4 / Error / 0.41±0.015
5 / 0.33±0.015
6 / 0.04±0.015 / 0.39±0.015
7 / 0.00±0.015
Feed / 0.027±0.015
Table 2: Liquid and vapor compositions from the second run of the distillation column under conditions of total reflux. Given in mole fractions of isopropanol in water. Calculated from density and refractive index data.
Tray / Liquid Composition / Vapor Composition6 / 0.496±0.015 / 0.612±0.015
7 / 0.383±0.015
8 / 0.374±0.015 / 0.524±0.015
9 / 0.091±0.015
10 / 0.027±0.015 / 0.505±0.015
11 / 0.016±0.015
12 / 0.020±0.015 / 0.103±0.015
Bottoms / 0.014±0.015
Feed / 0.027±0.015
Table 3: Liquid and vapor compositions from the distillation column with a reflux ratio of 2.22. Given in mole fractions of isopropanol in water. Calculated from density and refractive index data.
Tray / Liquid Composition / Vapor Composition1 / Error
2 / 0.603±0.015 / 0.225±0.015
3 / 0.573±0.015
4 / 0.549±0.015 / 0.232±0.015
5 / 0.504±0.015
6 / 0.450±0.015 / 0.160±0.015
7 / 0.028±0.015
8 / 0.026±0.015 / 0.121±0.015
9 / 0.025±0.015
10 / 0.022±0.015 / 0.092±0.015
11 / 0.020±0.015
12 / 0.018±0.015 / 0.070±0.015
Feed / 0.023±0.015
Bottoms / 0.017±0.015
The error points indicate samples that became tainted or otherwise inaccurate due to mistakes in the sampling or handling process. Further experimentation is necessary to determine the steady-state compositions at these points. Because the distillate (the liquid from tray 1) was improperly sampled, and because the composition at tray 2 is near the isopropanol-water azeotrope, it is assumed that the distillate for the partial reflux operation is near the composition of the tray 2 liquid.