Brown Industries
GGBL Building
Ann Arbor, MI 48109-2136
DATE: November 11, 2002
TO: Richard Wagner, Plant Supervisor, Brown Industries
FROM: Emma Wong, Team Leader, Brown Industries
Mario Fabiilli, Staff Engineer, Brown Industries
Jessica Garbern, Staff Engineer, Brown Industries
SUBJECT: Final Report for the Methanol Recovery Optimization via Distillation: Rotation 2
REF: Distillation work plan (10/21/02)
Rotation 1 final report (10/7/02)
Fogler memo (9/3/02)
Abstract
To investigate the feasibility of recovering of 97 vol% methanol from Brown Industries waste streams (waste streams: 5-10 vol% methanol), we needed to characterize our packed bed distillation column, assess and expand the flooding correlation provided by Rotation 1, and develop a model to predict the column performance. We quantified the effects of feed flow rate and feed temperature on product (methanol distillate) concentration (xD) and mass transfer coefficient (KOG). We determined that as the feed rate increased, the xD increased while KOG decreased. However, we found no correlation between feed temperature and xD or KOG. We determined that there was no correlation between reflux ratio and flooding and found that the correlation from Rotation 1 is accurate. To develop our model, we assumed the constant molar overflow assumption is valid for this system because the latent heats of vaporization of methanol and water are comparable (water: 35210 kJ/mol; methanol: 40657 kJ/mol). Our Excel model predicts column performance by taking inputs of reflux ratio, reboiler power, feed weight fraction, feed temperature, and distillate weight fraction and determining the optimum flow rate at which to run the column to satisfy these conditions. To account for changes in KOG at different column conditions, a correlation relating the KOG to feed flow rate and reflux ratio was determined using Polymath (KOG = 2123.5·(feed flow rate)-0.933·(reflux ratio)-0.552). The model also predicts whether or not flooding will occur using the Rotation 1 flooding correlation. This model can be used to determine the appropriate operating conditions to purify the methanol waste stream. Finally, we evaluated the online, tray distillation column designed by the University of Tennessee. We conclude that the simulation is good and relatively easy to use; however we also recommend a few changes we feel would improve the usability of the column.
Introduction
Manufacturing processes at Brown Industries currently produce aqueous waste streams containing 5-10 vol% methanol. If concentrated to at least 97 vol% methanol, these waste streams can potentially be sold to outside vendors for a profit. On September 3, 2002, Dr. Fogler requested that we investigate the feasibility of this methanol recovery process using the packed-bed distillation column in the Brown Industries Lab. To help us meet the objectives for Rotation 2, we used operational limits proposed by Rotation 1 on the pilot-scale methanol-water distillation column. Dr. Fogler also requested that we evaluate the online tray distillation column designed by the University of Tennessee-Chattanooga.
We set the following objectives to meet these goals:
· Quantify the effect of feed flow rate and feed temperature on product concentration (xD) and mass transfer coefficient (KOG). (See Appendix A for variable definitions.)
· Assess the effect of changing reflux ratio on flooding.
· Evaluate the proposed flooding correlation from Rotation 1.
· Determine whether the constant molal overflow assumption or the enthalpy-concentration method is valid for this distillation column.
· Develop a model to predict column performance that incorporates the following relevant parameters:
– Feed flow rate
– Reflux ratio
– Reboiler power
– Changes in the mass transfer coefficient
– Flooding
· Evaluate the online, tray distillation column designed by the University of Tennessee.
This report details our completed work for Rotation 2.
Theory and Data Analysis
Distillation is used to separate components in liquid solution. The separation is based on the different boiling points of the components in the mixture. When the liquid mixture is heated, a vapor forms, and the distribution of the components in the liquid and vapor phases will be different because of the different boiling points. Vapor leaving the top of the column, called distillate, is condensed. Part of the distillate is returned to the column as reflux, which allows liquid to flow back down the column and creates the opportunity for mass transfer between phases. The remaining distillate is removed as product. A reboiler also vaporizes some of the falling liquid while the remaining liquid exits the column as bottoms.
Feed Conditions
Feed flow rate can affect mass transfer by causing either laminar or turbulent flow in the column. At lower flow rates, the flow is laminar, which limits contact between the liquid and vapor phases, leading to relatively low mass transfer. However, at higher flow rates, the flow becomes turbulent, which improves mass transfer between phases.
Feed temperature affects the quality of the feed stream, where quality is the ratio of the amount of heat needed to vaporize one mole of feed at the entering conditions to the molar latent heat of vaporization of the feed. For a saturated liquid feed, q=1, while for a saturated vapor feed, q=0. The quality of the feed affects the slope of the operating lines, as the enriching and stripping operating lines intersect on the q-line (see “Constant Molal Overflow – McCabe Thiele Method” for more details).
To quantify the effect of feed flow rate and feed temperature on the product concentration and mass transfer coefficient, we constructed the following plots of our experimental data:
· Feed flow rate versus product concentration
· Feed flow rate versus mass transfer coefficient
· Feed temperature versus product concentration
· Feed temperature versus mass transfer coefficient
See Appendix B for detailed calculations involving KOG.
Flooding
Flooding occurs when the gas flow rate in the column is so great that the downward flow of liquid is hindered. If the gas flow rate is great enough, the liquid may actually rise up the column and exit in the distillate. We chose to examine reflux ratio because this value affects flow rates within the column, and may have an effect on flooding. Rotation 1 has shown that at small reflux ratios (i.e. producing a minimal increase in liquid flow compared to the feed rate), increased reboiler power is necessary to cause flooding at lower flow rates compared to higher flow rates. This may be because at low flow rates for a given reboiler power, there is less liquid falling down the column; thus an increased gas flow rate is necessary to result in a large enough liquid buildup to cause flooding.
There are several methods to develop a flooding correlation for a distillation column. We used an empirical correlation to predict flooding as a function of feed flow rate, similar to Rotation 1. See Appendix C additional flooding correlations and Appendix D for detailed flooding calculations. We plotted the reboiler power at which flooding occurred versus feed flow rate (see Results - Flooding). Above this line results in flooding, while below this line indicates normal (non-flooding) operation.
Constant Molal Overflow (McCabe-Thiele Method)4
We compared our data to predictions made by the McCabe-Thiele method to determine whether or not this method accurately simulates this distillation column. This method provides a graphical means of modeling the operation of a distillation column and can also be used to determine distillate composition or feed flow rate for a given set of conditions. The McCabe Thiele method assumes the following:
· Equal latent heats of vaporization for all components
· Negligible differences for the sensible heat
· Constant molal overflow in each section of the column
Because the McCabe-Thiele method assumes constant molar overflow (Equation 1), this results in straight operating lines.
Vn+1 + Ln-1 = Lv + Ln (1)
With this assumption, the following equations for the enriching section operating line (Equation 2), stripping-section operating line (Equation 3), and q-line (Equation 4) can be derived, respectively4.
(2)
(3)
(4)
For a liquid feed, the slope of the q-line increases as feed temperature increases until it reaches saturation (vertical q-line). If additional heat is added beyond saturated liquid, the feed becomes a mixture of liquid and vapor and the slope of the q-line ranges from 1 to 0. The q-line, enriching operating line, and stripping operating line all intersect at a single point.
Enthalpy-Concentration Method4
The enthalpy-concentration method accounts for differences in latent heats for each component, heats of solution or mixing, differences in sensible heats of the components, and does not assume constant molal flow. This results in curved operating lines. We determined whether this method is necessary for this system by comparing the latent heats of vaporization of methanol and water. This method would only me necessary if the latent heats are significantly different. Equation 5 gives the enriching-section operating line and Equation 3 gives the stripping-section operating line.
(5)
A trial and error solution must be used to simulate a column with this approach. This method is detailed in Appendix E4.
Distillation Model
We adapted a model provided by Rotation 1 to predict the performance of this column. This model is both experimentally and theoretically derived, where theory was used to construct the McCabe-Thiele plot, and empirical data was used to calculate the mass transfer coefficient. A detailed description of this model is found in Appendix F. This model allows us to account for changes in the mass transfer coefficient due to changes in column conditions, as well as predict when flooding will occur.
Equipment2
The distillation apparatus we used to perform pilot-scale separation processes is shown in Appendix G. The 30-inch tall packed-bed distillation column contains 1/4-inch Pro-Pack stainless steel random packing within a 3-inch inner-diameter glass pipe. The apparatus has two available feed tanks: a smaller tank for use during recycle-mode and a larger tank for use during production-mode. We used the column in production-mode to remove the possibility that the hot bottoms being returned to the recycle feed tank could change our results due to additional feed preheating. The preheater heats the feed, which then enters the column in the top, middle, or bottom feed port. The vapor exits the top of the column to be condensed then sent to a distillate receiver. The distillate is returned to the column as reflux, returned to the feed tank during recycle-mode, or collected in a distillate tank during production-mode. A bottoms waste stream is removed from the reboiler and is either returned to the feed tank during recycle-mode or held in a bottoms tank during production. Additionally, Appendix H contains a MSDS for methanol.
Throughout the distillation column there are seventeen thermocouples to record temperatures and four volumetric turbine flow meters to record feed, reflux, distillate, and bottoms flows. All of these values are recorded on a computer through LabTech Control software. For additional equipment operation instructions, see equipment manual2.
A gas chromatograph (GC) was used to analyze and determine the composition of various streams. Samples of approximately 1 ml were removed with a syringe from the feed tank or sample ports in the apparatus, transferred to a 1 ml Eppendorf tube, and quickly capped. Then, using an injection syringe, a 0.2 ml sample was placed in sample port B of the GC. The SOP for operating the GC was followed2. Because the GC has a high degree of error, it was calibrated before each use with the standard samples.
The online distillation column at the University of Tennessee – Chattanooga is a 12-tray column. The operators can change the feed pump setting, reflux percentage, and reboiler power. Four real time images of the column are present on the user interface. Additionally, real time temperature profiles are given for each of the trays, feed, distillate, and reflux.
Experimental Design
There are five independent variables with the Brown Industries distillation column:
· Feed flow rate
· Feed temperature
· Feed concentration
· Reboiler power
· Reflux ratio
Dependent variables consist of temperatures and compositions of the distillate and bottoms streams, as well as the properties calculated from these values, including the mass transfer coefficient.
Table 1 highlights the runs that we completed in order to determine the effects of feed flow rate and feed temperature on product concentration and hence the mass transfer coefficient.
Table 1 – Experimental Runs
Run / Feed Flow Rate (mL/min) / Feed Temperature (°C)1 / 131 / 25.0
2 / 326 / 23.6
3 / 530 / 24.2
4 / 146 / 35.0
5 / 330 / 35.0
6 / 533 / 35.0
7 / 150 / 45.0
8 / 350 / 45.0
9 / 550 / 48.7
Colored blocks indicate sets of runs, where the colored variables were varied while the others were held constant. The maximum feed rate and temperature were chosen based on maximum values reported by Rotation 1. During these runs, the feed concentration, feed location, reboiler power, and reflux ratio were set at 5.5 ± 0.9 vol%, middle, 50% and 4.7, respectively.
To continue the flooding study initiated by Rotation 1, we varied feed flow rate and reflux ratio according to the runs outlined in Table 2. For each run, the reboiler power was adjusted until flooding was observed, according to the methodology of Rotation 1.
Table 2 –Runs to Determine Flooding Conditions
Run / Feed Flow Rate (mL/min) / Reflux Ratio10 / 67 / 36.7
11 / 152 / 59.4
12 / 350 / 38.8
13 / 597 / 48.6
Since Rotation 1 conducted runs at a reflux ratio of 6.7, we decided to operate at higher reflux ratios. We predicted that this variable would most affect the gas flow rate inside the column, and hence cause flooding. The feed temperature, concentration, and location were set to 56.5 ± 12.1°C, 5.5 ± 0.9 vol%, and middle, respectively. Though there appears to be large variations in the feed temperature and reflux ratio between runs (due to difficulty maintaining steady state), we found that these fluctuations were insignificant and had no effect on determining a correlation for flooding or confirming the data found by Rotation 1. Further details can be found in the results and discussion sections.
The online distillation column at the University of Tennessee can test three independent variables: feed pump setting, percent reflux, and reboiler power. Table 3 outlines the runs we completed while evaluating this column. We chose to examine the effect of feed pump setting on the methanol distillate composition. Distillate composition was determined by using the temperature of the top tray of the column and assuming that the vapor is at the saturation point on each tray. Due to equipment difficulties, we were unable to complete additional runs we had planned to examine the effects of reflux and reboiler power on distillate composition.