Daniel R. Summers
Stone & Webster Engineering Corp.
Houston
Peter J. McGuire, Michael R. Resetarits
UOP
Tonawanda, N.Y.
Caryn E. Graves
Chevron Chemical Co.
Baytown, Tex.
Steven E. Harper
Chevron Chemical Co.
Kingwood, Tex.
Salvatore J. Angelino
Consultant
Tonawanda, N.Y.
In 1992, Chevron Chemical Company placed 325 of UOP's Enhanced Capacity Multiple Downcomer (ECMD) trays in its large C3 splitter at Port Arthur, Tex. The capacity of the splitter was increased by 40% to about 124,000 lb/hr.
Many times, engineers are faced with debottlenecking their fractionation trains. High-pressure and heavily liquid-loaded service is of particular interest because of the high capital cost to replace a vessel.
This recently patented high-capacity tray enabled Chevron to revamp its fractionation tower, thus avoiding costly tower replacement.
BACKGROUND
Chevron Chemical Co. is a world-scale producer of polymer- grade propylene. The propylene formed as a by-product of ethylene production is supplemented by refinery-grade propylene.
At Chevron's Port Arthur, Tex., plant, the propylene/propane stream first flows through a multiple-bed treatment system to ensure high-purity product. The stream then proceeds to a large propylene/propane fractionation unit that produces hundreds of millions of pounds per year of polymer-grade propylene while utilizing mostly waste heat to keep operating costs low.
This C3 splitter originally contained 250 four-pass valve trays in two series-operated 18-ft diameter vessels. Both towers are more than 220 ft tall because of the original 18-20 in. tray spacings. They were commissioned in July of 1970.
In 1988, Chevron initiated discussions with, then, Union Carbide regarding increasing the capacity of these towers by installing MD (Multiple Downcomer) trays. Although the MD trays would have allowed a 20% capacity increase, Chevron was looking to push its equipment even further.
An annual production rate of at least 725 million lb/year was targeted. Over the next 3 years, several scenarios were examined between Chevron and Union Carbide/UOP, including operating the towers in parallel and applying heat pump technology. Finally, in 1991, the ECMD tray technology was proposed.
A thorough review of the ECMD tray technology, first developed by UOP in 1989, was conducted by UOP and Chevron. Numerous cases were studied, including tray-for-tray and four-for-three tray replacements, and summer vs. winter operating conditions.
Feeds high in methyl acetylene and propadiene (MAPD) also were studied. Stone & Webster Engineering Corp. became involved at this time to study the possibilities for an overall ethylene plant expansion.
In June of 1991, a proposal was submitted to Stone & Webster regarding the placement of 325 UOP ECMD trays in two series- operated towers to increase propylene capacity by 35%. Fabrication and shipment of the trays were completed in January of 1992, and installation commenced in the summer of 1992.
ECMD TRAYS
UOP MD trays have already been described in the literature and are reasonably well known in the refining and petrochemical industries, having been employed in more than 400 columns worldwide.1 2
They are characterized by a large number of downcomers, the absence of receiving pans, and a very long outlet weir length. The trays handle high volumetric liquid-to-vapor-rate ratios.
With MD trays, columns can be revamped to increase capacity. With new columns, diameters and heights can be reduced.
MD trays scale up easily as a result of their unit cell construction. The multiplicity of downcomers yields very long outlet weirs. Short froths result from the long weirs. With short froths, the trays can be used at close spacings.
UOP's Slotted Sieve trays also have been described in the literature.3 4 With Slotted Sieve trays, slots are employed on the perforated decks of one-pass cross flow, one-pass parallel flow, and two-pass cross flow trays to achieve efficiency increases and pressure drop decreases.
Field data for Slotted Sieve trays have shown surprisingly high capacities. And laboratory testing revealed that, at high rates, increased slotting reduces froth heights and entrainment levels.
ECMD trays represent a combination of the MD and Slotted Sieve technologies. With ECMDs, slotting is employed on MD trays to reduce the froth heights of MD trays, especially at high rates. Capacities are increased.
In some cases, specially designed anti-jump baffles are required to prevent the slot-propelled liquid from shooting over and across the downcomer mouths. Fig. 1 (63360 bytes) compares the MD and ECMD technologies.
ECMD DEVELOPMENT
ECMD testing began in 1989 in UOP's Tonawanda, N.Y., tray- testing laboratory. Initially, three trays with square, 2-ft x 2- ft cross sections were studied in an air-water test column.
Capacities were found to be very high. The results encouraged UOP to study three larger trays, 8 ft in diameter, in another Tonawanda air-water column. From the 8-ft column tests, UOP engineers concluded that the capacity of ECMD trays is 15-22% higher than the already high-capacity MD trays.
To conclude the development program, UOP enlisted the services of the University of Manchester Institute of Science & Technology (UMIST) in 1991. ECMD tray efficiencies were studied across a broad range of vapor and liquid rates using a 2-ft diameter, five-tray, column that separates methanol from ethanol and methanol from water.
Tray spacings of 24 in. were employed. UOP and UMIST investigators concluded that ECMD tray efficiencies were sometimes 5% less than those of MD trays. In other cases, the ECMD and MD efficiencies were equal.
COMMERCIALIZATION
The first industrial application of ECMD trays occurred in a deethanizer owned by OMV Aktiengesellschaft, Schwechat, Austria. A total of 56 ECMD trays were employed to revamp that 9-ft (2,736 mm) diameter column.
Feedback from that revamp was very positive and was described by OMV and UOP engineers at a 1992 Institute of Chemical Engineering conference in Birmingham, U.K.5 Operating data showed that a capacity increase of at least 15% was achieved.
A condensation circuit bottleneck made it impossible to flood the tower.
TOWER, TRAY DESIGN
After many months of discussions, during which many process options were carefully considered, Chevron, Stone & Webster, and UOP engineers jointly decided to replace the existing 250 four- pass valve trays of Chevron's Port Arthur two-tower propylene production unit with 325 ECMD trays.
UOP employed its proprietary VLE package for process simulation of the unit. The 325 actual trays were simulated by UOP as 214 theoretical trays. The applied efficiency was thus 66%; the expected efficiency, based on the system's hydraulic and physical properties, was 71%.
Table 1 (19201 bytes) summarizes the mass balance associated with the final process design. The target of the revamp was the production of 99.6 mole % purity propylene product from a feed stream containing propylene and propane (respectively, 70% and 30%) flowing at a rate of 144,000 lb/hr.
The target propylene recovery was 98.6%; the target bottom product was 96.6 mole % propane (propylene, MAPD and C4s made up the balance). As with similar C3 splitters, when the purity requirement of the top product is high, the MAPD in the feed stream tends to leave with the bottoms product.
Fig. 2 (79887 bytes) is a rough process flow diagram of the unit.
A parallel processing configuration was considered at one time. Heat pumping also was considered. Chevron ultimately decided, however, to leave the towers in their existing series configuration.
The unit's vapor feed stream enters Tower DA-406. That tower strips and rectifies. Tower DA-407 completes the rectification. It also pasteurizes.
The final liquid propylene product is drawn off the side of DA-407, as shown in Fig. 2 (79887 bytes). The trays above the side draw pasteurize, i.e., they separate C2s and other light components from the final propylene product.
Five parallel reboilers drive the unit. The space above Tray No. 233 in the revamp design was chosen as the primary location for the vapor feed.
The heat in the overhead vapor stream from DA-407 is rejected to cooling water in four main condensers. Vapor left from these condensers goes to a refrigerant-driven vent condenser.
A small vent stream containing propylene and noncondensibles is recycled to the cracked gas compressor. The final propylene product leaves DA-407 as a side draw originating beneath Tray No. 11 in the revamped tower. Eleven ECMD trays were considered sufficient to effect the required pasteurization.
Fig. 3 (98168 bytes) shows how the 325 ECMD trays were apportioned between the two towers by the engineering team. The original 250 four- pass valve trays were replaced, roughly speaking, with a four- for-three revamp.
Tray spacings were reduced everywhere in the unit except in DA-407's pasteurization section. Close tray spacings (13.5 and 15 in.) are considered normal for MD and ECMD trays.
Table 2 (25986 bytes) compares the design of the original four-pass trays with the new ECMD trays.
The elimination of the receiving pans increased the bubbling area of the trays by 21%. The increased bubbling area, coupled with the deck slotting, gave the new trays their high capacity.
The downcomer areas of the old and new trays were virtually identical. Table 2 (25986 bytes) shows that the ECMD trays tripled the weir length compared to the old four-pass trays. The resultant reduced froth height made it possible to employ the ECMDs at closer spacings and increase the unit's tray count to 325 from 250.
Table 2 (25986 bytes) shows that three different tray designs were employed in the tower to account for the differences in the internal loadings above the draw, above the feed, and below the feed. Every tray, however, employed six downcomers.
The tray decks between the downcomers were 26 in. wide. In fact, they were wider than the tower manholes. It was therefore necessary to design the tray decks using a "tile deck" construction.
The decking between every pair of neighboring downcomers was comprised of pieces spliced together. Those pieces were no wider than the tower manholes. Each ECMD tray had a single centrally located manway.
The trays were fabricated by Denler Metal Products Manufacturing Inc., in Tonawanda, N.Y.
Fig. 4 is a top-view photograph of a two-tray trial assembly performed in Denler's shop. Engineers from Chevron and UOP inspected the assembly, during December of 1991.
The downcomers and the tray decks were fabricated using 12 U.S. gauge carbon steel.
REVAMP
All MD and ECMD trays require 360-degree rings for support. In the Port Arthur towers, two of every three rings were abandoned. Every third ring was filled in, starting with the segmental four-pass rings. For every two unused rings, three new rings had to be added.
All of the existing vertical bolting bars for the old four- pass downcomer curtains had to be cut back. New rings and ring pieces were welded directly to the shell.
Cana-Tex Corp., Houston, performed the ring work and installed the trays during June and July of 1992. All of this work in both towers was completed in 32 days.
START-UP, OPERATION
After start-up in August of 1992, UOP had the opportunity to visit the plant in early February 1993 to observe operation and gather data. Chevron was still in the process of tuning tower controls to accommodate the lower pressure drop and faster response time that a set of ECMD trays provides.
Tuning means adjusting the sensitivity of the various control loops to respond to the normal expected changes in process conditions., i.e., feed rate or composition changes.
During the plant visit, all the process instruments were jointly reviewed by UOP and Chevron, and recommendations were made. Operation improved, but the feed rate to the unit was not up to maximum conditions.
Operating data were collected for verification of tray efficiency. Values between 75 and 80% were calculated from the preliminary operating data, taken during February under light loading conditions.
The predicted tray efficiency, based on UOP's internal methods, indicated that a 71% tray efficiency should be expected.
In November of 1994, additional operating data at steady state conditions were obtained under higher loading conditions. Two periods of operation were studied: Nov. 7 and Nov. 22. On these 2 days, extended steady-state operation was recorded using two very different feed mixtures.
Operating data from computer-logged charts were used to collect the raw operating data shown in Table 3 (28441 bytes).
Compared to the original design, the Nov. 22 operating data are at about 93% of design production capacity, but the internal loads are slightly higher at about 95% of design. These internal loads are approximately 10% higher than a normal UOP MD tray would achieve.
The tower was recently pushed toward its flood point, but definitive capacity data have not yet been collected.
Figs. 5 (77351 bytes), fig 6 (76608 bytes) and fig 7 (69352 bytes) show operating diagrams for the three different trayed sections of this tower. Plotted on these diagrams are several lines that form a quadrilateral window. Operation within this window results in acceptable hydrodynamic behavior on the tray decks.
Also plotted on the windows is the Nov. 22 operating data point. As can be seen from the three figures, the column could have been turned up or down significantly after the data were collected without reaching the flood or weep point.
Both sets of steady state operating data were evaluated to examine tray efficiencies. Using UOP's proprietary VLE model on HYSIM, tray efficiencies between 72 and 75% were obtained for the November 1994 data? These values correspond well with the predicted efficiency of 71%.
To indicate the accuracy of this determination, the Nov. 22 operating data were used to generate Fig. 8 (40886 bytes). The curve shown in this figure represents constant feed and product purities.
Using computer simulation, UOP varied the number of theoretical stages below the pasteurization section and plotted them over Fig. 8 as a percentage of the actual number of trays, or tray efficiency. The resulting reflux rate was calculated and plotted. Matching the actual observed reflux rate to this curve revealed that the actual tray efficiency is 74.2%.
As can be seen from this figure, the tray efficiency determination is somewhat, but not overly, sensitive to the reflux rate. For example, a 3% higher reflux rate would have decreased the observed tray efficiency by 3.7 points to 70.5%.
The observed two-column pressure drop also is very close to design. The predicted pressure drop for the ECMD trays was 22 psi at design.
The ratio of the top pressure of the rectifying tower (DA- 407) to the top pressure of the second tower (DA-406), and the pressure drop in the second tower, were used to determine the observed two-tower pressure drop (including the line loss between the two towers). The observed pressure drops are within 10% of the design value.
ACKNOWLEDGMENTS
A very large team was required to revamp Chevron's propylene unit. The authors wish to thank the many engineers and technicians from Chevron, Stone & Webster, and UOP, who provided valuable contributions.
REFERENCES
1. Kirkpatrick, R.D., "MD Trays Can Provide Savings in Propylene Purification", Oil & Gas Journal, Apr. 3, 1978, p. 72.
2. Resetarits, M.R., Agnello, J., Lockett, M.J., Kirkpatrick, H.L., "Retraying Increases C3 Splitter Column Capacity," Oil & Gas Journal, June 6, 1988, p. 54.
3. Weiler, D.W., England, B.L., and Delnicki, W.V., "Flow Hydraulics of Large Diameter Trays," AIChE 74th National Meeting, Mar. 12-15, 1973, Paper 94-C, New Orleans, LA.
4. Weiler, D.W. and Lockett, M.J., "Design and Performance of Parallel Flow Slotted Sieve Trays," IChemE Symposium Series No. 94, April 1985, p. 141.
5. Resetarits, M.R., Miller, R.J., Navarre, J.L. Linskeseder, M., Reich-Rohrwig, P., "New Enhanced Capacity MD Tray Debottlenecks Deethanizer," I. Chem. E. Distillation and Absorption Conference, Birmingham, U.K., Sept. 7, 1992.
6. Hyprotech Ltd., "HYSIM Desktop Process Simulator," Version C2.54, November 1994.
Copyright 1995 Oil & Gas Journal. All Rights Reserved.
