ETHYLENE FRACTIONATOR REVAMP RESULTS IN 25% CAPACITY INCREASE

Daniel R. Summers
UOP
Tonawanda, N.Y.

Steven T. Coleman
Lyondell Petrochemical Co.
Channelview, Tex.

Ronald M. Venner
ABB Lummus Crest Inc.
Houston

Lyondell Petrochemical Co. and ABB Lummus Crest Inc. studied the revamp potential of Lyondell's two Channelview, Tex., olefins plants, setting a target ethylene capacity increase of 20%.

The target was exceeded, with each plant capable of producing 800,000 metric tons/year of polymer-grade ethylene. This represents an increase of more than 35%, compared to the original nameplate capacity.

The following modifications of the ethylene fractionation system resulted in a 25% capacity increase without adding a parallel splitter:

The flow scheme was modified for capacity increase via improved energy efficiency.
UOP High Flux tubing was retrofitted to the condensers.
The operating pressure and reflux rate were lowered.
valve trays were replaced with UOP Multiple Downcomer trays (also marketed as MD trays).
A new side reboiler was added.

The revamp of the ethylene fractionators was critical to the success of the olefins capacity expansion. The expansion project, completed in 1989, also increased the production of other key byproducts.

BACKGROUND

Lyondell's Channelview complex features two world-scale olefins plants. The two plants - originally designed as naphtha/gas oil crackers were commissioned in 1976 and 1977, each with nameplate capacities of 590,000 metric tons/year ethylene.

Since start-up, the plants have gone through a number of expansion projects including the addition of cracking heaters, a feedstock flexibility revamp, a product flexibility expansion, a propylene fractionator revamp, and an olefins capacity expansion.

The feedstock flexibility revamp, completed in 1981, expanded feedstock processing capability to include natural gas liquids. The propylene fractionator revamp increased polymer-grade propylene production capability by converting the existing conventional trayed system to MD trays. 1

The plant processing scheme, shown in Fig. 1, is that of a classic liquids cracking plant featuring a five-stage centrifugal compressor, high-pressure (31 bar) demethanization, back-end acetylene converters, super-fractionators for polymer-grade ethylene and propylene production, and a cascade refrigeration system using propylene and ethylene.

Prior to the revamp study, the known bottlenecks in the plant were the demethanizer, de-ethanizer, and the condensate stripper in the charge compression section.

The study indicated that numerous other plant sections would bottleneck at slightly higher capacity. These bottlenecks included the charge compressor, propylene refrigeration compressor, ethylene refrigeration compressor, and ethylene fractionator. The objective to simultaneously increase by-product production meant that the back-end systems (depropanization, debutanization, and DPG gasoline hydrogenation) would also bottleneck.

Lummus Crest recommended a revamp scheme for the plant based on its maximum-capacity expansion technology (MCET), which provides the most cost-effective expansion plan by employing some of these principles:

Modernize flow sheet for capacity gain
Minimize parallel processing
Introduce energy conservation to achieve target capacity
Maintain or improve plant operability and reliability
Minimize turnaround requirements.

In MCET, the processing scheme is modernized for capacity gain by incorporating state-of-the-art design features from grassroots or other revamp designs. Typically, parallel processing is minimized to lower investment costs. Heat-integration schemes and equipment efficiency improvements are used to debottleneck compressors.

The ethylene fractionator (C2 splitter) system presented a challenging problem. Modifications were made to the processing scheme surrounding the tower to help debottleneck the refrigeration compressors. These changes included adding a side reboiler, which condensed ethylene refrigerant but also increased the load on the tower.

The revamp objective was to debottleneck the tower without adding a substantial parallel column or series extension to the existing tower.

Various options were evaluated, including using different conventional tray designs, packing, and MD trays.

The MD tray option represented the most reliable and economical solution to meeting the target capacity while accommodating proposed flow-scheme modifications.

DESIGN CONSIDERATIONS

In mid-1988, Lummus Crest asked UOP to look at retraying the existing C2 splatters. The original design, shown in Fig. 2, contained 125 four-pass valve trays.

The top section of the tower served as the pasteurization section, with provision to sidedraw ethylene product.

UOP's experience with Lummus Crest in revamping similar units indicated that the requested 20% increase could easily be attained.

In UOP's experience, C2 splatters containing 110-130 conventional four-pass valve or sieve trays at a typical 457-mm tray spacing can be revamped so that a 20-25% increase in capacity results. This objective was accomplished through the use of MD trays.

The expanded C2-splitter processing scheme is shown in Fig. 3.

COLUMN DESIGN

Several design considerations were incorporated into the MD tray application for the Lyondell plant. The first consideration was to place more trays in the tower to reduce the reflux ratio and improve product purity.

Operation of the tower before the revamp resulted in ethylene concentrations as high as 2% in the column bottom and contaminants in the ethylene product as high as 1,000 ppm. Lyondell was concerned not only with ethylene capacity but also with their customers' increasing need for purer ethylene product.

Ethylene purity has constantly improved over the years (Table 1). Lyondell wanted to meet customer expectations with a product containing less than 400 ppm of combined ethane plus methane. A comparison of the original and proposed fractionation specifications is shown in Table 2.

The expansion design achieves tighter overhead and bottom specifications than did the original design or the pre-expansion operation.

A second consideration was to eliminate the pasteurization section and replace it with an external-vent recovery system. This idea enabled Lyondell to use additional column height and an extra 12 trays for the ethylene/ethane separation.

A third consideration in the design was to maintain or increase the temperature approach on the reboilers and condensers. This was accomplished through a low pressure-drop tray design and the use of High Flux tubing.

The MD trays were designed for considerably lower overall pressure drop. Even with approximately 25% more trays in the tower, the pressure drop was lowered from 104 kPa to 59 kpa-a 40% reduction.

The four condensing bundles of the main overhead condenser were replaced with High Flux tubing. This tubing provided additional condenser capacity and permitted a much smaller delta T.

Because the tubing allowed lower tower-pressure drop and a closer temperature approach, the tower was able to operate at lower pressure. Lower-pressure operation helped debottleneck the tower by both reducing the required reflux ratio and increasing delta T in the reboilers.

A fourth consideration was the need to add a new side reboiler to the splitter using condensing-ethylene refrigerant, thereby supplementing the charge-gas side reboiler that was already in place.

This new side reboiler debottlenecked the propylene refrigeration machine and enabled the bottom reboilers to handle the new capacity without modification. The total side reboiler duty is about half the total reboiler duty.

TRAY SPACINGS

The MD design incorporated 25% more trays than the original design, at closer tray spacings than the original 457 mm (18 in.). As has been shown, MD trays can be placed at spacings_ as close as 230 mm (9 in.).2

Typical tray spacings for MD trays incorporating a reasonable amount of turndown are in the range of 280-360 mm. Several tray spacing arrangements were considered to obtain the maximum column capacity within the energy guidelines put forth by Lummus Crest.

UOP considered the location of all existing nozzles, manways, and column girth welds when evaluating the various tray spacing options. In addition, it was advantageous to the timing of the revamp to reuse as many of the existing tray support rings as possible in the new tray arrangement.

Typical tray spacing considerations for an existing 457-mm tray spacing are listed in Table 3:

The chosen design was to use a 3-for-2 revamp below the feed point and a 6-for-5 revamp above it. This decision resulted in reusing 33 of the existing support rings, or 26% of the original 125.

This arrangement increased the number of theoretical trays and maximized capacity. The 381-mm (15-in.) tray spacing in the section above the feed is quite large for MD trays and provides high capacity. This was the most heavily loaded section of the tower.

Below the feed, especially below the side reboilers, the internal loads are much smaller, making a 305-mm (12-in.) tray spacing compatible. This smaller tray spacing enabled a considerable increase in theoretical trays below the feed.

Increasing the number of theoretical trays in a tower reduces the reflux ratio and improves capacity. However, at a certain point, usually when the minimum reflux ratio is approached, the reduction in tray spacing reduces capacity at a greater rate than the smaller reflux ratio improves it.

The chosen design optimized both tray spacing and reflux ratio.

SIDE REBOILERS

The feed point was also optimized to minimize energy requirements to the tower. With the addition of the new side reboiler, the feed point had to be moved to the uppermost alternate location. This move was necessary to provide sufficient theoretical trays below the feed point when the boil-up ratio in the bottom section was reduced.

The side reboilers provide a major portion of the energy to the tower. Based on this information alone, an attempt was made to place these reboilers as low in the tower as possible. However, the dew point of the vapor returning from the side reboilers prevented their liquid withdrawal from being situated very far below the feed point of the tower.

ln essence, the majority of the column's separation had to occur between the top of the tower and the side reboilers. This situation was ideally suited to MD trays, which are inherently compatible with small tray spacings.

The MD tray consists of sections of perforated tray decks separated by long, trough-like downcomers. 3 The bottoms of the downcomers are spouted to permit liquid to exit onto the deck of the tray below. The downcomers terminate in the vapor space above the froth of the tray below.

The key feature of the MD tray is the absence of a receiving pan. The elimination of this pan typically enables an additional 20% of the column cross-sectional area to be used as either active area or downcomer area.

For this application, approximately 10% more area was made available for vapor and liquid traffic by replacing the existing four-pass trays with MD trays.

Successive trays within an MD tray set are rotated 90 degrees to alleviate the liquid distribution problems associated with conventional two or four-pass trays. A long outlet weir, provided by the multitude of downcomers on an MD tray, leads to low froth heights.

Low pressure drops result from these low froth heights, which in turn result in small downcomer backups. Together, the low froth heights and small downcomer backups make possible the installation of MD trays at very small spacings.

Lyondell and Lummus Crest had considered using packing in the C2 splitter, but packing could not provide the required capacity within the existing tower shell. Other types of trays were not really considered because four-pass valve trays with sloped downcomers were already employed. A modified four-pass tray would not provide sufficient capacity increase.

PERFORMANCE GUARANTEE

The experience factor was important to Lyondell in the revamp of its tower. UOP had successfully incorporated MD trays in almost 30 previous designs in C, splitter operation (Table 1). As of early 1992, that number exceeded 40.

UOP also had the ability to guarantee the performance of the tower based on its experience and simulation capability. The expected performance of the C2 Splitter compared favorably to UOP's previous experience, and a performance guarantee was provided with the trays.

This performance guarantee stated that the tower would process the full feed rate and make the specified product without flooding, within a specified maximum condenser duty of 137 giga-joules/hr. UOP's guarantee on tower performance was further endorsed by Lummus Crest's guarantee on total plant capacity.

As can be seen in Fig. 4, showing the operation of the C2 splitter before the revamp, the split between ethylene and ethane is difficult. The McCabe-Thiele diagram shows two pinches, which result in a considerable number of theoretical trays being required.

Because these pinches are so tight as a result of energy minimization, the simulation tools necessary to model this system must be precise. UOP has invested considerable manpower into establishing design tools that accurately model the ethylene/ethane separation over a wide range of temperatures and pressures.

TRAY INSTALLATION

In May 1988, Lyondell made the decision to use MD trays in the C2 splatters in both ethylene units. The first unit was shut down in early 1989 for approximately 50 days for a normal scheduled maintenance turnaround. The MD trays were installed within that time period.

The revamp of the tower required roughly 28 days to complete. The start-up of the first unit (OP-1) was nearly flawless. During this time, Lyondell did not lose production because the second unit was still on line, and sufficient ethylene was stored in anticipation of the turnaround.

The revamp of the second unit commenced several weeks after the successful start-up of the first unit.

OPERATION

The operation of OP-1 has shown that the debottlenecked C, splitter could accommodate a 25% capacity increase (target was only 20%). Production levels as high as 2,450 metric tons/day have been demonstrated in both units since start-up.

Operating data on the C2 splatters were recently collected under steady-state conditions (Table 4). These data were simulated with 115 theoretical stages to match the reflux rate to the tower corresponding to an overall observed tray efficiency of 74%.

The data in Table 4 reflect typical operation without pushing the unit to its limits; the unit has demonstrate over 10% additional capacity with similar tray performance.

Fig. 5 shows a McCabe-Thiele diagram of the tower's current performance. Based on comparison to Fig. 4, the addition of the new hip reboiler and additional theoretical trays below the feed clearly show an improved bottoms purity. However, because the tower is "pinched" above the feed, this type of diagram shows that the separation is extremely difficult and close to minimum reflux.

The 25% increase in ethylene capacity was achieved without paralleling major pieces of equipment (for example, compressor and superfractionator)-a major factor in achieving a low-cost revamp. The fractionation efficiency improved, as shown by the improvement in ethylene product purity and recovery.

For the future, UOP is now studying a new enhanced-capacity MD tray, which promises greater capacity on the order of 1015%.

With no major new ethylene plant construction foreseen in the U.S., growth in ethylene production capacity will come from revamps of existing units.' To maximize the fractionation capability of existing towers, whether for capacity, for improved product purity or, as in Lyondell's case, for both, consideration should be given to using high capacity trays such as the MD tray.

REFERENCES

Summers, D.R., and Coleman, S.T., "Trayed Revamp Yields a Significant C3 Splitter Capacity Increase," paper presented at Session 129 AIChE meeting, Chicago, November 1990.
Ragi, E.G., Weiler, D.W., and Wolf, C.W., "Energy costs prompt improved distillation," OGJ, Sept. 1, 1975, pp. 85-88.
Lockett, M.J., and Wisz, M.W., "Use of Multiple Downcomer trays to Increase Column Capacity", paper presented at European Federation of Chemical Engineering Working Party meeting on Distillation, Absorption and Extraction, Winterthur, Switzerland, June 1983.
European Chemical News, 57 (1499), pp. 21-22.