ETHYLENE PLANT STEAM STRIPS WASTE WATER TO RECOVER BENZENE

Matthew A. Taylor
E.I. du Pont de Nemours & Co. Inc.
Orange, Tex.

Du Pont's Sabine River Works' ethylene plant in Orange, Tex., has selected a steam stripping process to remove benzene from its waste water stream.

Timely completion of the project plan diverted a possible cleanup cost in excess of $100 million.

The Du Pont plant was constructed in 1967 to crack feedstocks ranging from 50% ethane/50% propane to purity ethane. As shown in Fig. 1, the plant is currently cracking purity ethane in the presence of dilution steam to produce ethylene and various byproducts.

The dilution steam is added to improve selectivity and reduce fouling in the heater coils. The cracked gas exits the heaters and goes to the quench tower, where the steam and a portion of the heavier hydrocarbons such as benzene, toluene, and styrene are condensed.

After the cracked gas is cooled in the quench tower, it is compressed and then separated into the various products in the distillation train. The condensed steam and hydrocarbons are separated in the quench settler. The hydrocarbons are sent to the heavy aromatic distillate (HAD) storage for sale.

The condensed steam, with dissolved hydrocarbons, combines with the effluent from the API separator and goes to the dissolved gas flotation (DGF) unit, which was installed in 1988.

When the plant was started up, the quench water was recycled to make dilution steam. However, this system was plagued by plugging problems, and the quench water was rerouted to the bioponds. The AP] separator is the collection point for all of the oil and water blowdowns in the ethylene plant.

In addition to the blowdowns, large quantities of rainwater go to the API separator. The hydrocarbons leaving the DGF go to HAD storage, and the water goes to the bioponds.

BACKGROUND

The EPA Toxicity Characteristic (TC) was promulgated on Mar. 25, 1990, and became effective Sept. 25, 1990. The TC list, which contains 25 chemicals, is important because it defines a waste as Resource Conservation and Recovery Act (RCRA) hazardous if the waste contains a listed chemical at a concentration above the listed maximum. The chemical of interest to the ethylene plant is benzene, which has a maximum allowable concentration of 500 ppb.

Before the TC list became effective, the benzene concentration in the 450 gpm waste water stream going to the bioponds averaged about 360 ppm. Therefore, after the effective date, this stream would have become an RCRA hazardous waste.

Du Pont formed a project team in December 1988 to design, construct, and start up facilities to remove the benzene from the plant waste water. When the team began work, it was believed that benzene would be listed at 72 ppb, and that the TC list would be promulgated in March 1989, becoming effective in September 1989.

The major driving force for this project was to avoid contaminating the bioponds with a potential RCRA hazardous waste. The bioponds are not permitted for RCRA wastes, and it was believed that contamination of the bioponds would require their closing and clean up at a cost in excess of $100 million.

PROJECT

This project was schedule-driven from the beginning. Because of the belief that the TC list would become effective in September 1989, this project was managed in a way whereby the project scope development, the basic data generation, and the construction were performed concurrently.

The team was able to accomplish its goals because project objectives were developed and updated as information about the TC list changed.

The original project objectives were:

THE COMPLETED FACILITIES WILL MEET THE PROPOSED TC LIMITS. The plant was designed to meet a 72 ppb benzene limit. This limit was not changed when it became apparent that the limit was going to be 500 ppb. This change occurred late in the project and would not have affected the decision that steam stripping was the best process to meet the TC list requirement.
THE FACILITIES WILL START UP BY SEPTEMBER 1989. This date was moved to April 1990 when it became evident that the TC list was not going to be promulgated as early as had been expected.
THE PROJECT WILL COMPLY WITH THE VARIOUS REGULATORY STATUTES. The project team stayed in contact with the various regulatory agencies, and in particular, with the Texas Air Control Board (TACB), which issues construction permits. By keeping the TACB informed of the plans, Du Pont was able to obtain a permit which met its time requirements.
THE FACILITIES WILL HAVE NO SCHEDULED OUTAGES. This resulted in all critical equipment, including the stripper column, being spared. The electricity and steam were supplied from separate sources.
THE IMPLEMENTATION OF THE PROJECT WILL HAVE MINIMAL IMPACT ON THE ETHYLENE PLANT OPERATION. Requirements for plant tie-ins and equipment installation were identified early so that this work could be performed during unscheduled plant shutdowns.
THE PROJECT TEAM WILL MINIMIZE THE INVESTMENT REQUIRED TO MEET THE PROJECT OBJECTIVES. This was accomplished by obtaining used equipment from various plant sites. In addition to lowering our investment requirements, this reusage enabled the team to meet the construction schedule.
THE FACILITY WILL PRODUCE A MARKETABLE PRODUCT. This minimized the permitting that was required. If Du Pont had not produced a marketable product, an RCRA waste treatment permit would have been required before construction. This permitting procedure would have prevented the ethylene plant from being in compliance with the TC list when it became effective.

PROJECT DEVELOPMENT

The first task facing the project team was selecting a process that would meet the project objectives. In addition to the -process objectives, four other criteria were used in the selection of a process.

The initial process selection criteria included installation time, process reliability and operability, followed by operating cost and investment.

The project team considered four processes--activated sludge, Ultrox steam stripping, and carbon adsorption; but two of them were quickly eliminated.

An activated sludge unit was not seriously considered because several years would have been required to acclimate the bacteria, and also because a large area is required for the tankage.

The Ultrox process uses UV light and peroxides to reduce the hydrocarbons to CO2 and H2O. It was considered and rejected because the team felt that it had not been proven commercially.

After discussions with several vendors and Du Pont's engineering services department consultants, the project team decided that steam stripping and carbon adsorption were reliable and could be installed in the available time frame.

The team then agreed on the preliminary design for both processes, which is shown in Fig. 2.

This design was used to develop operating and investment costs. As mentioned earlier, the dissolved gas flotation unit (DGF) was existing. The DGF separates undissolved hydrocarbons from the water. This separation is improved by dissolving nitrogen in the water under pressure.

This gas effervesces in the DGF when the pressure is reduced. The oil leaving the unit contains small amounts of water and solids. A centrifuge produces three streams: a clean oil stream that goes to HAD, dry solids that are packaged for waste disposal, and an oil and water stream that is recycled to the API separator.

The water leaving the DGF goes to a set of sand filters where solids, and some oil, are removed to prevent pluggage in the steam stripper and carbon beds. The solids are backwashed to the API separator. The water then goes to a surge tank, which provides holdup for process upsets or heavy rains.

The water from the surge tank then goes to either a steam stripper or a set of carbon beds, then to the bioponds. The recovered hydrocarbons from the steam stripper go to HAD storage. The hydrocarbons removed in the carbon beds leave with the used carbon as a waste.

After these processes were developed, several investment options were calculated. These assumed that the equipment for the two processes was identical through the surge tank. Downstream of the surge tank, various equipment combinations were considered to improve the process reliability.

Investments were calculated for processes with two steam strippers, two sets of carbon beds and a steam stripper, and three sets of carbon beds.

Du Pont determined that the same investment would be required to install a steam stripper or a set of carbon beds.

The combination of a steam stripper and spare carbon beds was eliminated as an option based on a belief that if there were a problem with one process, there would not be time to learn how to operate the other process. After the process flow sheets and investment options had been developed, a process selection was made.

The steam stripping process had these six advantages over the carbon adsorption process:

Steam stripping would use a non-RCRA facility because the process would recover dissolved hydrocarbons for sale as HAD. Carbon adsorption would use beds to adsorb the hydrocarbons, then the hydrocarbons would be burned when the carbon was regenerated.
This facility would be considered an RCRA waste treatment facility and would require the appropriate permitting. This permitting would prevent the ethylene plant from being in compliance with the TC list when it became effective.
Steam stripping would not generate a new waste. However, the solids from the centrifuge would increase an existing solid waste stream. The use of carbon beds would have required the disposal of used carbon. Our estimates indicated that 20% of the activated carbon would be lost when the carbon was regenerated.
The operating cost of the steam stripper would be one tenth of that for the carbon beds. The major operating costs for the steam stripper would be in steam, and for the carbon beds, in carbon replacement.
The investment cost for steam stripping, with a spare steam stripper, was less than carbon adsorption because surplus equipment that could be reused was found at other Du Pont sites.
The availability of surplus equipment made it possible for the project team to meet project time requirements. The use of this equipment later made the selection of packing overtrays almost a foregone conclusion because one of the towers had been packed and could not be easily modified for trays.
The final advantage of steam stripping was, the project team believed, that the process scope could be optimized to lower the investment further.

One major advantage offered by carbon adsorption was that it was a proved process. Steam stripping had been used to remove hydrocarbons, but Du Pont did not have data to indicate it had been used continuously to reduce benzene to the goal of 72 ppb.

After comparing the advantages of steam stripping and carbon adsorption, it was obvious that steam stripping was the preferred process. A project scope was then written for the process shown in Fig. 3, and the detailed engineering design was begun.

PROCESS OPTIMIZATION

Immediately after steam stripping was selected for benzene recovery, testing began to define the system operability and reduce the required investment.

GLASS STRIPPER

The objective of the first test was to identify and solve any operating problems that might be present with steam stripping. Problems of particular interest were plugging, fouling, and foaming.

A 6-in. diameter, 8-ft tall glass column containing No. 15 Norton IMTP packing was installed. This column was fed an unfiltered 6 gpm slip stream from the water side of the DGF unit for 6 weeks. No plugging or foaming was seen. However, light fouling did occur, but was not a problem.

This test showed no operational problems and indicated that a lower quality filtration step could be used in the process.

At the completion of this program, two parallel tests were conducted. One test was aimed at eliminating the filtration investment completely, and the other at reducing this investment. The sand filters represented an investment of about $1.2 million.

STAINLESS STEEL STRIPPER

The objective of this test was to completely eliminate the filtration investment. For this test, a 43-ft tall, 6-in. diameter stainless steel column containing No. 15 IMTP was installed in place of the glass column. This column was installed to determine whether residual benzene in the solids should be a concern.

There was concern that the benzene would be stripped from the water and not the solids. If this happened, then the possibility existed that after the stream had left the stripper, the benzene in the solids would diffuse into the water and increase the benzene concentration above 72 ppb. If the benzene diffused from the solids, then some filtration would still be required.

These test results showed that the benzene concentration could be reduced below 50 ppb and that the benzene did not diffuse into the water. The success of this test eliminated the need for filtration.

LAKOS SEPARATORS

The objective of this test, which was performed in parallel with the stainless steel steam stripper test, was to reduce the investment required for filtration by demonstrating another separation technology. To conduct this test, a set of Lakos separators was installed at the inlet and exit to the DGF. A Lakos separator is shown in Fig. 4.

A feed stream containing several percent solids enters the separator tangentially and centrifugal force moves the solids to the outside of the separator as the flow moves to the bottom of the cone. When the flow stream then changes direction and travels up the center of the separator, the concentrated solids are left at the bottom of the unit.

The solids can then be blown down continuously or intermittently. The two separators were installed in series to increase the solids removal.

This test showed that Lakos separators could be used to reduce the filtration investment from $1.2 million to $300,000. It also showed that the DGF operation could be improved by installing the Lakos separators upstream of the DGF.

SOLIDS REMOVAL

The shortcoming of the process design shown in Fig. 3 is that the only point for solids to leave the system is with the oil from the DGF. Therefore, unless the solids float, they will not be removed from the system.

As soon as the project was formed, testing began to improve the method of removing solids from the system. Our initial testing of the DGF water effluent showed a high solids concentration, which indicated there was a large quantity of solids that did not float.

One solution to this problem was to combine the oil stream from the DGF with the sand filter backwash stream and send it to the centrifuge. If the sand filters were replaced with the Lakos separators, the Lakos underflow would be combined with the oil from the DGF and sent to the centrifuge. Because the underflow from the Lakos is much smaller than the backwash flow from the sand filters, a smaller centrifuge could be used to separate the sands, oil, and water.

Several test centrifuges were installed to make the different separations. Initially, a high-speed machine was installed to separate the water from the DGF oil stream. The machine provided excellent separation, but it lacked the reliability we needed.

The major problem with this machine was that the solids loading in the oil stream was much higher than expected, and was eroding the machine parts. After several months, it was decided that this machine was not suitable for this separation. Next, a centrifuge designed for three-phase separation and high-solids loading was installed. Several tests were run with this machine and it handled the different separations well.

This machine would have been used, except that during the design and installation of the stripper, two changes were made in the quench system that reduced the solids loading in the effluent from the DGF. These changes were the installation of basket strainers in the circulating quench water to remove large solids, and the installation of high-pressure nitrogen supply to the DGF.

Routine fluctuations in the existing nitrogen supply were adversely affecting the performance of the DGF. With the new high-pressure supply, the effectiveness of the unit was improved, and the finer solids (which tend to float) travelled overhead with the oil stream.

After the stainless steel steam stripper test was completed and the filtration step was eliminated from the process, the need to separate the solids lost importance. The only reason to retain the centrifuges on the DGF oil stream would have been to upgrade the stream by removing the small quantity of oil and solids.

Discussions with the HAD customers provided little incentive to remove the solids. Consequently, the centrifuges were removed from the process.

This successful testing program resulted in the project scope which is shown in Fig. 5. These scope reductions were worth about $5 million. The actual project implementation schedule was as follows:

December 1988: Formed project team
April 1989: Selected steam stripper
May 1989: Began construction
July 1989: Authorized project
March 1990: Completed construction; TC list promulgated
April 1990: Began operation
September 1990: TC list effective.

The project team began work in December 1988, and the equipment was installed and operating in April 1990. Operation could have started in September 1989, but by delaying the schedule to match the promulgation of the TC list, several hundred thousand dollars in overtime costs were saved.

This was a unique project for Du Pont because construction began before the project authorization, and the basic data for the process were developed as construction was proceeding. The project was completed on time, with the total investment under budget.

The estimator was asked to estimate what the project would have cost if the initial estimate had been done on the final project scope. This exercise showed that the actual project investment matched well with the estimate.

ENVIRONMENTAL IMPACT

Using the design basis of 450 gpm and 360 ppm benzene, this project reduced the biopond benzene emissions by 70 tons/year. This assumes that 20% of the benzene entering the bioponds was emitted into the atmosphere.

A TACB permit was obtained to begin construction in May 1989. This permit was modified in 1990. Originally, the tank and decanter emissions were vented to the atmosphere through a brine condenser and a set of carbon beds. After start-up, these vents were rerouted to the ethylene plant flare and will only be routed to the carbon beds when the flare is out of service.

By recovering approximately 400 tons of hydrocarbons per year from the water, the possible uses of this water have increased. Plans are currently being developed to use this water stream for cooling tower makeup.