Abu Dhabi NGL plant revamped to handle new feed composition

Adel Bamardouf Al-Amoodi
Abu Dhabi Gas Industries Ltd. (Gasco)
Ruwais, Abu Dhabi

Boris Ertl
M.W.Kellogg Ltd.
London

• Capacity study criteria [82,455 bytes]

Changes in 1996 to the stream composition feeding the NGL fractionation plant at Ruwais, Abu Dhabi, prompted operator Abu Dhabi Gas Industries Ltd. (Gasco) to revamp and debottleneck the plant.

The new feed-component distribution has led to increased propane and butane rates and to significantly increased sulfur component rates.

Gasco engaged Kellogg Construction Ltd., London, to lead the revamp project, from process debottlenecking studies through to construction and commissioning.

The project objective was to achieve a plant revamp accommodating the new feed, within the following constraints:

• Cost-effective and technically proven solution
• Minimum unit shutdown period for construction within a live plant
• Strict compliance with stringent quality, safety targets
• Minimum disruption to operating philosophy of existing plant.

Project objectives were met; the plant has operated well since project completion.

NGL from three plants

Originally built in 1979, the Ruwais plant was designed to process 21,800 metric tons/day (approximately 40,000 cu m/day) of NGL from three extraction plants at Asab, Bab, and Bu Hasa, and LPG from the Ruwais refinery.

The hydrocarbons are separated by distillation, and sulfur components are removed by molecular-sieve treating and by ADIP-D, a Shell International Oil Products (SIOP) licensed amine-treating process that uses di-iso-propanolamine (DIPA) as solvent. The molecular-sieve unit was supplied by UOP from its Antwerp site.

Fig. 1 [109,928 bytes] shows the Ruwais process facilities, prerevamp; one of two essentially identical process trains is shown.

Feed is a combination of streams from the upstream plants pumped to the fractionation unit. The NGL fractionation train consists of the de-ethanizer, depropanizer, and debutanizer with gas-fired reboilers and seawater-cooled condensers, except the de-ethanizer which requires refrigeration.

De-ethanizer overhead gas is treated with ADIP-D and molecular sieves for further sulfur removal and dehydration. Propane product is treated in the ADIP-D unit followed by molecular-sieve beds. Butane product is treated with molecular sieve.

In support of process operations, a three-level propane refrigeration cycle is installed in each process train to provide refrigeration for the de-ethanizer overhead condenser, condense boil-off gas from storage, and chill product rundown to storage prior to shipping.

Fig. 2 [91,444 bytes] shows the impact of developments to upstream plants on the pre-revamp design basis of the plant.

Design capacity is slightly increased, but the most significant difference is the changed component distribution. Propane and butane rates are increased by around 45%, and total design sulfur content in the feed is increased by around 800%.

Project execution

The project commenced in June 1993 with process studies whose objectives were to identify the ultimate capacity of the plant with potential new feeds and to identify the equipment items causing capacity limitations. Priced revamp options were developed for each bottleneck.

The study report issued in September 1993 enabled Gasco to define a "Basic Engineering" basis. The bulk of the revamp has been implemented, while revamps not required during initial years of Gasco's predicted production profile are deferred.

Under an aggressive schedule, the basic engineering package was completed in March 1994. Site construction began in November 1994, and successful commissioning and plant start-up of the two trains were achieved in December 1995 and March 1996, respectively.

Process modeling

Work commenced with simulation of the existing plant design.

The next stage was to adjust the model to match actual plant operating experience. Because the plant has two identical trains, Gasco was able to perform test runs of one train at a range of capacities, adjusting the second-train throughput to maintain the plant's production obligations.

Test runs were performed at the limits of train capacity with columns operating at stable, but close to flooding, conditions.

Results showed that the plant was capable of operating at greater-than-design capacity, and product specifications were achieved with better-than-predicted reflux ratios and utilities consumption.

Tray efficiencies used in the simulation model were adjusted to match the test-run data and found to be better than the values considered for the original design of the plant. Column performance exceeded design expectations, with higher vapor and liquid loads before reaching flooding than predicted by tray-rating calculations.

Test-run data were analyzed around all major equipment to build a "rating" model. Actual pump and compressor performances were logged to obtain operating machine efficiencies and to verify operating curves.

Measured heat-exchanger performance was used to build a heat-transfer model of each exchanger with actual heat-transfer coefficients and fouling resistance.

One significant area of concern was the distribution of sulfur components among the products because of the impact on design of downstream sulfur-removal units.

Proprietary Kellogg data for sulfur-component interaction were used in HYSIM to model this distribution. The model results were also tested against an Aspen simulation run, and alternative data sets for sulfur component interaction coefficients, to test sensitivity.

The result of this work, and Gasco's own experience with the existing plant, led to the basis for design distribution of sulfur components (Table 1) [29,098 bytes], including margin in some areas reflecting theoretical uncertainty and feed-gas composition uncertainty.

The outcome of this simulation work was a complete model of the plant, adjusted to predict actual operation. This tool enabled debottlenecking studies to be completed with sufficient accuracy to meet Gasco's objective to maximize re-use of existing equipment with tight or zero design margins, reflecting the high certainty obtained by having such complete data on actual plant performance.

A range of feed cases was defined, covering alternative operating conditions. For each of these cases, capacity studies were completed for all equipment to identify bottlenecks in each feed case. General guidelines for this study are found in the accompanying box (p. 44).

The outcome of these studies was a complete equipment list with each equipment item's potential capacity expressed as a percentage of the required capacity for each feed case. (A rating lower than 100% indicated a bottleneck.)

This gave an item-by-item "hit list," in order of priority, of equipment undersized for the new duty requirements.

Bottlenecks addressed

Engineering solutions to the bottlenecks identified were then developed.

Process work at this stage considered the practicalities of cost control, process control, construction, commissioning, and operations considerations.

This is particularly important for a complex revamp project in which much of the work needs to be completed within a plant that is on-line. The economics of total project cost make minimum shutdown duration a critical part of achieving profitable overall execution.

Success factors at this stage of the project included:

• Constructability reviews-Often the most elegant process solution to a bottleneck simply cannot be built. Early site surveys are critical to ensure process solutions are practical.

• Early construction planning-The cost and complexity of construction in revamp projects is such that an early estimate of the extent of works is important to achieve accurate cost estimates.

  Detailed information on construction planning and costs early give the design team the opportunity to find optimum process solutions which take account of the entire work process.

• Early shutdown planning-Project economics favor designs that minimize shutdown duration even at the expense of increased capital cost. Planning the shutdown period is best started during project definition because it can affect optimization of process solutions to revamp problems. An example would be choosing between making several new tie-ins to an existing system or building a new subheader with a single tie-in point.

• Value engineering studies-The process of analyzing project-cost break down must be started during project definition when there is the greatest opportunity to influence final cost. Significant cost savings were achieved on this project by reuse of existing equipment made redundant by the revamp.

• Control, operability reviews-Early involvement of Gasco operating staff in the design process ensured that the design team understood how the existing plant was operated. The project set itself an objective to minimize disruption to plant-operating strategy, and Gasco personnel were early involved in design reviews.

Solution reviews

Process solutions were proposed for each of the bottlenecks identified and subjected to the reviews previously outlined. Following are the final engineering solutions for each of the main process units.

Fractionation unit

Fig. 3 [90,464 bytes] shows the equipment revamps required in the fractionation unit.

The major bottleneck identified was the depropanizer and associated utilities, as a result of increased propane and butane throughput affecting traffic in this column.

Various options considered for debottlenecking included the following:

• Increase the number of theoretical stages by replacing trays with a packed section
• Improve column efficiency and capacity by installing a higher capacity proprietary tray design
• Add a new column in parallel with the existing duty
• Produce off-spec product in the existing column, with a polishing unit downstream
• Reduce operating pressure to improve distillation selectivity, redesign condenser system to operate at lower temperature, either by installing a colder utility or introducing a compression system in the column overhead
• Add inter-reboiler to redistribute loads
• Add feed preheat to redistribute loads
• Reconfigure fractionation strategy using Kellogg divided-wall column technology (KDWC).

The most cost-effective solution in this case was found to be retraying the column with a proprietary design of high-capacity trays enabling an increase in the column traffic.

Additional condenser and reboiler capacity is provided by a new exchanger shell in parallel with the existing shells and a new furnace in parallel with the existing reboiler.

The new condenser shell requires a separate structure to elevate it alongside the existing exchanger shells. Replacing existing shells with a compact heat-exchanger design to avoid the new structure was considered but rejected due to poor access for construction, an extended shutdown period, and risks inherent with seawater-cooling service.

Operating experience of the existing shells is that the seawater side needs frequent cleaning to maintain acceptable heat transfer in this unit. The revamp project re-balanced seawater flows around the plant with new restriction orifices to increase velocities through these exchangers, thus improving heat transfer and reducing scale build up.

Amine-treating unit

Fig. 4 [113,614 bytes] shows a flow diagram of the unit modified to accommodate the new duty, SIOP's licensed amine-treating process (ADIP-D) that uses di-iso-propanolamine (DIPA) as solvent.

De-ethanizer overhead (DEO) is treated in a trayed absorber column. Propane is treated in a packed bed liquid-liquid extractor column followed by a series of mixer/settlers.

Rich DIPA solution from the bottom of the absorber and extractor columns is regenerated by reboiler-generated steam stripping acid gases from the solution.

The major impact of the revamp design basis on the ADIP-D unit is increased DEO and propane rates and significantly higher levels of all sulfur contaminants, particularly H2S and COS.

To minimize the physical impact of these changes, the project maintained essentially the same solvent flow rate by approximately doubling the solvent concentration, allowing retention of much of the existing piping.

The DEO absorber column, its feed-gas knockout drum, and the propane extractor column are replaced with new, larger items to accommodate the increased flows.

The multi-stage mixer/settler system has been reinstated. Shortly after the start-up of the original plant, the mixers were disconnected following mechanical seal leakage.

These mixers have now been refurbished with a double seal system, using a continuous supply of lean DIPA seal solution. Deep COS removal from the propane would be impossible without this system.

First stage DIPA separation from the propane is a settler, which has been replaced with a larger vessel providing increased residence time.

Second-stage separation is a water-wash system followed by an activated carbon bed for polishing. The prerevamp, single-stage wash system has been replaced by a new two-stage counter-current system to reduce fresh water consumption.

The wash system re-uses the former settler vessel and the former coalescer vessel, both refitted with new internals.

The regeneration system treats the same volume of solvent, though at a higher concentration. The regenerator column has been refitted with new higher capacity trays of a proprietary design to accommodate the increased stripping-steam load.

Solvent filters were also relocated to a higher temperature point in the process, to mitigate the effects of increased viscosity at higher solvent concentration, and changes were required to the filter design to allow operation at higher temperature.

Molecular-sieve unit

Fig. 5 [102,161 bytes] shows a simplified process scheme and modifications made for the debottlenecking project. The unit treats three separate product streams:

• Propane pretreated in the ADIP-D unit requires further sulfur removal (traces of COS and mercaptans) and dehydration.
• Butane from the fractionation unit requires sulfur removal (mercaptans).
• DEO gas from the ADIP-D unit requires dehydration.

Each duty consists of a pair of beds, one absorbing, the other on regeneration.

The regeneration system uses dry DEO, with a furnace to provide the energy for the heating cycles and a cooler and knockout drum to recover spent regeneration gas before it is returned to the fuel-gas system.

The major impact of the revamp design basis on the molecular sieve is higher product rates and significantly higher levels of all sulfur contaminants, particularly mercaptans.

The existing molecular-sieve beds were significantly undersized for the new loads, necessitating new equipment throughout the system.

Alternative sulfur-removal technologies for the gas and liquid streams were considered, but the preferred revamp solution for this project was to maintain the existing configuration, replacing the molecular-sieve beds with new larger vessels, and rebuilding piping, control, and relief systems to accommodate higher loads.

Cost saving derived from reuse of the existing butane molecular-sieve vessels as the new DEO vessels, with modification of internals, refitting external insulation, and providing new molecular sieve.

This required careful stress calculations to analyze the impact of cyclic service on the viability of reusing these vessels at new conditions. Operating experience of these vessels reported problems with under-lagging corrosion from condensation; this is not a problem in the new service, which operated at greater than ambient temperature.

The new liquid molecular-sieve vessels are large, with 11-m high molecular-sieve beds. These required careful design to ensure safe operation in cyclic service.

Gasco required a simple depressurizing system to depressurize the treaters within 15 min. The design team had to find a solution that would achieve this requirement with liquid-full vessels, so that that if the same system were operated in a gas phase of the cycle, depressurizing velocities would not cause bed lifting.

The regeneration gas heater was already operating at design capacity. Replacement or revamp of the furnace was avoided, however, by specifying cycle times to prevent the need for simultaneous heating of two liquid beds.

Refrigeration unit

Fig. 6 [95,083 bytes] shows a simplified diagram of the system.

A closed-loop propane refrigeration cycle, with three pressure levels, is installed in each process train. This provides cooling for the de-ethanizer overhead condenser, rundown chilling of the three liquid products to storage, and condensation of boil-off gas from propane and butane storage.

The impact of the revamp project on this unit is increased propane and butane rundown rates and increased boil off gas from new storage tanks in the off-sites area.

The refrigeration unit was already operating at the limit of power available from the gas turbine drivers prerevamp. It was clear, therefore, that some modification would be necessary.

Options evaluated included the following:

• Add "helper motor" to gas turbine to increase power available to compressor
• Add booster compressor to existing machine to increase refrigeration capacity
• Reconfigure refrigeration system with new economizer to improve cycle efficiency
• Provide new, parallel closed-loop refrigeration system for some process duties and break these out of the existing cycle.

The preferred solution following design reviews and optimization was to provide a standalone refrigeration unit in the off-sites area to condense all the boil-off from existing and new storage tanks. This proved to be an effective solution, meeting the requirements of the revamp project.

The existing boil-off condensers operated at the two lowest pressure levels of the propane cycle. Unloading this end of the compressor achieved considerable gains in refrigeration available for the remaining process.

Providing a standalone package in the off-sites area reduced boil-off gas piping, heat leak, and compression needs. And significantly, this solution had no impact on shutdown duration and reduced revamp work in the live process plant.

Construction, precommissioning

The construction phase of this revamp project presented significant challenges. In particular, work was carried out within an operating plant, often in areas not completely gas free.

Many construction areas involved complex piping and structures, particularly where existing equipment was reused in new services. A large workforce was required, new to the site and new to Gasco regulations and standards.

The construction schedule was tight, and overall project economics were very sensitive to the timing and duration of the two shutdowns.

Precommissioning was complicated by working in a live plant and the schedule-driven requirement for activities to be carried out in parallel with construction work.

The two process trains were completed sequentially. Train 2 was completed considerably faster than Train 1 by implementation of lessons learned by the Kellogg and Gasco teams.

Success factors at this stage included the following:

• Planning-Rigorous advance planning was required to ensure adequate manning levels.
• Clear scope of work-A clear split of responsibilities between contractor and operations allowed the precommissioning team to move forward on a broad front and plan for concurrent work to accelerate progress.
• Relationship between construction and precommissioning teams-It was critical that each team have a full understanding of the others' work, to plan and prioritize work, with the precommissioning team seen as the "client" for construction output.
• Relationship between operations, precommissioning teams-Demarcation of responsibilities ensured precommissioning work progressed smoothly. For example, such activities as "spading" (or blinding: ensuring positive isolation at a flange in a piping segment) and despading required operating personnel's experience of the plant.
• Line blowing, drying, leak testing, and inerting-Gasco procedures required method statements to be approved for all precommissioning activities. The lesson learned here was the importance of agreeing to the method-statement format and handling procedure before starting work.
• Spading-Time was lost during the Train 1 shutdown because of a lack of resources. For example, insufficient scaffolding hampered progress, and the number of bolts in poor condition was underestimated.

Time was recovered during the Train 2 shutdown by implementing lessons learned; the equivalent spading of Train 2 was accomplished in one-third the time.

Construction and operation were considered at very early stages of the design process. The experience of the Train 1 shutdown showed that early involvement of the precommissioning team in the project is also advantageous. This project benefited by implementing lessons immediately on a second train.

Performance since start-up

The process trains were started up in December 1995 and March 1996, respectively. Formal test runs were carried out in October 1996 attended by representatives from the Gasco project team, Gasco operations, Kellogg, SIOP, and molecular-sieve vendor UOP.

Test runs were carried out with steady operation at 105% of the Phase 1 design feed rate. This overcapacity was achievable because of the time of year; lower-than-design seawater temperatures enable stable depropanizer operation at higher-than-rated flow. This is entirely consistent with process simulation predictions.

The test runs were completed after more than 3 days, enabling the test-run team to observe a full cycle of molecular-sieve bed regeneration. Stable production rates were maintained without interruption for the entire test period, with product quality within all specification requirements.

Operation since the revamp project has been stable and satisfactory. The sulfur contamination of the plant feed is currently below design capacity of the ADIP-D and molecular-sieve treating units. The molecular sieves are thus able to operate over a longer cycle time than their design, which is likely to extend bed life.

The sulfur-removal capacity of the molecular-sieve beds has been estimated by taking regular samples during an extended on-line period and extrapolating analysis results. This analysis showed the beds are operating as required.

Hydraulic checks showed that the piping design is satisfactory, and the sequence control for bed depressurization, pressurization, draining, and filling works well and within the required time periods.

The Authors