Shashi Dharmadhikari,In 1995, Eurolysine and Ing?rop worked together to design a safe ammonia-withdrawal system for Eurolysine's ammonia facility in northern France.
Jacques Kuhn Ing?rop
Paris
The new system uses a siphon device to withdraw ammonia from the top of the sphere.
Until 1995, ammonia was withdrawn from the bottom of the sphere and then pumped to the various users. To improve ammonia safety and minimize leaks, the French organization Drire (Direction R?gionale de l'Industrie, de la Recherche et de l'Environnement) recommended that Eurolysine find an alternative to bottom withdrawal of ammonia from the pressurized sphere.
Safety awareness was heightened by the construction of a new highway adjacent to Eurolysine's ammonia facilities.
Wanting to improve safety, Eurolysine invited Ing?rop to propose a technical solution.
The following case study of Eurolysine's modification of the ammonia storage facility shows that liquid withdrawal by a siphon system can operate successfully for pressurized storage. Careful attention, however, must be given during the process design to piping and instrumentation details and to the compressor selection.
The example shows that a compressor with built-in capacity control is necessary for successful operation of the new siphon system.
Eurolysine is a leading French manufacturer of animal-feed ingredients. The company receives liquid ammonia for various processes from pressurized road tankers and stores it in a 500 cu m sphere, which operates at about 4 bar. A refrigeration system condenses the vapor generated during tanker unloading.
Modification options
Table 1 [40,082 bytes] outlines the general specifications of the ammonia-storage facility. This installation has been operating since 1986. A diagram of the ammonia facility before its modification is presented in Fig. 1 [144,490 bytes].The simplest concept for top ammonia withdrawal would have been the installation of submerged transfer pumps inside the sphere, as is generally done for LPG products. Due to the toxic characteristics of ammonia, however, a better solution was needed.
Ing?rop's solution was modifying the existing ammonia sphere to include a siphon system to remove ammonia from its top.
Fortunately, during the initial design of the ammonia sphere a few spare nozzles were included on the top of the sphere.
The main disadvantage of a siphon system is the generation of vapor at points of reduced pressure in the siphon line. Because these vapors may not be vented to atmosphere, they must be otherwise handled. The following schemes were then considered:
- Scheme A-Injection of the compressed vapor back into the sphere (Fig. 2a [199,139 bytes])
- Scheme B-Use of an ejector (Fig. 2b)
- Scheme C-Compression and condensation of the vapor then injection of the subcooled liquid back into the sphere (Fig. 2c)
- Scheme D-Condensation of the vapor using refrigeration (Fig. 2d).
Although Scheme B looked simple and attractive, after consultation with different ejector vendors, it was discovered that there would be a risk of cavitation during transient conditions. This scheme, therefore, was also eliminated.
The remaining two schemes were analyzed in greater detail. Although the equipment costs for both schemes are close, Scheme C provides several more advantages than Scheme D (Table 2 [52,909 bytes]). These advantages are:
- Simpler
- Lower utility consumption and consequently lower operating costs
- More environmentally friendly because no refrigerant is used.
Process design
Most current simulation softwares can handle siphon calculations. Careful attention, however, must be given to the accuracy of the input data, particularly for the difference in elevation between the sphere and the siphon drum; the quantity of vapor generated is very sensitive to both piping elevation and length.The vapor quantity also depends upon:
- Liquid level in the sphere
- Liquid pressure and temperature in the sphere
- Sphere diameter.
The ammonia demand varies from one season to another as well as from different uses from time to time. The whole system must operate satisfactorily under all conditions.
If there is an inadequate amount of liquid ammonia available to condense the compressed vapor and subcool the liquid formed in exchanger E-3, uncondensed ammonia will accumulate in drum V-3. Increasing the pressure in V-3 will eventually result in a release of ammonia vapor to atmosphere. Furthermore, the recycled liquid will flash in the sphere.
The control system must also be able to handle all process scenarios. Manual bypasses around critical equipment are helpful to enhance operating flexibility.
Although the whole system operates above atmospheric pressure, care must be taken to prevent the entrance of air into the system. Ammonia diluted with air needs more compression power, and it leaks easily to atmosphere.
Equipment arrangement
For successful operation of the siphon and liquid injection system, the arrangement of equipment and piping must be done with care.At the modified facility, the siphon drum is installed above the sphere. This arrangement provides following process benefits:
- Good NPSH (net positive suction head) for the transfer pumps
- Better siphon operation (that is, no flashing to vapor)
- Easier piping installation
- Lower pressure losses.
The vapor/cooling water exchanger E-2 was installed above condenser E-3, and the liquid ammonia is subcooled in the condenser. The subcooled liquid flows by gravity to liquid drum V-3, and is finally injected back into the sphere, controlled by the level in V-3. As this liquid is subcooled, there is no flashing in the sphere.
For secure operations, two compressors (C-1A and C-1B), each with 100% operating capacity, were installed. These compressors are skid-mounted Corken HG 600, double-acting twine cylinder compressors, equipped with 6-in. diameter, nonlubricated pistons.
The compressor suction drum V-2 is equipped with a heating coil to vaporize any accumulated liquid. Furthermore, the compressor suction line is heat traced.
Plant start-up
The modified ammonia storage facility started up in September 1995. Initially, a stable siphoning operation was difficult to maintain; the siphon drum was filling too rapidly and causing frequent plant shutdowns.The compressor operation also was unstable; the machine was frequently shut down as a result of high discharge pressure.
Both these problems persisted despite various attempts to modify control set points. Finally, after detailed process analyses, the facility was shut down and the following revisions were carried out:
- Installation of a new control valve upstream of the siphon drum. Initially, the level of the siphon drum (V-1) was regulated by a control valve on the vapor line leaving the drum. The control system caused instability in the siphon system and allowed filling of the drum too rapidly. When this control valve was moved upstream of the drum, it considerably improved siphon operation.
- Addition of a minimum flow line for the compressors. The installed Corken compressors do not have built-in capacity control systems. During start-up, the compressor was frequently shut down as a result of low suction pressure, which was caused by low amounts of generated gases. A new external bypass line maintained the suction pressure at a preset value.
- Modification of the alarm control system for the compressor discharge. During start-up, the compressor shut down as a result of a high pressure alarm installed on the discharge line. Process investigations showed that there was an instrumentation problem and the logged discharge pressures were never too high. Two modifications remedied the problems: a control stabilizer was added on the high-pressure switch transmitter, and the set point of the high pressure was increased (remained below the pipe rating).
The facility was restarted in October 1995 and has been operating successfully ever since. It was found that the siphon operation is very sensitive to the operating pressure of the sphere. The liquid level in the sphere also has some influence on the quantity of vapor generated. The vapor quantity generated due to siphoning is nearly constant as the liquid level falls to 50% but then drops sharply (Fig. 3 [101,989 bytes]).
Process simulations carried out using data collected during the start-up period agree with the performance observed, confirming the need for inlet capacity control for the compressor.
Acknowledgment
The authors express their gratitude to the management of Eurolysine for permission to publish this article, and also to Odile Braesco of Eurolysine for her valuable contribution to the success of the project.The Authors
Shashi Dharmadhikari is process manager with Ing?rop, France. Previously, he worked with Technip and Air Products plc. He has over 25 years' experience in oil and gas, LNG, NGL, and energy fields in process activities, including plant operation.Dharmadhikari is a fellow member of the Institution of Chemical Engineers U.K., a member of Ashrae (the American Society of Heating, Refrigerating & Air-Conditioning Engineers), and a member of the International Institute of Refrigeration. He holds a PhD in chemical engineering from the University of Surrey in England.
Jacques Kuhn is technology director with Ing?rop, France, where he is responsible for the interface between basic and detailed engineering.Before joining Ing?rop, he worked with Lummus, Technip, and Sofresid. His prior experience includes 15 years in process design and 10 years in process evaluation, selection, and licensing in petroleum refining, oil and gas production, natural-gas treatment, and petrochemicals.
He is a member of AFTP (Association Fran?aise des Techniciens du P?trole).
Kuhn is a graduate of Ecole Nationale Sup?rieure des Mines de Paris and holds a masters degree in chemical engineering from Yale University.
Copyright 1998 Oil & Gas Journal. All Rights Reserved.