I.CHEM.E. SYMPOSIUM SERIES NO. 110 and loss... · I.CHEM.E. SYMPOSIUM SERIES NO. 110 ... where apparently so-called Boiling Liquid Expanding Vapor Explosions ... knock-out drums,

I.CHEM.E. SYMPOSIUM SERIES NO. 110

THE NEXT STEP IN ACCOMMODATING MAJOR CHEMICAL ACCIDENTS

Hans K. Fauske*

Physically based passive features are emphasized to prevent major explosions and to mitigate and/or contain non-explosive releases. Examples include possible means for BLEVE elimination, control of chemical reactors without direct venting, elimination of dense cloud forma­tion in connection with highly toxic liquid releases such as NH-, HF, etc., and quenching of short duration but very large two-phase releases in connection with venting of chemical reactors. The next step in accommodating major chemical accidents should include considerations of such features and would be an effective way to further decrease the potential for Seveso, Bhopal, and Mexico City type disasters.

INTRODUCTION

Infrequent, but major chemical accidents can be attributed to explosive, (i.e., catastrophic vessel failures) as well as non-explosive, (i.e., dis­charge through relief systems, pipe breaks, etc.) releases of toxic and flammable materials. Explosive releases can result from overpressure or overtemperature. The frequency of loss of pressure control as a result of inadequate vent size designs can be significantly reduced by adopting the methodology recently completed by the AIChE DIERS program [1]. However, for many gassy systems where the reaction is not cooled by latent heat of vaporization, adequately large vents may not be practicable, and for systems where large and costly effluent treatment systems are required due to large, short duration, two-phase discharges, the Inherently Safe Chemical Reactor (ISCR) concept introduced below may provide an alternative practical approach to pressure control. An example of overtemperature is the recent Mexico City disaster (1985) where apparently so-called Boiling Liquid Expanding Vapor Explosions (BLEVE) played a major role. A possible passive means for BLEVE elimination has been suggested recently [2].

Examples of non-explosive releases include the Seveso incident in Italy in 1976, in which dioxins were discharged into the air through the Emergency Relief System (ERS), the 1981 Farmland Industries release to the atmosphere of 40 tons of ammonia through the ERS on a storage tank, the Bhopal incident in 1984 which led to a large release of methyl isocyanate due to a largely inef­fective scrubbing system, and the recent accidental breach of piping to a 40,000 gallon hydrofluoric acid tank in Texas City resulting in a large toxic dense vapor cloud (October, 1987). The principal lessons learned from these incidents can be simply stated as follows.

*Fauske & Associates, Inc. 16W070 West 83rd Street, Burr Ridge, IL 60521.

185

I.CHEM.E. SYMPOSIUM SERIES NO. 110

0 Major accidents involving toxic materials will occur, and

G unlike the above noted incidents, they must be mitigated and/or contained.

Clearly, prevention of accidents must receive high priority, but preven­tion alone is not enough - major accidents do happen. It is for these infrequent, high consequence events that accommodation must be provided. In addition to the passive explosion preventive measures noted above, this paper will discuss means for largely eliminating dense cloud formation in connection with highly toxic liquid releases such as NH3, HF, etc., and quenching of short duration but very large two-phase releases in connection with venting of process vessels. The next step in preventing and accommodating major chemical accidents should include considerations of such features and would be the most effective way to further decrease the potential for Seveso, Bhopal, and Mexico City type disasters.

PREVENTION OF EXPLOSIVE RELEASES

Major explosive releases can result from overpressure as well as overtempera-ture.

Overpressure

The frequency of overpressure scenarios leading to explosions in connec­tion with runaway reaction and/or fire exposure to chemicals in process and storage vessels can be significantly reduced by adopting the methodology completed by the AIChE DIERS program. For the first time, runaway reactions in the DIERS bench scale apparatus can approximate the severity of those in industrial vessels [3] and simple analytical methods accounting for two-phase flow allow vent sizing directly from such data [4]. The simplified models together with the bench scale equipment is now commercially available under the tradename VSP (Vent Sizing Package). The VSP methodology has recently been extended to include the effect of vapor disengagement and laminar flash­ing flow conditions [2].

Application of state-of-the-art (DIERS) vent sizing methodology, however, may in some cases lead to requirements for impractically large vents and/or downstream effluent control equipment. This is particularly true for many gassy systems where there would be little or no "tempering", (i.e., little or no cooling by latent heat of vaporization) when the emergency relief system is activated. These systems may be subjected to rapid decomposition with very high gas generation rates when the materials reach moderately high tempera­tures. The Inherently Safe Chemical Reactor (ISCR) concept has the potential to fulfill the requirement for passive safety for such systems by Brresting the runaway reaction within the process vessel at a relatively low tempera­ture.

The ISCR system illustrated in Figure 1 consists of an immersed coil (ISCR Coil) attached to the gravity driven emergency heat sink. The heat sink fluid on the tube side of the ISCR coil is not the same as the reactor fluid. During normal reactor operation, there is no flow in the ISCR coil and hence, no effect on the normal energy balance on the reactor. In case of an upset,

186

I.CHEM.E. SYMPOSIUM SERIES NO. 110

the flow of heat sink fluid is activated by rupture of the ISCR rupture disk by rising vapor pressure of the heat sink fluid. The rupture disk burst pressure is set to limit the reactor temperature to a preselected limit. This temperature limit is selected at some level higher than the normal process temperature. Upon activation, the ISCR coil becomes a boiling heat pipe which extracts more energy than is being generated by the chemical reaction and leads to a turn-around in the reactor temperature thus preventing uncontrolled runaway condition. The system is activated passively and does not require the agitator or other active devices to be in operation. Performance data ob­tained to date suggest practical operation is feasible for batch type reactors [5]. With the highly reliable passive design of the ISCR and the benign consequences of ISCR activation, the concept would appear to provide a practi­cal and cost effective means for controlling runaway accidents without providing large relief devices, piping systems, knock-out drums, scrubbers, flares, etc.

Overteaperature

Explosions of high pressure storage vessels occur frequently due to overtemperature conditions resulting from extended fire exposure. The worst case to date was the Mexico City disaster. These events may occur under conditions shown in Figure 2. The failure comes about from unwetted parts of the vessel becoming hot and lose strength. As a result the vessel may burst and produce a BLEVE, even though it is below its design pressure [6].

A remedy to the BLEVE problem is illustrated in Figure 3, where baffles are constructed inside the normal walls of a storage vessel providing flow channels between the baffles and the external wall. A fire or other external heat source causes boiling of the liquid near the wall and the resulting vapor will "pump" liquid throughout the flow channel. This inherently passive process provides cooling of the wall, preventing wall overheating, and thus preventing a BLEVE. Construction of the proposed BLEVE Eliminator would appear to be practical, requiring simple stand-off support from the vessel wall.

MITIGATION AND CONTAINMENT OF NON-EXPLOSIVE RELEASF.S

The growing realization that most emergency releases involve high momentum two-phase discharges even in cases of non-explosive releases, has led to increased interest in mitigating and/or containing such releases. In case of venting runaway chemical reactors it is no longer acceptable to dispose of the exhaust outside the building, or onto the floor or onto the roof. Furthermore, it is important to recognize that many emergency releases such as the Farmland Industries release of ammonia and the recent Texas City release of hydrofluoric acid happen too fast to allow completely effective evacuation. Examples of passive means for mitigating and containing such releases are discussed below.

Dense Vapor Cloud Mitigation

High momentum two-phase discharges of materials such as NH3, HF, etc. in the absence of mitigation can transform into dense vapor clouds. The complete

187

I.CHEM.E. SYMPOSIUM SERIES NO. 110

vaporization of the liquid released as well as cooling of the vapor to tem­perature well below its boiling point resulting in heavier than air behavior, can be related to the inherent entrainment process following the initial depressurization of the high momentum jet [7]. The various stages of interest during the release of high momentum two-phase jet releases are illustrated in Figure 4.

The length of the depressurization region (where air entrainment is absent) is typically of the order of several leak (hole) diameters. In this region, only adiabatic flashing occurs, with the major portion of the release remaining in the liquid state. However, due to aerodymamic and nucleation breakup effects, the liquid exits the depressurization region in the form of small droplets of the size ranging from 10 to 100 n, i.e., the droplets can remain airborne due to the high momentum of such jets. These droplets are subsequently completely vaporized in the entrainment region due to the inflow of warm air. It is in this region that the character of the release is changed to that of a dense cloud due to the tremendous cooling effect as­sociated with the evaporation process.

The physical understanding of such jet behavior suggests that significant mitigation of such jet releases would appear possible by preventing the onset of the entrainment region. On the other hand, mitigation based on providing active water curtains in the entrainment zone is at best only partially effec­tive [8]. It follows that consequences of such releases can perhaps be drastically reduced by stagnating and separating the flashing fluid prior to the onset of significant entrainment as illustrated in Figure 5. By these means the majority of the release can rain out while the remaining vapor can escape and quickly disperse on the basis of remaining buoyant relative to the ambient air. The effectiveness of this approach has been demonstrated by stagnating sonic low quality flashing steam-water mixtures [9]. Both foamy and non-foamy steam-water mixtures were evaluated for their liquid rainout effectiveness. Following impact on the stagnation shield plate, rainout of the liquid droplets consistent with the adiabatic flash fraction were noted for both non-foamy and foamy mixtures.

Containment of Two-Phase Discharges

As briefly described above, the DIERS program has provided methodology for both system characterization and sizing vents for accommodating two-phase flows in connection with runaway chemical reactions. The DIERS activity, however, did not address the equally important aspects of what to do with a reacting, two-phase eaergency release.

While many chemical companies are now actively employing the DIERS methodology, they are increasingly faced with the problem of what to do with the emergency release. Assessment suggests that traditional relief system trains involving components such as separators, scrubbers, flare systems, etc. are not cost effective and that large two-phase emergency releases are better accommodated by largely utilizing a single-step design approach based upon quenching [10]. Examples of such passive designs are illustrated in Figures 6 and 7 for gassy and vapor systems, respectively.

In the case of gassy systems, (i.e., the presence of noneondensable gases), separation of liquid and gases is desirable prior to quenching of the still reacting liquid mass. A possible passive means for assuring effective

188

I.CHEM.E. SYMPOSIUM SERIES NO. 110

quenching of the liquid reactants is illustrated in Figure 6. This approach provides a passive means of mixing the poison material in the separated liquid mass. When the pressure reaches PRD1 , the rupture disk (RD,) on the small

line to the poison vessel will burst, pressurizing the poison vessel. The pressure in the reactor will continue to rise. When the pressure reaches P R n

and PRD3_ (which are approximately equal), rupture disks RD2 and RD3 will

burst. The vented material will be a two-phase mixture of gas and liquid, which is subsequently separated in the separator. Simultaneously, the poison material will flow from the poison vessel, through the passive poison arm into the separator. The vigorous jetting of poison into the separator should provide excellent passive mixing of poison and reactants, thus assuring that the reaction is terminated.

For pure vapor systems, the two-phase discharge can be vented directly into the quench liquid as illustrated in Figure 7. Based upon analysis and data of vapor jet condensation in subcooled liquids, the condensation effec­tiveness for low quality, low viscosity liquids is projected to be excellent, i.e., complete condensation is projected to occur over relatively short dis­tances for low quality jet releases with an appropriately designed quencher arm [9]. The condensing effectiveness in connection with low quality, high viscous flashing liquids may be influenced by potential plugging problems in the quencher. For such systems small-scale laboratory experiments simulating a single opening in the quencher arm design are recommended for each specific application.

CONCLUDING REMARKS

Several passive features have been illustrated for preventing explosions as well as mitigating and containing non-explosive releases. Consideration of these features should constitute an important part of the next step to accom­modate major chemical accidents in order to further decrease the potential for severe consequences like the Seveso, Bhopal and the Mexico City disasters.

REFERENCES

1. Fauske, H. K., "Relief System Design for Runaway Chemical Reactions", Proc. Int. Symp. on Preventing Major Chemical Accidents, February 3-5, 1987, Washington, D.C.

2. Fauske, H. K., "Emergency Relief System Design for Reactive and Non-Reactive Systems: Extension of the DIERS Methodology", paper presented at the AIChE 1987 Summer National Meeting, August 16-19, 1987, Minneapolis, Minnesota, also to appear in Plant/Operations Progress.

3. Fauske, H. K. and Leung, J. C., "New Experimental Technique for Characterizing Runaway Chemical Reactions", Chem. Eng. Prog., 81, No. 8, (August, 1985).

4. Leung, J. C. and Fauske, H. K., "Runaway System Characterization and Vent Sizing Based on DIERS Methodology", Plant/Operations Progress, Vol. 8, No. 2, (April, 1987).

189

I.CHEM.E. SYMPOSIUM SERIES NO. 110

5. Grolmes, M. A. and de Leon, M., "Experiments on the Performance of aPrototype Heat Exchanger Device for the Mitigation of Runaway ChemicalReactions", Fauske & Associates, Inc. Report No. FAI/87-3, (February,1987).

6. Lees, F. P., Loss Prevention in the Process Industries, Vol. 1,Butterworth & Company, (1980).

7. Fauske, H. K. and Epstein M., "Source Term Considerations in Connectionwith Chemical Accidents and Vapor Cloud Modeling", Proc. Int. Conf. onVapor Cloud Modeling, November 2-4, 1987, Cambridge, Massachusetts.

8. Blewitt, D. N., et al., "Effectiveness of Water Sprays on MitigatingAnhydrous Hydrofluoric Acid Releases", Int. Conf. on Vapor CloudModeling, November 2-4, 1987, Cambridge, Massachusetts.

9. Fauske & Associates, Inc., "Prevention, Mitigation and Containment ofChemical Systems", FAI/87-100, (December, 1987).

10. Fauske, H. K., "Disposal of Two-Phase Emergency Releases", 4th Int. Symp.on Multi-Phase Transport and Particulate Phenomena, December 15-17, 1986,Miami Beach, Florida.

190

I.CHEM.E. SYMPOSIUM SERIES NO. 110

191

I.CHEM.E. SYMPOSIUM SERIES NO. 110

192

I.CHEM.E. SYMPOSIUM SERIES NO. 110

193