Alternative Alkylation Technologies in a Refinery - A Case Study in Risk-Based Decision Making

Geoffrey Kaiser, Ph.D., SAIC, Reston, Virginia, USA
Steven T. Maher, PE CSP, Risk Management Profesionals, Mission Viejo, California, USA

1) Introduction

        Alkylation processes in refineries are a key element in the production of valuable, high octane gasoline blending components that allow refiners to meet the stringent requirements of the United States' Clean Air Act. The use of anhydrous hydrogen fluoride (AHF) as a catalyst in these alkylation processes has become very controversial. For example, in the Clean Air Act Amendments of 1990, Congress required the Environmental Protection Agency (EPA) to undertake a study of the hazards associated with the use of AHF [1]. There has been considerable interest in developing alternative technologies, including some that seek to make the use of AHF less risky by mixing it with an additive that reduces the effective vapor pressure of the AHF, thereby lessening the potential severity of any accidental release.

        The purpose of this paper is to discuss a Quantitative Risk Assessment (QRA) that was used in the decision making process for a particular mixture of additive and AHF - Modified HF or MHF - that has been chosen as a catalyst for use in the alkylation unit at Mobil's Torrance, California refinery. In 1987, there was a widely publicized accidental release of AHF at the refinery. For this and other reasons, in 1989 the City of Torrance filed a lawsuit against Mobil. Subsequently, the two parties entered into a Consent Decree [2a] which was filed with the Superior Court of the State of California for the County of Los Angeles. The Consent Decree established a Safety Advisor (SA) with the authority to investigate and make recommendations on a wide range of safety-related issues, including the use of AHF. Section 4 of the Consent Decree permitted Mobil to "commit to a modified HF (MHF) catalyst by December 31, 1994 only if it has demonstrated to the satisfaction of the Safety Advisor that the catalyst as modified would not form an aerosol or dense vapor cloud upon release." A subsequent Stipulation and Order was filed in September 1994 [2b] which allowed Mobil to use risk-based criteria to demonstrate that "the modified HF catalyst (including mitigation) presents no greater risk than a sulfuric acid alkylation plant producing a comparable amount of alkylate."

        During 1994, Mobil performed a comparative, quantitative risk assessment (QRA) of MHF and sulfuric acid alkylation technologies. The Safety Advisor's team (which, for this task, consisted of the two authors of this paper) reviewed the risk assessment and concluded in a report to the Court [3] that Mobil had satisfactorily demonstrated that "the modified HF catalyst (including mitigation) presents no greater risk than a sulfuric acid plant producing a comparable amount of alkylate." The Superior Court subsequently approved the SA's evaluation [2c]. The purpose of this paper is to summarize the Comparative QRA and associated phenomenological evaluations from the perspective of the SA.

2) Key Evaluation Criteria and Issues

        The QRA and all contributing information was reviewed by the SA. The specific areas reviewed included (1) accident scenario identification; (2) accident scenario frequency quantification; (3) consequence analysis; (4) risk assembly; (5) uncertainty characterization and simplifying assumptions; (6) transportation and regeneration risks; (7) comparison with other risk assessments and (8) emergency response. Some of the key issues are summarized below. In addition, the SA's team visited an MHF demonstration unit in Paulsboro, NJ. The unit had been successfully operated for nearly a year.

2.1) The Risk Assessment

        Mobil performed a QRA of an assumed 22,000 BBL/day MHF alkylation unit and a representative 20,600 BBL/day sulfuric acid unit at the Torrance refinery. Figure 1 shows the comparative societal risks as presented in the form of an F-N line (or Complementary Cumulative Distribution Function, CCDF) [3]. The horizontal axis shows the number of people (N) predicted to be exposed above the ERPG-3 (the Emergency Response Planning Guideline, published by the American Industrial Hygiene Association). The vertical axis shows the predicted frequency F with which N will be equalled or exceeded.

        On Figures 1 and 2, the estimate of 4% for the amount of sulfuric acid left airborne after an accidental release of a mixture of sulfuric acid and hydrocarbons corresponds to the low end of a range (4%-7%) suggested by a local agency, the South Coast Air Quality Management District (SCAQMD) while 2.8% is at the lower end of a range (2.8-3.0%) obtained from experiments on accidental releases of sulfuric acid/hydrocarbon mixtures [4]. The prediction that as much as 2.8% of the initially released sulfuric acid will remain airborne is critical to the results of the Comparative QRA. Sulfuric acid has a very low vapor pressure at typical alkylation unit operating temperatures and would be expected to fall to the ground like water from a spigot, leading to predictions of very small airborne release fractions. Indeed, the experiments show that, for pure sulfuric acid, no sulfuric acid droplets become airborne. That is, there can be no significant source term from sulfuric acid storage and transportation. The experiments show that it is only if sulfuric acid is mixed with hydrocarbons, as is the case in an alkylation unit, that sulfuric acid droplets can become and remain airborne. One possible explanation of this phenomenon is that the presence of surfactants in the sulfuric acid promotes the formation of small bubbles of sulfuric acid surrounding hydrocarbon vapor. The collapse of these bubbles could potentially leave a fine sulfuric acid mist. By contrast, it is known that the same surfactant phenomenon does not exist in HF/hydrocarbon mixtures.<

        Figure 2 shows a comparable result to that of Figure 1, the difference being in the neglect of active mitigation of MHF releases. Active mitigation systems for the MHF alkylation unit consist of water sprays, water deluge, AES (acid evacuation system), pump isolation/shutdown, and firewater application. Figure 2 shows that, even with such an extreme assumption, the predicted MHF risks are still comparable to the predicted sulfuric acid risks. Credit for the success of mitigation systems and emergency response is highly dependent on scenario and response timing, emergency preparedness and many other issues of human reliability. In addition to the results shown on Figures 1 and 2, the SA examined the scenarios that are dominant contributors to MHF risk and performed sensitivity studies on human error probabilities, leading to an MHF F-N line that lies about halfway between the MHF F-N lines on Figures 1 and 2. This result is supportive of the SA's conclusions (see Section 3) and proved adequate to address both the legal and technical issues of the consent decree.

        The SA also considered various uncertainties and simplifying assumptions. These included but are not limited to (1) exclusion of transportation risks; (2) exclusion of sulfuric acid regeneration; (3) use of generic equipment failure rates; (4) mitigation systems and strategies in sulfuric acid alkylation units; (5) ignition probabilities for scenarios that include releases of hydrocarbons and (6) various issues associated with dispersion modeling and HF and sulfuric acid toxicology. These issues were addressed either by showing that they favor sulfuric acid alkylation, or that they make little difference, or by performing sensitivity studies.

2.2) Phenomenological Issues

        Experimental work on MHF releases was performed at the Quest Consultant's test site near Norman, Oklahoma [5]. Approximately 150-500 pounds of MHF were released per test through a circular orifice into a specially designed flow chamber. The amount of airborne hydrogen fluoride and the amount of hydrogen fluoride falling downward into test pans to form a liquid pool together with the additive were measured. Test were conducted for a range of compositions, temperatures, pressures and orifice sizes. Reference [6] shows airborne HF reduction percentages (ARPs) in the range 62.8% to nearly 100%, depending upon the conditions of release, compared with an ARP of near zero observed in pure AHF experiments [7]. For operational reasons, Mobil has chosen to design its MHF unit to maintain an ARP of 65%. Because maintaining an ARP of 65% is crucial to the conclusions of the comparative QRA, two important recommendations arose from the SA's study: (1) Mobil should produce operational practices and procedures to ensure that the additive concentration does not fluctuate outside the range required to achieve the desired ARP of 65% and (2) there should be an agreed plan to audit additive concentrations periodically.

        Mobil has also developed a two-phase release model [6] that is intended to provide the initial conditions and release for subsequent runs of atmospheric dispersion models. Reference [6] accurately reproduces the experimental results. Mobil performed comparative dispersion analyses of the consequences of a sulfuric acid or MHF release from an assumed 2" orifice in the bottom of an alkylation unit settler. Within the expected range of uncertainties, these evaluations showed that the consequences of such MHF releases at the nearest fenceline or nearest public road are comparable to the consequences of such sulfuric acid releases.

        The comparative QRA was based on calculations of the number of people who might be affected above the ERPG-3, which is 50 ppmv for HF and 7.5 ppmv for sulfuric acid over a duration of exposure of one hour. It was also assumed that the effect of differing exposure times can be accounted for by assuming Haber's Law, namely that a constant dose produces the same health effect. Mobil performed experiments on rats that verified that, for the HF ERPG-3, Haber's law is valid to a low exposure time of 2 minutes [3].

3) Conclusions

        The results of the SA's evaluation included the following: (1) "For Mobil's Torrance Refinery, both [MHF and sulfuric acid] alkylation processes represent a significant improvement [i.e. reduction in predicted risk] over the current alkylation process utilizing an anhydrous HF catalyst;" (2) "The best estimate risk and phenomenology results clearly identify MHF and sulfuric acid as being of comparable risk, with alkylation using a modified HF catalyst showing lower calculated best estimate risk values;" (3) "The Consent Decree Criteria for not forming an aerosol has been explicitly met" and (4) "The specific criteria of the September, 1994 Stipulation and Order will be explicitly met with a conversion to Mobil's stated MHF technology." It is also pertinent to note that the Comparative Risk Evaluation shows that the risks associated with the operation of either a MHF or a sulfuric acid alkylation unit are very small when compared with other risks such as traffic fatality, structural fire fatality, drowning and fatality due to a fall.

        The primary vehicle for communication of the results was the SA's written evaluation [3]. In addition to being documented in Reference [3], the results of the SA's evaluation have been scrutinized in peer reviews, televised public forums, industry discussions and a Court hearing to adopt the results as acceptable for meeting the requirements of the Consent Decree.

        The QRA summarized above was performed in a situation in which there were a number of parties with an interest in the outcome. The stakeholders included Mobil, with its wish to continue using an HF based catalyst to produce alkylate at the Torrance refinery; the Administering Agency (AA), namely the City of Torrance Fire Department; the people of Torrance and their representatives, the City Council; and the Superior Court of the State of California for the County of Los Angeles. The SA's evaluation of Mobil's QRA was performed in close consultation with both Mobil and the AA. The AA's concerns were fully taken into account and the SA and Mobil ensured that all key issues were addressed in reference [3]. As a result, the AA accepted the SA's analysis and communicated it to the City Council with a recommendation that Mobil be allowed to proceed with MHF alkylation.

        In conclusion, the QRA proved to be a means of defusing disagreements between various stakeholders while preserving Mobil's ability to use HF alkylation and addressing citizen's concerns about safety. This represents an ideal use of QRA in that it formed the framework for a rational dialog that led to a decision that was accepted by all parties.

4) References

1. United States Environmental Protection Agency, Hydrogen Fluoride Study, EPA550-R-93-001, September, 1993.

2. (a) Consent Decree, October 19, 1990; (b)Stipulation and Order, September 30, 1994; (c) Order Approving the Safety Advisor's Evaluation of the Modified Hydrogen Fluoride Alkylation Catalyst: Superior Court of the State of California for the County of Los Angeles, People of the State of California vs Mobil Corporation, Case No. C 719 953.

3. Maher, S.T. and G.D. Kaiser, "Torrance refinery Safety Advisor Project - Evaluation of Modified HF Alkylation Catalyst," 59111.01, EQE International, Irvine, CA, May 1995.

4. Johnson, D.W., Sulfuric Acid Release Report. In; National Petroleum Refiner's Association Conference, San Antonio, TX, March, 1994.

5. Jersey, G.R., et al., Large-Scale Release Testing of a Modified HF Catalyst. In: 1993 AIChE National Summer Meeting, American Institute of Chemical Engineers, New York, August 1993.

6. Muralidhar, R., et al., A Two Phase Model for Subcooled and Superheated Liquid Jets. In: International Conference and Workshop on Modeling and Mitigating the Consequences of Accidental Releases of Hazardous Materials, Mew Orleans, LA, September, 1995, pp 189 -. Center for Chemical Process Safety, New York.

7. Blewitt, D.N., et al., Conduct of Anhydrous Hydrofluoric Acid Spill Experiments. In: International Conference on Vapor Cloud Modeling, Boston, Ma, 1987 pp 1-. Center for Chemical Process Safety, New York.