The following is a PDF Digital Document Re-creation of the Original Report that was Filed with the Superior Court of the State of California for the County of Los Angeles - Some Variations in Page Numbers or Format May Occur Mobil TORRANCE REFINERY SAFETY ADVISOR PROJECT EVALUATION OF MODIFIED HF ALKYLATION CATALYST (ANALYSIS OF PROPOSED ADDITIVE CONCENTRATION CHANGES) Final Report, Rev. 0 October 1999 (In Support of Consent Decree Section 4) Primary Contributors: Steven T. Maher, PE CSP, Risk Management Professionals, Inc. Dr. Geoffrey D. Kaiser, Science Applications International Corp. Dr. Mardy Kazarians, Kazarians & Associates Approved: Steven T. Maher, PE CSP Process Safety Risk Management Professionals, Inc. 25108-B Marguerite Parkway, #156 Mission Viejo, California 92692 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst CONFIDENTIALITY STATEMENT Some information provided by Mobil Oil Corporation, which formed the basis for this evaluation by the Safety Advisor, contained confidential, proprietary information that should not be made available or disclosed to or discussed with anyone outside Mobil Oil Corporation, its divisions, and wholly owned affiliates nor used for any purpose except as authorized by Mobil Oil Corporation under duly executed agreements permitting such disclosure or use. In the same manner, this Safety Advisor evaluation may contain potentially commercially sensitive information that should be considered confidential and proprietary to Mobil Oil Corporation. The publicly available copy of the document will contain { } which delineate stricken confidential information. Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst EXECUTIVE SUMMARY Background, Objectives, and Key Activities As a result of the settlement of a public-nuisance lawsuit filed by the City of Torrance against Mobil Oil Corporation in 1989, a Consent Decree (Ref. 1) was entered into between the two Parties and was filed with the Superior Court of the State of California for the County of Los Angeles. One of the primary objectives of the Consent Decree was the establishment of a Safety Advisor (SA) for the Mobil Oil Corporation Refinery in Torrance, California. Among other responsibilities, the Safety Advisor has the authority to investigate and make recommendations on a wide range of safety-related issues for the refinery. Section 4 of the Consent Decree permits Mobil "to commit to a modified HF 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 (Reference 2) which allows the SA, as an alternative, to use risk criteria as a facet of its review, i.e., Mobil may demonstrate that "the modified HF catalyst (including mitigation) presents no greater risk than a sulfuric acid alkylation plant producing a comparable amount of alkylate." The Consent Decree and subsequent Stipulation and Order provide the objective, as well as the authority, for the Safety Advisor to evaluate and ascertain the basis and validity of both the phenomenological and quantitative risk comparisons provided by Mobil. Reference 4 summarizes the SA evaluations of Mobil's 1994 Quantitative Risk Assessment (QRA) and the unit design. Following this research, design, analysis, and approval phase; detailed design work and field implementation began. Following modification of the Alkylation Unit to accommodate the MHF process (Fall 1997), operation at the Court-approved target HF wt% concentration (see Table ES.1), although satisfying Consent Decree requirements, resulted in unit operational instabilities. This provided the operations staff with a significant challenge, as well as compromising product quality and preventing unit operation at full production capacity. Supporting the undesirability of this operating condition is the general truism that stable refinery unit operation is the safest operating mode. Additional background information regarding operational and safety issues, which stimulated a need to modify the Alkylation Unit design and operating parameters is provided in the March 12, \TorMHF.doc i Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst 1998 submittal from Mobil to the City and to the Safety Advisor (Ref. 5). Alkylation Unit operational stability issues prompted the innovative reconsideration of the balance of additive concentration vs. application of other mitigation features. These other mitigation features were focused on various types of encapsulation mechanisms to decrease the momentum of the stream from a potential accidental release. A favorable property of the additive used in the MHF process is a high affinity for HF, which results in a significant increase in the Airborne Reduction Factor (ARF), i.e., a decrease in the fraction of HF released to the atmosphere and remaining airborne, if the initial momentum of the release is decreased and aeration of the stream is minimized. Thus, the application of barriers at more credible release locations (or those of potential high risk contribution) can significantly increase the HF ARF at those locations. It is clear, from the analysis presented in this report, that the installation of these encapsulation mechanisms more than compensates for a decrease in additive concentration and has the potential for resulting in improved Alkylation Unit stability along with the same or a net decrease in risk to the Torrance Community. In January 1998, Mobil identified its desired target additive concentration decrease, and on February 1, Mobil began to provide an updated QRA for the purpose of justifying operation under this new configuration (final): • Increased HF wt% in the alkylation catalyst (see Table ES.1) • Installation of Flange Shrouds (polymer shields surrounding flanges in portions of the Alkylation Unit process containing significant concentrations of MHF catalyst) • Installation of Settler Pans (metal shields around the portions of the Acid Settlers containing significant concentrations of MHF catalyst) • Installation of Acid Circulation Pump Barriers (complete metal barriers around the Acid Circulation Pumps) Since the engineering and installation of all these mitigation features could not be completed in a timely fashion, an interim operation mode was proposed that involved a target HF wt% concentration (greater than the Court-approved target, but less than the desired final target), along with the installation of the following mitigation features: • Settler Pans • Acid Circulation Pump Shrouds (polymer shields around the Acid Circulation Pump seals) Both the interim and final operating modes are the subject of this Safety Advisor evaluation. \TorMHF.doc ii Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst The following provides a brief timeline of some of the key events following the initial February 1, 1998 definition of strategy: • February 1 - March 12 - Multiple submittals of preliminary proposals and QRA updates to the SA. Frequent feedback from the SA. Performance of independent justifications and verification calculations by the SA. • March 12 (Ref. 5) - Mobil proposed a final operating configuration, and separately, an interim configuration (see Table ES.1 for the proposed interim HF catalyst concentration). • March 16 (Ref. 843.00) - The Safety Advisor's validation and concurrence with the final and interim operating strategies, and summary list of recommended action items - At that time, the SA had "completed independent verification calculations, which sufficiently characterize the various risk contributors, such that we can definitively state that the risk of operation associated with the" interim "operating configuration, for HF Concentration Targets at 76 weight percent or lower, is equivalent to or less than the risk associated with the current configuration approved in 1995." • March 24 - The Safety Advisor and the Torrance Fire Department inspected the added mitigation features (interim), which provided decisive safety improvements for the Community and offset potential risk increases associated with decreased additive concentration. • March 25 (Ref. 844.00) - The City concurred with the final and interim operating strategies (with necessary follow-up actions specified). • May 15 - Mobil completed the installation of the Acid Circulation Pump Barriers and Alkylation Unit Flange Shrouds. • May 15 - TFD and the Safety Advisor performed a detailed inspection of the added mitigation features (final) that enhanced the margin of safety for the Community. • May 17 - The Safety Advisor identified that Mobil met its Consent Decree obligations (with the enhanced mitigation features) and recommended allowing the Torrance Refinery Alkylation Unit to use an increased target HF concentration (see Table ES.1). Since March 25, when Mobil achieved its interim operating strategy safety objectives and began to increase HF concentration, significant operational enhancements have been achieved, along with improved unit stability and increased alkylate production. The key objectives of the Safety Advisor's evaluation are: • to validate the calculations and conclusions of Mobil's update to the 1994 QRA \TorMHF.doc iii Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst • to correlate any new technical knowledge associated with potential releases from sulfuric acid alkylation units that may impact the risk comparison basis of the 1994 Stipulation and Order • to provide a focused, independent check of Mobil's analysis to ensure accuracy and completeness • to identify any new developments in alternative alkylation unit processes that may obviate any of the Safety Advisor's conclusions from the SA evaluation documented in Ref. 4 • to verify the installed configuration with the parameters and assumptions utilized in the analyses • to identify any potential vulnerabilities that might exist in transitioning to the proposed operating configuration • to recommend any necessary modifications to equipment, procedures, or training to address these potential vulnerabilities Although not specific objectives, as necessary, the Safety Advisor: • provided additional technical justifications and bases • correlated Mobil's results and conclusions with an independent QRA quantification This report is offered to address these objectives and summarize the Safety Advisor's evaluation of the proposed additive concentration changes at the Torrance Refinery (an extension of Task M of the Safety Advisor Project). The key activities involved in support of this evaluation effort included: • Review of the Mobil QRA Update to verify calculation accuracy and the defensibility of methods and assumptions • Determination if the analytical models are sufficiently detailed and appropriate for their intended use, and for characterizing or comparing risk or phenomenological results • Correlation of Mobil's analysis and results to Mobil's 1994 QRA, the SA's previous MHF evaluation activities, and the SA's previous MHF risk analyses • Performance of benchmark calculations to verify the accuracy of the Mobil QRA Update • Performance of SA calculations to ensure that there was a clear correlation between potential contributors to risk and the calculated risk values and to verify the risk assembly process • Providing information, justifications, and bases for select technical issues • Performance of select, independent modeling of specific release scenarios \TorMHF.doc iv Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst • Performance of an independent QRA quantification to correlate Mobil's results and conclusions • Performance of walkdowns and field inspections to verify the correlation of the installation with key analytical parameters • Review of available technical literature • Discussions, interviews, and interaction with Torrance Refinery, Torrance Fire Department, and industry personnel on alkylation unit mitigation system issues • Participation in meetings and teleconferences • Review all information to determine if there is an adequate and appropriate basis for decision-making and demonstrating the acceptability of the application of the proposed additive concentration changes and mitigation system enhancements for the Torrance Refinery Alkylation Unit • Providing specific recommendations to address potential weaknesses or additional requirements The above activities involved Mobil, the Torrance Fire Department, and the Safety Advisor. The Safety Advisor Team included representatives of Risk Management Professionals, Science Applications International Corporation, and Kazarians & Associates. A summary of key observations and conclusions are contained within this Executive Summary, and Table ES.2 contains a summary listing of recommendations and suggestions. Evaluation criteria, methods, observations, and results are contained within the main body of the report. \TorMHF.doc v Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst { } Results and Recommendations The Consent Decree empowers the Safety Advisor to “conduct investigations” and “make recommendations”. Thus, the Safety Advisor's results and conclusions focus on improvements to address any perceived potential vulnerabilities associated with the application of the proposed additive concentration changes and mitigation system enhancements for the Torrance Refinery Alkylation Unit. Specific recommendations and suggestions identified within the report are summarized in Table ES.2, Summary of Specific Recommendations and Suggestions, which provides follow-up actions for Mobil, the Torrance Fire Department, or the Safety Advisor, necessary for closure of this Consent Decree activity. In some cases, these are offered to further refine quality programs, which had already been initiated by the City or Mobil. As appropriate, recommendations are provided for: • specific changes which enable Mobil to meet acceptance criteria • verification items necessary for validating the SA's conclusions • necessary follow-up SA actions \TorMHF.doc vi Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Summary of Key Observations In addition to the improvement areas identified in Table ES.2, the following provides some general insights, perspectives, and observations, which precipitated from the SA’s evaluation: • The recommendations and suggestions summarized in Table ES.2 reflect several areas where important improvements were identified by the Safety Advisor to address specific issues identified within this evaluation. Although considered important and necessary, none of these improvement areas represent “a situation involving the safety of Refinery operations which immediately threatens the environment or the health of on-site or off-site personnel,” subject to the provisions of Paragraph 19 of the Consent Decree (Ref. 1). Please note that the use of the term “on-site or off-site personnel” in the quote from the Consent Decree refers to “on-site workers or off-site individuals.” To support subsequent prioritization and scheduling by the City and Mobil, Table ES.2 identifies priorities for each Safety Advisor recommendation. • The following results/conclusions were derived from the evaluations in the identified subsections: IV.A.1 - Materials Used for the Flange Shrouds ♦ The SA finds the materials of construction of the flange shrouds suitable for their application at the Torrance Refinery Alkylation Unit. ♦ The SA finds the results of the shroud material testing to be credible and consistent with an ability to meet functional requirements. ♦ A seismic event with a potential for resulting in concurrent leakage and possible damage to flange shrouds and a fire with a potential for resulting in concurrent leakage and possible damage to flange shrouds were considered to be negligible contributors to risk. ♦ See Recommendation M-11. ♦ See Recommendation M-12. ♦ See Recommendation M-13. IV.A.2 - Other Flange Shroud Design Issues ♦ The SA finds the following design and construction issues for the flange shrouds, e.g.: Flange Shroud Drain Port Direction Flange Leak Aligned with Drain Port Need for Piping Drain Port Release to Grade Demister Pad Construction and Materials \TorMHF.doc vii Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Flange Shroud Piping Tie-offs Ability of the Demister Pad to Prevent Aerosolization and Decrease Momentum to have been suitably addressed in the final application at the Torrance Refinery Alkylation Unit. ♦ The SA finds the results of the testing of the configuration that included demister pads and stainless steel clamps to be credible and consistent with an ability to meet functional requirements. IV.A.3 - "Unlikely"/Non-Barriered Release Scenarios ♦ For the final configuration, i.e.: Flange Shrouds Settler Pans Acid Circulation Pump Barriers release barriers were installed to address credible release scenarios that represented dominant contributors to risk (in the processing area of the Alkylation Unit). Thus, use of barriered consequence analysis results for these credible accidental release scenarios is appropriate for other future risk-based applications, e.g.: Risk Management Programs and Emergency Preparedness Issues (such as the Community Alert Siren and Tone-Activated Radio Receivers). IV.A.4 - Comparisons with Sulfuric Acid Alkylation ♦ During the course of the SA evaluation, it was verified that no new published (or un-published) information (familiar to the SA) contradicted the sulfuric acid alkylation unit configuration or airborne release fraction data used by the SA in the 1995 evaluation (Ref. 4). IV.A.5 - SRI Contribution of Releases Mitigated by Acid Settler Pans ♦ The SA finds the calculations performed for the QRA Update (Ref. 5) associated with the SRI contribution of releases mitigated by Acid Settler Pans to be accurate. ♦ The Acid Settler Pans and the Acid Circulation Pump Barriers are primarily structural in nature. Thus, maintaining their integrity is analogous to maintaining the integrity of other important structural elements (e.g., steel support structures), which is typically accomplished by Operator visual inspection, rather than a specific Mechanical Integrity Program element. Thus, the inclusion of the Acid Settler Pans and the Acid Circulation Pump Barriers in \TorMHF.doc viii Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst the Torrance Refinery Mechanical Integrity Program was considered, but deemed unnecessary. IV.A.6 - SRI Contribution of Releases Mitigated by Acid Circulation Pump Barriers ♦ Ref. 5 (Pages 8 & 9) assigned “hole sizes of 50mm or less” “to the shrouded inboard section” of the pump. “The remaining hole size of 100mm was assigned to the outboard external casing section of the pump.” To provide a correlation between Acid Circulation Pump release failure modes to their calculated risk contribution, the SA used a technique detailed in Exhibit IV.A.61. The results of this exhibit demonstrated that approximately one-half of the total risk contribution associated with the Acid Circulation Pumps was mitigated by the Acid Circulation Pump Shrouds. Using this approach, it can be shown that a significant amount of the improvements that are achieved by the Acid Circulation Pump Barriers are also achieved by using the Acid Circulation Pump Shrouds. ♦ Based on the results of Exhibit IV.A.6-1, the Ref. 5 approach for identifying the portion of releases might be "mitigated" by pump shrouds appears to yield a conservatively high result (45.37% of breaches subject to barrier mitigation, resulting in approximately a 17% change in the risk contribution). Therefore, the Ref. 5 results associated with the Interim Operating Configuration are very conservative and are acceptable. ♦ It should be noted that the high percentages of the total failure rates calculated to be attributable to failures in the vicinity of moving parts (and possibly mitigated by the pump shrouds) is consistent with an intuitive understanding of potentially dominant pump subfailure modes. IV.A.7 - Flange Failure Contributions to the QRA ♦ Mobil’s commitment to the installation of the piping flange shrouds is commendable, and its exclusion of the risk mitigation credit associated with piping flange shrouds in Ref. 5 is conservative. IV.A.8 - Impeded vs. Unimpeded Release Scenarios ♦ The SA reviewed the basis for the application of “impeded” vs. “unimpeded” release concepts in the 1994 QRA and the application of Airborne Reduction Factors (ARFs) for the 1998 QRA Update, and found the following approach to provide a consistent and accurate basis for correlating the results of the two studies: \TorMHF.doc ix Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Apply the same ratio (of calculated number of exposures) for previously “impeded” or “unimpeded” cases, for those releases that would be mitigated by barriers: { } {} {} Apply the same ratio (of calculated number of exposures) for previously “unimpeded” and “impeded” cases, for those releases that would not be mitigated by barriers. {} {} {} To address a minor conservatism in these calculations, the base value for the calculated numbers of exposures for “unimpeded” cases could be changed to that value for an “impeded” release, for those releases that would be mitigated by barriers, if this analysis is updated in the future. It is not expected that the removal of this conservatism would have a significant impact on the results. IV.A.9 - Use of Un-Weighted Average PIRs for the Mobil Calculations ♦ The SA found the specific averaging methods employed in Ref. 5 to calculate change in SRI (using an un-weighted average PIR) to be difficult to defend. Thus, when comparing the risk tradeoffs associated with changes in acid concentration and barrier application and effectiveness (to the 1994 QRA), the use of a similar discretized approach is necessary. ♦ As an extension of its review efforts, the SA utilized a detailed, discretized QRA approach that was benchmarked to the 1994 QRA to investigate the risk tradeoffs associated with changes in acid concentration and barrier application. IV.A.10 - Geometric Assessment of Changes in Isopleth Size to Population Impact ♦ The results of a theoretical correlation of changes in impact distance to Population Impact Ratio (PIR) show that it does not follow the correlation identified on Page 4 of Ref. 5, i.e., the PIR is not directly proportional to isopleth area ratio (i.e., square of endpoint distance). However, from the geometrical evaluation provided in Exhibit IV.10-1, for small variations (i.e., small changes in \TorMHF.doc x Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst risk, for cases where there is an interest in showing only a small impact on risk), linearity between the isopleth area ratio and PIR may be assumed as an approximation. The linearity improves as the proportion of area outside the fenceline to the total area improves. ♦ Although the assumption made on Page 4 of Ref. 5 regarding the effect of changes in isopleth area ratio to PIR may yield acceptable results, it should be realized that it is acceptable for this application, because the objective of the proposed operational and design changes to the Alkylation Unit is to not significantly change risk (i.e., Ref. 5 and this evaluation is essentially performing a hypothesis to test that the net risk changes for the proposed changes to the process are near zero). Therefore, the results appear to be unaffected by this simplifying assumption and can be supported. ♦ Based on the results of representative dispersion modeling calculations (documented in Exhibit IV.A.10-2), it was shown that the general assumption that the isopleth area is proportional to the release rate for a continuous release is acceptable for use for this updated MHF QRA. ♦ These results will be used for the SA’s discretized QRA to investigate the risk tradeoffs associated with changes in acid concentration and barrier application. IV.A.11 - Assessment of Population Characteristics in the Vicinity of the Torrance Refinery ♦ The Safety Advisor specifically assessed the Torrance population characteristics in the vicinity of the Torrance Refinery. ♦ From a theoretical perspective, a linear proportional extrapolation is valid only when the population density is the same for all points of a hazard footprint. However, based on the above analysis (i.e., given the result of the linear regression), it can be stated that a linear extrapolation applied to this QRA is supportive of accurate conclusions (i.e., the percentage of changes are likely to cancel each other out). It should be noted that this conclusion is based on the assumption that the QRA includes a statistically significant number of release scenarios. Thus, although the effect of changes in release rates to PIR may not be entirely consistent with the Torrance population, the results appear to be unaffected by this simplifying assumption and can be supported. ♦ The population impact associated with small to medium increases in isopleth area does tend to balance out, as long as one is working under the context that there is no net change in risk for the Alkylation Unit. It should be noted that for \TorMHF.doc xi Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst the assessment of operational or design changes that would yield a calculated change in risk, this simplifying assumption would NOT be appropriate to use. IV.A.12 - ARF Extrapolations Beyond the Designated Concentration Range ♦ Figures IV.A.12-1a & IV.A.12-1b illustrate the correlation of Airborne Reduction Factor (ARF) to the wt% HF and the wt% additive within the process. Figure IV.A.12-1b clearly illustrates that significant gains in ARF are achieved at relatively low additive concentrations, and that the effectiveness curve for additional additive flattens out. IV.A.13 - Use of Airborne Reduction Factors in SA QRA Calculations ♦ A discrepancy was identified between the definition of a barriered-case Population Impact Ratio for Ref. 5 for the calculations performed for the 1994 QRA. Barriered-case PIRs used for the SA’s 1998 QRA Update were adjusted to provide a direct comparison of the results of the 1998 QRA Update to the 1994 QRA. ♦ Table IV.A.13-1 contains a summary of the estimated Population Impact Ratio (PIR) values used for the various calculations performed by the Safety Advisor for the QRA Benchmark Quantifications (Section IV.A.14), as well as the discretized QRA validation calculations (Section IV.A.15). IV.A.14 - Results of QRA Benchmark Quantification ♦ The QRA benchmark quantifications provided a relatively accurate indication of the relative changes to risk associated with the following specific options for Alkylation Unit design and operation. These approximations were also used to validate the more detailed check calculations described in Section IV.A.15, and it was observed that these benchmark calculations provide a reasonably good correlation to the more comprehensive discretized QRA described in Section IV.A.15. ♦ The results for the QRA benchmark quantifications (see Table IV.A.14-1) show configurations in bold that would be considered acceptable based on the results of this QRA. It also shows that as long as Acid Circulation Pumps are either completely barriered or shrouded around the seals, the results of the 1998 QRA Update identify a lower risk than that calculated and identified as acceptable for the 1994 QRA. ♦ The results of these QRA benchmark quantifications were suitable (and were used) for addressing acceptability issues associated with the interim operation configuration. \TorMHF.doc xii Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst IV.A.15 - Discretized QRA of the Dominant Contributors ♦ The discretized QRA of the dominant contributors yielded results that could be readily correlated to the 1994 QRA. Table IV.A.15-1 contains a summary of those results. ♦ This discretized SRI calculation (based on the impact of changes in additive concentration and barrier configuration for each specific scenario) yields results that are more accurate, but which correlate well to the calculations performed in Section IV.A.14, thus, validating the acceptability of the approach used by the Safety Advisor for determining the risk impact for a variety of different configurations. ♦ Although Ref. 4 identifies the more appropriate basis for the probabilities to be used for mitigation system availability, the original benchmark cases chosen for the 1994 QRA in Ref. 508.00 were chosen for this 1998 QRA Update. Although the absolute value of this benchmark Societal Risk Index (SRI) of 7.85E-3 was discussed and corrected in Ref. 4 (Sections IV.B.5.e/f and IV.D), this value is acceptable to use for evaluating relative changes to risk for this 1998 QRA Update. IV.A.16 - Walkdown Results and Length of Time for Interim Operation ♦ The findings of the March 24, 1998 walkdown were that the field installation was consistent with the assumptions used as a basis for the Safety Advisor's letter of March 16 (Ref. 843.00). Ref. 843.00 validated the feasibility of the Spring 1998 Proposed Interim HF Concentration. ♦ Consistent with the intent of the March 24 installation to satisfy the needs for an interim operating configuration, the Safety Advisor agreed that the observed installation of the Acid Circulation Pump Shrouds would not be suitable for longterm operation or as a final operating strategy. However, it would certainly be suitable for interim operation (4-8 week time period), with the following additional condition: During this interim operation period, the Acid Circulation Pump Shrouding should be inspected (to ensure that it is secure) by Alkylation Operation Staff twice per week. ♦ This condition was implemented by Torrance Refinery personnel during this interim operating period (March 24 - May 15). ♦ It should be noted that, although the specific calculations performed for the QRA had a definitive scientific/numerical basis that reflected the risk associated \TorMHF.doc xiii Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst with the Alkylation Unit in a specific configuration, the allowable length of time to allow for interim operation was based on a qualitative understanding of potential risks and the SA's inspection of the installation on March 24, 1998. ♦ Based on the May 15, 1998 Walkdown, the installation of the Acid Circulation Pump Barriers and Alkylation Unit Flange Shrouding met or exceeded the bases for the 1998 QRA Update and the Safety Advisor’s calculations (that provide a basis for justifying an increase in acid strength to the May 1998 Final HF Concentration). Therefore, as of May 15, beginning to ramp up to this acid strength was considered appropriate and justifiable. IV.B.1 - Aerosolization ♦ It is acceptable to use Mobil’s two-phase model to provide predictions of ARFs for the May 1998 Final HF Concentration. The experimental data and modeling are consistent with there being no aerosolization at this concentration and the operating temperature of the Acid Settlers. IV.B.2 - Re-Volatilization of HF (from ground) Following a Release ♦ For a significant release event, evaporation from the pool of catalyst collecting on the ground (and running into drains) can be neglected in comparison to the initial vaporization from airborne droplets of catalyst, even though the barrier may only be a short distance away from the source. IV.B.3 - Water Spray Effectiveness ♦ Within the framework of Mobil’s use of PIRs, not specifically modeling water sprays would overpredict risk and is therefore considered acceptable. IV.B.4 - HF Reduction Versus Barrier Distance ♦ The SA concurs with Mobil that 89% is a reasonable value of ARF for the barriered May 1998 Final HF Concentration case releases of MHF. IV.B.5 - Acid Circulation Pump Barrier Thickness ♦ Mobil has used established engineering approaches in confirming that the Acid Circulation Pump barrier thickness is adequate. IV.B.6 - Pseudo SRI, Treatment of F-N Pairs, and Related Issues ♦ The final Torrance Refinery proposal (Ref. 5) did not use a pseudo SRI approach, but instead explicitly analyzed specific, risk-dominant scenarios from the 1994 QRA and evaluated the effect on each scenario of increasing the catalyst HF concentration to the May 1998 Final HF Concentration and of installing various types of barriers. Moving from the pseudo SRI approach to a scenario-specific analysis yielded results that were more scrutable, more \TorMHF.doc xiv Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst defensible, and more closely tied to the original QRA. In addition, the questions raised regarding the defensibility of using a PIR approach resulted in an augmentation of the SA’s evaluation (see Section IV.A.9) that provided a defensible basis for validating the use of the PIR approach in lieu of specific population distribution calculations. IV.B.7 - Impact on QRA Results of Not Applying Haber’s Law ♦ If Mobil had explicitly used the Haber’s-law modified version of the ERPG-3, correspondence with the 1994 QRA would have been preserved, and the calculated risk values would have been more meaningful. However, the evaluation outlined in Section IV.B.7 gives confidence that Mobil’s approach will not introduce large errors. • Several issues have been addressed via the recommendations listed in Table ES.2. Conclusions The SA has thoroughly reviewed the proposed modified HF alkylation catalyst additive concentration changes and mitigation system enhancements for the Torrance Refinery. Recommendations and suggestions are offered by the SA to provide additional improvements. The SA concludes that: • the scope and level of detail of this evaluation, • implementation of recommended actions precipitating from the SA’s review, and • the review of this implementation by the SA will satisfy the stated requirements and objectives of the Consent Decree and subsequent stipulation and order. The final configuration of additive concentration changes and mitigation system enhancements for the Alkylation Unit allow for the Torrance Refinery to achieve the design operating conditions and alkylate production, while maintaining a margin of safety for the Community that continues to meet or exceed that of sulfuric acid alkylation. Furthermore, the findings of the Safety Advisor’s evaluation also show that the risk of operation associated with the final operating configuration is equivalent to or less than the risk associated with the configuration approved by the Court in 1995. In other words, Mobil's analysis and the Safety Advisor's analysis show that the final operating configuration would provide an \TorMHF.doc xv Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst improvement to the level of safety to the Community. In fact, the Safety Advisor's analysis currently shows that sufficient safety margin exists, such that the installation of flange shrouds and the use of the complete barriers in lieu of the Acid Circulation Pump Shrouds may represent features that provide added layers of protection beyond the minimum required to maintain the same degree of safety as the configuration approved by the Court in 1995. (Ref. 843.00) \TorMHF.doc xvi Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst TABLE ES.2 SUMMARY OF SPECIFIC RECOMMENDATIONS AND SUGGESTIONS ACTION CLASSIFICATION IMPLEMENTATION RESPONSIBILITY REFERENCE (Basis) M-11) Given the inherent uncertainties in this new application for flange shrouds, and the existence of possible discrepancies between technical literature and this particular application, the following recommendation is considered appropriate to ensure flange shroud functionality and reliability. Mobil may exceed or supplement this minimum requirement. Every six months starting on January 1, 1999, and annually beginning on January 1, 2000; on a rotating basis, four flange shrouds (representing a full spectrum of shroud sizes) should be removed, firewater-tested using the January 1998 test apparatus (at least the maximum potential process system pressure, e.g., the 200 gpm @ 260 psig test apparatus, using a gasket gap of 40/1000" for a 2" flange), and replaced with new shrouds. It should be noted that this flowrate is greater than the flowrate for the MHF alternative release scenario chosen for the Risk Management Plan. The intent of rotating the shrouds tested is to ensure that all shrouds are periodically tested and replaced, with a priority for testing being those shrouds that may be exposed to light and/or heat. Tests should be videotaped (or equivalent) (with the video-medium archived by Torrance Refinery EHS for at least 5 years), a brief write-up (e.g., e-mail identifying the shroud and the test results) forwarded to TFD (possibly on a quarterly basis), and TFD invited to observe the tests (at their option). Other test criteria include: ensuring that the calibration of the water supply pressure gauge is current (or that the pressure readings are correlated to another source (e.g., the fire truck water pressure gauge)) and running the test for at least 2 minutes. Although actual gasket failures would likely involve some sort of “ramp-up” of pressure and flowrate during the onset of the breach, one improvement area for the test would be the use of a “fast-acting” valve (e.g., one of the pilot-operated firewater monitor valves) for the initial portion of the test (up to 150 psig). After that, using the engine to “ramp-up” to 250 psig is acceptable, exceeding the maximum system pressure and providing a representative challenge to worst-case failure types. Recommendation (H) Mobil Section IV.A.1 M-12) To provide a design-basis documentation reference point for future flange shroud testing, Mobil should document its initial testing and verification of the shrouds (e.g., field tests at higher pressure with firewater system, use of a demister pad instead of a hose on the drainage end of the shroud, materials tests, etc.) as a referencable package submitted to the City and the SA. This package should also document the testing that indicated that the demister pad worked to decrease aeration and momentum, and withstood full system pressure. This package should also document that the test fluid material characteristics (i.e., water) were sufficiently similar to MHF to reach this conclusion. Recommendation (M) Mobil Section IV.A.1 \TorMHF.doc xvii Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst ACTION M-13) Mobil should document its inspection and testing practices (for ensuring shroud integrity and functionality) as part of its Mechanical Integrity Program. This might include a checklist of items to periodically verify through visual and tactile inspection. [*] \TorMHF.doc CLASSIFICATION IMPLEMENTATION RESPONSIBILITY REFERENCE (Basis) Recommendation (M) Mobil Section IV.A.1 Priorities: H - High, M - Medium, L- Low xviii Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst TABLE OF CONTENTS Page EXECUTIVE SUMMARY....................................................................................................... I. i INTRODUCTION A. Background......................................................................................................... B. Consent Decree Objectives & Evaluation Scope................................................. I-1 I-1 II. EVALUATION CRITERIA .......................................................................................... II-1 III. EVALUATION METHODS UTILIZED ........................................................................ III-1 IV. OBSERVATIONS AND RESULTS ............................................................................ IV.A-1 A. Quantitative Risk Comparison Observations and Results ................................... 1) Materials Used for the Flange Shrouds ........................................................ 2) Other Flange Shroud Design Issues ............................................................ 3) "Unlikely"/Non-Barriered Release Scenarios ................................................ 4) Comparisons with Sulfuric Acid Alkylation .................................................... 5) SRI Contribution of Releases Mitigated by Acid Settler Pans ...................... 6) SRI Contribution of Releases Mitigated by Acid Circulation Pump Barriers .. 7) Flange Failure Contributions to the QRA...................................................... 8) Impeded vs. Unimpeded Release Scenarios................................................ 9) Use of Un-Weighted Average PIRs for the Mobil Calculations ..................... 10) Geometric Assessment of Changes in Isopleth Size to Population Impact... 11) Assessment of Population Characteristics in the Vicinity of the Torrance Refinery ........................................................................................ 12) ARF Extrapolations Beyond the Designated Concentration Range .............. 13) Use of Airborne Reduction Factors in SA QRA Calculations ........................ 14) Results of QRA Benchmark Quantification................................................... 15) Discretized QRA of the Dominant Contributors ............................................ 16) Walkdown Results and Length of Time for Interim Operation ...................... B. Phenomenological Observations and Results ..................................................... 1) Aerosolization............................................................................................... 2) Re-Volatilization of HF (from ground) Following a Release .......................... 3) Water Spray Effectiveness........................................................................... 4) HF Reduction Versus Barrier Distance......................................................... 5) Acid Circulation Pump Barrier Thickness ..................................................... 6) Pseudo SRI, Treatment of F-N Pairs, and Related Issues ........................... 7) Impact on QRA Results of Not Applying Haber’s Law .................................. C. Safety Advisor Observations and Results Photos ............................................... IV.A-1 IV.A-1 IV.A-5 IV.A-7 IV.A-8 IV.A-9 IV.A-12 IV.A-16 IV.A-19 IV.A-21 IV.A-24 V. CONCLUSIONS ........................................................................................................ V-1 VI. RECOMMENDATIONS AND SUGGESTIONS .......................................................... VI-1 VII. REFERENCES .......................................................................................................... VII-1 \TorMHF.doc xix IV.A-29 IV.A-33 IV.A-36 IV.A-38 IV.A-42 IV.A-50 IV.B-1 IV.B-1 IV.B-4 IV.B-5 IV.B-6 IV.B-9 IV.B-9 IV.B-11 IV.C-1 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst TABLE OF CONTENTS Page (continued) APPENDIX A - Glossary of Terms ........................................................................................ A-1 TABLES ES.1 ES.2 IV.A.11-1 IV.A.11-2 IV.A.13-1 IV.A.14-1 Summary of Select Operational Parameters............................................ Summary of Specific Recommendations and Suggestions...................... Sub-regionalization of Select Census Tracts ........................................... Population Change as a Function of Changes to Isopleth Area ............... Estimated PIR Values for Varying HF Concentrations ............................. QRA Benchmark Calculations - Resultant SRI as a Percentage of 1994 QRA Value............................................................................................... IV.A.15-1 Discretized QRA - Resultant Societal Risk Index (SRI) as a Percentage of 1994 QRA Value.................................................................................. \TorMHF.doc xx vi xvii IV.A-30 IV.A-31 IV.A-37 IV.A-39 IV.A-44 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project I. INTRODUCTION A. BACKGROUND Evaluation of Modified HF Alkylation Catalyst The key objectives for the Safety Advisor's evaluation are: • to validate the calculations and conclusions of Mobil's update to the 1994 QRA • to correlate any new technical knowledge associated with potential releases from sulfuric acid alkylation units that may impact the risk comparison basis of the 1994 Stipulation and Order • to provide a focused, independent check of Mobil's analysis to ensure accuracy and completeness • to identify any new developments in alternative alkylation unit processes that may obviate any of the Safety Advisor's conclusions from the SA evaluation documented in Ref. 4 • to verify the installed configuration with the parameters and assumptions utilized in the analyses • to identify any potential vulnerabilities that might exist in transitioning to the proposed operating configuration • to recommend any necessary modifications to equipment, procedures, or training to address these potential vulnerabilities Although not specific objectives, as necessary, the Safety Advisor: • provided additional technical justifications and bases • correlated Mobil's results and conclusions with an independent QRA quantification B. CONSENT DECREE OBJECTIVES & EVALUATION SCOPE The September 30, 1994, Stipulation and Order, agreed to by the Parties, provides the latitude of demonstrating "to the satisfaction of the Safety Advisor that its modified HF catalyst meets the phenomenological standard originally set forth in the Consent Decree or that the modified HF catalyst (including mitigation) presents no greater risk than a sulfuric acid alkylation plant producing a comparable amount of alkylate." This evaluation focused on release characteristics, potential risks, and general safety issues associated with the use of a modified HF catalyst within the Alkylation Unit at the Torrance \TorMHF.doc I-1 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Refinery. In general, the scope of the SA review encompassed a review of all information (addressing both phenomenology and quantitative risk comparison) provided by Mobil as well as key industry studies and test results cited as references in Section VIII, to determine if Mobil: a) "... 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." (Ref. 1) - or b) has demonstrated to the satisfaction of the SA “... that the modified HF catalyst (including mitigation) presents no greater risk than a sulfuric acid alkylation plant producing a comparable amount of alkylate." (Ref. 2) Specific standards and resources used in the evaluation are listed in Section II, Evaluation Criteria, and Section VIII, References. The above criteria do not require comparison of the quantitative results to any absolute risk acceptance criteria. \TorMHF.doc I-2 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project II. Evaluation of Modified HF Alkylation Catalyst EVALUATION CRITERIA The criteria identified in this section are considered to be reasonable, to be technically appropriate, and to address the key requirements and acceptance criteria of the Consent Decree (Ref. 1) and subsequent Stipulation and Order (Ref. 2). General Issues: • if specific regulatory requirements were met, • consistency with known industry standards, guidelines, and practices, and • if the specific requirements of CA Civil Codes 3479 and 3480 (Ref. 12 and Ref. 13) were met. Quantitative Risk Assessment Issues: • criteria "b" in Section I.B (criteria also used in 1995 Safety Advisor Evaluation), and • a lower calculated risk value than the Court-approved Alkylation Unit configuration and HF wt% concentration - Although more stringent than the previous item, this provides for Alkylation Unit operation that meets or exceeds the margin of safety committed to in 1994. Phenomenological Issues: • criteria "a" in Section I.B (criteria also used in 1995 Safety Advisor Evaluation) In addition to the above criteria, other criteria identified in Section II of the 1995 Safety Advisor Evaluation (Ref. 4) may have been used. \TorMHF.doc II - 1 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst III. EVALUATION METHODS UTILIZED A. KEY ACTIVITIES The Executive Summary (pages iii-v) summarizes key activities involved in this evaluation of the application of the proposed additive concentration changes and mitigation system enhancements for the Torrance Refinery Alkylation Unit. B. SPECIFIC INFORMATION REVIEWED BY THE SAFETY ADVISOR Section VII identifies all of the specific references, which are applicable to this evaluation. Where non-written (e.g., verbal or visual) information was obtained, a reference to the date and source is provided. \TorMHF.doc III - 1 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project IV. Evaluation of Modified HF Alkylation Catalyst OBSERVATIONS AND RESULTS IV.A QUANTITATIVE RISK COMPARISON OBSERVATIONS AND RESULTS The following evaluations documented in Reference 4 for the 1994 design are also applicable to the 1998 Torrance Refinery Alkylation Unit design and operating conditions (and therefore do not require additional evaluation by the SA): • General Issues and Assumptions Associated with Accident Scenario Identification • General Issues and Assumptions Associated with Accident Scenario Frequencies • General Issues and Assumptions Associated with Accident Scenario Quantification and Risk Assembly • Uncertainty Characterization and Simplifying Assumptions • Transportation and Regeneration Risks • Comparison with Other Published and Unpublished Risk Results • General Issues Associated with Emergency Response Other important issues are addressed in the following subsections. IV.A.1 Materials Used for the Flange Shrouds Key Issue(s): • Material Compatibility - Flange shroud materials must be compatible with potential shortterm accidental releases of Alkylation Unit process materials (especially HF) and long-term environmental factors (trace concentrations of HF, sunlight exposure, etc.), without compromising shroud strength. • Clarification of Functional Requirements - Although there are several operability requirements (e.g., transparency of shroud side materials to allow for surveillance of the covered flange), the key Consent Decree requirement is to maintain the ability to impede a release of process materials. • Manufacturer Literature - “The RAMCO Safety Shield Manual” indicated that some of the shroud materials were incompatible with HF. However, industry literature, industry practices, and equipment design characteristics are consistent with the materials chosen for this shroud application (e.g., these materials are utilized in valve/flange seals in industrial applications involving HF exposure). \TorMHF.doc IV.A - 1 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Reference Information: • Ref. 831.00 - Feb94, "The RAMCO Safety Shield Manual" - Identifies incompatibility of some flange shroud materials with HF. • 22Apr98, M.P. Younis, E-mail, "HF Barrier Compatibility Test, Pump Seal Shrouding" Indicates good compatibility for the PolyVinylChloride (PVC) shroud materials with hydrofluoric acid. • Ref. 832.00 - January 5, 1998 and January 8, 1998 meetings with Mobil and the City of Torrance - During the meetings, a video tape was shown and two photographs provided to meeting attendees depicting a test using a 160 psig firewater supply to feed a 2” flange leak (240 gpm) impinging on an installed flange shroud. See Recommendation M-12. • Photo IV.1 - Typical Flange Shroud Installation (depicting piping clamps and demister pad) • Photo IV.2 - Flange Shroud Test Configuration (shroud unmounted) • Photo IV.3 - Flange Shroud Test Configuration (shroud mounted) • Photo IV.4 - June 30, 1999 Flange Shroud Test • See Recommendation M-13. Observations: • Materials of Construction: ♦ Clamps - Stainless Steel (used as an upgrade to the polymer ties that are commonly used for the RAMCO Shrouds) ♦ Black Sidepiece - PVC (Younis, 22Apr98) ♦ Threads - PVC ♦ Side-Shielding Materials - Teflon ♦ Drain - PVC • Initial field tests indicated that, during a release, some materials might exit through the seams. Seams were overlaid with butyl rubber to ensure integrity. • Even though the RAMCO literature did not endorse the use of the shroud materials for HF exposure, the shroud materials are consistent with related industrial applications, and SA knowledge and experience with polymers indicates that these materials are suitable for this application. Use of these materials for alkylation unit applications is considered acceptable, with the provision of the periodic testing recommended below. • Although there may be some uncertainties with respect to this new application, the implementation of diligent operations personnel field surveillance of flange shroud integrity \TorMHF.doc IV.A - 2 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst and periodic materials testing can ensure integrity and ability to meet functional requirements. • One issue that had been raised was the potential for shrapnel or pieces of gasket tearing the shroud. Although certainly a consideration, it was gauged that the flange shroud test procedure was adequate for the following reasons: ♦ The momentum of the process fluids during a release at system pressure would already have an ability to challenge the integrity of the shroud, and gasket pieces are not expected to significantly increase the potential for shroud damage. ♦ The use of a polymer for the shroud material means that the shroud is a fairly “tough” (materials engineering use of the term) and able to readily withstand shock. ♦ Small tears in the shroud are not likely to significantly compromise the ability of the shroud to perform its momentum-reduction function. Results/Conclusions: • The SA finds the materials of construction of the flange shrouds suitable for their application at the Torrance Refinery Alkylation Unit. • The SA finds the results of the shroud material testing to be credible and consistent with an ability to meet functional requirements. • A seismic event with a potential for resulting in concurrent leakage and possible damage to flange shrouds and a fire with a potential for resulting in concurrent leakage and possible damage to flange shrouds were considered to be negligible contributors to risk. • Recommendation - Given the inherent uncertainties in this new application for flange shrouds, and the existence of possible discrepancies between technical literature and this particular application, the following recommendation is considered appropriate to ensure flange shroud functionality and reliability. Mobil may exceed or supplement this minimum requirement. Every six months starting on January 1, 1999, and annually beginning on January 1, 2000; on a rotating basis, four flange shrouds (representing a full spectrum of shroud sizes) should be removed, firewater-tested using the January 1998 test apparatus (at least the maximum potential process system pressure, e.g., the 200 gpm @ 260 psig test apparatus, using a gasket gap of 40/1000” for a 2” flange), and replaced with new shrouds. It should be noted that this flowrate is greater than the flowrate for the MHF alternative release scenario chosen for the Risk Management Plan. The intent of rotating the shrouds tested is to ensure that all shrouds are periodically tested and replaced, with a priority for testing being those shrouds that may be exposed to light and/or heat. Tests \TorMHF.doc IV.A - 3 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst should be videotaped (or equivalent) (with the video-medium archived by Torrance Refinery EHS for at least 5 years), a brief write-up (e.g., e-mail identifying the shroud and the test results) forwarded to TFD (possibly on a quarterly basis), and TFD invited to observe the tests (at their option). Other test criteria include: ensuring that the calibration of the water supply pressure gauge is current (or that the pressure readings are correlated to another source (e.g., the fire truck water pressure gauge)) and running the test for at least 2 minutes. Although actual gasket failures would likely involve some sort of “ramp-up” of pressure and flowrate during the onset of the breach, one improvement area for the test would be the use of a “fast-acting” valve (e.g., one of the pilot-operated firewater monitor valves) for the initial portion of the test (up to 150 psig). After that, using the engine to “ramp-up” to 250 psig is acceptable, exceeding the maximum system pressure and providing a representative challenge to worst-case failure types. • Recommendation - To provide a design-basis documentation reference point for future flange shroud testing, Mobil should document its initial testing and verification of the shrouds (e.g., field tests at higher pressure with firewater system, use of a demister pad instead of a hose on the drainage end of the shroud, materials tests, etc.) as a referencable package submitted to the City and the SA. This package should also document the testing that indicated that the demister pad worked to decrease aeration and momentum, and withstood full system pressure. This package should also document that the test fluid material characteristics (i.e., water) were sufficiently similar to MHF to reach this conclusion. • Recommendation - Mobil should document its inspection and testing practices (for ensuring shroud integrity and functionality) as part of its Mechanical Integrity Program. This might include a checklist of items to periodically verify through visual and tactile inspection. \TorMHF.doc IV.A - 4 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project IV.A.2 Evaluation of Modified HF Alkylation Catalyst Other Flange Shroud Design Issues Key Issue(s): a) Flange Shroud Drain Port Direction - A concern was raised regarding the impact of the Drain Port not being pointed at the pavement (typical configuration); however, this concern was obviated by the addition of a secure demister pad in the drain port. b) Flange Leak Aligned with Drain Port - A concern was raised regarding the likelihood of a flange leak aligned in that direction and directly exiting the drain port; however, this concern was obviated by the addition of a secure demister pad in the drain port. In addition, based on the orientation of hole and dimensional characteristics, the potential for the stream directly exiting the hole is minimal, and would not measurably impact the results of the QRA. c) Need for Piping Drain Port Release to Grade - A suggestion was raised regarding potential improvements in accidental flange release mitigation by directing material from the drain port to grade through a hose or pipe; however, this potential need was obviated by the addition of a secure demister pad in the drain port. The SA feels that sufficient momentum is removed by the demister pad, such that process materials exiting to ground would not have a big impact on the size, duration, or concentration of the downwind HF plume. d) Demister Pad Construction and Materials e) Flange Shroud Piping Tie-offs - Integrity (in the event of an accidental release) and operability (being able to perform proper maintenance and surveillance of the unit) concerns were initially raised regarding the base manufacture design that included cord tieoffs at the interface edges with the piping. f) Ability of the Demister Pad to Prevent Aeration and Decrease Momentum Reference Information: • http://idealgraphics.com/ramco contains a description and diagram for a Vue-Drain-Gard Safety Shield, which is similar to the flange shroud applications within the Torrance Refinery Alkylation Unit. • Ref. 857.00 • Photo IV.1 - Typical Flange Shroud Installation (depicting piping clamps and demister pad) • Photo IV.5 - Flange Shroud Demister Pad Configuration Detail • See Recommendation M-12. \TorMHF.doc IV.A - 5 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Observations: d) The demister pad is securely wired into the drain port and is constructed of stainless steel mesh that is designed to more than adequately withstand a flange gasket failure accidental release. e) Specific tests were performed on the design adopted for the Torrance Refinery Alkylation Unit (stainless steel clamps at the interface edges with the piping) that demonstrated that they would be sufficient to maintain their integrity during an accidental release. f) The tests performed on the flange shroud with the demister pad demonstrated that momentum was adequately decreased (consistent with the QRA consequence modeling assumptions for impeded releases) and aerosolization did not occur as the stream exited the demister pad. Results/Conclusions: • The SA finds the following design and construction issues for the flange shrouds, e.g.: ♦ Flange Shroud Drain Port Direction ♦ Flange Leak Aligned with Drain Port ♦ Need for Piping Drain Port Release to Grade ♦ Demister Pad Construction and Materials ♦ Flange Shroud Piping Tie-offs ♦ Ability of the Demister Pad to Prevent Aerosolization and Decrease Momentum to have been suitably addressed in the final application at the Torrance Refinery Alkylation Unit. • The SA finds the results of the testing of the configuration that included demister pads and stainless steel clamps to be credible and consistent with an ability to meet functional requirements. \TorMHF.doc IV.A - 6 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project IV.A.3 Evaluation of Modified HF Alkylation Catalyst "Unlikely"/Non-Barriered Release Scenarios If catalyst HF wt% is increased, it is clear that an accidental release of catalyst from a nonbarriered portion of the process would result in an increased quantity of HF becoming and remaining airborne. Concerns had been raised regarding some initial calculations that had credited reductions in airborne release fractions for all releases (barriered and non-barriered). However, all current analyses (Mobil and SA) credit these release fraction reductions only for barriered accidental release cases. For the final configuration, i.e.: • Flange Shrouds • Settler Pans • Acid Circulation Pump Barriers release barriers were installed to address credible release scenarios that represented dominant contributors to risk (in the processing area of the Alkylation Unit). Thus, use of barriered consequence analysis results for these credible accidental release scenarios is appropriate for other future risk-based applications, e.g.: Risk Management Programs and Emergency Preparedness Issues (such as the Community Alert Siren or Tone-Activated Radio Receivers). \TorMHF.doc IV.A - 7 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project IV.A.4 Evaluation of Modified HF Alkylation Catalyst Comparisons with Sulfuric Acid Alkylation During the course of the SA evaluation, it was verified that no new published (or un-published) information (familiar to the SA) contradicted the sulfuric acid alkylation unit configuration or airborne release fraction data used by the SA in the 1995 evaluation (Ref. 4). Regarding the installation of additional mitigation features on a representative sulfuric acid alkylation unit; as discussed in the Executive Summary and Section IV.D.6 of Ref. 4, for this evaluation, it is appropriate to use a contemporary, commercial sulfuric acid alkylation unit as the reference point for the quantitative risk comparison. While it is correct that the addition of mitigation features such as settler pans and flange shrouding could also achieve safety improvements for a sulfuric acid alkylation unit, they are not used at any sulfuric acid alkylation unit application that the SA is aware of. In addition, this analysis and the sensitivity calculations performed in Ref. 4 partially address this concern by using conservative and defensible assumptions about MHF mitigation system availabilities. This still leaves a margin of approximately a factor of three in favor of the MHF societal risk estimate compared with the sulfuric acid alkylation best estimate. This margin allows for potential improvements to the sulfuric acid alkylation unit design without altering the SA’s conclusion about the acceptability of MHF at the Torrance Refinery. In addition, the margin is even greater due to the numerous conservative assumptions identified in Ref. 4. \TorMHF.doc IV.A - 8 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project IV.A.5 Evaluation of Modified HF Alkylation Catalyst SRI Contribution of Releases Mitigated by Acid Settler Pans Key Issue(s): • Settler Pan Strength and Design Characteristics • Risk Credit for Release Scenarios - Given the pan height and settler surface area covered, credits for a barriered release should be consistent with the bases for the release categories. Reference Information: • 12Mar98, H.A. McVeigh, "Torrance Refinery MHF Alkylation Unit Process Enhancements" (Ref. 5) and related submittals. • Photo IV.6 - Acid Settler Pan Installation (single Acid Settler) • Photo IV.7 - Acid Settler Pan Installation (both Acid Settlers) • Ref. 828.00 - 20Dec97, Job#026973C1, "Belly Pan for 5C-37 & 38," Design Details Identifies Settler Pan height as 4'0". • 05Jan98, Meeting with M.P. Younis and B. Clementson - Identifies Acid Settlers as typically operating with a 20-24" catalyst level or a maximum of 48" for a single-settler operating mode (infrequent). Observations: • Inspection of the Settler Pans (24Mar98) identified that the pan height and design characteristics were consistent with providing the functionality credited in the QRA Update. Section IV.A.16 contains additional information from the Safety Advisor and TFD inspections. During the installation of the Acid Settler Pans, some additional holes had to be added (approximately 24, 6" by 1.5" holes); however, based on hole orientation and dimensional characteristics, the potential for a release stream directly exiting the hole is minimal, and would not measurably impact the results of the QRA. • Concerns had been raised regarding some initial calculations that had credited reductions in airborne release fractions for all settler releases and all settler spool releases. However, all current analyses credit these reductions only for accidental release cases that are mitigated by the Acid Settler Pans. • Currently, ARF changes consistent with barriers are applied to the following release scenarios: ♦ 25mm releases from #1 Settler Spool \TorMHF.doc IV.A - 9 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst ♦ 25mm releases from #2 Settler Spool ♦ 25mm, 50mm, and 100mm releases from #1 Settler ♦ 25mm, 50mm, and 100mm releases from #2 Settler • A review of Page 32 of the 1994 QRA (Ref. 508.00) indicates that the various piping segments associated with the #1 and #2 Settlers are categorized as #1 Settler Spool and #2 Settler Spool, and Page 2.1-3 of the Appendices validates this categorization by identifying 18 meters of piping for the spools. • The application of barrier-release ARF values to the 25mm releases from the settler spools is based on the application of flange shrouds. The application of the 25mm, 50mm, and 100mm releases (and exclusion of rupture events) from the Acid Settlers is based on the height of the Acid Settler Pans (4'0") being greater than the nominal alkylation catalyst level in the Acid Settlers. • QRA calculations were performed by the Safety Advisor to validate the QRA Update and also to perform additional sensitivity analyses. Although a scenario-based QRA Update was performed by the SA, checking calculations and sensitivity analyses were performed using the previously derived category contributions documented in Appendix 3.1/Page1 of Ref. 508.00, adjusted for current modifications to the Alkylation Unit. For this case, the fractional contribution of those releases mitigated by Acid Settler Pans to the total 1994 calculated Societal Risk Index (SRI) is calculated as follows (Appendix 3.1/Page 1 of Ref. 508.00): ♦ #1 Settler (25mm, 50mm, 100mm) = 0.0037+0.0271+0.0092 = 0.0400 ♦ #2 Settler (25mm, 50mm, 100mm) = 0.0171 This fractional contribution (0.0571 from both Acid Settler Pans) may be used to determine the impact on SRI (risk) associated with the improved mitigation of the Settler Pans. It is worth noting that the same fractional contribution to the SRI is also calculated in Mobil’s Ref. 5, Table 1 by first calculating the scenario-specific contributions to settler and spool releases in the vicinity of the settlers (25mm, 50mm, 100mm - summing to 0.1888) and then deriving the settler’s contribution to all of these various releases (0.3030). The net contribution of those releases mitigated by Acid Settler Pans to the total SRI is the product of these two values or 0.0572 (correlating simply to the sum of the values from Ref. 508.00). \TorMHF.doc IV.A - 10 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Results/Conclusions: • The SA finds the calculations performed for the QRA Update (Ref. 5) associated with the SRI contribution of releases mitigated by Acid Settler Pans to be accurate. • The Acid Settler Pans and the Acid Circulation Pump Barriers are primarily structural in nature. Thus, maintaining their integrity is analogous to maintaining the integrity of other important structural elements (e.g., steel support structures), which is typically accomplished by Operator visual inspection, rather than a specific Mechanical Integrity Program element. Thus, the inclusion of the Acid Settler Pans and the Acid Circulation Pump Barriers in the Torrance Refinery Mechanical Integrity Program was considered, but deemed unnecessary. \TorMHF.doc IV.A - 11 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst IV.A.6 SRI Contribution of Releases Mitigated by Acid Circulation Pump Barriers Key Issue(s): • Acid Circulation Pump Barrier Strength and Design Characteristics • Risk Credit for Release Scenarios (Final Configuration) - Given the characteristics of the Acid Circulation Pump Barriers, credits for a barriered release should be consistent with the bases for the release categories. • Risk Credit for Release Scenarios (Interim Operating Configuration) - Given the characteristics of the Acid Circulation Pump Shrouds used for interim operation, credits for a barriered release should be consistent with the bases for the release categories. Reference Information: • Ref. 5 • Ref. 836.00 • Ref. 838.00 • Ref. 848.00 • Ref. 849.00 • Photo IV.10 - Acid Circulation Pump Barrier (side view) • Photo IV.11 - Acid Circulation Pump Barrier (seal area configuration) Observations: • A review of initial submittals had identified some concerns regarding a 27% risk credit for the original design, which involved the application of flexible shrouds around the seals of the Acid Circulation Pumps. The 27% had been based on a failure frequency percentageof-total pump failure rate (for releases of process materials), rather than risk contribution. However, the 1994 QRA assigned Acid Circulation Pump seal failures only to the 5mm leak category. Thus, seal failures did not contribute (analytically) to any of the pump release scenarios that measurably contributed to risk (e.g., 20.5% of the total calculated risk in the 1994 QRA, see Page 2.1-2, which identifies that pump seal failures did not contribute to release risks above a 5mm breach). Thus, from a comparative risk perspective, the seal failures that were being protected by these shrouds were negligible risk contributors. It should be noted that this was a challenge for the 1998 QRA primarily due to the categorizations that had been chosen for the 1994 QRA. \TorMHF.doc IV.A - 12 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst • The Safety Advisor checked Ref. 838.00 data on pumps. It substantiated that approximately 20% of critical pump leaks would involve seals/gaskets; however, there was no indication of severity or leak size. The Safety Advisor also checked CCPS Equipment Reliability Data (Ref. 17) taxonomy details for pump failures; however, it was not sufficiently detailed to characterize a failure rate for pump seal leaks. • For the Final Configuration, the Airborne Reduction Factor (ARF) for a barriered release was applied to all Acid Circulation Pump release scenarios. During the course of the Safety Advisor’s evaluation, Torrance Refinery personnel identified that one of the design criteria for the Acid Circulation Pump Barriers was that a direct impingement would not result in any barrier deformation or release potential. • For the Interim Operating Configuration, to address the lack of a correlation between Acid Circulation Pump Seal failure and risk-dominant release scenarios, Ref. 5 (Pages 8 & 9) assigned “hole sizes of 50mm or less” “to the shrouded inboard section” of the pump. “The remaining hole size of 100mm was assigned to the outboard external casing section of the pump.” Table 2 of Ref. 5 uses a detailed event scenario breakdown to calculate a value of “45.37 percent of the off-site risk attributable to recirculating pumps“ “associated with 50mm or less hole sizes.” This is readily verified from Ref. 508.00, Page 3.1-1 of Appendix 3.1: ♦ (0.0367+0.0736)/0.2431 = 0.4537 • Section IV.A.14 provides benchmark calculations utilizing this fraction of off-site risk that would be mitigated by the Acid Circulating Pump Shroud design proposed for the Interim Operating Configuration. • Exhibit IV.A.6-1 is offered to provide an improved basis for determining Acid Circulation Pump potential leakage rates outside of the proposed shrouded area. Ref. 5 does not provide a definitive correlation between Acid Circulation Pump release failure modes to their calculated risk contribution. Therefore, the exhibit provides a basis for the mitigation credits for the Acid Circulation Pump Shroud by subtracting known failure modes for locations outside the shroud from the total failure frequencies for the specific category. Where uncertainties exist, the failure frequencies of releases not mitigated by the pump shrouds were made conservatively high. Results/Conclusions: • Ref. 5 (Pages 8 & 9) assigned “hole sizes of 50mm or less” “to the shrouded inboard section” of the pump. “The remaining hole size of 100mm was assigned to the outboard \TorMHF.doc IV.A - 13 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst external casing section of the pump.” To provide a correlation between Acid Circulation Pump release failure modes to their calculated risk contribution, the SA used a technique detailed in Exhibit IV.A.6-1. The results of this exhibit demonstrated that approximately one-half of the total risk contribution associated with the Acid Circulation Pumps was mitigated by the Acid Circulation Pump Shrouds. Using this approach, it can be shown that a significant amount of the improvements that are achieved by the Acid Circulation Pump Barriers are also achieved by using the Acid Circulation Pump Shrouds. • Based on the results of Exhibit IV.A.6-1, the Ref. 5 approach for identifying the portion of releases might be "mitigated" by pump shrouds appears to yield a conservatively high result (45.37% of breaches subject to barrier mitigation, resulting in approximately a 17% change in the risk contribution). Therefore, the Ref. 5 results associated with the Interim Operating Configuration are very conservative and are acceptable. • It should be noted that the high percentages of the total failure rates calculated to be attributable to failures in the vicinity of moving parts (and possibly mitigated by the pump shrouds) is consistent with an intuitive understanding of potentially dominant pump subfailure modes. \TorMHF.doc IV.A - 14 Final Report, Rev. 0 EXHIBIT IV.A.6-1 Contribution of Releases Mitigated by Acid Circulation Pump Barriers The following is offered to provide an improved basis for determining Acid Circulation Pump potential leakage rates outside of the proposed shrouded area. Ref. 5 does not provide a definitive correlation between Acid Circulation Pump release failure modes and their calculated risk contribution. Therefore, the exhibit provides a basis for the mitigation credits for the Acid Circulation Pump Shroud by subtracting known failure modes for locations outside the shroud from the total failure frequencies for the specific category. Where uncertainties exist, the failure frequencies of releases not mitigated by the pump shrouds were made conservatively high. Failure Rate Data from the 1994 QRA: 9.00E-06 Flange Failure Frequency (5 mm leaks) 1.00E-06 Flange Failure Frequency (25 mm leaks) 2.76E-06 1.15E-06 4.60E-07 2.30E-07 Per Meter Piping Failure Rate for a 5 mm Leak in 100 mm piping Per Meter Piping Failure Rate for a 25 mm Leak in 100 mm piping Per Meter Piping Failure Rate for a 50 mm Leak in 100 mm piping Per Meter Piping Failure Rate for a 100 mm Leak in 100 mm piping Note that the use of 100 mm piping yields conservatively high failure rates. Rupture 1994 QRA Total 100mm 50mm 25mm 5mm TOTAL 0 1.00E-04 2.50E-04 7.50E-04 7.47E-03 8.57E-03 Less 15 Flange Equivalents 0 0 0 3.00E-05 2.70E-04 (multiplied by 2 for vibration-susceptability of failures) Less 10 Meters of Piping 0 2.30E-06 4.60E-06 1.15E-05 2.76E-05 (equivalent & conservative in terms of addressing weld failures) Net Value for Upper Limit of Failure Rates Associated with Leaks Encompassed by Pump Shrouds: 0 9.77E-05 2.45E-04 7.09E-04 7.17E-03 % of 1994 QRA 97.7 98.2 94.5 96.0 Risk % of Total A) 1994 QRA (from Appendix Page 3.1-1) B) Ref. 5 Approach (assignment of "hole sizes of 50 mm or less" "to the shrouded inboard section" of the pump. C) Relative Risk Contribution for Shrouded Configuration (Interim Operating Mode) 25mm 50mm 100mm TOTAL Frac of 1994 QRA Case A 3.67 7.36 13.28 24.31 1.00 Case B 1.98 4.93 13.28 20.19 0.83 Case C 2.08 4.98 5.81 12.87 0.53 Conclusions Based on the above results, the Ref. 5 approach for identifying the portion of releases might be "mitigated" by pump shrouds, appears to yield a conservatively high result (Case B) compared to what could be calculated from the failure mode basis described above (Case C). Therefore, the Ref. 5 results associated with the Interim Operating Configuration are very conservative and are acceptable. It should be noted that the high percentages of the total failure rates calculated to be attributable to failures in the vicinity of moving parts (and possibly mitigated by the pump shrouds) is consistent with an intuitive understanding of potentially dominant pump sub failure modes. Torrance Refinery Safety Advisor Project IV.A.7 Evaluation of Modified HF Alkylation Catalyst Flange Failure Contributions to the QRA Key Issue(s): • Risk Credit for Release Scenarios - Given the characteristics of flange shroud installation, credits for a barriered release should be consistent with the bases for the release categories. Reference Information: • Ref. 5 • Ref. 507.01 Observations: • A review of initial 1998 QRA Update submittals identified some concerns regarding the risk credit for the flange shrouds. The initial risk credit (27%) had been based on a failure frequency percentage-of-total piping failure rate, rather than risk contribution. However, the 1994 QRA assigned flange failures only to a fraction of the 5mm leak category, and to a smaller fraction for 25mm releases (Ref. 507.01, Appendix 2.1, Page 12). Exhibit IV.A.7-1 contains the check calculations identifying the fraction of total piping failures associated with flange failures. Thus, flange failures did not contribute (analytically) to any of the riskdominant release scenarios (100mm and 50mm leaks). Thus, from a comparative risk perspective, the flange failures that were being protected by these flange shrouds were negligible risk contributors. • Although early versions of the 1998 QRA Update Submittal assigned some flange failures to the 50mm release category and to the 100mm category, the actual risk credits associated with applying shrouds to piping flanges are calculated in Section IV.A.14 to be approximately 1.5%. • Thus, for the final report (Ref. 5, Page 9), “it was determined that rigorous pursuit of flange shrouding credits could not be justified. Mobil is, however, committed to installation of these flange shrouds, as they will have a beneficial impact in reducing on-site worker risks.” \TorMHF.doc IV.A - 16 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Results/Conclusions: • Mobil’s commitment to the installation of the piping flange shrouds is commendable, and its exclusion of the risk mitigation credit associated with piping flange shrouds in Ref. 5 is conservative. \TorMHF.doc IV.A - 17 Final Report, Rev. 0 EXHIBIT IV.A.7-1 Release Fractions Reaction Area Circulating Pumps (Source = 1994 QRA, Appendix 2.1, Pg 2) Fractional Contributions Size Total Fr Seals Casing Dominant 5 mm 7.47E-03 Seal 25 mm 7.50E-04 0 1 Casing 50 mm 2.50E-04 0 1 Casing 100 mm 1.00E-04 0 1 Casing Note: "Seal" is identified as dominant for 5 mm leaks, but numbers indicate that casing failure rates are larger. 2.17E-3/yr seal failure rate was used. Piping/Flanges mm hole 20" 9.30E-07 18" 1.00E-06 16" 1.20E-06 6" 3.10E-06 4" 4.60E-06 3" 6.30E-06 2" 9.30E-06 5 25 50 100 500 5 25 50 100 450 5 25 50 100 400 5 25 50 100 150 5 25 50 100 100 5 25 50 100 75 5 25 50 50 50 Percent Freq/m-yr 0.36 3.35E-07 0.24 2.23E-07 0.25 2.33E-07 0.1 9.30E-08 0.05 4.65E-08 0.36 3.60E-07 0.24 2.40E-07 0.25 2.50E-07 0.1 1.00E-07 0.05 5.00E-08 0.36 4.32E-07 0.24 2.88E-07 0.25 3.00E-07 0.1 1.20E-07 0.05 6.00E-08 0.36 1.12E-06 0.24 7.44E-07 0.25 7.75E-07 0.1 3.10E-07 0.05 1.55E-07 0.6 2.76E-06 0.25 1.15E-06 0.1 4.60E-07 0.5 2.30E-06 0 0.00E+00 0.6 3.78E-06 0.25 1.58E-06 0.1 6.30E-07 0.05 3.15E-07 0 0.00E+00 0.73 6.79E-06 0.23 2.14E-06 0.05 4.65E-07 0.1 9.30E-07 0.05 4.65E-07 SCENARIO FAILURE RATES Scenario m 20" m 18" m 16" m 6" m 4" 0.00E+00 13.37 30.33 2.18E+01 5 4 11 1.50E+01 Flanges Freq 5mm Freq 25 mm Freq 50 mm Freq 100 mm Freq Rupt 0 0 0 0 0 4.81E-06 3.21E-06 3.34E-06 1.34E-06 6.69E-07 1.31E-05 8.74E-06 9.10E-06 3.64E-06 1.82E-06 2.44E-05 1.62E-05 1.69E-05 6.77E-06 3.38E-06 m 3" m 2" 0 1.62 3 0 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 6.12E-06 2.55E-06 1.02E-06 5.10E-07 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 Contribut. From Contribut. From Flanges Piping 2.97E-04 3.30E-05 0 0 0 4.84E-05 3.07E-05 3.04E-05 1.23E-05 5.87E-06 Sum 3.45E-04 6.37E-05 3.04E-05 1.23E-05 5.87E-06 Fractional Contribution Flanges 0.86 0.52 0.00E+00 0.00E+00 0.00E+00 Flange treatment matches exactly, but is slightly different than the writeup on Page 12 of Appendix 2.1; however, differences are not significant, and are actually conservative. 0.00E+00 25.98 9 8.94 2 1.51E+00 3.00E+00 0 0 0 0 0 0 0 0 9.35E-06 6.24E-06 6.50E-06 2.60E-06 1.30E-06 3.86E-06 2.57E-06 2.68E-06 1.07E-06 5.36E-07 1.69E-06 1.12E-06 1.17E-06 4.68E-07 2.34E-07 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 10 Flanges Freq 5mm Freq 25 mm Freq 50 mm Freq 100 mm Freq Rupt 1.26E-04 1.40E-05 0 0 0 1.49E-05 9.93E-06 1.03E-05 4.14E-06 2.07E-06 1.41E-04 2.39E-05 1.03E-05 4.14E-06 2.07E-06 0.89 0.58 0.00E+00 0.00E+00 0.00E+00 Torrance Refinery Safety Advisor Project IV.A.8 Evaluation of Modified HF Alkylation Catalyst Impeded vs. Unimpeded Release Scenarios Key Issue(s): • The 1994 QRA categorized a certain fraction of all dominant release scenarios as “impeded” or “unimpeded”, whereas the 1998 QRA Update assigned a single value to the Airborne Reduction Factor (ARF) for all “barriered releases” and a single ARF for all “unbarriered releases”. • Thus, the SA Evaluation addressed whether these different treatments correlate the same parameters as the basis for the 1998 QRA Update calculations, so that the results of the 1998 QRA Update can be accurately compared to the 1994 QRA. The SA also investigated whether barriers provide the same degree of improvement for releases that were previously categorized as having their momentum “impeded” as those that were previously categorized as having their momentum “unimpeded”. • After verifying that the bases are consistent, the appropriate ratios (of calculated numbers of exposures) to use for the Safety Advisor’s verification calculations and for the SA’s Update of the 1994 QRA were identified. Reference Information: • Ref. 5 Observations: • The definition of “impeded” releases for the 1994 QRA was in the context of SAFETI/PHAST dispersion modeling usage. In that context, the effect of impeding the release is to significantly reduce the initial momentum. Although the momentum changes, it does not cause any reduction in the amount of HF that becomes airborne. The reduction in momentum leads to a decrease in the amount of air that is entrained close to the source by jet entrainment. Thus, for the 1994 QRA the difference between “impeded” and “unimpeded” releases is in the momentum associated with the release and in the resultant shape and size of the isopleth (and in the calculated number of off-site exposures). In the 1994 QRA, a 65% ARF was applied to both, resulting in the same quantity of HF becoming airborne. • The 1994 QRA used an “impeded/unimpeded ratio” of “0.5 for relatively open process areas, 0.75 for relatively congested areas and 0.9 for highly congested areas.” \TorMHF.doc IV.A - 19 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst • In the 1994 QRA, for each scenario, “unimpeded” consequences (i.e., calculated number of off-site exposures) are measurably higher than the “impeded” consequences. Results/Conclusions: • The SA reviewed the basis for the application of “impeded” vs. “unimpeded” release concepts in the 1994 QRA and the application of Airborne Reduction Factors (ARFs) for the 1998 QRA Update, and found the following approach to provide a consistent and accurate basis for correlating the results of the two studies: ♦ Apply the same ratio (of calculated number of exposures) for previously “impeded” or “unimpeded” cases, for those releases that would be mitigated by barriers: { } {} {} ♦ Apply the same ratio (of calculated number of exposures) for previously “unimpeded” and “impeded” cases, for those releases that would not be mitigated by barriers. {} {} {} ♦ To address a minor conservatism in these calculations, the base value for the calculated numbers of exposures for “unimpeded” cases could be changed to that value for an “impeded” release, for those releases that would be mitigated by barriers, if this analysis is updated in the future. It is not expected that the removal of this conservatism would have a significant impact on the results. \TorMHF.doc IV.A - 20 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project IV.A.9 Evaluation of Modified HF Alkylation Catalyst Use of Un-Weighted Average PIRs for the Mobil Calculations Key Issue(s): • Validity of the use of a un-weighted average Population Impact Ratio (PIR) as the basis for calculating net risk impact on the public due to a change in alkylation catalyst concentration Reference Information: • Ref. 5 Observations: • In Ref. 5, PIR was determined by calculating the square of the ratio of the “calculated impact distance” for a certain release size and catalyst concentration to the “calculated impact distance” for the 1995 Court-Approved HF Concentration reference case (Ref. 5, Page 4). The PIRs for each release category (i.e., 5 mm, 25 mm, 50 mm, and 100 mm) were then numerically (linearly) averaged to determine average PIRs for a barriered release and for an unbarriered release. From these values a “(Mod)PIR” was calculated by weighting the PIRs for barriered and unbarriered releases to the previously calculated risk contribution of ALL barriered releases and for ALL unbarriered releases. This “Mod(PIR)” was then applied to all “N” values for the previous F-N curve to calculated the modified Societal Risk Index (SRI). These calculations are detailed in Mobil’s spreadsheet, Far76fin.xls, and the results are presented on Ref. 5, Page 10. • Potential issues with the approach used in Ref. 5 include: ♦ The baseline risk calculations performed in the 1994 QRA discretized specific release scenarios, determined the frequency and consequences for each scenario, and multiplied them to determine risk. For the 1994 QRA, the risk associated with all identified scenarios was then summed to determine the total potential risk associated with releases within the Torrance Refinery Alkylation Unit. ♦ For Ref. 5, PIRs for unbarriered releases were measurably larger for the larger release sizes { }. Thus, if the PIRaverage=1.081 is used, it would underpredict risk, if the original risk was dominated by larger release sizes. A review of the results of the 1994 QRA (Ref. 4) did indeed identify that larger release sizes dominated the risk of unbarriered releases. \TorMHF.doc IV.A - 21 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst ♦ The value of the “Mod(PIR)” does not have any direct correlation to risk since the PIRaverage does not have any direct correlation to risk. Thus, the adjusted “N” values and the modified SRI in Ref. 5 cannot be defended as being accurate. ♦ The “N” values in the F-N curve and the SRI are a composite of the risk associated with many discrete scenarios. Thus, by applying parameters (e.g., “Mod(PIR)”) that are associated with an average of a wide range of release scenarios, risk could also be underpredicted if the scenarios used to calculate “N” were dominated by larger release scenarios. • Thus, the application of the non-weighted average PIRs at several locations in the calculations performed for Ref. 5 may be providing overlapping reductions to the actual risk impact associated with changes to additive concentration. • Using the square of the ratio of the “calculated impact distance” for a certain release size and catalyst concentration to the “calculated impact distance” for the 1995 Court-Approved HF Concentration reference case (Ref. 5, Page 4) to calculate the PIR for specific release sizes also relies upon two important assumptions: ♦ Ref. 5, Page 4 implicitly assumes that the populated impact area for an ellipticallyshaped isopleth, with the release at one end of the isopleth and a large portion of the isopleth (within the fenceline) with nearly zero population, varies as the square of the linear distance to the endpoint value. There are no known “first principles” that support this assumption. ♦ It is being implicitly assumed that the population density within the impact area is homogeneous at all locations from the release point, i.e., the population density within the boundaries of the Torrance Refinery is identical to that outside the fenceline. This assumption may apply for other refinery locations; however, for the Torrance Refinery, population density within the fenceline and outside of the fenceline are significantly different. The population density within the majority of the impact area (primarily refinery property) is zero. However, there is a high population density outside the fenceline, and relatively small changes in affected areas outside the refinery boundary can have a large impact on the population affected. Sections IV.A.10 and IV.A.11 of this report evaluate these assumptions and provide a defensible basis for their use. \TorMHF.doc IV.A - 22 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Results/Conclusions: • The SA found the specific averaging methods employed in Ref. 5 to calculate change in SRI (using an un-weighted average PIR) to be difficult to defend. Thus, when comparing the risk tradeoffs associated with changes in acid concentration and barrier application and effectiveness (to the 1994 QRA), the use of a similar discretized approach is necessary. • As an extension of its review efforts, the SA utilized a detailed, discretized QRA approach that was benchmarked to the 1994 QRA to investigate the risk tradeoffs associated with changes in acid concentration and barrier application. \TorMHF.doc IV.A - 23 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project IV.A.10 Evaluation of Modified HF Alkylation Catalyst Geometric Assessment of Changes in Isopleth Size to Population Impact Key Issue(s): • Section IV.A.9 identified several key assumptions utilized for the Ref. 5 calculation of a Population Impact Ratio (PIR) for specific release sizes. Ref. 5, Page 4 implicitly assumes that the populated impact area for an elliptically-shaped isopleth, with the release at one end of the isopleth and a large portion of the isopleth (within the fenceline) with nearly zero population, varies as the square of the linear distance to the endpoint value. There are no known “first principles” that support this assumption. Reference Information: • Ref. 5 Observations: • To provide a basis for the assumptions made in Ref. 5, Exhibit IV.A.10-1 provides a derivation of isopleth areas (approximated by an ellipse) as a function of impact distance and fenceline location. The results of this assessment were: ♦ a validation of isopleth area varying as a function of the square of the impact distance ♦ using a base example, a demonstration that, although not directly proportional to isopleth area, for small changes in isopleth area, linearity between the isopleth area ratio (a function of release rate and quantity) and area outside the fenceline (PIR) may be assumed as an approximation • To support the above basis for the assumptions made in Ref. 5, the following hypothetical example was used to demonstrate that for a homogeneous population outside the fenceline, for small changes in the distribution of the same total release quantities (i.e., different release scenarios that sum to the same total risk), the net risk is sufficiently similar to address the requirements of this evaluation. This section bridges the gap between release rate assumptions used in Ref. 5 and PIR. Therefore, for a discretized QRA, where some release scenarios increase and some decrease, and if the change in risk is not too large, the use of a linear relationship between release rate and PIR is acceptable. ♦ Using Exhibit IV.A.10-1, consider a simple Alkylation Unit with only three release points of identical breach size, location, isopleth area, and likelihood (probability) of failure. If the risk associated with a single release is 100 exposures over a given time period, than the risk of all three releases is 300 exposures over the same time period. \TorMHF.doc IV.A - 24 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst ♦ Consider the same case with one of the three release points shrouded (with a 20% example isopleth area decrease) and the other two unshrouded (with a 10% example area increase). This is an example of a set of values that is consistent with a zero net change in total release quantity. From Exhibit IV.A.10-1, the two unshrouded releases increase 24% each, yielding a risk of 124 exposures over a given time period. The shrouded release decreases to 0.55, yielding a risk of 55 exposures over that same time period. The total risk of exposure is 303 (over this time period), nearly identical to the original value of 300. ♦ Consider the same case with two of the three release points shrouded (with a 10% example isopleth area decrease) and the other one unshrouded (with a 20% example area increase). This is an example of a set of values that is consistent with a zero net change in total release quantity. From Exhibit IV.A.10-1, the one unshrouded release increases 49%, yielding a risk of 149 exposures over a given time period. The shrouded releases decrease to 0.77, yielding a risk of 77 exposures over that same time period. The total risk of exposure is 303 (over this time period), nearly identical to the original value of 300. • Another minor missing link to the key assumption in Ref. 5 is the variation of isopleth area as a function of release rate. A general assumption is that the isopleth area is proportional to the release rate for a continuous release. To verify this assumption, Exhibit IV.A.10-2 was provided. ♦ Exhibit IV.A.10-2 summarizes several simplified dispersion computations (using the SLAB software) for six HF, horizontal jet release scenarios of identical conditions except for the release rate. The footprint dimensions at a 20 ppm concentration are identified. ♦ Based on these results, a correlation was made between release quantity and the area under the hazard footprint. Assuming the footprint is an ellipse with its major axis as the downwind distance (provided by SLAB) and the minor axis as the width provided by SLAB, it was found that the area of the ellipse is linearly correlated with the release rate (correlation coefficient r=0.999). \TorMHF.doc IV.A - 25 Final Report, Rev. 0 Torrance Refinery Safety Advisor Project Evaluation of Modified HF Alkylation Catalyst Results/Conclusions: • The results of a theoretical correlation of changes in impact distance to Population Impact Ratio (PIR) show that it does not follow the correlation identified on Page 4 of Ref. 5, i.e., the PIR is not directly proportional to isopleth area ratio (i.e., square of endpoint distance). However, from the geometrical evaluation provided in Exhibit IV.10-1, for small variations (i.e., small changes in risk, for cases where there is an interest in showing only a small impact on risk), linearity between the isopleth area ratio and PIR may be assumed as an approximation. The linearity improves as the proportion of area outside the fenceline to the total area improves. • Although the assumption made on Page 4 of Ref. 5 regarding the effect of changes in isopleth area ratio to PIR may yield acceptable results, it should be realized that it is acceptable for this application, because the objective of the proposed operational and design changes to the Alkylation Unit is to not significantly change risk (i.e., Ref. 5 and this evaluation is essentially performing a hypothesis to test that the net risk changes for the proposed changes to the process are near zero). Therefore, the results appear to be unaffected by this simplifying assumption and can be supported. • Based on the results of representative dispersion modeling calculations (documented in Exhibit IV.A.10-2), it was shown that the general assumption that the isopleth area is proportional to the release rate for a continuous release is acceptable for use for this updated MHF QRA. • These results will be used for the SA’s discretized QRA to investigate the risk tradeoffs associated with changes in acid concentration and barrier application. \TorMHF.doc IV.A - 26 Final Report, Rev. 0 EXHIBIT IV.A.10-1 Geometric Evaluation of Isopleth Area vs. Area Beyond the Fenceline b -a a d -b Assume that a release isopleth is characterized by an ellipse with release origin at x=-a, farthest apex at x=a, width (y) of 2b, and location (d) as the fenceline (the distance between d and -a is the plant boundary). Between the release location and the fenceline, a zero population density is assumed, i.e., we are looking at the area outside the fenceline and the ratio of the change in population as a function of the change in overall isopleth area. Using a standard ellipse equation of x^2/a^2+y^2/b^2=1, it can be derived that for the area to double (with the aspect ratio remaining the same), the new a = a*sqrt(2) and the new b = b * sqrt(2). The following equation for area can be derived from the standard equation for an ellipse and integrating from x=d to x=a to determine the area of the ellipse beyond the fenceline. The total ellipse area = pi*b*a. Range of Applicability: -a