DET NORSKE VERITAS TM

DET NORSKE VERITAS TM Report QRA FOR SKANGASS LNG PLANT - FERRY BUNKERING PROJECT SKANGASS AS REPORT NO./DNV REG NO.: 2013-4091 / 17TLT29-7 REV.1, 2013-06-11

Rev. No. Date Reason for Issue Prepared by Verified by Approved by A 2013-06-03 Draft issue signed and verified BERTHOM RISVIK ELLING Reference to part of this report which may lead to misinterpretation is not permissible. DNV Reg. No.: 17TLT29-7 Date : 2013-06-11 Page iii of iv

Table of Contents Page EXECUTIVE SUMMARY...1 1 INTRODUCTION...3 2 FACILITY AND OPERATION DESCRIPTION...4 2.1 Facility description... 4 2.2 Facility operation... 4 3 QUALITATIVE COMPARISON WITH CONVENTIONAL BUNKERING SYSTEMS...6 4 COMPARISON OF RIGID LNG LOADING ARMS AND FLEXIBLE LOADING HOSES...7 5 RISK RESULTS...8 5.1 Risk contours... 8 5.1.1 Introduction... 8 5.1.2 Risk contours generated by the Train 1... 8 5.1.3 Risks contours generated according to the bunkering facility operational phases... 11 5.1.4 Risk contours generated by the plant with the bunkering facility... 14 5.2 Risk 1 st party and 2 nd party... 16 5.2.1 Introduction... 16 5.2.2 Hours of presence... 16 5.2.3 Risk Results... 17 5.3 Societal and Risks 3 rd party... 21 5.3.1 Societal Risk for 3 rd party... 21 5.3.2 Hours of presence... 22 5.3.3 Risk for 3 rd party... 23 6 SENSITIVITIES...28 6.1 Risk 1 st party and 2 nd party... 28 6.2 Societal and 3 rd party... 31 6.2.1 Societal... 31 6.2.2... 33 6.3 Sensitivities discussion... 35 7 SHIP MANIFOLD LOCATION...36 8 CONCLUSIONS AND RECOMMENDATIONS...37 9 REFERENCES...39 DNV Reg. No.: 17TLT29-7 Date : 2013-06-11 Page iv of iv

EXECUTIVE SUMMARY Skangass is in the process to install an LNG bunkering station at the Fjordline ferry in Risavika, located adjacent to Skangass LNG Base Load Plant. Skangass would like to take benefit of the close proximity to ferry (approximately 600 m), by feeding LNG directly from the plant, through an underground pipeline. The Skangass ferry bunkering project is currently in the detailed engineering phase. In order to provide to the reader and the decision maker a overview as close as possible to the reality, some of the assumptions have been reviewed, such as, for example, differentiation of criteria for ferry workers (2 nd party). All the assumptions are available in the Appendix A of this report. The main objective of this QRA is to: Assess the combined for the ferry area, taking into consideration both the new LNG bunkering station and the updated results for the Base Load Plant, Compare the s with the acceptance criteria. Based on experience in LNG bunkering, DNV assesses that the time of detection is likely to be close to 90s in total (60s for detection and initiation, 30s for completion (time to close the valve)). While Skangass estimates, based on its experience, that the total time can be reduced down to 36s (30s for detection and initiation, 6s for completion). Even if a closing time of 6 seconds could be possible if Skangass documents that the facility uses a valve in compliance with this requirement, DNV s experience shows that time to detect and initiate the ESV is often extended to 60 seconds. Therefore this report contains three sets of results: - The picture based on DNV recommendation corresponding to ESD total time of 90s, presented in section 5. - Two sets of sensitivities, presented in section 6: o Risk picture based on a mixed inputs from DNV recommendation and Skangass input, 66s ESD total time, o Risk picture based on Skangass estimation, 36s ESD total time. DNV considers the results based on the first alternative (total ESD time of 90s) as the valid results. The two sensitivities are there to help the reader to understand the impact of a shorter duration of ESD completion Risk calculated for the planned LNG bunkering station and LNG Base Load Plant (Train 1), combined, is presented in the report: Date : 2013-06-11 Page 1 of 39

The 1 st /2 nd party individual is found to be within the ALARP region for all personnel categories. The calculated 3 rd party societal is acceptable or within the ALARP area for all events. 3 rd party Specific Risk is acceptable or in the ALARP area for all individuals except. The passengers on board the ferry are at lower than the passengers at the. As a consequence of this boarding of most of passengers is planned to take place before bunkering, ref. Skangass Design Basis /4/. Late-comers will be able to board once the bunkering operations are completed. According to the same reference, no passengers are allowed in the passenger tube during bunkering and only late-comers will board through the tube after bunkering. The sensitivities show a reduction in the picture for 1st party (only for the group Skangass operators at the ferry jetty ), 2 nd party and 3 rd party. However, the results remain in the same order of magnitude. The detailed engineering of Risavika ferry station is regarded as safer than standard practice. As the calculated is in ALARP area, Skangass should consider taking measures to reduce the as the one presented below as example. Indeed, ALARP stands for As Low As Reasonably Practicable, meaning that the is tolerable if reduction is impracticable or if its cost is grossly disproportionate to the improvement. Recommendations for mitigation provided in the plant s QRA /1/ are valid also for the bunkering station: Focus on ignition source control to reduce the ignition probability In general, reduction of LNG volume in the bunkering pipeline (e.g. either by segmentation or by depressurizing part of the pipelines in between bunkering operation), would reduce the fire and explosion. Liquid bunds/trays around the LNG loading arm where an LNG leak could occur would help limiting the pool size in the event of an LNG leak. DNV recommends that ESD equipment is qualified for compliance with the requirements to short response time (from gas detection to isolation) assumed in the sensitivities in this analysis. DNV recommends to apply all the points for manual local supervision as presented in section 8. Date : 2013-06-11 Page 2 of 39

1 INTRODUCTION Fjordline is converting ferries for use of natural gas (LNG) instead of conventional fuels. In light of this, Skangass is evaluating to install an LNG bunkering station at the Fjordline ferry in Risavika, located adjacent to Skangass LNG Base Load Plant. Skangass would like to take benefit of the close proximity to ferry (approximately 600 m), by feeding LNG directly from the plant, through an underground pipeline. The objective of this QRA is to: Calculate the for the ferry area, taking into consideration both the new LNG bunkering station and the updated results for the Base Load Plant, Compare the s with the acceptance criteria, described in appendix C. Risk levels are reported in terms of: for on-site manning (1 st /2 nd party) Societal for off-site population (3 rd party) Report Structure The main part of this report presents background, objectives, acceptance criteria and results of the analysis. Mitigating measures are provided as recommendations. The content of the 5 appendices is described below: Appendix Description Appendix A Assumptions Includes all major assumptions for the analysis. Appendix B HAZID HAZID review of the ferry bunkering station in Risavika. Appendix C Risk in hazardous activities and criteria Risk definition and criteria for the 1 st party, 2 nd party personnel, and 3 rd party population. Appendix D Leak frequency Basis and results from LEAK. calculation Appendix E PHAST RISK software package General description of the software used to estimate the Risk and to 3 rd party. Date : 2013-06-11 Page 3 of 39

2 FACILITY AND OPERATION DESCRIPTION 2.1 Facility description The bunkering facility is designed to complete bunkering of Fjordline ferries in about 1 hour with vapour return. Furthermore the system is designed with no need for cooling down of the bunkering line prior to bunkering operation. The LNG Bunkering facility will consist of the following major equipment components: LNG Bunkering Pump located in the LNG plant Fiscal metering located in the LNG plant 8 LNG Pipeline 4 Vapour Return Line LNG Loading arm If needed, the details of equipment for the LNG bunkering facility are available in the appendix D of this report. The LNG bunkering pump and fiscal metering package will be located inside the Skangass LNG plant, and the loading arm will be located at the bunkering jetty. LNG transfer pipeline between the LNG plant and the bunkering station will be vacuum insulated and installed below ground to protect it from external damage, ref. Figure 1. 2.2 Facility operation As mentioned above, the bunkering facility is designed to be completed in 1 hour. During the operation, at least two operators will be present at the ferry jetty respectively on the shore side and on the ship side to supervise locally the operation. Beside the fact that the operators help the operation to get going, the operators will be special trained to take actions in case of detection of any type of leak. This continuous local supervision will come in addition to the automatic fire & Gas detection and will help to initiate the ESD system as soon as possible and reduced the leak inventory. Based on experience in LNG bunkering, DNV assesses that the time of detection is likely to be close to 90s in total (60s for detection and initiation, 30s for completion (time to close the valve)). While Skangass estimates on its experience, that the total time can be reduced down to 36s (30s for detection and initiation, 6s for completion). Even if a closing time of 6 seconds could be possible if Skangass documents that the facility uses a valve in compliance with this requirement, DNV s experience shows time to detect and initiate the ESV is often extended to 60 seconds. Therefore this report contains three sets of results: - The picture based on DNV recommendation corresponding to ESD total time of 90s, presented in section 5. - Two sets of sensitivities, presented in section 6: Date : 2013-06-11 Page 4 of 39

o Risk picture based on a mixed inputs from DNV recommendation and Skangass input, 66s ESD total time, o Risk picture based on Skangass estimation, 36s ESD total time. DNV considers the results based on the first alternative (total ESD time of 90s) as the valid results. The two sensitivities are there to help the reader to understand the impact of a shorter duration of ESD completion. - Figure 1 LNG Base Load Plant and Fjordline ferry area. LNG pipeline route (indicative) as red line. Date : 2013-06-11 Page 5 of 39

3 QUALITATIVE COMPARISON WITH CONVENTIONAL BUNKERING SYSTEMS Skangass intention with its current design is to provide an LNG bunkering system for ferries which is safer than conventional systems, e.g. by truck-to-ship, ship-to-ship or -toship. By conventional is meant the small scale LNG filling stations in Norway, which is found acceptable to DSB and the Norwegian Maritime Authority (Sjøfartsdirektoratet), and therefore may be regarded as standard practice. Naturally, these stations vary in technical features but are typically characterized by Flexible filling hose Bolted flange connections requiring manual disconnection Manual operation (e.g. truck driver, ship crew) Ventilation lines / relief to atmosphere No automatic/instrumented shut down upon gas detection Reasons for stating that Skangass proposed design is safer are basically due to the simplicity of the bunkering due to the benefit of the close proximity to the Skangass LNG Plant and selection of equipment with low leak frequency. Some examples are: Simpler fewer possible leak sources: o No ventilation lines or pressure safety relief valves to atmosphere o All equipment components are selected based on low leak frequency (LNG bunkering pump, LNG bunkering arm, vacuum insulated piping, etc.) Reduced LNG inventory in the proximity of the bunkering system: o No LNG buffer tank o Pipeline in underground tunnel Use of best available equipment: o Loading arm instead of hose o ESD link between ship/shore according to the SIGTTO standard o Stainless steel, double wall, vacuum insulated piping gas detection between the walls instead of PUR piping outside the plant o Can pump without external seals instead of a centrifugal pump Automatic, in addition to manual, ESD system F&G detection and alarm system triggering automatic shutdown of the bunkering system located on shore side and ship side. Bunkering operation always manned with at least two operators. One on shore side and one on ship side. Date : 2013-06-11 Page 6 of 39

A thorough assessment of the actual improvement of Skangass design as opposed to conventional bunkering systems goes beyond the scope of this report, as this would require a detailed assessment of each of the points above. The following is however noted: Each of the points in the list above does not implicitly represent an improvement compared to conventional design, in terms of safety, but as a whole, the technical design of Risavika bunkering station is regarded as safer than standard practice. 4 COMPARISON OF RIGID LNG LOADING ARMS AND FLEXIBLE LOADING HOSES The memo developed by DNV /5/ presents a qualitative assessment of the difference between rigid LNG loading arms and flexible hoses used in the truck-to-ship operations for instance. Prior to this assessment, Skangass have decided upon rigid LNG loading arms as a base case solution for LNG bunkering at Risavika. In case of maintenance or failure of these loading arms, Skangass are considering using flexible LNG loading hoses as a backup solution for their Ferry Bunkering Station at Risavika. The conclusion presented below summarizes the assessment done. All the details leading to this conclusion can be found in the memo /5/: DNV agree with Skangass prioritization on using rigid loading arms as the base case for the LNG bunkering operations at Risavika. If hoses shall be used as a backup solution, stringent operation and maintenance procedures need to be in place to control for loss of containment. DNV suggest to further analyze the picture of bunkering operations with flexible hoses. However the available data are not sufficient and consistent enough to develop a possible acknowledged failure frequency for flexible LNG hoses at this stage. This will only be possible after a qualification program, such as the program TNO 1 initiated last year. 1 Dutch Organization for Applied Scientific Research (TNO), who has an independent position that allows to give objective scientifically founded judgments and acknowledged as valid source of information for technical safety subjects, have recently focused on the lack of sufficient qualification programs and proven track records for flexible LNG hoses Date : 2013-06-11 Page 7 of 39

5 RISK RESULTS 5.1 Risk contours 5.1.1 Introduction The contours show location specific result. This is the for a hypothetical individual assumed to be continuously present at the specific location. Continuously present correspond to an exposition of 8760 hours or one year period. For instance, to be exposed to the level at the ferry presented in Figure 6, an individual must live there 24 hours/7 days during one year. Normally, individuals will not be continuously present throughout the year, therefore more realistic estimate of the for an individual is provided by the individual-specific figures, taking the actual exposure time into account, presented in section 5.2 and section 5.3. The is compared against the criteria presented in the Appendix C of this report. Note that the contours shown are valid for 1 st, 2 nd and 3 rd parties. Below, contours are presented for different configurations as follows: - Risk contours generated by the Train 1 of the plant or as it is today in section 5.1.2 ; - Risk contours generated by Train 1 and bunkering facility or as it could be in section 5.1.4. To help the reader to differentiate the due to the different operations phase of the bunkering facility, the contours generated by the plant and the bunkering facility for the two operational phases are presented in section 5.1.3. But note that these are considered as snapshots and are only valid during the two specific periods: - Risk contours generated by Train 1 and bunkering facility between the bunkering operations (23 hours per day); - Risk contours generated by Train 1 and bunkering facility during the bunkering operations (assumed taking place 1 hour per day); 5.1.2 Risk contours generated by the Train 1 Figure 2 and Figure 3 show contours for the peninsula today due to the existing LNG Base Load Plant (Train 1). The highest contour identified is 4E-4 per year covering the northwest part of the LNG plant. The contours are comparable as the ones in the update of the 2009 QRA, ref. /1/ Date : 2013-06-11 Page 8 of 39

Figure 2 Risk contours today: LNG Base Load Plant (Train 1) without ferry bunkering installation Date : 2013-06-11 Page 9 of 39

Figure 3 Risk contours today: LNG Base Load Plant (Train 1) without ferry bunkering Date : 2013-06-11 Page 10 of 39

5.1.3 Risks contours generated according to the bunkering facility operational phases The contours generated by the LNG Base Load Plant (Train 1) and the ferry bunkering facility presented below in the Figure 4 and Figure 5 are there to represent the variation between the two different phases system during bunkering and between bunkering operations. Note that theses contours are only valid during the two specific operational phases and do not represent the contours a hypothetical individual continuously present in the area. 5.1.3.1 Risk contours between bunkering operations Figure 4 shows contours. Note that these contours represent the case when no LNG bunkering is taking place. I.e. the figure is a snapshot showing the a person, present at any point outdoors, is exposed to between bunkering operations. Date : 2013-06-11 Page 11 of 39

Figure 4 Risks contours (Train 1 and ferry bunkering system) between bunkering operations Date : 2013-06-11 Page 12 of 39

5.1.3.2 Risk contours during bunkering operations Figure 5 shows contours for the peninsula due to the LNG Base Load Plant (Train 1), with the ferry bunkering facility installed. These contours however are valid during bunkering of LNG. I.e. the figure is a snapshot showing the a person, present at any point outdoors, is exposed to during bunkering operations. Comparison with Figure 4 shows that the 1E-4 and 1E-5 contours are drifted slightly southwards, due to the activity at the jetty during the bunkering. However, according to Figure 5 the for passengers outdoors at the ferry is slightly increasing, but remains in the same order of magnitude. Figure 5 Risk contours (Train and ferry bunkering system) during bunkering Date : 2013-06-11 Page 13 of 39

5.1.4 Risk contours generated by the plant with the bunkering facility Figure 6 and Figure 7 show contours for the peninsula due to the LNG Base Load Plant (Train 1) with the ferry bunkering system installed. The contours show individual, for a person present at any point outdoors, continuously exposed 8 760 hours per year, at any time of the day, an average of both bunkering and no bunkering phases. Therefore, these figures are the most appropriate for comparing with today s picture, i.e. Figure 2 and Figure 3. The presence of the underground pipeline increases the overall level. The additional pipeline raises the around the main pipe rack located at the south of the plant; the areas covered by 2E-4 and 1E-4 contours appear to be extended southwards. Apart for this increase, the new installation does not change the coverage of s contours 4E-4, 3E-4 as shown in Figure 2. Concerning the 1E-5 contour; it now covers a greater part of the peninsula including the ferry and its surroundings. Figure 6 Risks Contours for the combination of Train 1 and bunkering facility, average of both bunkering and no bunkering phases. Date : 2013-06-11 Page 14 of 39

Figure 7 Risks contours combination of Train 1 and bunkering facility, average of both bunkering and no bunkering phases. Date : 2013-06-11 Page 15 of 39

5.2 Risk 1 st party and 2 nd party 5.2.1 Introduction This chapter presents the calculated individual for 1 st and 2 nd parties. For better seeing what contributes to the, the results are presented in the following way: 1. Risk for the area today, due to the existing LNG Base Load Plant (Train 1). The calculated values are compared with the acceptance criteria, ref. Appendix C. 2. The added contribution due to the planned ferry bunkering facility, when no LNG bunkering is taking place. I.e. with 7 barg in the pipeline upstream the ESV at the loading arm, 23 hours a day. The calculated values thus do not represent total during this operational phase, and cannot be compared with the acceptance criteria. 3. The added contribution due to the planned bunkering facility, during bunkering. I.e. 10 barg operating pressure for 1 hour a day. The calculated values thus do not represent total during this operational phase, and cannot be compared with the acceptance criteria. 4. Risk for the area after installation of the ferry bunkering facility, equals to the sum of 1-3 above. The calculated values are compared with the acceptance criteria, ref. Appendix C. 5.2.2 Hours of presence To calculate the different specific exposure to the s expressed in in Table 3 and Table 2, the different hours of presence per group has been assumed as presented in Table 1. Conservatively it has been assessed that the each group is present to all the bunkering taking place during their working days assumed to be 225 days per year. Table 1 Hours of presence per year of 1st party personnel Control room building Operator/Maintenance Location Control room building (indoor) Most exposed process point Control room building (indoor) 1. LNG Base Load Plant (Train 1) Hours of presence (per year) 2. Ferry bunkering facility no bunkering Hours of presence (per year) 3. Ferry bunkering facility during bunkering Hours of presence (per year) 1800 1800 225 360 360 225 1440 1440 225 Total 1800 1800 225 Truck loading (1 person per truck per 1.2h) Ship loading (jetty only during connection and disconnection) Ship deck (during loading only) Truck loading Ship loading Ship loading 438 438 225 122 122 122 671 671 225 Date : 2013-06-11 Page 16 of 39

Ship bridge (indoor fraction 0.75, during loading only) Skangass operators at the ferry jetty Ship loading Ferry loading 671 671 225 450 450 225 Table 2 Hours of presence per year of 2nd party personnel Ferry office workers Location Ferry indoor 1. LNG Base Load Plant (Train 1) Hours of presence (per year) 2. Ferry bunkering facility no bunkering Hours of presence (per year) 3. Ferry bunkering facility during bunkering Hours of presence (per year) 1800 1800 225 Ferry industry workers Ferry outdoor 1800 1800 225 5.2.3 Risk Results Table 3 presents calculated Specific Risk () per working group at the Skangass LNG Base Load plant and ferry bunkering facility (1 st party). During bunkering, 2 Skangass personnel are assumed to be at the jetty inside the bunkering station area, and are therefore included as 1 st party. : - for all groups in all locations is according to Table 1 within the ALARP area or acceptable. No values exceed the maximum criterion of 1E-03 per year. Operators, who are assumed spending 20% of their working time in the process plant and 80% of their time in the control room building, have the highest individual of 2.58E-5 per year. - Adding the ferry bunkering system to the plant involves a slight augmentation of each, especially for the personnel around the loading bay (loading truck area), closest location to the future pipeline s location. Table 5 presents calculated Specific Risk () per working group at the Ferry (2 nd Party): - for all groups in all locations is according to Table 5 within the ALARP area. No values exceed the maximum criterion of 1E-04. - Adding the ferry bunkering system to the plant involves a net augmentation of both 2 nd party. However, when comparing the contribution to the according to the activities running, the phase no bunkering is contributing with about 80%. This is due to the high fraction of time (23/24) for this phase compared to the fraction of time bunkering is taking place (1/24). Date : 2013-06-11 Page 17 of 39

Table 3 specific s for most exposed persons (1st party). Note that s in column pairs 2. and 3. are additional contributions, i.e. they do not represent total during this operational phase, and cannot be compared with the acceptance criteria, ref. appendix C. Control room building Operator/Maintenance Location Control room building (indoor) Most exposed process point Control room building (indoor) 1. LNG Base Load Plant (Train 1) 2. Ferry bunkering facility no bunkering Added contribution 3. Ferry bunkering facility during bunkering Added contribution 4. Combined Risk y 1.18E-05 2.42E-06 1.04E-05 2.14E-06 2.00E-05 5.13E-07 4.22E-05 5.07E-06 4.16E-04 6.50E-06 7.97E-06 5.38E-07 4.47E-05 4.50E-07 4.94E-04 7.49E-06 1.18E-05 1.52E-05 1.04E-06 2.47E-06 2.00E-05 8.48E-07 4.22E-05 1.83E-05 Total 2.17E-05 3.01E-06 1.30E-06 2.58E-05 Truck loading (1 person per truck per 1.2h) Ship loading (jetty only during connection and disconnection) 1 Ship deck (during loading only) Ship bridge (indoor fraction 0.75, during loading only) Truck loading Ship loading Ship loading Ship loading 1.36E-04 6.80E-06 8.00E-05 4.00E-06 9.38E-05 2.41E-06 3.10E-04 1.32E-05 4.45E-05 6.20E-07 9.53E-06 1.33E-07 9.93E-06 1.38E-07 6.40E-05 8.91E-07 4.92E-05 3.77E-06 1.09E-05 8.35E-07 1.33E-05 3.42E-07 7.34E-05 4.95E-06 4.92E-05 2.05E-06 1.09E-05 2.99E-07 1.33E-05 1.28E-07 7.34E-05 2.48E-06 Skangass operators at the ferry jetty Ferry loading *Represent level for 2 hours of exposure to the LNG Load Base plant when operators are at the ferry jetty. 4.45E-05 3.80E-07 * 9.53E-06 4.32E-07 1.08E-04 2.78E-06 1.32E-04 3.59E-06 Date : 2013-06-11 Page 18 of 39

In Table 4 the AIR is calculated based on the Specific Risk () for each working group and the percentage of the total number of people each group constitutes. The AIR is calculated from the formula AIR ( Number of personnel) Number of personnel As shown in Table 4 AIR is still in the ALARP area, compared with the Skangass acceptance criterion (ref. Appendix C), after installation of the ferry bunkering facility. Table 4 Average for 1 st /2 nd party Criteria Total number of exposed individuals LNG Base Load Plant (Train 1) LNG Base Load Plant (Train 1) + ferry bunkering facility AIR 1 st /2 nd party 59 4.56E-06 6.83E-06 Date : 2013-06-11 Page 19 of 39

Table 5 specific s for most exposed persons (2 nd party). Note that s in column pairs 2. and 3. are additional contributions, i.e. they do not represent total during this operational phase, and cannot be compared with the acceptance criteria, ref. appendix C. Ferry office workers Ferry industry workers Location Ferry indoor Ferry outdoor 1. LNG Base Load Plant (Train 1) (per year) 2. Ferry bunkering facility no bunkering Added contribution (per year) 3. Ferry bunkering facility during bunkering Added contribution (per year) 4. Combined y (per year) 1.02E-06 1.7E-07 1.68E-05 8.20E-07 2.32E-05 5.96E-07 2.82E-05 1.63E-06 7.39E-06 1.2E-06 3.99E-06 3.45E-06 1.08E-04 2.77E-06 1.32E-04 7.74E-06 Date : 2013-06-11 Page 20 of 39

5.3 Societal and Risks 3 rd party 5.3.1 Societal Risk for 3 rd party Figure 8 presents the calculated societal for 3 rd party, in terms of F-N curve, before and after having installed the ferry bunkering facility. For better seeing what contributes to the, the results are presented in a similar manner as in the previous chapter: 1. Risk due to the existing LNG Base Load Plant (Train 1) is shown as the light blue, fully drawn line. The curve may be compared with the acceptance criteria, shown as green and red straight lines, ref. Appendix C. 2. The added contribution due to the planned ferry bunkering facility, no LNG bunkering, is shown as the violet, dotted line. The curve may not be compared to the acceptance criteria. 3. The added contribution due to the planned bunkering facility, during bunkering, is shown as the red, dotted line. The curve may not be compared to the acceptance criteria. 4. Risk for the area after installation of the ferry bunkering facility, equal to the sum of 1-3 above, is shown as the dark blue, fully drawn line. The curve may be compared with the acceptance criteria. As can be seen from the figure, the societal is still within the ALARP or acceptable area after having installed the ferry bunkering facility. Figure 8 F-N curves for societal 3 rd party, DNV inputs Date : 2013-06-11 Page 21 of 39

5.3.2 Hours of presence To calculate the different specific exposure to the s expressed in intable 7, the different hours of presence per group has been assessed as presented in Table 6. Conservatively it has been assessed that the groups Energiveien+Risavika both, Container area both, Rest companies both are present to the entire bunkering operation duration of 2 hours happening during their working days assumed to be 225 days per year, cf. Appendix A of this report. Also, conservatively it has been assumed that the groups Peninsula and Hiking track are present during the bunkering operation and therefore exposed to the. In addition, it has been assessed that the passengers on the ferry or boarding the ferry can be present only during 30% of the bunkering operations, as they cannot be present each evening (cf. Appendix A of this report). Table 6 Hours of presence per year for 3rd party Location 1. LNG Base Load Plant (Train 1) Hours of presence 2. Ferry bunkering facility no bunkering Hours of presence 3. Ferry bunkering facility during bunkering Hours of presence Peninsula Peninsula 280 280 280 Hiking track Hiking track 70 70 70 Ferry Terminal passengers Ferry indoor 219 110 110 Energiveien+Risavika office workers Energiveien indoor 1800 225 225 Energiveien+Risavika industry workers Energiveien outdoor 1800 225 225 Container area office workers Container area indoor 1800 225 225 Container area industry workers Container area outdoor 1800 225 225 Rest companies office workers Rest indoor 1800 225 225 Rest companies industry workers Rest outdoor 1800 225 225 Living quarters (indoor fraction of 0.75) Tananger population (indoor fraction of 0.75) Rest indoor 8760 365 365 Tananger indoor / Tananger outdoor 8760 365 365 Ferry deck Passengers indoor 219 109.5 109.5 Parking area - passengers Passengers outdoor 219 109.5 109.5 Date : 2013-06-11 Page 22 of 39

5.3.3 Risk for 3 rd party Table 7 shows the calculated Specific Risk for 3 rd party, presented similarly as in Table 3. For better seeing what contributes to the, the results are presented in the following way: 1. Risk for the area today, due to the existing LNG Base Load Plant (Train 1). The calculated values are compared with the acceptance criteria, ref. Appendix C. 2. The added contribution due to the planned ferry bunkering facility, when no LNG bunkering is taking place. I.e. with 7 barg in the pipeline upstream the ESV at the loading arm, 23 hours a day. The calculated values thus do not represent total during this operational phase, and cannot be compared with the acceptance criteria. 3. The added contribution due to the planned bunkering facility, during bunkering. I.e. 10 barg operating pressure for 1 hour a day. The calculated values thus do not represent total during this operational phase, and cannot be compared with the acceptance criteria. 4. Risk for the area after installation of the ferry bunkering facility, equals to the sum of 1-3 above. The calculated values are compared with the acceptance criteria, ref. Appendix C. The highest combined is calculated for people present at the North West of the plant (peninsula and hiking track) and in the ferry. However note that all the values are within the ALARP area. The combined values for people on the peninsula, at the hiking track, in the parking area, passengers inside the ferry and the ferry are within the ALARP area. The rest of the categories are in the acceptable area. It must be noted that the main contributor to the for each population is the duration of their presence. Indeed the picture for the different categories except the passengers has been assessed with the conservative assumption made that the same people will be present to each bunkering operations all over the year during their working days (225 working days over a year). It is more likely that the different groups will work based on a shift and therefore could work in the morning instead and not be exposed to the due to the bunkering activity. It appears that the is higher between the bunkering operations than during the bunkering operations for the location parking area, Rest companies, Living quarters. The reasons are twofold: - In an event of a leak between the bunkering operations, the total inventory of the system has been assumed to be released. No leak control such as ESD system is taken into account. - Also, the ignition sources are more present between the bunkering operations in the vicinity of this location. For example, between the bunkering operations, the passengers will have to board the ferry with their cars. Therefore, contribution to the ignition probability from the traffic on the parking and for passengers boarding have Date : 2013-06-11 Page 23 of 39

been taken into account. Thus the ignition probability is higher, generating a greater of fire event. For the Tananger population, the increase, due to the additional ferry bunkering activity, is not significant. The related to the passengers are in the ALARP area but vary according to their location. The passengers on board the ferry are exposed of a lower than the passengers at the. As a consequence DNV recommends to keep the boarding of passengers before the bunkering is taking place as suggested in the ref. Skangass Design Basis /4/. According to the same reference, no passengers are allowed in the passenger tube during bunkering and only late-comers will board through the tube after bunkering. Date : 2013-06-11 Page 24 of 39

Table 7 Risk for 3 rd party, Base Load Plant and ferry bunkering station combined. Note that s in column pairs 2. and 3. are additional contributions, i.e. they do not represent total during this operational phase, and cannot be compared with the acceptance criteria, ref. appendix C. Location 1. LNG Base Load Plant (Train 1) 2. Ferry bunkering facility no bunkering Added contribution 3. Ferry bunkering facility during bunkering Added contribution 4. Combined y Peninsula Peninsula 3.48E-05 1.11E-06 2.61E-06 8.34E-08 3.30E-06 1.06E-07 4.07E-05 1.30E-06 Hiking track Ferry Terminal passengers Energiveien+Risavika office workers Energiveien+Risavika industry workers Container area office workers Container area industry workers Rest companies office workers Rest companies industry workers Living quarters (indoor fraction of 0.75) Tananger population (indoor fraction of 0.75) Hiking track Ferry indoor Energiveien indoor Energiveien outdoor Container area indoor Container area outdoor 1.95E-04 1.56E-06 3.06E-05 2.45E-07 4.32E-05 3.45E-07 2.69E-04 2.15E-06 1.02E-06 6.53E-08 3.99E-06 8.98E-08 2.32E-05 5.55E-07 2.82E-05 7.10E-07 Negl. Negl. 8.26E-10 1.70E-10 Negl. Negl. 1.58E-09 3.25E-10 7.55E-09 1.55E-09 8.26E-09 1.70E-09 Negl. Negl. 1.58E-08 3.25E-09 1.03E-08 2.12E-09 2.04E-08 4.19E-09 1.58E-07 4.07E-09 1.89E-07 1.04E-08 1.03E-07 2.12E-08 1.16E-07 2.38E-08 1.11E-06 2.86E-08 1.33E-06 7.36E-08 Rest indoor 2.33E-09 4.79E-10 3.98E-10 Negl. Negl. Negl. 2.80E-09 5.62E-10 Rest outdoor 2.32E-08 4.77E-09 3.98E-09 8.18E-10 7.41E-10 Negl. 2.79E-08 5.60E-09 Rest indoor 2.33E-09 7.55E-09 3.98E-10 1.29E-09 Negl. Negl. 2.80E-09 8.85E-09 Tananger indoor / Tananger outdoor 1.78E-08 (indoor) / 1.78E-07 (outdoor) 5.80E-08 2.25E-09 (indoor) / 2.25E-08 (outdoor) 7.31E-09 1.97E-09 (indoor) / 1.97E-08 (outdoor) Negl. 2.21E-08 (indoor) / 2.21E-07 (outdoor) (per year) 6.55E-08 Date : 2013-06-11 Page 25 of 39

Ferry deck Parking area passengers Location Passengers indoor Passengers outdoor 1. LNG Base Load Plant (Train 1) 2. Ferry bunkering facility no bunkering Added contribution 3. Ferry bunkering facility during bunkering Added contribution 4. Combined y 4.00E-07 2.18E-08 2.51E-06 5.16E-08 1.29E-04 6.12E-07 3.19E-05 6.85E-07 2.38E-06 5.95E-08 3.54E-06 4.43E-08 1.89E-06 2.37E-08 7.81E-06 1.27E-07 (per year) Date : 2013-06-11 Page 26 of 39

In Table 4 the AIR is calculated based on the Specific Risk () for each working group and the percentage of the total number of people each group constitutes. The AIR is calculated from the formula: AIR ( Number of personnel) Number of personnel As shown in Table 8 AIR is up from acceptable to ALARP area, compared with the Skangass acceptance criterion (ref. Appendix C), after installation of the ferry bunkering facility. Table 8 Average specific 3 rd party ferry bunkering station Criteria Total number of exposed individuals LNG Base Load Plant (Train 1) LNG Base Load Plant (Train 1)+ ferry bunkering system AIR 3 rd party 10 531 4.67E-08 1.37E-07 Date : 2013-06-11 Page 27 of 39

6 SENSITIVITIES The results provided in section 5 are based on DNV recommendation corresponding to a total duration of 90 seconds (60s to detect and initiate ESD and 30s for the ESD to close). However, Skangass estimates that it is possible to activate the ESD system in a shorter time of 36s (30s to detect and initiate ESD and 6s for the ESD valve to close). Even if a closing time of 6 seconds could be possible if Skangass documents that the facility uses a valve in compliance with this requirement, DNV s experience shows that the detection, initiation time of the ESV is often extended to 60 seconds. Therefore two sensitivities have been developed to investigate the picture for: - A duration of 66 seconds (60s to detect and initiate ESD, as recommended by DNV and 6s for the ESD to close, as suggested by Skangass) - A duration of 36 seconds as estimated by Skangass (30s to detect and initiate ESD and 6s for the ESD to close) For both cases it has been assumed that the consequences of a leak on the pipeline between the bunkering operations will remain the same. Thus only the picture during the bunkering operation will be impacted. The results of these two sensitivities are compared against the results based on the time recommended by DNV. 6.1 Risk 1 st party and 2 nd party For 1 st and 2 nd party the reduction in ESD time reduce the for the Skangass operators and workers at the ferry, For 1 st party, Table 9 and Table 10 show the different individual according to the ESD time. The individual for 2 nd party is presented in Table 11 and Table 12. For the most exposed 1 st party which are the Skangass operators at the ferry, the individual contribution from bunkering is reduced by 62% with initiation and closing time according to Skangass inputs, and the overall LNG-related is reduced with 39%. For the most exposed 2 nd party, which are the workers at the ferry, the individual contribution from bunkering is reduced by approximately 50% with initiation and closing time according to Skangass inputs, and the overall LNG-related is reduced with 36% for outdoor workers and 18% for indoor workers. Table 9 Sensitivity, specific s for most exposed persons (1st party). Note that s in column 3.. are additional contributions, i.e. they do not represent total during this operational phase, and cannot be compared with the acceptance criteria, ref. appendix C. Location 3. Ferry bunkering facility during bunkering Added contribution DNV inputs Date : 2013-06-11 Page 28 of 39 3. Ferry bunkering facility during bunkering Added contribution DNV and Skangass inputs 3. Ferry bunkering facility during bunkering Added contribution Skangass Inputs

Control room building Operator/Maintenance Control room building (indoor) Most exposed process point Control room building (indoor) (per year) 2.00E-05 5.13E-07 1.99E-05 5.12E-07 1.98E-05 5.09E-07 4.47E-05 4.50E-07 4.47E-05 4.50E-07 4.47E-05 4.50E-07 2.00E-05 8.48E-07 1.99E-05 8.46E-07 1.98E-05 8.37E-07 Total 1.30E-06 1.30E-06 0.00E+00 Truck loading (1 person per truck per 1.2h) Ship loading (jetty only during connection and disconnection) 1 Ship deck (during loading only) Ship bridge (indoor fraction 0.75, during loading only) Skangass operators at the ferry jetty Truck loading Ship loading Ship loading Ship loading Ferry loading 9.38E-05 2.41E-06 9.36E-05 2.40E-06 9.30E-05 2.39E-06 9.93E-06 1.38E-07 9.93E-06 1.38E-07 9.93E-06 1.38E-07 1.33E-05 3.42E-07 1.33E-05 3.42E-07 1.33E-05 3.42E-07 1.33E-05 1.28E-07 1.33E-05 1.28E-07 1.33E-05 1.28E-07 1.08E-04 3.69E-05 1.05E-04 2.71E-06 5.40E-05 1.39E-06 Table 10 Sensitivity, specific s for most exposed persons (1st party). Base Load Plant and ferry bunkering station combined. Control room building Operator/Maintenance Location Control room building (indoor) Most exposed process point Control room building (indoor) 4.Combined Risk DNV inputs 4.Combined Risk DNV and Skangass inputs 4.Combined Risk Skangass Inputs (per year) 4.22E-05 5.07E-06 4.22E-05 5.07E-06 4.20E-05 5.07E-06 4.94E-04 7.49E-06 4.94E-04 7.49E-06 4.69E-04 7.49E-06 4.22E-05 1.83E-05 4.22E-05 1.83E-05 3.26E-05 1.83E-05 Total 2.58E-05 2.58E-05 2.58E-05 Truck loading (1 person per truck per 1.2h) Ship loading (jetty only during connection and disconnection) 1 Ship deck (during loading only) Ship bridge (indoor fraction 0.75, during loading only) Skangass operators at the ferry jetty Truck loading Ship loading Ship loading Ship loading Ferry loading 3.10E-04 1.32E-05 3.10E-04 1.32E-05 3.09E-04 1.32E-05 6.40E-05 8.91E-07 6.40E-05 8.91E-07 6.40E-05 8.91E-07 7.34E-05 4.95E-06 7.34E-05 4.95E-06 7.34E-05 4.95E-06 7.34E-05 2.48E-06 7.34E-05 2.48E-06 7.34E-05 2.48E-06 1.32E-04 3.59E-06 1.30E-04 3.52E-06 7.82E-05 2.20E-06 Date : 2013-06-11 Page 29 of 39

Table 11 Sensitivity, specific s for most exposed persons (2 nd party). Note that s in column pairs 2. and 3. is additional contributions, i.e. they do not represent total during this operational phase, and cannot be compared with the acceptance criteria, ref. appendix C. 3. Ferry bunkering facility during bunkering Added contribution Location DNV inputs Ferry office workers Ferry industry workers Ferry indoor Ferry outdoor (per year) 3. Ferry bunkering facility during bunkering Added contribution DNV and Skangass inputs 3. Ferry bunkering facility during bunkering Added contribution Skangass Inputs 2.32E-05 5.96E-07 2.26E-05 5.81E-07 1.19E-05 3.04E-07 1.08E-04 2.77E-06 1.05E04 2.70E-06 5.39E-05 1.38E-06 Table 12 Sensitivity, specific s for most exposed persons (2 nd party). Base Load Plant and ferry bunkering station combined. Ferry office workers Ferry industry workers Location Ferry indoor Ferry outdoor 4.Combined Risk DNV inputs 4.Combined Risk DNV and Skangass inputs Individua l 4.Combined Risk Skangass Inputs 2.82E-05 1.63E-06 2.76E-05 1.61E-06 1.69E-05 1.33E-06 1.32E-04 7.74E-06 1.29E-04 7.67E-06 7.81E-05 4.96E-06 Date : 2013-06-11 Page 30 of 39

6.2 Societal and 3 rd party 6.2.1 Societal Figure 9, Figure 10 and Figure 11 show the societal according to the different ESD time. If the detection and closing time is reduced then the societal due to the bunkering operation is slightly reduced. The limited reduction observed is because the LNG Base Load Plant, the bunkering facility and the LNG pipeline to the jetty are the main contributors, and the change from the loading operation itself is relatively small. Figure 9 F-N curves for societal 3 rd party, DNV inputs Date : 2013-06-11 Page 31 of 39

Figure 10 Sensitivity F-N curves for societal 3rd party, Mixed inputs Figure 11 Sensitivity F-N curves for societal 3rd party, Skangass inputs Date : 2013-06-11 Page 32 of 39

6.2.2 Table 13 and Table 14 show the different individual according to the different ESD total time. We can see that the individual is reduced when the ESD initiation and closing time is reduced. The different remain in the same order of magnitude. When comparing between the DNV recommendations (total time of 90s) and a shorter time for closing the valve as in the mixed inputs, a reduction in level is observed. Applying the Skangass data for initiation and closing, the total reduction in LNG related is 39% for passengers inside the ferry and about 47% for passengers on board the ferry. There is hence a significant reduction potential in in actions to reduce the initiation of the ESD system by for example designing a manifold location (cf. section 8) allowing an easy activation by the operators supervising the operation. The reduction in the related to bunkering itself is reduced with app. 50% with the Skangass data for initiation and closing of ESD. Table 13 Sensitivities - Risk for 3 rd party, Note that s in column 3. is additional contributions, i.e. they do not represent total during this operational phase, and cannot be compared with the acceptance criteria, ref. appendix C. Location 3. Ferry bunkering facility during bunkering Added contribution DNV inputs Date : 2013-06-11 Page 33 of 39 3. Ferry bunkering facility during bunkering Added contribution DNV and Skangass inputs 3. Ferry bunkering facility during bunkering Added contribution Skangass Inputs Peninsula Peninsula 3.30E-06 1.06E-07 3.30E-06 1.06E-07 3.30E-06 1.06E-07 Hiking track Ferry Terminal passengers Energiveien+Risavika office workers Energiveien+Risavika industry workers Container area office workers Container area industry workers Rest companies office workers Rest companies industry workers Living quarters (indoor fraction of 0.75) Tananger population (indoor fraction of 0.75) Ferry deck Hiking track Ferry indoor Energiveien indoor Energiveien outdoor Container area indoor Container area outdoor 4.32E-05 3.45E-07 4.32E-05 3.45E-07 4.32E-05 3.45E-07 2.32E-05 5.55E-07 2.26E-05 5.41E-07 1.19E-05 2.80E-07 Negl. Negl. Negl. Negl. Negl. Negl. Negl. Negl. Negl. Negl. Negl. Negl. 1.58E-07 4.07E-09 1.10E-07 2.82E-09 1.65E-08 4.24E-10 1.11E-06 2.86E-08 8.70E-07 2.23E-08 1.27E-07 3.26E-09 Rest indoor Negl. Negl. Negl. Negl. Negl. Negl. Rest outdoor 7.41E-10 Negl. 2.26E-10 Negl. 2.25E-10 Negl. Rest indoor Negl. Negl. Negl. Negl. Negl. Negl. Tananger indoor / Tananger outdoor Passengers indoor 1.97E-09 (indoor) / 1.97E-08 (outdoor) 2.67E-10 1.97E-09 (indoor) / 1.97E-08 (outdoor) 2.67E-10 1.97E-09 (indoor) / 1.97E-08 (outdoor) 2.67E-10 1.29E-04 6.12E-07 2.83E-05 6.05E-07 1.50E-05 2.91E-07

Parking area passengers Location Passengers outdoor 3. Ferry bunkering facility during bunkering Added contribution DNV inputs 3. Ferry bunkering facility during bunkering Added contribution DNV and Skangass inputs 3. Ferry bunkering facility during bunkering Added contribution Skangass Inputs 2.07E-06 2.59E-08 2.07E-06 2.59E-08 1.14E-06 1.42E-08 Table 14 Sensitivities - Risk for 3 rd party, Base Load Plant and ferry bunkering station combined. Location 4.Combined Risk DNV inputs 4.Combined Risk DNV and Skangass inputs 4.Combined Risk Skangass Inputs Peninsula Peninsula 4.07E-05 1.30E-06 4.07E-05 1.30E-06 4.07E-05 1.30E-06 Hiking track Ferry Terminal passengers Energiveien+Risavika office workers Energiveien+Risavika industry workers Container area office workers Container area industry workers Rest companies office workers Rest companies industry workers Living quarters (indoor fraction of 0.75) Tananger population (indoor fraction of 0.75) Ferry deck Parking area passengers Hiking track Ferry indoor Energiveien indoor Energiveien outdoor Container area indoor Container area outdoor 2.69E-04 2.15E-06 2.69E-04 2.15E-06 2.69E-04 2.15E-06 2.82E-05 7.10E-07 2.76E-05 6.96E-07 1.69E-05 4.35E-07 1.58E-09 3.25E-10 1.58E-09 3.25E-10 1.58E-09 3.25E-10 1.58E-08 3.25E-09 1.58E-08 3.25E-09 1.58E-08 3.25E-09 1.89E-07 1.04E-08 1.40E-07 9.12E-09 4.72E-08 6.73E-09 1.33E-06 7.36E-08 1.09E-06 6.73E-08 3.46E-07 4.83E-08 Rest indoor 2.80E-09 5.62E-10 2.75E-09 5.61E-10 2.75E-09 5.61E-10 Rest outdoor 2.79E-08 5.60E-09 2.74E-08 5.59E-09 2.74E-08 5.59E-09 Rest indoor 2.80E-09 8.85E-09 2.75E-09 8.84E-09 2.75E-09 8.84E-09 Tananger indoor / Tananger outdoor Passengers indoor Passengers outdoor 2.21E-08 (indoor) / 2.21E-07 (outdoor) 6.55E-08 2.21E-08 (indoor) / 2.21E-07 (outdoor) 6.55E-08 2.21E-08 (indoor) / 2.21E-07 (outdoor) 6.55E-08 3.19E-05 6.85E-07 3.12E-05 6.78E-07 1.79E-05 3.65E07 7.99E-06 1.30E-07 7.99E-06 1.30E-07 7.06E-06 1.18E-07 Date : 2013-06-11 Page 34 of 39

6.3 Sensitivities discussion As stated in section 6.2.2, the reduction in ESD involves a reduction of the level. This is most notably observed for the workers, both Skangass jetty operators and ferry dedicated workers, and the 3 rd party passengers on the ferry and at the. The for these populations are in the ALARP area, and reducing measures shall hence be evaluated a cost benefit perspective. DNV therefore recommends taking action to reduce as much as practicable the ESD total initiation and closing time to reduce the inventory released. And as presented in the section 8, Skangass will get benefit if the ignition sources are controlled during the bunkering as it will reduce the ignition probability of a drifting cloud and is likely to reduce the overall level. It should be noted that level reduction is not linearly proportional to the ESD total time reduction, which again is proportional to the volume of LNG released. Indeed, while the time is reduced by 60%, the of the passengers in the ferry is reduced by 39%. Hence, the observed reduction is smaller than the actual reduction in closing time for the ESD valve. This result is considered as normal. The reasons are multiple. - In order to ignite the gas cloud has to be within the Lower Flammability Limit (LFL) and Upper Flammability Limit (UFL) when it is exposed to the ignition source. A larger gas cloud will also represent a larger extension of the cloud which is above UFL and hence not ignitable. Hence, some of the ignition sources that for a smaller cloud were exposed to concentration below UFL may now be outside this envelope. - If the larger gas cloud that is built up by longer initiation time and closing time does not expose additional ignition sources, the increase in ignition probability will be less. - The model is based on modelling of selected representative scenarios and a representative wind direction and drifting patterns. If the wind direction and release scenarios result in gas dispersion that have a tendency to expose areas with few strong ignition sources, a growth in gas cloud size will not result in a proportionally equivalent increase in level.. Date : 2013-06-11 Page 35 of 39