Stacy Nelson, Total E&P UK, Margaret Caulfield, Vectra Group, and Andrew Cosham, Atkins Boreas Summary The management of onshore-pipeline safety in the United King- dom is governed by the Pipelines Safety Regulations, 1996. This requires pipeline operators to design, build, and operate pipelines to ensure that they are safe, so far as is reasonably practicable. This is achieved, in part, by applying good practice and designing the pipeline to recognized onshore-pipeline design codes. Where the proposed design falls outside code guidance, a quantitative risk assessment (QRA) provides an effective means of demonstrating a safe design. This paper describes a study that was carried out to define the design requirements for the onshore section of a multiphase pipeline. The proposed pipeline will have a design pressure of 380 barg. However, onshore design codes PD 8010-1 (2004) and IGE/TD/1 (Steel Pipelines 2001) do not provide guidance on the design of pipelines with pressures exceeding 100 barg. Experience from other projects indicates that it is important to consider the implications of such high design pressures early in a project. The QRA considered a number of different design options. The safety risks of the proposed pipeline options were compared with those from a typical 100-barg-gas-transmission pipeline of the same diameter. This means that the risk of the proposed pipeline can be compared in a relative manner, to an acceptable design, and in an absolute manner to limits on individual and societal risk. The results of QRAs of the different design options are sum- marized and discussed. The effects of mitigation measures, such as reducing the pipeline-design factor (the ratio of the hoop stress to the specified minimum yield strength), increasing the wall thick- ness, and incorporating a pressure-limiting system at the landfall, are illustrated. It is shown that the proposed design is feasible. The individual and societal risks for the proposed design were lower than that of a typical 100-barg-gas-transmission pipeline. The individual risk was in the broadly accepted region of risk, as defined by the UK Health and Safety Executive (HSE), and the societal risk was below the acceptance criterion given in IGE/TD/1 (Steel Pipelines 2001). Introduction In the United Kingdom, the management of pipeline safety is governed by the Pipelines Safety Regulations, 1996. These regu- lations consist of goal setting and encompass a risk-based ap- proach to safety. They require pipeline operators to design, build, and operate pipelines to ensure that they are safe, so far as is reasonably practicable (Chatfield 2005). The safety of the pipeline is achieved in part by applying recognized onshore-pipeline design codes. However, in some cases, there is a need to demonstrate the safety of the pipeline by quantifying the risks associated with the pipeline installation and ensuring that the risks in the vicinity of the pipeline are as low as reasonably practicable. This paper describes a case study that was conducted at the early stages of a development project to identify the risks associ- ated with a high-pressure onshore pipeline. The development in- volved a gas field located approximately 125 km northwest of the Shetland Islands, UK. The produced gas, wet natural gas of ap- proximately 86% methane, was assumed to be transported to a gas-processing plant on the Shetland Islands by two new 16-in. import pipelines. The pipelines were assumed to have a design pressure of 380 barg, equal to the wellhead shut-in pressure from the gas field. However, the normal operating pressure was ex- pected to be 63 barg. Two main import routes were investigated by the project. One route involved a mainly offshore route, with a short onshore section where it came into the gas plant. The other route, which is the subject of this paper, included a 19-km section of onshore pipeline. The UK has approximately 7,000 km of high-pressure onshore gas-transmission pipelines operating in the range of 70 to 85 barg. There are several precedents for short onshore gas pipelines with design pressures greater than 100 barg [e.g., the Central Area Transmission System (CATS) pipeline, the Interconnector Pipe- line from Belgium to the UK, and the BBL Pipeline from The Netherlands to the UK]. (The definition of the boundary between onshore and subsea pipelines in PD 8010-1 means that all subsea pipelines that go to an onshore terminal have a short onshore section.) However, the proposed pipeline is atypical because of the high pressure (up to 380 barg) and the relatively long onshore sections (up to 19 km) that will traverse through areas that are not within, or adjacent to, a gas-processing terminal. In view of the potentially sensitive nature of such a high- pressure pipelinethe history of the Corrib development (Corrib Gas Development 2005) is instructive in this regardit was con- sidered important to investigate the implications of the proposed onshore pipeline in the early stages of the project. The results of this study would inform the decision on which potential routes to take forward to detailed design. Consequently, a QRA was con- ducted of various options for the proposed onshore pipeline. It is normally unusual to conduct this level of risk analysis at such an early stage of a project. The main aspects of the case study are described in the paper: Identification of the main requirements of the pipeline- design codes Definition of the proposed pipeline-design cases Description of the methodology used for conducting the QRA Discussion of the results obtained for the individual and so- cietal risk Onshore-Pipeline Design Codes. In securing compliance with the regulations, the UK HSE expects duty holders to apply relevant good practice in design, and safe management principles through- out operation. Within the UK, the main pipeline codes recognized by the HSE as being good practice are IGE/TD/1 Edition 4: 2001, Steel Pipelines for High Pressure Gas Transmission (Steel Pipelines 2001) PD 8010-1: 2004, Code of Practice for PipelinesPart 1: Steel Pipelines on Land (PD 8010-1 2004) BS EN 1594: 2000, Gas Supply SystemsPipelines for Maximum Operating Pressure Over 16 barFunctional Require- ments (BS EN 14161 2003). The PD 8010-1 and IGE/TD/1 pipeline codes were identified as being most relevant for application to the proposed pipeline. These codes specify a system of location class, limits on the hoop stress (design factor), and minimum distances from normally occupied buildings to control the risk imposed by the pipeline on the sur- rounding population. IGE/TD/1 is applicable to the design of onshore pipelines for the transmission of dry natural gas (predominantly methane) at a Copyright 2008 Society of Petroleum Engineers This paper (SPE 111672) was accepted for presentation at the Eighth SPE International Conference on Health, Safety & Environment in Oil and Gas Exploration and Production held in Abu Dhabi, UAE, 24 April 2006. Original manuscript received for review 16 No- vember 2007. Paper peer approved 28 January 2008. 1 September 2008 SPE Projects, Facilities & Construction maximum operating pressure not exceeding 100 barg. IGE/TD/1 may be applied to other gases, provided that the characteristics and consequential effects are taken into account (e.g., internal corro- sion for wet gas and fracture propagation for rich gas). PD 8010-1 is applicable to the design of onshore pipelines for the transmission of oil, gas, and other substances that are hazard- ous by the nature of being explosive, flammable, toxic, reactive, or liable to cause harm to persons or the environment. The scope of PD 8010-1 does not quote a maximum pressure, but reference is made to the 100-barg limit in IGE/TD/1. Because PD 8010-1 is applicable to a wide range of different fluids (not just natural gas), it classifies fluids to be transported in pipelines with respect to their potential hazard to public safety (e.g., Category Dnontoxic, single-phase natural gas; and Category Eflammable and/or toxic fluids that are gases at ambient temperature, and at atmo- spheric pressure conditions, and are conveyed as gases or liquids). The closest classification relevant to the proposed (wet) gas de- velopment is Category E. Location classification, population density, and proximity requirements are specified for Category-D and -E fluids. ASME B31.8, a gas-pipeline design code used in the USA and internationally, also specifies a system of location class and limits on the hoop stress, but it does not specify a minimum distance to occupied buildings (ASME B31.8-2003 2003). A supplement to the design code, ASME B31.8S, introduces the concept of a po- tential impact area, which is similar to the proximity distances in PD 8010-1 and IGE/TD/1, but is not a minimum distance (ASME B31.8S 2004). QRA. Risk is influenced by two main factors: (1) the probability of failure and (2) the resulting consequences. The magnitude of the risk is dependent on how likely the event is to happen and how severe the consequences would be if that event were to happen. A risk may be high if one of these factors is high; however, although the consequence of an event may be severe, the resulting risk may be low because of the low probability of that event occurring. The pipeline-design codes aim to reduce the failure probability by stipulating requirements for the mechanical design such as the design factor, wall thickness, and depth of burial. Failure conse- quences are managed by prescribing the separation distance for the pipeline on the basis of the location and population density. The use of QRA provides a means of demonstrating that the risk associated with the pipeline design is acceptable, so far as is reasonably practicable. In a QRA, both the probability of pipeline failure and the consequences of failure are quantified. The risk to the surrounding population is then expressed numerically as an individual risk, in the form of a risk transect, and as a societal risk, commonly in the form of a frequency/consequence (FN) curve. Both individual and societal risks need to be considered when demonstrating that the risk of the proposed onshore pipeline is acceptable or is as low as reasonably practicable (ALARP). The QRA was an integral part of the pipeline-design process. The main drivers for the use of this approach were The pipeline is atypical, and there is no specific guidance in the design codes for pressures in excess of 100 barg. The risk-assessment methodology serves as a useful tool to compare different pipeline-design options and provides a frame- work for deciding on the most appropriate design solution. There are a number of risk-mitigation measures that could be applied to the onshore pipeline to reduce the risk to the population further such as modification to the route or a lower design factor. These also have been considered as part of the case study. A number of assumptions and simplifications were made for the purposes of this QRA because it was conducted in the early stages of project development. These assumptions and simpli- fications would need to be revised in the detailed-design stage of the project. Pipeline-Code Requirements It was acknowledged in the early days of the pipeline industry that the most significant factor contributing to the damage of pipelines was external interference. The original intent of pipeline-design codes was, therefore, to route the pipeline away from the popula- tion to reduce the probability of damage from activities in the vicinity of the pipeline. The requirements of the PD 8010-1 and IGE/TD/1 codes for pipelines conveying Category D and E fluids are in terms of the pipeline design factor; population density, or number of persons per unit area in the vicinity of the pipeline and proximity to occupied buildings. Design Factor. PD 8010-1 and IGE/TD/1 both define the maxi- mum permitted design factor, with respect to hoop stress, on the basis of the location or area through which the pipeline is routed. PD 8010-1 refers to the location class, while IGE/TD/1 refers to the area type. The classification depends on the population density along the route of the pipeline and is similar for both pipeline codes. The area classification and corresponding design factor are presented in Table 1. The design factor is limited to 0.3 in suburban areas because, at this stress level, it is very unlikely that a pipeline will fail as a rupture. The maximum permitted design factor is increased to 0.5 in IGE/TD/1, if the wall thickness is at least 19.05 mm, be- cause it has been shown that it is unlikely that significant damage could be introduced into a pipeline of this thickness. Population Density. The population density is defined in terms of the number of persons per hectare. The population density for each area type in IGE/TD/1 and PD 8010-1 is defined in Table 1. The population density is the average number of persons per unit area, within a 1.6-km strip, centered on the pipeline. PD 8010-1 states that if individual-risk contours are available, then the width of this strip should be based on the 0.3-chance-per-million (cpm) indi- vidual-risk contour. In the absence of individual-risk contours, PD 8010-1 uses a building-proximity distance (BPD) in a manner similar to IGE/TD/1. Here, the population density is defined as the average number of persons per unit area along a 1.6-km strip, centered on the pipeline, with a width equal to eight times the BPD for a Type-R-area pipeline; see Fig. 1. 2 September 2008 SPE Projects, Facilities & Construction Proximity to Occupied Buildings. The design codes require that there be a minimum distance between the pipeline and occupied buildings. PD 8010-1 states that if individual-risk contours are available, then the distance between the pipeline and occupied buildings should not be less than the distance to the 10-cpm indi- vidual-risk contour. In the absence of individual-risk contours, PD 8010-1 uses the BPD to define the requirements. The BPD speci- fies the minimum distance from the pipeline to normally occupied buildings (i.e., there should be no normally occupied buildings within 1 BPD of the pipeline); see Fig. 1. Proximity distances are different for different fluids, to reflect their potential to cause a hazard. For pipelines transporting Cat- egory-D fluids, PD 8010-1 refers to the BPD in IGE/TD/1. This gives BPDs for pipelines in rural areas and suburban areas. The BPDs for suburban areas are smaller than those for rural areas, even though the population density around the pipeline is higher, because the maximum permitted design factor in suburban areas is lower, so the probability of failure is lower. For Category-E fluids, PD 8010-1 defines the equivalent of a BPD with an equation that includes pressure, diameter, and a substance factor. The substance factor is related to the hazard potential of the fluid. The BPD depends on the maximum operating pressure and the diameter of the pipeline; the BPDs for a single 16-in.-diameter natural-gas pipeline in a rural area are shown by Line F in Fig. 2. The BPDs are based on considerations of risk and can be related to the thermal radiation from an ignited full-bore release. The scope of IGE/TD/1 is limited to a maximum operating pressure not ex- ceeding 100 barg because BPDs have not been calculated for higher pressures. Pipeline-Design Definition The development of the case study for the pipeline design first involved an initial definition of the onshore-pipeline routes. This is an iterative process because the pipeline-design parameters (e.g., pressure and design factor) have implications for the route, and vice versa. To summarize, The diameter and design pressure of the proposed pipe- line were used to estimate the minimum BPD required by the design code. Potential routes were identified by looking at maps and undertaking a site survey, with regard to the BPD and the popu- lation density. A QRA was conducted to determine if the proposed routes for the pipelines impose an acceptable risk on the surround- ing populations. Fig. 1Definition of the area for determining the population density along the route of a pipeline. Fig. 2Minimum proximity distance from normally occupied buildings of pipelines in Type-R areas (Steel Pipelines 2001). 3 September 2008 SPE Projects, Facilities & Construction The BPD for the proposed pipeline was determined by use of the equation given in PD 8010-1 (2004). The proposed pipeline is a multiphase pipeline, although the fluid is mostly methane. There- fore, conservatively, the substance factor for natural-gas liquids was used. The proposed pipeline would consist of two 16-in. pipe- lines laid in a common trench (to minimize construction impact). Therefore, the proximity distances were calculated for a single pipeline and for the equivalent of two adjacent pipelines. The estimated minimum proximity distances for a single pipeline are given in Fig. 3. The distances follow a trend similar to that of the radius of the potential impact area, as defined in ASME B31.8S (2004), and are conservative with respect to extrapolating the curves in IGE/TD/1. The publics perception of risk is as influential as engineering calculations and the results of a QRA. It was, therefore, considered prudent, where practically possible, to find a route that used the proximity distances defined in PD 8010-1 because these were believed to be conservative. Pipeline-Route Options. Two route options were considered for the onshore import pipeline. Both onshore routes have the landfall at Sand Wick, on the west coast of the Shetland Islands, with approximately 19 km of onshore pipeline to the Sullom Voe Ter- minal (SVT). The first route option is by way of Urafirth (Route A). The second option is routed further north by way of Upper Urafirth (Route B). The routes are shown as the black lines in Fig. 4. Fig. 3Estimated minimum proximity distance from normally occupied buildings for the proposed pipeline. Fig 4Map showing part of the onshore section of the import pipeline, Sand Wick to SVT by way of Urafirth. 4 September 2008 SPE Projects, Facilities & Construction Note that the red lines signify roads. The red circles on the maps represent clusters of dwellings, and the numbers shown against each of the circles are the number of dwellings when there is more than one. For each route, the most-populated 1-mile section along the pipeline was identified for input to the risk assessment. The worst- case 1-mile onshore section of Route A lies along the section of the pipeline indicated by the bold blue line in Fig. 4. The worst- case 1-mile onshore section of Route B was also found to lie along this section of the pipeline and is indicated by the bold green line in Fig. 4. Pipeline-Design Cases. The definition of the pipeline-design cases was based on the requirements of the PD 8010-1 pipeline code, with reference to IGE/TD/1. The pipeline route is mainly through rural areas; therefore, the design factors were chosen on the basis of that area classification. It was assumed in the study that the depth of cover of the pipelines would be at least 1.1 m, in accor- dance with IGE/TD/1. The options considered in the case study are given in Table 2. The design of the pipeline includes a corrosion allowance of 3 mm (the design factors in Table 2 do not include the corrosion allowance). For the risk assessment, the corrosion allowance was included to reflect the true pipeline proposed for the study. This corrosion allowance will be valid at the start of the lifetime of the pipeline, but as the pipeline ages, there may be internal corrosion, which reduces the wall thickness. To consider the effect of the re- duced wall thickness on the failure frequency, a sensitivity study was undertaken, and the impact on failure frequency and subsequent risk was investigated. This is discussed in QRA Methodology. The gas-transportation system in the UK consists of national and local transmission systems. The national transmission system has pipelines with diameters of 18 to 48 in. that operate at pres- sures typically from 70 to 85 barg. These large-diameter pipelines typically operate at a design factor of 0.72. The approach to the risk assessment was to compare the risk of each of the proposed pipeline-design cases with a typical pipeline of the same diameter that is unambiguously within the design code. The MAOP of the reference pipeline was chosen at 100 barg because this was the highest pressure covered by IGE/TD/1. This then means that the risk of the proposed pipeline can be compared in a relative manner to an acceptable design, and in an absolute manner to limits on individual and societal risk. The installation of a pressure-limiting system at the landfall was considered as part of the design for inclusion in the risk assessment. The onshore pipeline would be designed to withstand the full wellhead shut-in pressure of 380 barg. The pressure- limiting system would provide additional protection by preventing the downstream onshore pipeline from being exposed to pressures higher than a predetermined value. Two set pressures were con- sidered for the risk assessment: 233 barg (Case 3) and 140 barg (Case 4). The reduction in onshore pressures for these cases re- sulted in a decrease in the design factor because the wall thickness was maintained the same as it was for the 380-barg-design case. QRA Methodology The methodology employed for the QRA involved a number of discrete steps that are illustrated in Fig. 5. The first step in the QRA process requires the identification of the complete set of applicable hazards and hazardous events for subsequent analysis. It is then necessary to perform the following assessments for these identified hazards: Determine the frequency of occurrence of the initiating event. Develop potential scenarios showing the possible outcomes of each initiating event. Analyze consequences (e.g., release rates, vapor-dispersion distances, fire effects) for each of the hazardous events. Assess impact in terms of the effects of these consequences on people and occupied buildings. Summarize risk, combining the outputs of the above steps to procure estimates of individual- and societal-risk levels. Compare the assessed risk levels against established risk cri- teria, and where necessary, demonstrate ALARP. Consider additional risk-reduction measures where neces- sary, as indicated in the feedback loop. The application of these steps to the onshore-pipeline assess- ment is described in the subsequent five sections. To summarize, the QRA considered the risk by thermal radiation from the imme- diate ignition of ruptures caused by external interference. Hazard Identification. A hazard-identification study was carried out as part of the development project to identify all possible hazards and hazardous events for the onshore pipeline. The dom- inant type of hazard for the pipeline is the occurrence of loss of containment, resulting in a release of a pressurized hydrocarbon fluid. Natural gas is flammable, and ignition of a release would lead to fire hazards, which could impact buildings and personnel within the hazard zone because of thermal radiation. In theory, dispersing gas could enter buildings along the route, leading to gas buildup to flammable concentrations and possible explosion hazards should an ignition source be present. The gas conveyed by the pipeline is nontoxic, and as such, the risk asso- ciated with the release of a toxic vapor has been dismissed. Frequency Analysis. The probability of failure of a pipeline is determined by considering all credible failure modes or threats to the integrity of the pipeline. These failure modes could include External interference (mechanical damage) Corrosion Fatigue Weld defects Pipe-body (manufacturing) defects Ground movement Natural events (e.g., earthquakes) Human error The main causes of pipeline failures in the UK and Europe are external interference and corrosion (Browne and Hicks 2005; 6 th EGIG-Report 2005). External interference is more likely to cause a rupture than corrosion (Browne and Hicks 2005; Pipeline Performance in Alberta 1998). Therefore, for the purposes of the QRA, it was assumed that other failure modes would not contrib- ute significantly to the overall risk, and that only failures caused by external interference need to be considered. This assumption would need to be revisited in the detailed-design stage. Ruptures are rare events and difficult to use in a predictive manner. However, where damage data have been recorded, the failure frequency can be calculated using a combination of histori- cal data with statistical techniques or detailed structural-reliability 5 September 2008 SPE Projects, Facilities & Construction analyses. An analysis of the United Kingdom Onshore Pipeline Operators Association (UKOPA) failure data was used as the basis for obtaining the failure frequency of the proposed import pipeline (Mather et al. 2001; Browne and Hicks 2005). This analy- sis subdivides the data to a greater level of detail than is normally published and uses statistical techniques to fill in gaps for pipeline geometries for which failure has not, to date, been reported. Factors were then applied to take account of wall thickness, depth of cover, and location type. These factors were obtained from the analysis of the UKOPA failure data and by use of a predictive method that combines historical data with defect failure models (Corder and Fearnehough 1987). A comparison with fail- ure frequencies derived by use of structural-reliability analysis showed that the frequencies derived from a statistical analysis of the historical data were reasonable; see Fig. 6. The latter were used in the QRA. Fig. 5QRA methodology. Fig. 6Estimated failure frequency of the proposed pipeline. (Pipe diameter given in inches, wall thickness given in millimetres.) 6 September 2008 SPE Projects, Facilities & Construction The proposed pipeline comprises two 16-in. pipelines. There- fore, the calculated failure frequency has been doubled. Consequence Analysis. The consequence analysis involves the identification of the potential types of release, determination of the release rates for each scenario, and an assessment of the type of hazardous event that might occur subsequent to the release. Gas Releases. Releases from leaks can occur only up to some critical size before they propagate to produce a full-line rupture. The analysis methodology calculated equivalent circular-hole sizes for critical-length leaks. Smaller-leak and pinhole failures would result in a steady-state jet fire. Because of the depth of cover, it is considered that any leak or pinhole failure will result in a vertical release. Following a major failure such as a rupture, the pipeline will depressurize as gas is released. The release will initially form a toroidal, or mushroom-shaped, cloud rising at the head of a tran- sient jet, which quickly disappears, to leave the jet fed from the continuing release from the two pipe ends. The pressure in the pipes will fall, rapidly at first, until a steady-state condition is reached where gas continues to be supplied, or it falls to zero where the gas supply is stopped. The time at which steady-state- flow conditions are reached will depend on the position of the isolation valves and, thus, the lengths of pipe involved. The transient gas outflow from the pipeline was calculated with a consequence-assessment package over the range of cases to be considered in the risk assessment. The import pipeline was taken to be 121 km in length, with the position of the rupture at 100 km from one end. Crater Formation. If an underground pipeline fails as a rup- ture, a crater will be formed as a result of soil being ejected as debris while the stored pressure in the pipeline is released. It is possible that some items of debris could be ejected as missiles and travel for considerable distances. The effect of these would, how- ever, be localized. The shape of the crater, which acts as the source for the sub- sequent release, is a function of the pipeline diameter, depth of cover, operating pressure, soil type, failure length, and alignment of the pipe ends. The standard assumption for a modern high- toughness pipeline is that the failure will be arrested within ap- proximately two pipe lengths. It is also assumed that the pipe ends at the point of failure will remain anchored in the ground and remain aligned. The crater shape for the normal range of depth of cover will then approximate to a semi-ellipse with steep, or even incurved, sides at the ground surface. The jets issuing from the two pipe ends will interact within the crater and exit the crater as a vertical jet that can then be affected by the wind. Gas Dispersion. The gas released from the crater will form a high-velocity momentum-driven jet initially directed vertically into the atmosphere, where it will be influenced by the wind. The effective diameter of the source of this vertical jet has been taken as 10 pipeline diameters, which is consistent with data from incidents and crater-formation models (Hopkins et al. 1993). The dispersion of this release was simulated with the consequence- assessment package. All the simulations indicated that the disper- sion is momentum-dominated in such a way that the gas will mix with the surrounding atmosphere and disperse to the lower flammable limit as a jet. The simulations also indicate that the flammable envelope does not extend significantly downwind at ground level. Therefore, it is considered extremely unlikely that gas would enter, or be present in the vicinity of, buildings at hazardous concentrations. Ignition. The hazards from a pipeline failure are dominated by those associated with a locally ignited release. Ignition of a release from a pipeline rupture can result from a range of sources, both close to and remote from the release, and could occur immediately or following a delay. Because of the nature of the release, with an initial highly transient phase followed by a slowly decaying (quasi steady-state) release, the longer ignition is delayed, the fewer the consequences. If ignition occurs immediately, the gas released during the initial transient (mushroom-cloud) phase will burn as a fireball rising rapidly into the air and burning out within approxi- mately 30 seconds after the failure (Acton et al. 1998). If ignition is delayed, the fireball will not be seen. For the purposes of this study, ignition is assumed to be immediate because this is the more-pessimistic option. An ignition probability of 0.5 was assumed for rupture releases, and a value of 0.1 was assumed for leaks. These values are higher than the European-gas-pipeline-incident-data-group (EGIG) inci- dent data, which suggest 0.3 for ruptures and 0.02 for leaks (6 th EGIG Report 2005). The ignition probability is, however, likely to increase with increasing pressure, and it is considered that the values chosen are appropriate for the proposed import-pipeline pressure and diameter. Impact AssessmentThermal-Radiation Effects. The assess- ment of the impact of a fire is based on two hazard distances derived from a prediction of the thermal-radiation hazards of an ignited release: The building-burning distancethe distance to which build- ings could be ignited, assuming that this will occur as a result of piloted ignition The escape-to-shelter distancethe distance at which people out of doors could reach shelter beyond the building- burning distance Ignition of combustible material on buildings or structures can also be caused by intense thermal radiation, although this again is dependent on the thermal-radiation flux level and the time of ex- posure. The threshold for buildings exposed to thermal radiation is taken as the thermal-radiation dose, at which secondary fires may be started by piloted ignition of combustible materials (this incor- porates a minimum threshold of 12 kW/m 2 ) (Acton et al. 1998). To evaluate the escape-to-shelter distance, it has been assumed that an individual outdoors will be a casualty if he/she receives a total thermal dose of 1,060 thermal dose units (tdu), representative of a 1% lethality rate (Hymes et al. 1996). The duration of expo- sure for an individual outdoors was 30 seconds, which was con- sidered to be a reasonable time for a person to reach shelter or escape from the incident. For the immediate-ignition cases, the thermal dose was calculated as a combination of the dose received during the fireball, plus the dose received by the jet fire, up to a total exposure time of 30 seconds. The worst-case assumption of immediate ignition was made for each scenario. The hazard distances computed from the simulation of a rup- ture in the proposed pipeline are given in Table 3. Table 4 gives 7 September 2008 SPE Projects, Facilities & Construction the hazard distance for a leak for the case with the largest flow rate. These are the hazard distances for the failure of a single pipeline. The proposed pipeline layout comprises two 16-in. pipelines. In the QRA, it has been assumed that the failure of one pipeline will not cause the failure of the adjacent pipeline. On the basis of an average escape velocity for an able-bodied person of approximately 2.5 m/s, it is clear that, within 30 seconds, such an individual will be able to escape beyond the distances quoted in Table 4. An additional analysis was performed for the leak corresponding to the maximum escape distance for credible puncture fires, to determine the dose received by an individual escaping from the release at 2.5 m/s. This analysis indicated that the individual would receive only 339 tdu, escaping to a radiation level of 4 kW/m 2 (i.e., significantly less than the 1,060 tdu used for the 1% lethality level). Therefore, it can be seen that at the escape velocity, the thermal dose received by an able-bodied person would not be sufficient to contribute to the risk from the pipeline (i.e., those exposed would be able to escape before becoming casualties). Therefore, it is also the case that for buildings located within the building-burning distance, their occupants would be able to escape to safety before becoming casualties. There are no buildings within the burning distances given in Table 4. The impact assessment indicates that leaks do not contribute significantly to the risk at the populated locations, and hence, the risk assessment has been restricted to ruptures. Calculation of Individual and Societal Risk. Individual Risk. Individual risk is defined as the risk of becoming a casualty to an individual who is present at a given location (either indoors or outdoors, or a combination of the two) for 365 days/yr. The indi- vidual risk will be a function of the distance of the location from the pipeline and the proportion of time that the occupant spends indoors/outdoors at the location. It has been assumed that a house occupant is present 100% of the time with 10% of the time out- doors, the standard used in the UK (Reducing Risks 2001). The variation of individual risk at locations in the vicinity of the pipeline is determined primarily by the length of pipeline that could potentially impact that location following a failure, usually referred to as the interaction length. Fig. 7 illustrates the concept of the interaction length for building occupants based on the build- ing-burning distance. Where occupants are outdoors, the inter- action length would be based on the escape distance. The pipeline-failure frequency at each step is combined with the interaction length and ignition probability to produce the in- dividual risk per year at a particular distance from the pipeline. Individual-risk transects can be generated by considering various distances from the pipeline. Societal Risk. Societal risk is defined as the relationship be- tween the frequency of an incident and the number of casualties that may result. The societal risk can be related to a section of pipeline through particular population clusters, as in the case ex- amined in this paper, or it can be representative of the pipeline as a whole. Where a range of numbers of casualties, N, with their associated frequencies, F, can be calculated, the risks typically are shown in the form of an FN curve where the frequency, F, of N or more casualties is plotted against N. Values of F and N are determined by stepping along each pipeline section of interest and recording the number and classification of buildings that are located within the building-burning and escape-to- shelter distances for each step, as illustrated in Fig. 8. For each step, the number and classification of buildings within the building-burning and escape-to-shelter distances determines N, using specific rule sets and assumptions regarding the occupancy for each building classification. The corresponding frequency, F, for each failure is given by the product of the step length, the rupture frequency per unit length, and the ignition probability. To determine the corresponding values of F and N or more casualties and to plot the FN curve, these values of F have to be summed up, starting with the frequency for the maximum predicted number of casualties, up to and including the frequency for the minimum predicted number of casualties. Risk-Assessment Results and Discussion Individual Risk. The individual risk was calculated with the methodology described in the Calculation of Individual and Societal Risk section for the standard case. This represents an upper limit. It should be noted that these individual-risk calculations are applicable to the pipelines, regardless of the pro- posed route because the individual-risk values are calculated for a hypothetical individual located at a given distance from the re- spective pipeline. The variation of the individual risk with distance from the pipeline for each case is shown in Fig. 9. For all of the pipeline cases, the individual-risk levels varied very little close to the pipe- line because of the geometry governing the reduction in the inter- action length with distance from the pipeline. Beyond the building- burning distance, this value quickly fell to zero. Fig. 9 also dem- onstrates that close to the pipeline, the risks associated with the proposed import-pipeline design (Cases 24) were less than that for the 100-barg-pipeline reference case (Case 1). This is attributed Fig. 7Building-interaction length. 8 September 2008 SPE Projects, Facilities & Construction to the greater wall thickness and, therefore, to the lower design factors of the import pipelines and the subsequent effect on de- creasing pipeline-failure frequency. Farther away from the pipe- line, the risk levels for the reference case become lower in all but one of the proposed design cases because of the lower pressure associated with the reference case. The individual-risk-acceptability criteria were based on the UKs HSE document, Reducing Risks, Protecting People (2001). The individual risk is said to be broadly acceptable if it lies below a level of 110 6 per year, and on the basis of individual risk alone, no further risk-reduction measures would need to be con- sidered as long as current relevant standards are met. If the indi- vidual risk lies between 110 6 and 110 4 per year, then the risk is said to lie in the ALARP region. Further risk-reduction measures need to be considered, and the risk-reduction benefits of the mea- sures need to be compared to their cost. If the individual risk is greater than 110 4 per year, then the risk is said to be unaccept- able, and risk-reduction measures must be undertaken. The indi- vidual-risk values associated with the reference case fall within the ALARP region. However, the individual-risk values obtained for the proposed pipeline cases all lie in the broadly acceptable region of risk. Societal Risk. In assessing the major hazards posed by onshore pipelines, the societal risk is normally of more concern than the individual-risk assessment. However, it is more difficult to define an acceptance criterion for societal risk. An approach to societal- risk criteria for pipelines has been proposed in IGE/TD/1 based on a unit length of pipeline. This is consistent with the risk-control requirements in the design code that determines the location classes on the basis of the population density over a defined length of pipeline (e.g., 1 mile). Fig. 8Site-interaction length. Fig. 9Individual-risk transects for import-pipeline cases. 9 September 2008 SPE Projects, Facilities & Construction The criterion envelope included in Edition 4 of IGE/TD/1 is based on societal-risk calculations for a wide range of pipelines designed to IGE/TD/1 standards, using the risk-assessment meth- odology developed by National Grid (Acton et al. 1998; Hopkins et al. 1993). The QRA described here has been conducted using a methodology similar to that adopted by the National Grid. There- fore, the criterion envelope in IGE/TD/1 can be used to assess the results of the QRA. The societal risk calculation considered only the worst-case 1-mile length of the pipeline for each of the proposed import routes as indicated previously in Fig. 4. The FN curves for each of the cases along Route A (Urafirth) are shown in Fig. 10. The FN curves for pipeline cases along Route B (Upper Urafirth) are shown in Fig. 11. For all the design cases considered, the FN curves lie within the IGE/TD/1 acceptable-risk-criterion curve and can be classified as broadly acceptable. For Route A, the societal risk FN values associated with the proposed import-pipeline-design cases were less than that for the 100-barg-pipeline reference case for casualty values up to N12. Above this N value, the F value of the reference curves dramati- cally falls; however, FN curves continue for the design cases. It is, however, noted that above the reference FN-curve limit, values of the design cases have frequencies only up to 210 7 for the highest design pressure of 380 barg. Reducing the pressure reduces the risk. For Route B, there were no buildings within the building- burning or escape distances for the reference case. For Case 4, the proposed design-case option having the onshore pressure lim- ited to 140 barg, it was not possible to generate an FN curve because the numbers of casualties and associated frequencies were very small. Pressure-Limiting System. The installation of a pressure-limiting system at the landfall was considered as part of the design and included in the risk-assessment process. Two set pressures were considered as part of the risk assessment233 barg (Case 3) and Fig. 10Societal risk FN curve for import-pipeline cases for Route A (Urafirth). Fig. 11Societal risk FN Curve for import-pipeline cases for Route B (Upper Urafirth). 10 September 2008 SPE Projects, Facilities & Construction 140 barg (Case 4). There is a finite probability of the system failing to operate on demand (i.e., when called on to isolate the source of pressure excursion, all or part of the system may not function to achieve the required action). Such a failure would lead to a high pressure in the onshore pipeline downstream of the isolation valves up to the maximum wellhead shut-in pressure of 380 barg. How- ever, since the onshore pipeline is designed to withstand the well- head shut-in pressure, it is not anticipated that pipeline failure will occur because of increased pressure alone. The worst-case conse- quence associated with the pipeline failure would be a rupture of the pipeline as assessed in Case 2. The integrity of the pressure-limiting system is determined by its safety-integrity level (SIL). A SIL-3 system, for example, would have a probability of failure on demand of 10 4 and <10 3 . The probability of a pipeline failing as a result of failure of the pressure-limiting system is dependent on a number of factors: The probability of an overpressure per year The probability of the failure on demand of the pressure- limiting system The probability of there being a defect in the pipeline that will fail when the pressure increases The probability of failure of the pressure-limiting system has not been considered in the QRA. If it were to be included, then the risks associated with Cases 3 and 4 would be slightly higher than quoted. Corrosion-Allowance Sensitivity. The pipeline wall thickness in- cludes a corrosion allowance, and this will be valid at the start of the lifetime of the pipeline. However, as the pipeline ages, there may be internal corrosion that reduces the wall thickness. A re- duction in the wall thickness implies an increase in the failure frequency. This was considered in a sensitivity study for Cases 2, 3, and 4. The sensitivity study concluded that the risks remain broadly acceptable (i.e., below an individual risk of 110 6 per year and less than the FN criteria). Uncertainties Because of High Pressure. There are uncertainties in both the consequences and the risks calculated for pipelines operating at pressures up to 380 barg. These uncertainties will tend to increase with increasing pipeline pressure. The main reason for the uncertainty is that the underlying consequence models have not been validated at such high pressures. Few large-scale fire experi- ments have been performed at pressures greater than 100 barg. There is also uncertainty regarding the behavior of mixtures of hydrocarbons on release at such high pressures. The consequence models may perform satisfactorily for releases of pure methane at pressures outside the validated range, but this may not be the case for mixtures containing heavier hydrocarbons. These uncertainties support the adoption of a cautious approach to the design of the proposed pipeline. This will make it easier to dem- onstrate that the design is safe, so far as is reasonably practicable. Conclusions A QRA of the onshore section of a proposed high-pressure pipe- line has been conducted in the early stages of a development project. A number of design options were considered: different routes, design factors, and onshore-pipeline pressures. Individual and societal risks were calculated for each option. The QRA provided the project with a robust framework for assessing the pipeline-design options and deciding on the most appropriate design solution early in the development of the project. By comparing the risks with those of a typical onshore pipeline, the project was better able to demonstrate that the risk to the onshore population was broadly acceptable and below that of a typical onshore pipeline. Acknowledgments The authors would like to thank Total E&P UK Ltd. and its project partners for their permission to publish this paper. The authors would also like to acknowledge colleagues at Total E&P UK Ltd., Vectra Group Ltd., and Atkins Boreas who have contributed their ideas and technical expertise to the paper, specifically: Dean Henderson, Steve Hall, Tim Turner, Will Sutton, and Harry Hopkins of HH Risk. References 6 th EGIG-Report 19702004, Gas Pipeline Incidents 19702004. 2005. Annual report, Document No. EGIG 05.R.0002, European Gas Pipeline Incident Data Group, The Netherlands (December 2005). Acton, M.R., Baldwin, P.J., Baldwin, T.R. and Jager, E.E.R. 1998. The Development of the PIPESAFE Risk Assessment Package for Gas Transmission Pipelines. Presented at the International Pipeline Confer- ence (IPC 98), Calgary, 711 June. ASME B31.8-2003, Gas Transmission and Distribution Piping Systems. 2003. New York: ASME. ASME B31.8S 2004, Managing System Integrity of Gas Pipelines. 2005. New York: ASME. Browne, D. and Hicks, R. 2005. Pipeline Product Loss Incidents (1962 2004), 4 th Report of the UKOPA Fault Database Management Group. Advantica Report R 8099, FDMG, Derbyshire, UK. BS EN 14161: 2003, Petroleum and Natural Gas Industries. Pipeline Transportation Systems. 2003. London: British Standards Institution (BSI). Chatfield, S. 2005. Decision-making on assessment of high pressure gas transmission pipelines. Health & Safety Executive, http://www.hse. gov.uk/gas/supply/gsmrassess/symposium.pdf. Corder, I. and Fearnehough, G.D. 1987. Prediction of Pipeline Failure Frequencies. Paper ERS E576 presented at the Second International Conference on Pipes, Pipelines, and Pipeline Systems, Utrecht, The Netherlands, June. Corrib Gas DevelopmentOnshore Pipeline Project. 2005. Quantified Risk Assessment (QRA) documents. Dublin, Ireland: Department of Communications, Energy and Natural Resources, http://www.dcenr. gov.ie/Press+Releases/Corrib+Gas+Development.htm. Released 24 May 2005. Hopkins, H.F., Lewis, S.E., and Ramage, A.D. 1993. The development and application of the British Gas TRANSPIRE Pipeline Risk Assessment Package. Presented at the IGEM Midlands Section Meeting, Lough- borough, UK, 19 October. Hymes, I., Boydell, W., and Prescott, B. 1996. Thermal Radiation 2: The Physiological and Pathological Effects. Warwickshire, UK: Major Hazards Monograph Series, Institute of Chemical Engineers. Mather, J., Blackmore, C., Petrie, A., and Treves, C. 2001. An assessment of measures in use for gas pipelines to mitigate against damage caused by third party activity. CRR 372/2001, Health and Safety Executive, London. PD 8010-1: 2004, Code of Practice for Pipelines. Steel Pipelines on Land. 2004. London: British Standards Institution. Pipeline Performance in Alberta 19801997. 1998. Report 98-G, Alberta Energy and Utilities Board, Calgary, Alberta. Reducing Risks, Protecting People: HSEs Decision-Making Process. 2001. London: HSE Books. Steel Pipelines for High Pressure Gas Transmission. 2001. Recommenda- tions on Transmission and Distribution Practice, IGE/TD/1 Edition 4 Communication 1670, Institute of Gas Engineers and Managers, Loughborough, Leicestershire, UK. SI Metric Conversion Factors bar 1.0* E+05 Pa in. 2.54* E+00 cm mile 1.609 344* E+00 km *Conversion factor is exact. Stacy Nelson is a safety engineer with Total E&P UK Ltd. and provides safety engineering expertise to the operations and project functions within the company. She is a member of the Institute of Chemical Engineers and the Safety and Reliability Society. Nelson graduated from the University of the West In- dies, Trinidad & Tobago, with a BS in chemical and process engineering. She later attended the University of Aberdeen, UK, where she completed an MS in safety engineering, reliabil- 11 September 2008 SPE Projects, Facilities & Construction ity and risk management. This was followed by a PhD in inves- tigating the safety and integrity issues relating to sand erosion of process pipe work. Over the last nine years, Nelson has worked with the oil and gas industry, focusing on the manage- ment of major accident hazards for both onshore and offshore gas processing facilities. Margaret Caulfield is currently the se- nior consultant at Vectra Group Ltd. She earned a BS in math- ematics and an MP in mechanical and process engineering at the University of Strathclyde. Caulfield has almost 20 years experience working in hazards in the oil and gas industry. She has previously worked for Advantica (formerly British Gas Re- search and Technology), primarily working on fire and gas dispersion hazards and the associated risks. She also has ex- tensive experience of pipeline risk assessment, safety case preparation, and safety management systems. Her current areas of work include quantified risk assessments and safety cases of onshore and offshore oil and gas facilities. Andrew Cosham is a principal consultant at Atkins Boreas, based in the UK. He has more than 14 years of experience in the oil and gas industry, involved in fracture mechanics, pipeline and pres- sure vessel integrity, and defect assessment. Cosham previ- ously worked at Andrew Palmer and Associates and British Gas. He holds a PhD from the University of Cambridge for his research on the effect of prestrain on the toughness of line pipe steel and an ME in mechanical engineering from the Uni- versity of Bristol. 12 September 2008 SPE Projects, Facilities & Construction