Advertisement

Multi-Level Explosion Risk Analysis for VCEs in Super-Large FLNG Facilities

  • Guowei MaEmail author
  • Yimiao Huang
  • Jingde Li
Chapter
  • 235 Downloads

Abstract

This chapter illustrates a multi-level explosion risk analysis method for super-large oil and gas facilities, so as of the FLNG platform. Three levels of risk analyses, i.e., a qualitative risk screening, a semi-quantitative risk classification and a quantitative risk assessment, are implemented. The CFD method is applied for detailed risk quantification, and an as low as reasonably practical (ALARP) method is adopted as a calibration tool. Safety barriers are introduced as extra risk indicators and a case study is conducted based on a cylindrical FLNG model.

11.1 Introduction

This chapter develops a multi-level explosion risk analysis (MLERA) procedure for the super-large FLNG platforms as described in Chap.  6. Figure 11.1 shows the world’s first FLNG facility designed by Shell Global, the Prelude FLNG. It is 488 m long and 74 m wide, weighing more than 600,000 tons when fully ballasted. It is roughly six times the weight of the largest aircraft carrier (Shell Global, 2016).
Fig. 11.1

Shell prelude FLNG (Shell Global, 2016)

(permission from Elsevier)

Explosion risks are governed by three critical conditions, i.e., confinement, congestion and ventilation. Since a FLNG facility processes and stores large amount of flammable gas in a relatively small and congested area compared to onshore LNG plants, much higher explosion risk is expected on FLNG platforms. Compared to other congested offshore structures, explosion events may yield much more severe consequences due to the super-large space on board, which allows for a large volume of gas cloud to be accumulated. Therefore, for this kind of large and highly congested structure, explosion risks should be addressed during the design process and restrained to an acceptable level.

Among all the explosion safety assessment methods, explosion risk analysis (ERA) is one of the most prevailing approaches to derive the accidental loads for design purposes. Description of ERA could be found in Vinnem (2011) and detailed guidelines on how to perform ERA are given in NORSOK Z013 (2001) and ISO 19901-3 (2014). Due to the complex geometry and obstacles of the offshore structures, computational fluid dynamics (CFD) tools, such as FLACS (GEXCON, 2011), are usually adopted in ERA. On the other hand, Hocquet from Technip (2013) pointed out that one critical issue in applying ERA to FLNGs is the numerous CFD dispersion and explosion calculations. It will render intractable computational intensity due to the large size, complex structures of FLNGs and various potential uncertainties.

This chapter introduces a multi-level explosion risk analysis method (MLERA) for FLNGs, which classifies the FLNG into different subsections of different risk levels before the detailed CFD simulations are conducted. The advantage of this method lies in the highly reduced CFD computational intensity since detailed computation is limited to the identified areas with the highest risks. The MLERA includes three levels: qualitative risk screening, semi-quantitative risk classification and quantitative risk assessment. Through the three levels of analyses, an exceedance curve of frequency versus overpressure will be developed. An as low as reasonably practical (ALARP) method is adopted to ascertain if the explosion risk is acceptable (NOPSEMA, 2015). Risk mitigations are required until the explosion risk of the target area is reduced as low as reasonably practical.

Another challenge in assessing explosion risks for an FLNG facility is that there are neither design rules nor industry standards available as references and benchmarks because FLNG is a new technology (Paris & Cahay, 2014). Current standards, such as UKOOA (2003), HSE (2003) and API (2006), provide detailed guidelines on how to perform offshore explosion analyses with a specific procedure. Most of these guidelines are proposed based on fixed platforms; it may not be appropriate to comply in conducting an explosion risk analysis for FLNG platforms. For example, if the risk screening process used for fixed platforms is extended straightforwardly to FLNG facilities, all FLNG platforms are identified at the highest risk level, which implies the inefficacy of the risk screening process.

To amend the aforementioned improficiency, other than the usual contributors from current standards, such as confinement, congestion and ventilation, safety barriers are engaged in the risk screening and classification procedures in the currently proposed method. Safety barriers are usually applied for both likelihood reduction and consequence mitigation. Some of the safety barriers used in the MLERA are listed and briefly introduced in the following section.

11.2 Multi-Level Explosion Risk Analysis

A multi-level explosion risk analysis (MLERA) method (Huang, Ma, & Li, 2017) is proposed in this chapter by integrating a multi-level risk assessment method into the traditional ERA method for offshore platforms.

The multi-level risk assessment method is an extension from the corresponding framework of the Department of Planning & Infrastructure of New South Wales Government (2011) to formulate and implement risk assessment and land-use safety planning processes. It is a balanced trade-off between the derivation cost and the quality of the results. To achieve that, both qualitative and quantitative approaches are engaged. Some key factors of the three levels of analysis from NOPSEMA (2012) are shown in Table 11.1.
Table 11.1

Key factors of multi-level risk analysis

Level 1 of preliminary qualitative risk screening

• Likelihood and consequence are expressed on a scale and described in words

• There is no numerical value for risk output

• Often used as a preliminary risk assessment or screening tool

• Rapid assessment process and relatively easy to use

Level 2 of semi-quantitative risk classification and prioritization

• Generate a numerical value, but not an absolute value of risk

• Provides greater capacity to classify hazards on the basis of risk

• Better for evaluating cumulative risk

Level 3 of detailed quantitative risk assessment

• Provides a calculated value of risk based on estimates of consequence (usually software modelling) and likelihood (estimates based on failure rate data—site or industry)

• Appropriate for complex decision making or where risks are relatively high

• More intensive and expensive than other prevailing methods

The traditional ERA for offshore platforms is one of the most prevailing approaches to derive the accidental loads for design purposes. As mentioned above, one of the critical issues in applying ERA to FLNG platforms is the intractable computational intensity. Due to the huge size of the FLNG facilities, numerous CFD dispersion and explosion simulations are required to acquire sufficient data to derive realistic design explosion loads.

Therefore, the multi-level method is developed in this chapter to improve the ERA to modulate the computational cost to a manageable level. The proposed MLERA method is a systematic risk analysis approach that includes three assessment stages, i.e., qualitative explosion risk screening as the first level, semi-quantitative explosion risk classification as the second level and quantitative explosion risk analysis as the third level. It aims to provide an efficient risk analysis method for explosion accidents on offshore super-large structures, such as FLNG facilities.

The key aspects in multi-level risk analysis, as given in Table 11.1, are then briefly described for the proposed MLERA for FLNG platforms. Related analysis features of each level are listed, and detailed explanations of each step are discussed in the following context.

Level 1 of qualitative risk screening:
  • qualitative description of critical risk contributors,

  • taking the overall FLNG facility as the analysis target,

  • using a risk matrix diagram to rank the risk level of a FLNG platform.

Level 2 of semi-quantitative risk classification:
  • using a score and weight system to quantify each risk contributor,

  • estimating the risk of each FLNG subsection,

  • classifying the subsections using a cumulative density function diagram.

Level 3 of quantitative risk assessment:
  • combining ERA and FLACS to derive the quantitative results of explosion frequency and consequences,

  • assessing the subsections with the highest risk levels, which are identified from the previous two analyses,

  • the final result is indicated by an overpressure versus frequency exceedance curve,

  • the ALARP concept is applied to check if the explosion risk of the corresponding subsection is as low as reasonably practical.

Since the proposed MLERA considers also the safety barriers, some of the safety barriers that are engaged in the proposed method are briefly introduced in Table 11.2.
Table 11.2

Explosion safety barriers

Blast relief panels

The overpressure can be diverted away from potential escalation sources by blast relief panels. Blast relief panels will open quickly during an explosion in order to reduce peak overpressures

Emergency shutdown systems (ESD)

An effective ESD system will limit the inventory released in an incident, and thus, the size and duration of any resulted fire. The location of the ESD valves is usually determined based on the judgement where each particular inventory could be released

Isolation and blowdown

A leak may be reduced by isolating it manually or using the ESD system and depressurizing the leaking section using the blowdown system. Damage or fatality risk in escalation can be reduced by isolation and blowdown, so that evacuation may be avoided

Blast wall

Blast walls have long been used to protect adjacent areas from the impingement of overpressure. These walls are designed to absorb blast energy through displacement

Water deluge

Deluge has been found suitable for reducing overpressure in congestion-generated explosions. If explosion mitigation is considered critical, a deluge flow-rate of at least 13–15 L/min/m2 is recommended for general area coverage

Artificial vent

Artificial ventilation is defined as the ventilation not supplied from the action of the environmental wind alone. Upon detection of flammable gas, the standby fan(s) should be started to give maximum possible ventilation in order to aid dilution of the leak to prevent or limit the generation of an explosive cloud

Inert gas

Inert gas can be used to dilute the flammable mixture by flooding the volume within which the gas has been detected. For example, CO2 or N2 is typical inert gas. The explosive gas can then be diluted below its lower explosive limit

Detection device

Detection measures can be used to identify hazardous conditions on the plant, such as excess process pressure, an unignited release of flammable gas or a fire. Detection devices enable controlling or mitigating measures and emergency response to be activated

Alarm

The alarm system may allow operators to mitigate leaks before they ignite or to, at least, evacuate the area

Soft barriers

Progress has been made in devising soft barriers such as the micro-mist device, which consists of a cylinder of superheated water that is released quickly as a fine mist in response sirening pressure or flame sensors during an explosion. This device suppresses the explosion and significantly reduces overpressures

Safety gap

In the process industry, the safety gap is an open space with no congestion, deliberately placed in between congested process areas. The absence of obstacles in a safety gap eliminates the fluid–obstacle interaction, thereby preventing the generation of turbulence. It can be very effective in reducing pressures prior to the onset of detonation

11.2.1 First Level of Qualitative Risk Screening

The first-level risk screening defines the total qualitative risk level of a FLNG platform and provides guideline for the next two levels of explosion risk analyses. At this stage of risk screening, general risk indicators, as well as safety barriers, design, operation and maintenance measures, are engaged to define a relative risk level for FLNG in view of that unreasonably high explosion risk will be yielded if traditional risk screening methods are applied for such super-large and highly congested structure. Based on API (2006) and UKOOA (2003), most of the qualitative risk indicators for the traditional risk screening process are listed in Table 11.3.
Table 11.3

Traditional risk screening indicators from explosion risk standards

Consequence

Low consequence

• Low congestion level due to the low equipment count, being limited to wellheads and manifold with no vessels (i.e., no associated process pipework)

• No more than two solid boundaries, including solid decks

• Unattended facilities with low maintenance frequency, less frequent than 6-weekly

Medium consequence

• Medium congestion level due to the greater amount of equipment installed compared to those of the low consequence cases

• Higher confinement level than that for the low consequence cases

• Unattended facilities with a moderate maintenance frequency, more frequent than 6-weekly

• A processing platform necessitating permanent manning but with low escalation potential to reach quarters, utilities and control areas located on a separate structure

High consequence

• High congestion level due to the significant processing on board, which leads to a high equipment count

• High confinement level of the potential gas release point

• Permanent manning with populated areas within the consequence range of escalation scenarios

Likelihood

Low likelihood

• Low equipment and inventory count, which align closely with the consequence scenarios

• Low frequency of intervention, less frequent than 6-weekly

• No ignition sources within the potential gas cloud

Medium likelihood

• Greater amount of equipment installed than those for the low likelihood

• Medium frequency of intervention, more frequent than 6-weekly

• Weak ignition sources, such as a hot surface, exist within the potential gas cloud

High likelihood

• A high equipment and inventory count

• Permanently manned installations with frequent processing on board

• Strong ignition sources exist within the potential gas cloud

Table 11.4 describes exhaustively risk screening process that uses safety barriers, design, operation and maintenance measures as screening contributors. A corresponding modified risk matrix diagram is illustrated in Table 11.5. From the modified diagram, only a relative risk category is defined. The results from this category will be applied as guideline for further assessment in the next stage of the proposed MLERA.
Table 11.4

Risk indicators based on safety barriers

Consequence

No.

Risk level

Description

A

Moderate

• Safety barriers covering most or all parts of the FLNGs

• High design capacity of the structure to counteract dynamic pressure, overpressure, missiles and strong shock response. No or minor structural damages would occur

B

Major

• Safety barriers covering the structural critical elements only

• Medium design capacity of the structure to counteract dynamic pressure, overpressure, missiles and strong shock response. A medium level of structural damages would occur without affecting the overall structural integrity

C

Catastrophic

• No or only safety barriers for human living quarters

• Low design capacity of the structure to sustain dynamic pressure, overpressure, missiles and strong shock response. Significant structural damages would occur and undermine the structural integrity

Likelihood

No.

Risk level

Description

1

Almost certain

• No or only safety barriers for human living quarters

• Low level of operation and the maintenance measure corresponding to a level considerably lower than industry average

2

Likely

• Safety barriers covering only critical potential release points

• Medium level of operation and maintenance philosophy corresponding to the industry average

3

Possible

• Safety barriers covering all or most of potential release points of the FLNG structures

• High level of operation and the maintenance philosophy corresponding to the best standard in industry

Table 11.5

Risk matrix diagram for further risk screening of FLNGs

11.2.2 Second Level of Semi-Quantitative Risk Classification

The analysis at this level estimates the risk level of each subsection of a FLNG facility to identify assessment prioritization for the third-level ERA. A score and weight system is applied to each selected risk contributor, so that the subsections can be classified according to the respective accumulated value.

Only some of the main risk contributors for offshore explosion events are selected and briefly described in Table 11.6. Each contributor is evaluated by two elements, i.e., weight and score. The weight of each risk factor is subjectively specified by the authors based on the relevant standards as shown in below Eq. (11.1) (API, 2006; Bjerketvedt, Bakke, & Van Wingerden, 1997; UKOOA, 2003). This may be adjusted by the safety engineers according to their own experience and the practical conditions of their projects. Flammability limits for fuel mixtures can be calculated by Le Chatelier’s law:
Table 11.6

Weight and score of explosion risk contributors

Risk Contributor

Description

Weight

Score

Equipment count

Leak frequency is proportional to the amount of process equipment on the platform

3

= number of equipment count

Ignition

In general, the main ignition sources are welding/hot work, compressors, electrical equipment and engines/exhausts. A weak, continuous ignition source can stay and wait for the gas cloud to reach its flammable range

7

= 3 if continuous ignition source exists

= 2 if only discrete ignition source exists

= 1 if no or few ignition source exists

Flammable limit of process material

The higher the upper flammable limit of a certain fuel, the easier it is usually to get a flammable cloud in the air Eq. (11.1).

4

= 3 if upper flammable limit >40%

= 2 if upper flammable limit is between 10 and 40%

= 1 if upper flammable limit <10%

Congestion

Explosion events are most likely to occur in congested areas, and therefore, avoiding congestion in the modules can reduce both of the probability and overpressure of an explosion event. Table 11.7 defines the congestion level based on the congestion classification of Baker–Strehlow–Tang model (Baker et al., 1996)

10

= 3 if congestion is defined as high

= 2 if congestion is defined as medium

= 1 if congestion is defined as low

Fuel reactivity

The higher the laminar burning velocity, the higher the explosion loads will be

4

= 3 if laminar burning velocity >75 cm/s

= 2 if laminar burning velocity is between 45 cm/s and 75 cm/s

= 1 if laminar burning velocity <45 cm/s

Confinement

The ignition probability depends on the gas concentration and the ignition sources in the area. Low confinement dilutes the gas concentration

8

= 3 if the flame expansion is defined as 1D

= 2 if the flame expansion is defined as 2D

= 1 if the flame expansion is defined as 2.5D or 3D

Distance to target area

Distance to the target area significantly affect the consequent load impinging on the target area

7

= 3 if distance is shorter than 1/3 total length of the structure

= 2 if distance is longer than 1/3 and shorter than 2/3 total length of the structure

= 1 if distance is larger than 2/3 total length of the structure

Table 11.7

Blockage ratio classification

$$ {\text{LFL}}_{\text{Mix}} = \frac{100}{{C_{1} /{\text{LFL}}_{1} + C_{2} /{\text{LFL}}_{2} + \cdots + C_{i} /{\text{LFL}}_{i} }} $$
(11.1)
where C1, C2, …, Ci are the volumetric proportions of each gas in the fuel mixture without air (Kuchta, 1985).
Safety barriers are employed as extra risk contributors in the semi-quantitative risk classification procedure. All safety barriers are divided into three categories: likelihood reduction, consequence mitigation and those for both functions. Based on the classifications, safety barriers are given different weights as shown in Table 11.8. Score is specified by the quantity of each barrier deployed in each module.
Table 11.8

Weight of barriers based on function classification

Barrier

Classification

Weight

Emergency shutdown (ESD) system

Likelihood reduction

6

Detection device

Likelihood reduction

6

Water deluge

Likelihood reduction

4

Inert gas

Likelihood reduction

4

Safety gap

Consequence mitigation

3

Blast wall

Consequence mitigation

3

Blast relief panels

Consequence mitigation

3

Soft barriers

Consequence mitigation

3

Artificial vent

Both

9

Isolation and blowdown

Both

9

Alarm

Both

9

From Table 11.8, for safety barriers to reduce likelihood, two different weights, 6 and 4, are defined. It is because although water deluge and inert gas are able to reduce the flammable limit of the cloud and consequently prevent the explosion, they may simultaneously aggravate the consequence if the explosion eventually occurs. Inert gas can pose a significant asphyxiation risk to personnel and water deluge without proper design may increase turbulence of the affected area, and thus, the blast loads. Therefore, these two barriers are given lower weight than normal prevention barriers unless more persuasive design is presented.

The total weighted score of each subsection can then be calculated as
$$ S_{T} = S_{C} - S_{B} $$
(11.2)
$$ S_{C} = \mathop \sum \limits_{i = 1}^{n} w_{ei} s_{ei} $$
(11.3)
$$ S_{B} = \mathop \sum \limits_{j = 1}^{n} w_{bj} s_{bj} $$
(11.4)
wherein \( S_{t} \) refers to the total weighted score for each subsection; \( S_{C} \) and \( S_{B} \) are the weighted scores of risk contributors and barrier functions of each subsection, respectively.
After the total score of each subsection is calculated, the total weighted scores are derived with a cumulative density function and the identified the level in the risk category as shown in Fig. 11.2. The cumulative percentage is calculated from the total weighted scores of all the subsections from the target FLNG platform.
Fig. 11.2

Cumulative density function (CDF) for total risk score

(permission from Elsevier)

Figure 11.3 illustrates the analysis procedure of the proposed MLERA and identifies the subsections requiring third-level risk quantification.
Fig. 11.3

Implementation procedure of MLERA

(permission from Elsevier)

The first-level risk screening process divides the qualitative results into three risk levels, i.e., relatively low, medium and high risks. If the FLNG facility is categorized with a relatively low explosion risk level, only the subsections with the highest risks, which belong to category S1 (top 10%), are required for additional detailed quantitative explosion risk assessment. From Fig. 11.2, for an FLNG facility with relatively low explosion risks, the number of category S1 subsections is two. The numbers of subsections with relatively medium risk in categories S2 (50%) or high risk in S3 (90%) are 10 and 18, respectively, which require risk quantification assessment. If all the subsections in one category fail the ERA, then the next level subsections require further ERA as well.

11.2.3 Third Level of Quantitative Risk Assessment

Third-level QRA is a CFD software-based quantitative analysis procedure. The process includes four main steps: leak frequency analysis, flammable gas dispersion simulation, ignition probability modelling and flammable gas explosion simulation. Figure 11.4 shows the detailed quantitative analysis procedure applied to offshore structures using CFD tools, such as FLACS.
Fig. 11.4

Quantitative assessment procedure

(permission from Elsevier)

From the quantitative ERA analysis, an overpressure versus frequency exceedance curve is deduced. The risk calibration method ALARP is adopted to define the risk acceptance criteria. The ALARP framework of risk criteria is divided into three regions as shown in Fig. 11.4.
  • An unacceptable region: in this region, risks are not acceptable except for extraordinary circumstances. Risk reduction measures should be deployed.

  • A tolerable region: it is usually identified as an ALARP region, which implies that the risks are considered acceptable providing that they have been made as low as reasonably practicable. In this region, risk reduction measures are desirable but may not be necessary if a cost–benefit analysis shows that their cost is disproportionate to the benefit to be achieved.

  • A broadly acceptable region: risks in this region are acceptable and no more risk reduction measures are required.

Figure 11.5 shows an example of application of ALARP to the overpressure versus frequency exceedance curve.
Fig. 11.5

Application of the ALARP to final results of MLERA

(permission from Elsevier)

As shown in the diagram, if the design strength of the primary components of the FLNG comes to cross the predicted explosion load in the unaccepted zone, more risk reduction measures are required to be deployed until the design strength is proved to be sufficient to resist the explosion loads. No further reductions are required if the design strength falls in the accepted zone. For the ALARP zone, reduction measures should be deployed unless the cost is proved to be disproportionate to the potential expectant benefit.

11.3 Case Study

For FLNG structures, cylindrical FLNG vessels introduced in Chap.  6 are applied to improve hydrodynamic stability. Figure 11.6 shows the geometry of a cylindrical FLNG platform in FLACS. The cylindrical platform has a smaller area than a usual rectangular one. All the highly congested subsections are concentrated on the board, which may increase the explosion risks. Little research has been reported on the gas explosion risk analysis for cylindrical platforms. In this section, a cylindrical FLNG structure proposed by Li, Ma and Abdel-jawad (2016) is adopted as the basic model to illustrate the proposed MLERA.
Fig. 11.6

Geometry of cylindrical FLNG

(permission from Elsevier)

Figure 11.7 shows the arrangement of topside modules including totally 12 modules. A brief introduction of each module is listed as below.
Fig. 11.7

Topside arrangement of modules

(permission from Elsevier)

  • Module 1: power generation,

  • Module 2: Trent gas turbines and two essential diesel generators,

  • Module 3: nitrogen package, hot oil, mono-ethylene-glycol (MEG) processing and inlet facilities,

  • Module 4: boil-off gas compressor and fuel gas system,

  • Module 5: acid gas removal unit and end flash gas compressor,

  • Module 6: dehydration and mercury removal,

  • Module 7–Module 12: liquefaction modules.

11.3.1 Qualitative Risk Screening of Cylindrical FLNG

For the first level of the risk screening process, based on the concepts from API (2006) and UKOOA (2003), the selected FLNG module is specified as a high-risk platform because it is a permanently manned and highly congested offshore structure with a large amount of equipment and inventories. The second level of risk screening analysis is then conducted directly. The conditions of safety barriers, design, operation and maintenance philosophies are defined as below.
  • Safety barriers: as shown in Fig. 11.7, safety gaps are applied to every module. In view of grossly insufficient information about other safety barriers, such as alarms, detection devices, ESDs and water deluges on this FLNG model, a medium level for the condition of the safety barriers on this FLNG platform is assumed.

  • Design philosophy: this FLNG is a recently designed offshore structure to have a high level of design philosophy under the most recent design standards.

  • Operation and maintenance philosophy: the standard of operation and maintenance philosophy is assumed to be medium implying that the average industry standard is employed because that no such FLNG facility has yet been operated throughout the world.

Based on the aforementioned conditions, this cylindrical FLNG is specified as of a medium risk, which indicates that all subsections belongs to category S2 from the second level of semi-quantitative risk classification. Thus, it is subject to detailed assessment in the third step.

11.3.2 Semi-Quantitative Risk Classification

The subsections of the selected model are defined by 12 modules. Each module is assessed in this risk classification process by specifying the explosion and safety barrier contributors, which are defined in Tables 11.6 and 11.8. The target area of consequence analysis is the human living quarters. As shown in Fig. 11.8, a cumulative density function diagram can be calculated based on the final scores from Table 11.9.
Fig. 11.8

Cumulative density function diagram of subsections

(permission from Elsevier)

Table 11.9

Scores and weights of risk contributors

Subsections

1

2

3

4

5

6

7

8

9

10

11

12

Total score of explosion contributors

93

64

66

66

81

69

101

101

108

108

115

115

Total score of safety barrier contributors

30

30

30

30

27

30

27

30

30

30

30

30

Final score

63

34

36

36

54

39

74

71

78

78

85

85

During the second level of risk classification process, some of the contributors have the same score for different subsections. For instance, as in Table 11.9, the final scores of safety barriers are the same for most of the modules. It is probably because this second level of risk classification process is still an abbreviated assessment of each module. Thus, it may not be sufficiently proficient to differentiate the modules by one particular contributor. It is understood that lack of detailed information also contributes to this improficiency. For example, in this case study, module 5 and 7 have less safety gaps than the other modules based on design drawings of the proposed model. This is the only difference of safety barriers that can be defined and the other scores of barriers for each module are assumed to be the same due to the insufficiency of data. Therefore, the total scores of barriers for most of the subsections remain the same. It is expected that more detailed information of the target structure will lead to a higher level of accuracy for this classification.

From the cumulative density function diagram, the S2 category includes six subsections, i.e., modules 7–12. Therefore, six subsections require further detailed assessment.

11.3.3 Detailed Quantitative Risk Assessment

As a medium risk level is identified for the target FLNG during the first level of the qualitative risk screening process, modules 7–12, which belong to category S2, require further detailed assessment. Therefore, detailed quantitative risk assessment for FLACS is conducted in this section. However, due to the limitation of computer capacity, a simplified analysis model, which is assumed to be sufficient to demonstrate the proposed method, is built and applied.

In this model, three leak locations on subsections 7, 9 and 11 are specified for assessment. Final results of this assessment are deduced by combining the analyses on these three locations. The three selected locations are shown in Fig. 11.9. Other specific assumptions for this model are described as below.
Fig. 11.9

Selected leak locations on cylindrical FLNG platform

(permission from Elsevier)

  • Four leak rates (12, 24, 48, 96 kg/s) are simulated to study the potential gas volume build-up in order to compare the blast wall configurations.

  • In the simulations of dispersion leaks and explosion gas clouds, the inventory of the gas composition inside the cylindrical FLNG platform is summarized as in Table 11.10.
    Table 11.10

    Gas composition for dispersion and explosion study

    Component

    Export gas (%)

    Methane

    27

    Ethane

    33

    Propane

    15

    Hexane

    19

    CO2

    6

  • In this study, the assessment focuses on the living quarters with a protective blast wall on the west side. The living quarters are located at the very east side of the FLNG (Fig. 11.7).

  • Wind speed and wind direction are specified to be constant of +4 m/s in the due east direction to examine the worst gas dispersion scenarios.

  • Leak directions are modelled in both eastern and western directions.

  • Dispersion analysis

Based on the aforementioned assumptions, the overall leak cases adopted in this chapter are listed in Table 11.11. The gas monitoring region for dispersion analysis covers all the modules on the cylindrical FLNG platform.
Table 11.11

Leak cases identified for dispersion study

Case

Wind direction

Wind speed (m/s)

Leak rate (kg/s)

Leak position

Leak orientation

1

Due east

4

12, 24, 48, 96

West end

Along and opposite wind

2

Due east

4

12, 24, 48, 96

Middle

Along and opposite wind

3

Due east

4

12, 24, 48, 96

East end

Along and opposite wind

Figure 11.10 demonstrates examples of dispersion simulation outputs for gas releases with a leak rate of 48 kg/s. The releases are simulated from both release directions and leak locations are set on the ground centre of modules 7, 9 and 11.
Fig. 11.10

Gas dispersion simulations for leaks with leak rate of 48 kg/s

(permission from Elsevier)

To investigate all the potential leak rate cases, the overall cumulative curve of gas cloud sizes within the gas monitor region is depicted in Fig. 11.11. The cumulative curve is derived by sorting the gas cloud size from small to large with assumption of equal leak frequencies.
Fig. 11.11

Cumulative curve of gas cloud sizes for all leak rate scenarios

(permission from Elsevier)

  • Explosion simulations

Explosion simulations are performed using gas cloud data from dispersion simulations with leak rates of 12 to 96 kg/s. The gas clouds are specified in four different locations covering the entire platform. So that the overall gas explosion consequences for all modules can be analysed. For all the gas clouds, the plan view sizes are all fixed at 100 × 80 m2, while the heights of the clouds are variable in consistency to the corresponding gas dispersion results. For each gas explosion simulation, the gas cloud is ignited at the ground centre of each module.
Figure 11.12 gives an overview of the gas cloud coverage and ignition locations. From Fig. 11.12, each gas cloud covers four modules. About 200 monitor points are homogeneously deployed on the ground to record the overpressures in a gas explosion simulation. Totalling all the different gas leak rate scenarios, gas cloud sizes and locations, more than 3000 VCE overpressures are monitored in this probabilistic study on gas explosion. Since the main purpose of this study is to assess the condition of the living quarters, ten monitor points are assigned near the living quarter to record the overpressures for each gas explosion scenario.
Fig. 11.12

Overview of gas cloud coverage and ignition locations

(permission from Elsevier)

Figure 11.13 shows three explosion examples simulated based on different leak rates: 96, 48 and 24 kg/s. The explosive gas clouds are set at the north and east ends of the model. The ignition is assigned at the centre of the gas cloud located at the east and north ends of the platform. The gas explosion blast is observed to spread from the ignition point to all the surrounding objects. The maximum overpressures are found near the edge of the gas cloud in the congested region.
Fig. 11.13

Gas explosion simulation examples with different leak rates

(permission from Elsevier)

To consider the influence of blast walls, a blast wall is modelled in front of the west end of the living quarter. Two monitors are set at both sides of the blast wall. A large overpressure of approximately 1.8 bar is detected at the left side of the wall under the leak rate of 96 kg/s as shown in Fig. 11.14. However, after reduction by the blast wall, the overpressure at the right side of the wall is about 0.2 bar to prove the effectiveness of the blast wall.
Fig. 11.14

Demonstration of function of blast wall

(permission from Elsevier)

To consider all the gas dispersion outputs as input for the gas explosion simulations, 120 explosion cases are numerically simulated, which are commensurating with the former dispersion simulations for four leakage rates, two leakage directions, three gas release locations and five different series of blast wall layout designs (Li, 2017). The overpressure of each case is calculated by FLACs and the overall cumulative curve of gas explosion simulations is summarized in Fig. 11.15. Equal frequencies are allocated to all monitored overpressures for the living quarters, which are sorted from small to large.
Fig. 11.15

Cumulative curve of overpressure for living quarters

(permission from Elsevier)

  • Frequency analysis

A simple illustrative explosion frequency calculation is then conducted in this subsection. The exceedance curve of frequency against overpressure at the living quarters is specified using the monitored overpressures for over 1000 scenarios.

To simplify the analysis process, the leak frequencies of different leak rates are assumed to be the same. Based on the data from the Purple Book (Uijt & Ale, 2005), the leak frequency is taken as 3.33 × 10−1 per year. Based on the ignition intensities and the previously performed dispersion simulations, the ignition probability is determined to be 0.36%. The explosion frequency is consequently calculated by multiplying the leak frequency with the ignition probability. Therefore, the total explosion frequency is derived as approximately 1.2 × 10−3 per year.

Consequently, the explosion risk regarding the living quarters subject to overpressures of VCE from the liquefaction modules is evaluated and the probability of exceedance curves with a frequency of 10−4/year is shown in Fig. 11.16.
Fig. 11.16

Exceedance curve of overpressures around living quarters for all leak rate scenarios

(permission from Elsevier)

From Fig. 11.16, the maximum overpressure is about 0.4 bar in the living quarters. In view of the ALARP criterion, the acceptable zone starts from 0.2 bar, which implies that no further risk reduction method is required if the maximum strength of the primary components of the FLNG is designed to be larger than 0.2 bar with a corresponding frequency of 10−11. Otherwise, risk reduction measures should be deployed until the design strength demonstrates greater than 0.05 bar with corresponding frequency of 10−4 and also proves to be as low as reasonably practical.

11.4 Summary

In summary, a more efficient multi-level explosion risk analysis method (MLERA) is proposed in this chapter. This method includes three levels of assessment, i.e., qualitative risk screening for an FLNG facility at the first level, semi-quantitative risk classification for subsections at the second level and quantitative risk calculation for the target area with the highest potential risks at the third level.

Since the current design standards for normal offshore platforms are not sufficiently robust for effectively assessing explosion risks of super-large offshore structures, during the risk screening and risk classification processes, safety barriers are adopted as extra risk indicators besides the traditional ones such as congestion, confinement and ventilation. From the aforementioned analyses, with only traditional standards, FLNG platforms will always be identified as high risk. However, with the extra contributors of safety barriers, the target FLNG facility is distinguished into relatively low, medium or high risks for different subsections, which identifies the subsections for further assessments.

For detailed quantitative risk assessment, a CFD software, FLACS, is applied to model and simulate the target FLNG platform. The results are shown as an exceedance curve, which describes the possibilities of overpressure at the target area. An ALARP method is then adopted as a calibration tool to determine if the explosion loads from the exceedance curve can be accepted. If the overpressure exceeds the acceptable limitation, more safety barriers should be installed. Further assessments are required until the final results show that the risk is reduced to an acceptable level or/and as low as is reasonably practical.

Through the three levels of risk assessments, the areas with the highest level of potential risks are identified to be further assessed. From the case study, only half of the subsections on the selected model require detailed assessment using FLACs with regard to the living quarters. It implies the potential computational intensity is reduced to half by the proposed method. In the next chapter, a more detailed CFD-based ERA is introduced with consideration of blast wall effect to explore potential risk reduction methods.

References

  1. API. (2006). Recommended practice for the design of offshore facilities against fire and blast loading (1st ed.). United States: American Petroleum Institude.Google Scholar
  2. Baker, Q. A., Tang, M. J., Scheier, E. A., & Silva, G. J. (1996). Vapor cloud explosion analysis. Process Safety Progress, 15(2), 106–109.CrossRefGoogle Scholar
  3. Bjerketvedt, D., Bakke, J. R., & Van Wingerden, K. (1997). Gas explosion handbook. Journal of Hazardous Materials, 52(1), 1–150.CrossRefGoogle Scholar
  4. Department of Planning & Infrastructure of New South Wales Government. (2011). Assessment Guideline Multi-level Risk Assessment. New South Wales, Australia: Department of Planning & Infrastructure.Google Scholar
  5. GexCon. (2011). FLACS v9.1 user’s manual. Norway: Doxygen.Google Scholar
  6. HSE. (2003). Fire, explosion and risk assessment topic guidance (Issue 1). United Kingdom: Health and Safety Executive.Google Scholar
  7. Hocquet, J. (2013). Explosion risk analysis ‘ERA’ for FLNG facilities: The main challenges. Chemical Engineering Transactions, 31, 595–600.Google Scholar
  8. Huang, Y., Ma, G., & Li, J. (2017). Multi-level explosion risk analysis (MLERA) for accidental gas explosion events in super-large FLNG facilities. Journal of Loss Prevention in the Process Industries, 45, 242–254.CrossRefGoogle Scholar
  9. ISO 19901-3. (2014). Petroleum and natural gas industries—specific requirements for offshore structures—part 3: Topsides structure (2nd ed.). Switzerland.Google Scholar
  10. Kuchta, J. M. (1985). Investigation of fire and explosion accidents in the chemical, mining and fuel-related industries—A manual. United States Department of the interior, Bureau of Mines, Bulletin 680.Google Scholar
  11. Li, J., Ma, G., & Abdel-jawad, M. (2016). Gas dispersion risk analysis of safety gap effect on the innovating FLNG vessel with a cylindrical platform. Journal of Loss Prevention in the Process Industries, 40, 304–316.CrossRefGoogle Scholar
  12. Li, J., Ma, G., Hao, H., & Huang, Y. (2017). Optimal blast wall layout design to mitigate gas dispersion and explosion on a cylindrical FLNG platform. Journal of Loss Prevention in the Process Industries. Google Scholar
  13. NOPSEMA. (2012). Risk assessment guidance note. Australia: National Offshore Petroleum Safety and Environmental Management Authority.Google Scholar
  14. NORSOK Z-013. (2001). Risk and emergency preparedness analysis. Norway: Norwegian Technology Centre.Google Scholar
  15. Paris, L., & Cahay, M. (2014). Challenges in a multi-disciplinary approach for explosion design of FLNG. Paper presented at the Offshore Technology Conference, Houston, Texas, USA.Google Scholar
  16. Shell Global. (2016). Prelude FLNG—Overview. Retrieved from http://www.shell.com/about-us/major-projects/prelude-flng/prelude-flng-an-overview.html.
  17. Uijt, P., & Ale, B. (2005). Guidelines for quantitative risk assessment. VROM: Ministerie van Verkeer en Waterstaat.Google Scholar
  18. UKOOA. (2003). Fire and explosion guidance part 1: Avoidance and mitigation of explosions (Issue 1). United Kingdom: UK Offshore Operation Association.Google Scholar
  19. Vinnem, J. E. (2011). Offshore risk assessment principles, modeling and applications of QRA studies (2nd ed.). Springer Series in Reliabilities Engineering, London, UK.Google Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2019

Authors and Affiliations

  1. 1.School of Civil and Transportation EngineeringHebei University of TechnologyTianjinChina
  2. 2.Department of Civil, Environmental and Mining Engineering, School of EngineeringUniversity of Western AustraliaPerthAustralia
  3. 3.Centre for Infrastructural Monitoring and ProtectionCurtin UniversityPerthAustralia

Personalised recommendations