2 Aims and principles of fire and explosion hazard management management
EMERGENCY RESPONSE
3 Assessment of and protection from fires and explosions
3.4 Methods and approaches to structural analysis .1 General .1 General
The risk level for the installation as defined by the risk matrix given in Section 2.8.2 determines the level of sophistication required for the fire assessment.
If the conservatism of simplified methods of analysis can be guaranteed, then these could be used at an early project phase or as a first step in a sequence of analyses of increasing sophistication.
For High or Medium risk installations, a ‘Structural Assessment’ should be performed for a representative range of fire scenarios. The process and Safety disciplines will generally define these Scenarios.
A Structural Assessment may be performed at three levels of increasing complexity starting with a ‘Screening Analysis’. Should a structure fail the ‘Screening analysis’ then a ‘Strength level analysis’ will be required. If it fails the Strength level analysis then ‘Ductility level analyses must be performed. If the Ductility level analysis indicates failure then mitigation measures are required. These could involve measures for elimination or reduction of the frequency of exceedance of the initiating event, reduction of the severity of the consequences of the event or structural modification.
‘Failure’ in the context of fires means failure to satisfy the performance standards for the installation. High-level performance standards for the installation and safety critical components are defined in terms of allowable peak temperature, strain or time to collapse depending on the method of analysis used.
Screening Analysis, Strength level or Ductility level performance characteristics from an assessment of one installation may be used to infer the fitness for purpose of other similar installations, provided the framing, foundation support, service history, structural condition, blast and fire barriers and payload levels are not significantly different. In cases where one platform’s detailed performance characteristics are used to infer those of another similar platform, documentation should be developed to substantiate the use of such generic data.
The required initial level of analysis depends on the Risk level assigned to the installation or the risk level associated with a representative set of fire scenarios.
The risk category of the installation does not preclude the use of more sophisticated methods of assessment which may result in reductions in conservatism and hence cost, if they are considered more appropriate.
Table 3.5 - Appropriate method of analysis – fires
Risk level Analysis
method Load calculation bases Response calculation Low Screening
analysis Allowable temperature (yield strength reduction to 60 %
Past experience.
Design basis checks.
Past experience for demonstrably similar platforms.
Strength level
analysis Calculate peak temperature member by member, from nominal fire loads and fire extent.
Strength level analysis, Redundancy analysis Medium or
High
Ductility level analysis
Calculate temperature - time history of primary members from fire loads time history and flame extent.
Redundancy analysis, Ductility level analysis
A structural ‘Redundancy Analysis’ of a topside structure will indicate which members can be removed without collapse of the structure. In addition those members not supporting the TR, muster areas, escape routes or safety critical equipment must survive during and after the fire event for sufficient time to allow personnel on board to escape, allowing for the possible need to assist injured colleagues. The results of a fire response analysis will then indicate which structural members and SCEs must be protected to achieve the fire performance standards for the installation.
• Simple fire response analyses are usually performed based on the following assumptions [3.5].
• Unprotected structural members and panels have no variation of temperature through thickness or along their length. In practice the critical sections of the member are considered from the point of view of resistance.
• Fire protected structural members and panels have a constant steel temperature, the thermal insulation has a linear variation of temperature through thickness.
• Each member may be considered to have reached a steady-state variations of temperature are due entirely to changes in boundary conditions and incident heat fluxes.
• Conduction between members need not be considered (except when considering coat-back requirements).
• Thermal stresses due to restraint may generally be neglected as supports also soften during fire loading.
The methods of analysis identified above are discussed in the following sections.
3.4.2 Screening analysis
A screening analysis for an existing installation consists of a condition assessment which may involve a survey followed by design basis checks.
Design basis checks consist of checking the basis of the existing design for the installation and determining if the methods used for the design are currently acceptable in the context of fire events.
For a Screening Analysis, the Zone method may be used. The Zone method assigns a maximum allowable temperature that a steel member can sustain. This method does not take into account the stresses present in the member before the fire. The maximum allowable temperature may be read from Table 3.6.
These temperature values correspond to a yield strength reduction to 0.6 of the ambient temperature values. The fact that a fire is an accidental load will mean that the allowable stress is the full yield stress value as opposed to about 60 % of the yield stress allowed for in the conventional design load cases. The yield stress corresponding to 0.6 of the ambient yield stress will then give an allowable stress the same as that for the structure before the fire.
Higher strain levels than 0.2 % may give a proportionately higher decrease in Young’s Modulus giving an unmatched reduction in yield strength with the reduction in Young’s modulus exceeding the reduction in yield strength. The Zone method may then not be applicable [3.6].
Fire barriers must perform according to their required rating.
A blanket critical temperature for all members may be postulated as in the Zone method. This critical temperature is chosen typically to be 400 °C as this requires no modification to the normal code checks if strains are limited to 0.2 % in an elastic Design Level analysis. This approach may result in unnecessary protection and may be unconservative locally to areas of high strain.
It will not usually be necessary to protect every vulnerable member unless the scenario performance standards demand that the installation is required to re-start after a few days.
The above method will indicate the protection of non-essential members from the point of view survival of the installation.
The temperature calculation for each member is performed and measures are taken to restrict the temperatures to values below the critical temperature usually by the application of PFP (Passive Fire Protection).
It will also be necessary to check that radiation levels on escape ways remain at acceptable levels (i.e. below 2.5 Kw m-2), to allow for personnel on board to escape.
3.4.3 Strength level analysis
Strength level analyses are conventional linear elastic analyses as used in design against environmental, operating and gravity loads. The loads used in such an analysis should be in a form which could be interpreted as a load case as used in the design process.
The transfer of conclusions and load characteristics from the analysis of a geometrically similar platform with similar structural and process characteristics is acceptable for strength level analyses.
In a strength level analysis a maximum allowable temperature in a steel member is assigned based on the stress level of the member prior to the fire such that the utilization ratio remains less that the corresponding value given in Table 3.6 below.
The maximum allowable temperature in a steel member as a function of utilization ratio is given in Table 3.6 below.
Table 3.6 - Maximum allowable steel temperature as a function of Utilisation Ratio [3.5]
Maximum member temperature Yield strength reduction factor
Member UR at 20 °C to give UR = 1 at max. temperature
°C °F
400 752 0.60 1.00
450 842 0.53 0.88
500 932 0.47 0.78
550 1022 0.37 0.62
600 1112 0.27 0.45
If the primary structure on the installation have been designed for all credible fire scenarios to then the primary structure will be acceptable from a fire resistance point of view.
Higher strain levels than 0.2% may give a proportionately higher decrease in Young’s Modulus giving an unmatched reduction in yield strength and Young’s modulus. Strength level checks may then give utilization ratios above unity. A Ductility level or ‘elastic-plastic’ method of analysis may be required.
Table 3.7 gives the maximum allowable steel temperature as a function of strain. The peak temperature of each member needs to be determined for the fire event to check if the structural response remains elastic.
Table 3.7 -Maximum allowable temperature of steel
Maximum Allowable Temperature of Steel Strain %
°Celsius °Fahrenheit
0.2 400 752
0.5 508 946
1.5 554 1029
2.0 559 1038
3.4.4 Scenario or performance based strength level analysis
In a performance or scenario based approach, the first task involves the definition of credible fire scenarios from the failure probabilities of vessels and piping, the inventory pressures local to the release point, the material released, the emergency shut down systems available and the ventilation conditions. This information will normally be supplied by the Safety and Process disciplines.
A scenario based Strength level analysis is performed in the following general stages:
1. Definition of a fire scenario. (See Section 3.2.1.
2. The time history of the rate of release is calculated.
3. If required, the probability of occurrence of the release can be estimated from published failure statistics and the numbers of past failures which are available for most types of vessels, flanges and process equipment generally [3.7, 3.8, 3.9, 3.10, 3.11 and 3.12].
4. Calculation of the burning rate for the fire geometry enables the extent and duration of the fire to be determined.
5. The heat output from the fire in terms of radiation and the convection of hot air and gases may then be determined (see also Sections 5.3, 5.4 and 5.5).
6. The heat incident on structural members and panels may then be used to calculate the temperature/time history of the member.
7. For load bearing members, the temperature determines the appropriate values for the yield stress and Young’s modulus of the material of the member to be used in the structural analysis performed.
8. For panels and firewalls, which are usually non load bearing, the important parameters are the temperature of the cold face and the time to reach certain limiting temperatures which determine the walls’ rating.
9. It will also be necessary to check that radiation levels on escape ways remain at acceptable levels where immediate injury will not caused (i.e. below 2.5 kW m-2).
10. A utilization ratio of up to 1.7 will be acceptable for members loaded in bending if a small amount of plastic deformation is acceptable. A different utilization ratio will be appropriate for detection of buckling. Shear stresses should be kept within the yield stress for the material at that temperature. Alternatively the yield stress may be enhanced by a factor of 1.5 to take account of the fact that fire is an accidental load.
11. Modified code checks may be made on the structural members and if load re-distribution is neglected then the material effects and isolated plasticity may also be taken into account in the analysis.
12. The occurrence of plastic hinging may be taken into account by factoring the acceptable utilization ratio by the ratio of the plastic ‘Zx’ to elastic section modulus ‘Sx’.
This factor is generally greater than 1.12 and will be in the range 1.1 to 1.5. The
13. Use of the critical temperature approach may give an efficient scheme for the application of PFP in combination with a full non-linear elastic-plastic (progressive collapse) structural analysis. This type of analysis is referred to as a ductility level analysis for fire or explosion response calculations.
3.4.5 Redundancy analysis
An elastic linear analysis is performed to determine the minimal structure which will fulfil the requirement that escape ways will remain available for sufficient time to allow escape and that the TR integrity is maintained during and after a fire event.
A structural redundancy analysis will determine which members are essential for the above.
Protection of these members with PFP will complete the assessment on the basis of a redundancy analysis.
The determination of the critical structural members may be performed using an elastic structural frame model as follows:
1. Eliminate all non-critical structural elements by inspection, with due regard to escalation potential;
2. Remove all members identified in step 1;
3. Modify the static loading to represent the probable load at the time of the fire 75% of the loads associated with process contents and storage may be used as suggested for Earthquake analysis [3.6];
4. Remove safety factors in the code check, enhance the yield stress by a 1.5 factor, or allow a correspondingly higher utilization;
5. Identify those members with the highest utilization ratios - particularly relating to stability using the frame model;
6. Remove these members from the geometry;
7. Repeat step 6 and assess the remaining structure at each stage.
3.4.6 Ductility level analysis
A ductility level analysis may be required for Medium or High Risk installations. This method of analysis can take into account the load re-distribution which takes place when structural components fail and the time to failure of the structure considered.
A number of options for the linearization of stress/strain relationships at elevated temperatures exist. If the software used for the ductility level analysis allows temperature dependent stress/strain curves to be input then the linearization will not be necessary.
1. The levels of heat radiation and convection from the selected fire scenarios are calculated.
2. The time history of the increase of temperature of the structural components is derived.
3. Conduction of heat from neighbouring structural components will also occur but may generally be ignored in the primary framing analysis.
4. Once the steel temperature at a given time is known, the reductions in yield stress and Young’s modulus may be calculated.
5. Failure of a structural member is defined as collapse (where increasing displacement results in no net increase of capacity) under the imposed static gravity and operating loads.
6. In investigating the effect of a fire the ‘live’ loads such as contained liquids and storage may be taken as 75 % of their maximum values as is the case for the consideration of Earthquakes. Alternatively, live loads may be taken as the values used in the fatigue analysis performed for the installation if these have been properly derived.
7. Optimization of PFP (passive fire protection) thickness is rarely worthwhile as application of PFP to a given thickness is not sufficiently controllable. The thickness of PFP is controllable at best to within about 3 mm.
8. The scenario based strength level analysis method will not detect failures at intermediate or later times caused by thermal restraint from cold members. This is, in any case, an unlikely event in the context of offshore topside structures. Imperfections or deflections for example due to a previous explosion will not be taken into account. It will be necessary to use a ductility level analysis to take these effects into account.
9. In view of the fact that a single scenario is only one among many, the spatial variation of thermal loading is not generally meaningful. It is unlikely that this level of analysis will be necessary unless a single extreme event such as a riser failure or blow-out which puts the whole installation at risk is being considered.
10. It will also be necessary to check that radiation levels on escape ways remain at acceptable levels (i.e. below 2.5 kW m-2).
3.4.7 Assessment of fire barriers
Fire barriers are given a ‘rating’ derived from the SOLAS (Safety of Life at Sea) [3.4]
classification system for use on ships. Originally they were developed for cellulosic fires as opposed to hydrocarbon fires, which are more severe. The type of fire is represented in a furnace test where the firewall is in contact with a furnace with a well-defined temperature/time relationship. The hydrocarbon fire curve has a higher rate of temperature rise and attains a higher peak temperature.
The three main ratings used offshore are:
B Maintains stability and integrity for 30 minutes when exposed to a cellulosic fire. The temperature rise of the cold face is limited to 140 °C for the period in minutes specified in the rating. i.e. A30 has a 30 minute time period during which temperature rise is below 140 °C.
A Maintains stability and integrity for a period of 60 minutes when exposed to a cellulosic fire. The temperature rise of the cold face is limited to 140 °C for the period specified in the rating.
H Maintains stability and integrity for a period of 120 minutes when exposed to a hydrocarbon fire. The temperature rise of the cold face is limited to 140 °C for the
J Currently proposed one in latest draft version of the ISO (22899-1) [3.13]; identifies Type of application / Critical temperature rise (°C) / Type of fire / Period of resistance (minutes)
Here retaining ‘stability and integrity’ means that the passage of smoke and flame is prevented and that the load bearing components of the barrier do not reach a temperature in excess of 400 °C.
Insulation failure is also deemed to occur when the average temperature rise on the unexposed face of a separating element exceeds 140 ºC or the maximum temperature rise exceeds 180 ºC, whichever occurs first. These limits are to prevent combustion of any material which may be close to the unexposed face. Their origins are unknown and, in many cases, the limits may be excessively conservative.