Gas Explosions in Compartments, Buildings and Offshore Modules
- 10.1 Gas Explosions and Venting
- 10.2 Shape of Compartment
- 10.3 Types of Vent Areas
- 10.4 The Effect of Congestion and Obstructions
- 10.5 Ignition
- 10.6 Gas Cloud
- 10.7 Deflagration to Detonation Transition
- 10.8 Explosion Outside a Compartment
- 10.9 Mechanical Ventilation System
- 10.10 Fire, a Common Event after a Gas Explosion
- 10.11 Water Deluge
- 10.12 How to Estimate the Loads from Gas Explosions in Compartments
- 10.13 Guidelines
If fuel is accidentally released inside a partly confined area or if combustible gas is drifting into such an area, serious explosions may occur. The consequences of such explosions will depend on several parameters, such as type of fuel, size and concentration of the gas cloud, ignition and geometrical layout, i.e. confinement and obstructing objects. In consequence analyses all these factors have to be taken into account. Variations of these parameters may result in large changes in peak explosion pressure.
As discussed in Chapter 5, confinement and obstructing objects are key factors for the development of high explosion pressures in accidents. In buildings, offshore modules and partly confined areas containing process equipment, there will be confinement and obstacles. Walls, roofs, floors and decks will confine the gas cloud. The process equipment and piping engulfed by the cloud will act as obstructing objects during an explosion. By following simple guidelines while designing or modifying compartments, one can reduce the hazard potential significantly.
In this chapter we will mainly focus on simple guidelines for improving gas explosion safety and discuss methods for predicting gas explosions in compartments. The objectives of this chapter are to:
- Describe how a gas explosion behaves in a compartment and explain which factors are important for the pressure build-up.
- Present and discuss simple guidelines and possible methods for mitigating gas explosions in compartments.
- Discuss why it is difficult to predict the consequences of gas explosions.
Present a review of FLACS results as a database for evaluation of expected explosion pressure in compartments.
The objective of this section is to illustrate how geometrical conditions such as confinement, obstacles and venting influence explosion pressure in accidental explosions.
Gas explosions are a result of liberation of chemical energy due to flame propagation (i.e. combustion) through premixed fuel-air clouds. As discussed in Chapters 5 and 6, in a premixed cloud the flame can propagate in two distinct modes; the deflagration and the detonation. The flame speeds of deflagrations range from a few m/s up to 500 -1000 m/s, resulting in overpressures from close to zero up to several bar (see Figure 5.5). The detonation is a supersonic combustion wave causing explosion pressure in the 20 bar range. When discussing gas explosions and flame propagation in this chapter we refer to the deflagrative mode of flame propagation if nothing else is stated.
The pressure build-up during a gas explosion is governed by the balance between pressure generation by the flame, and relief of the pressure through venting. In an unconfined situation or in a compartment with large explosion vent areas, a flame speed in excess of 100 m/s is required in order to obtain damaging pressure waves (see Figure 5.5). However, if a fuel-air cloud explodes within a compartment with no or very little venting, even slow burning can cause pressure build-up. In extreme cases, a slow flame can in a closed compartment cause pressures up to 8 bar, if the compartment does not disintegrate.
In an accidental gas explosion the flame will normally start out as a slow laminar flame with a velocity of the order of a few m/s. If the cloud is truly unconfined and unobstructed (i.e. no equipment or other structures are engulfed by the cloud) the flame is not likely to accelerate to velocities of more than 20-25 m/s, and the overpressure will be negligible.
Figure 10.1. Gas explosion in a partly confined area containing process equipment.
In a partly confined area with obstacles as shown in Figure 10.1, the flame may accelerate to several hundred meters per second. The main mechanism of flame acceleration under such conditions is turbulent mixing due to generation of turbulent flow fields ahead of the flame.
Figure 10.1 shows a compartment filled with a premixed combustible fuel-air cloud. The cloud is ignited in the centre of the compartment. When the flame consumes the fuel-air cloud, the gas expands. This expansion can be up to 8-9 times the initial volume. Due to the expansion, unburned gas is pushed ahead of the flame, and a flow is generated in the compartment. Some of the unburned gas will in the early phase of the explosion be pushed outside the compartment through the vent openings (i.e. open parts of the compartment). Inside the compartment the gas has to flow through and around process equipment, piping etc. The process equipment and piping will obstruct the flow and generate turbulence ahead of the flame.
In Figure 10.1 the explosion starts in the centre of a compartment. Since the vent area is located only on the right hand side of the compartment, the dominant direction of the flow and flame propagation will be towards the vent area. The flow in this area will be turbulent due to the obstructing effect of process equipment etc. and this turbulence will support the flame acceleration. The location of the ignition point relative to the vent opening is a key factor for how the flow field or turbulent flame acceleration develops during a gas explosion.
A consequence of the above statement is that the ignition point location relative to the location of the vent opening is also very important for the effectiveness of the venting. When the flame front reaches the vent opening, combustion products will start to flow out from the vent opening. Since the hot combustion products have a much higher sound speed (~900 m/s) than the unburned fuel-air mixture (~340 m/s), the flow velocity through the vent will increase when the hot combustion products start to vent. In experiments and FLACS simulations we have often seen that the pressure starts to drop immediately after the combustion products have reached a major vent area. In addition to the enhanced venting, the venting of hot combustion products may also influence turbulence generation and flame acceleration. If hot combustion products are vented out of a compartment, the flow and the turbulence can be reduced since the driving pressure is relieved and less gas is pushed ahead of the flame. Venting of combustion products as a way of minimising the positive feed-back mechanism that causes flame acceleration is illustrated in Figures 5.7 and 5.9. However, in some cases we will find that pressure will continue to rise even if the combustion products are vented. This is typical for compartments with small vent areas, high density of process equipment, piping etc., or when the gas explosion reach high pressures before the combustion products start to vent. For very high-speed deflagrations venting may not be effective at all.
Many studies have aimed at identifying the necessary vent area to relieve overpressures for a confined volume. Unfortunately there is a wide spread between simple model predictions and experimental results. Harrison and Eyre (1987b) claim that the models do not properly take account of:
i) turbulence inside the enclosure
ii) acoustic resonance inside the enclosure, and
iii) combustion of gas outside the enclosure.
Figure 10.2. Maximum overpressure observed in a 50 m3 tube, Moen et al. (1982), compared with the upper pressure bound (solid curve) based on vent areas for central ignition in near-spherical vessels as proposed by Bradley and Mitcheson (1978).
In Figure 10.2 a comparison is shown between the maximum overpressure from CMR experiments (Moen et al., 1982) and the commonly used Bradley and Mitcheson safe recommended vent area (Bradley and Mitcheson, 1978). The CMR experiments were performed in a 50 m3 vessel (a tube 2.5 m in diameter and 10 m long, open at one end) with regularly spaced obstacles in the form of orifice plates. The gas mixture was methane-air. The variation in the experimental results for a constant vent area corresponds to different internal geometries in the tube. More details from these experiments can be found in section 5.3. The experimental results show that simple models are inadequate for such conditions. This has further been confirmed in the review report by British Gas for the Department of Energy (1989) "Review of the Applicability of Predictive Methods to Gas Explosions in Offshore Modules".
A gas explosion in a compartment is a very complex process strongly depending on several parameters. In the following sections we will discuss these parameters separately.
The objective of this section is to discuss the influence of the shape of the compartment on flame acceleration and pressure build-up, and to point out what is the optimum compartment shape to keep the explosion pressure as low as possible. The shape of the compartment and location of vent areas are closely linked and will therefore depend on each other.
There are mainly three principles to apply when optimising the shape of a compartment.
i) From the ignition point the flame should be able to propagate in a spherical mode for as long as possible (see Figure 5.5).
ii) An ignition point anywhere in the compartment should be as close as possible to the major vent areas, so hot combustion products can be vented out in an early phase of the explosion.
iii) Avoid strong turbulence in the unburned gas ahead of the flame and long flame travel distances
Figure 10.3. For a given volume a cubical compartment gives the best explosion venting in the case of vent areas on the two end walls.
For a compartment with explosion venting on two end walls the ideal shape is a cubical box. In such a configuration a relatively low explosion pressure can be expected.
If the module is elongated and vent openings are only located on the two ends most explosion scenarios will give high pressures. The situation in an elongated module is in principle the same as the channel in Figure 5.7. The flame can travel over a long distance and the conditions, i.e. limited venting, will support the flame acceleration. The flame will propagate in a planar propagation mode (one-dimensional propagation) in the main part of the module.
For the cubical module the flame will propagate in a spherical mode. A spherical propagation mode requires higher flame velocity than a planar mode to generate the same explosion pressure. The pressure wave can expand more "freely" in the spherical mode (three-dimensional propagation). (See Figure 5.18).
If the compartment has a vent opening only in one of the side-walls, it is even more important to avoid an elongated shape. In case of ignition at a closed end wall, the flame can accelerate over a long distance and venting has no beneficial effect since it only leads to flow past obstacles and thereby turbulence generation.
Figure 10.4. The height of the module can be important.
We have seen in our work that the height of the compartment is often important, as illustrated in Figure 10.4. By increasing the height of the module the explosion pressure can in some cases be reduced. However, the advantage of increasing the height of a compartment or the smallest side of the compartment, depends also on how densely packed the compartment is with obstructing objects (i.e. process equipment and piping). In compartments with a lot of obstructing objects, there may be little or no advantage in increasing the height. In such situations, the obstructing objects are controlling the flame propagation and the shape of the compartment is less important.
In compartments with low density of obstructing objects it may be beneficial to replace solid decks with grated decks and thereby create a more cubical shape of confinement. Such an action should be viewed in relation to gas dispersion and fire hazards.
For an elongated compartment it is necessary to open up at least one of the long sides for venting if we want to have venting close to a randomly chosen ignition point. In a building the possibility may exist of venting through the roof. In discussing flame acceleration in Chapter 5, it was pointed out that ignition close to a vent area will cause venting of hot combustion products in the early phase of an explosion. This is an effective way of minimising flame acceleration and high explosion pressures. This effect can be utilised by venting through three walls of the module as shown in Figure 10.5.
Figure 10.5. Elongated module with explosion venting on three sides.
In some FLACS simulations and experiments we observed a factor of 10 reduction in explosion pressure by opening one of the long side walls in a module. The roof, deck or a side wall should be considered as possible venting areas.
Unfortunately we have also seen examples of flames that have been able to accelerate and cause high explosion pressures even with venting on three sides. These are cases where the compartment is large and contains many obstructing objects. One way of mitigating the consequences of gas explosions in such cases may be to introduce a solid blast wall in the central part of the compartment as shown in Figure 10.6.
Figure 10.6. Elongated module with venting on three sides and a solid wall in the central part of the compartment.
The solid wall will prohibit a strong turbulent flow ahead of the flame and guide the flow out of the compartment. The length of flame travel will be reduced. The disadvantage of introducing a solid wall is the reduced natural ventilation. Build-up of large homogeneous gas clouds is therefore more likely with a solid wall. This concept will be investigated further.
With vent openings we mean areas where gas can be relieved from the compartment during a gas explosion (pressure relief). The important factors for effective venting are:
i) Size of the vent area.
ii) How the vent area is distributed.
iii) Direction of explosion relief.
iv) For explosion relief panels: how quickly are they activated?
It is very simple to make guidelines for the size of the vent area. The vent area should simply be as large as possible.
In section 5.4 and Figure 5.15 some experimental results from CMR's 10 m long wedge-shaped vessel are presented. For distributed venting in the top plate (i.e. along the long side-wall) the explosion pressure was low when the top plate was 50% and 80% open. However, the explosion pressure increased from a few mbar to more than 1 bar when the same vent area was located in the far end of the vessel with respect to the ignition point. The consequences of a gas explosion are strongly linked to both the size and the distribution of the vent area. Since we normally do not know where the ignition point is located, the general recommendation for locating vent areas is to distribute the vent areas around the side-walls of the compartment. It is important that vent areas guide the flow ahead of the flame away from obstacles!
As a general principle, the gas explosion venting should be directed into open areas with a minimum of obstructions. If one is venting into another compartment or a congested area, combustible gas clouds may be pushed into this area and a violent multi-compartment explosion may occur. This can be investigated by using the FLACS code.
In many situations it is not practical to have open walls. Weather conditions, noise reduction and fire protection may require closed walls or partly closed walls. If these initially closed walls shall act as vent openings during an explosion, they must be lightweight and designed to open quickly.
Figure 10.7 shows the displacement versus distance for wall elements from 10 kg/m2 to 300 kg/m2 when subjected to a triangular pressure pulse with a duration of 100 msec and a peak pressure of 1 barg.
The 100 kg/m2 and the 300 kg/m2 walls will move very short distances within the duration of the pressure pulse. Walls in this weight range will not act effectively as vent panels. They start to move after the explosion and their only contribution will be to act as dangerous projectiles. The 10 kg/m2 panel is moving away fairly quickly . Panels of this weight or lighter, will normally be effective vent panels if the opening mechanism is carefully constructed. Our experience is that panels of 30 - 50 kg/m2 are too heavy to have a significant effect on peak pressure. But, some reduction in the duration of the explosion pressure may be expected. The vent areas that are initially closed should not be heavier than 10 kg/m2, preferably lighter to be effective.
Figure 10.7. Displacement distance for wall elements subjected to a triangular pulse of 1 barg peak pressure and 100 msec duration.
In the last part of this section, we will discuss the different types of explosion vents commonly used. They can be listed as follows:
i) open walls
ii) louvered walls
iii) solid wall /cladding
iv) relief walls (also called wind walls or weather cladding) opening during an explosion
v) glass windows (not recommended)
The open wall is normally the best solution from an explosion point of view. If a large part of the module is open, the natural ventilation will be good and explosive cloud formation will be less likely. If an explosion should occur, the open wall will relieve the explosion pressure as well as is possible. However, due to weather conditions, fire protection and noise reduction requirements, fully open walls are often impractical or unacceptable.
A louver wall will also act as a vent area, but the effective vent area will be about half that of the open wall. Louver walls for offshore applications weigh typically 40-50 kg/m2. Even though these walls may be designed to open up during an explosion, i.e. break loose at low static overpressure, they will normally not open up fast enough to improve venting. Bjørkhaug (1988b) tested experimentally the behaviour of louver panels. FLACS simulations of explosions in offshore modules of approximately 10,000 m3 with louvers on three sides indicate that the opening of the louver wall does not reduce the peak pressure, but may reduce the duration of the positive pressure phase.
A relief wall, also called wind wall or weather cladding, is an alternative combining the requirements for acceptable working environment and gas explosion safety. A relief wall is mainly a frame with a thin metal plate covering the frame. If an explosion should occur, the metal plate will break loose on the rim and collapse. The principle of a relief wall is shown in Figure 10.8. This panel opens at 50 mbar overpressure and is fully open after about 40 msec.
Figure 10.8. Relief panel type Stord Industrier AS
Critical parameters for selecting relief walls should be:
- acceptable natural ventilation must be ensured
- the opening mechanism must allow for fast release
- the weight of the panel (kg/m2) must be low
dangerous fragments (projectiles) should be avoided
The use of relief walls (wind walls) should be limited so that acceptable natural ventilation is obtained under normal operation. Without natural ventilation even a small gas leak can build up a hazardous gas cloud (unless sufficient forced ventilation is available).
A relief wall should open as early as possible during an explosion, but not open due to wind. Our experience is that the design of the opening mechanism is not straightforward. Experimental testing with dynamic loads (i.e. explosion testing) appears to be required. Static testing of the opening mechanism may not produce relevant information. A panel that has a static opening pressure of 50 mbar, may not open before the pressure reaches 100-200 mbar if the load is a dynamic load from a gas explosion.
Figure 10.9. Explosion pressure versus opening pressure of relief walls from FLACS simulations. (Note: Different results will be obtained with different geometries).
Figure 10.9 shows some results from FLACS simulations in an offshore module with relief walls on three sides. The simulations were performed with different opening pressures of the relief walls. In this geometry the explosion pressure increases with a factor of two when the relief wall opening pressure goes from 0 (i.e. open wall) to 150 mbar.
The weight of the panel (kg/m2) will indicate how fast the panel will move after it has started to open. One should select relief walls with low panel weight. Relief walls with a weight of 510 kg/m2 are commercially available today.
When the panels are blown open by an explosion they should be designed so that no dangerous fragments are generated. Flying fragments or even the flame jet can cause damage to piping or equipment and also hurt personnel.
The FLACS code is capable of handling explosion relief walls with different opening pressures and weights. In addition to that FLACS can discriminate between hinged panels and "pop-out" panels.
In existing facilities, such as old process plants, buildings etc. we may find large window areas that were intended for, or will act as vent areas during an explosion. Ordinary glass windows will break when exposed to explosion pressures of 2070 mbar (Harris 1983). But the dangerous fragments generated when glass windows break, is a very serious problem. Accidents, such as in Flixborough in 1974 and in Pernis in 1968 show that fragments from windows can cause both serious injuries and a large number of injuries (Lees,1980). It is not recommended to have ordinary glass windows in areas where gas explosions can occur. Therefore apply the information given above to determine how to replace windows intended for explosion relief with more proper relief walls.
A compartment will contain process equipment, pipework, rooms etc. During a gas explosion these objects will obstruct the flow and thereby cause turbulence. These objects will also interfere with the explosion venting. As discussed in Chapter 5, turbulence and venting are very important for the flame acceleration and pressure build-up in gas explosions. In this section we will discuss how to arrange obstructing objects in order to keep the explosion pressure at a minimum. The main principle is to arrange the obstructing objects so that:
i) minimum turbulence is generated and
ii) explosion venting is not blocked.
Figure 10.10 shows the top view of two different layout arrangements in a compartment. The compartment has venting on the two end walls. The obstructing objects consist of two vessels and a room.
Figure 10.10. Top view of two compartments. In the layout on the left side the room blocking the vent area and the vessels are generating turbulence by acting as repeated obstacles. The right side show an improved layout.
In the first layout, the room will block the main parts of the vent area on the right hand side and the vessels in the left part of the compartment will cause reduced venting and flame acceleration, respectively. In Chapter 5 the effect of repeated obstacles is discussed. The vessels arranged on the left side in Figure 10.10 will act as repeated obstacles for a centrally ignited explosion. In an explosion turbulence will be generated in the wake of the obstacles, as shown in Figure 5.7. In the turbulent wake the flame will burn very fast and the positive feedback mechanism for flame acceleration will be activated. The result may be high explosion pressure.
The flame acceleration due to turbulence depends on the arrangement of the equipment and on the turbulence level in the flow field. It is very important to arrange the equipment in such a way that a minimum of turbulence is generated during an explosion. This is normally obtained when the longest side / dimension of the equipment is parallel with the flow direction during an explosion, i.e. pointing in the direction of the vent area. In Figure 10.10, the right side shows the vessels pointing in the direction of the vent area. This is a better arrangement than the layout on the left hand side.
Layout of different types of equipment should not be viewed in isolation. We may find situations where relocating major equipment will add to piping, hence pressure may rise to a higher level with the new equipment layout, even though the situation for the major equipment appears to be improved. It is important to avoid sub-optimisation.
In Figure 10.10 the room blocks the vent area on the right side. In this case, a much better solution would be to rotate the room 90° and if possible move it to the central part of the compartment. The vent area on the right hand side will then become more effective.
Even in a compartment with venting on three sides, the location of rooms can be very important. Figure 10.11 shows a bad and a good example of how to arrange a room in a compartment. In the bad example the solid wall and the room will form a compartment. In FLACS simulations of similar geometries we have seen high pressures predicted in the area confined by the solid wall and the room. The passage between the room and the solid wall will act as a funnel, the flow is forced through this passage and high pressure is generated.
The layout to the right in Figure 10.11 is a much better solution. In this case ignition anywhere in the compartment will be fairly close to the vent areas and confinement is at a minimum.
Figure 10.11. Top view of two compartment layouts. In the layout on the left side the room blocks the vent area.
Venting can also be hindered when obstructions are placed outside the protected room. An example is obstructions placed on a laydown area in front of a vent opening. Hence the presence of a laydown area in front of a vent opening should be avoided. Intermodular gaps should be as wide as possible (see also section 5.8).
Both the strength and the location of the ignition source can be important factors for the consequences of the gas explosion.
In section 5.9 it was shown that jet flame ignition of the cloud could cause very strong explosions even for unconfined situations. If a cloud is ignited by detonating a high explosive charge within the cloud a detonation could be initiated directly.
Even though extreme ignition scenarios exist, the most likely scenario is a weak ignition source like a hot surface or a spark. In consequence analyses it is common to assume a weak ignition source.
Various experiments and FLACS simulations have shown that explosion pressures can be very sensitive to the location of the ignition point. In many scenarios the peak explosion pressure can be changed by an order of magnitude if the ignition is moved from a worst case location to a more favourable place. In general the lowest pressure is obtained if the ignition point is :
i) close to the vent area or
ii) at the edge of the cloud
but as we will come back to in the end of this section, there are exceptions to this.
Figures 5.7 and 5.10 show how repeated obstacles generate turbulence, while venting of combustion products reduces the turbulence generation. By igniting near the vent opening the combustion products will be vented and the flow velocity and the turbulence in the unburned mixture will be low. Figure 10.12 shows how different flow regimes will be generated in the same geometry with different ignition locations. In case a) the flow velocity ahead of the flame will be low if the compartment is not too long. In case b) a high flow velocity will be generated ahead of the flame which will support a high burning rate and cause high explosion pressure. For simplicity obstacles have been omitted from the figure.
Figure 10.12. Differently located ignition points in a compartment. In case a) the flow velocity ahead of the flame will be low, and low explosion pressure can be expected. Case b) is a " worst case" scenario since the flow ahead of the flame will be turbulent and therefore cause rapid burning and high pressure.
However, if the venting of combustion products is not sufficient to keep the flame speed at a low level, ignition at the edge may cause higher explosion pressures than central ignition. Figure 10.13 shows an example of this. In this case the length of flame travel is a more important factor than the venting of the combustion products. By increasing the length of flame travel, the flame will have the possibility to accelerate over a longer distance, by passing a greater number of obstacles. This effect will be most pronounced for reactive fuels, high density of obstructions, small vent areas and large scale.
Figure 10.13. Flame speed versus distance for centrally and edge ignited explosions in a double-configuration (i.e. solid top plate) with obstacles (van Wingerden and Zeeuwen, 1986).
The practical implication of this is that one should try to locate potential ignition sources away from worst-case locations.
In an accident situation the combustible gas cloud in an obstructed and/or partly confined area may only fill a part of the volume at the time of ignition. The filling ratio is, of course, an important parameter. But in some situations 30-50 % filling ratio may cause the same explosion pressure as a 100 % filled compartment. The reason for this is that during an explosion the gas that burns will expand and push the unburned gas ahead of the flame. Thereby air or fuel-air is pushed out of the compartment. As discussed in section 4.9 the expansion of the combustible cloud on burning can be up to 8-9 times the initial volume. Figure 10.14 illustrates how a small cloud upon burning is pushing out air from a compartment and thereby fills the whole compartment with a combustible cloud.
Figure 10.14. During an explosion of a small cloud air can be pushed out through the vent area and thereby the whole volume can be filled with a combustible cloud.
Pappas (1983) made some simple calculations on the effect of having only a part of the compartment filled with a gas cloud. He is assuming that the ignition point and the gas cloud is far from the vent opening. His results are shown in Figure 10.15. The explosion pressure starts to drop at about 30% filling ratio.
Figure 10.15. Pressure reduction in a partly confined compartment as function of gas filling ratio. Gas cloud and ignition away from the vent opening. (Pappas 1983).
An explosion in a partly filled compartment can in some instances cause the same explosion pressure as in a 100% filled compartment. It should be added that when the cloud is only filling a portion of the enclosure, the ignition point location will be a much more sensitive parameter. If the ignition occurs at the edge of the cloud and/or close to the vent area we can expect lower pressure for the partly filled, than for the 100% filled case.
A deflagration is a sub-sonic combustion wave with respect to the unburned gas ahead of the flame, and is a common mode of flame propagation in an accidental gas explosion. The deflagration pressure goes from zero to several bars depending on confinement and flame speed. A detonation is a supersonic combustion wave propagating at 1500-2000 m/s in fuel-air and the pressure is 15-20 bar. In an accidental explosion the ignition is normally a weak ignition source, e.g. a hot surface or a spark. In this situation the explosion will start out as a slow burning deflagration. Due to obstructing objects and confinement, the deflagration can accelerate and become a fast burning deflagration. When a deflagration becomes sufficiently rapid, a sudden transition from deflagration to detonation can occur. Presently there is no theory which can predict transition from deflagration to detonation. There are therefore great uncertainties related to the transition process and in practical situations it can be hard to evaluate the possibility of deflagration to detonation transition.
It is very important however, to know if transition to detonation can occur. If it occurs, very high pressure loads, in the order of 50 bar, can be reached locally and severe damage can be expected within the compartment. If a detonation has been established in the compartment it may also propagate into the unconfined cloud outside. The existing criteria for propagation and transmission of detonations are described in section 6.4.
A deflagration propagating into a large truly unconfined and unobstructed cloud will slow down and the pressure generation will normally be negligible. A detonation, however, will propagate through the entire cloud at a high velocity and cause severe blast waves.
The possibilities of transition to detonation will mainly depend on i) type of fuel, ii) size of cloud and iii) geometrical conditions, such as obstructing objects and confinement.
As shown in Figure 5.23 the flame acceleration will depend strongly on the type of fuel. Fuels like hydrogen, acetylene, ethylene-oxide and ethylene, are the most likely fuels to detonate. There are several examples of accidental explosions where hydrogen has detonated.
Fuels like propane and butane may also detonate, but a strong deflagration is required to initiate the detonation.
In methane it is difficult to initiate detonations. It is still uncertain whether it is possible to get a detonation in an accidental explosion with methane. Full scale tests in large volumes, like an offshore module, would be required to test this out.
For natural gas which mainly contains methane and various amounts of higher hydrocarbons, the content of higher hydrocarbons is important. Bull et al. (1984) have shown that even small quantities of higher hydrocarbons increase the sensitivity and thereby the likelihood for transition to detonation considerably.
In a practical situation, presently the most effective way of mitigating the occurrence of detonations is to avoid situations where a deflagration can accelerate to a condition where transition from deflagration is possible, i.e. a high pressure deflagration.
An explosion inside a compartment may lead to strong turbulent jets of fuel-air shooting out from the compartment's vent openings. In some situations the explosion in the jet can have significant strength and it may cause pressures as high as, or even higher than inside the compartment. Explosion in such jet flames is discussed in section 5.7. In this section it is shown that a jet from a long pipe can cause a strong explosion outside. This has also been shown in FLACS-simulations (Moen et al., 1989). Our experience from experiments with a 1:5 scale offshore module and FLACS simulations is that an explosion outside a compartment is not dominant as long as the compartment is filled with obstructing objects, or if the compartment is not very long and narrow (tunnel-like).
In our consequence analyses we therefore normally neglect the contribution from the external explosion. But this must not be confused with the blast wave outside the compartment. As discussed in Chapter 7, a strong explosion inside a compartment may result in strong blast waves propagating a long distance away from the explosion compartment.
There is no doubt that mechanical ventilation systems can counteract the formation of explosive gas clouds, if the release rates are small. However, for a massive release, a ventilation system may transport gas from one area to another. Furthermore, if ignition occurs and the explosion propagates into the ventilation channels, a violent explosion may occur within the ventilation system.
The emerging flame jet from an explosion in a ventilation channel can act as a very strong ignition source. If such a flame ignites a cloud in a module very high explosion pressure can be expected. This flame jet ignition has been experimentally observed to cause very violent explosions, and even transition to detonation for sensitive fuels.
To avoid this type of hazard the reliability and response time of shut down systems for the ventilation systems are critical.
Figure 10.16. Front elevation of a compartment. Ventilation ducts should not block the vent areas. Placing them behind the I-beams will be more beneficial.
In design of offshore modules we have seen that ventilation ducts have been implemented in the later phase of the detailed engineering and that these ducts have blocked significant parts of the vent areas. Figure 10.16 shows a bad and a good location of a ventilation duct in an offshore module or another building. By locating the ventilation ducts behind the I-beams they will not lead to additional blocking of the vent openings of the module.
In design one should try to locate ventilation ducts in such a way that they do not block the vent openings. They should also be taken into account as early as possible in the design.
A gas explosion in a compartment will often be followed by serious fires.
The source of the fire can either be
i) the initial leak source that caused the formation of the explosive cloud, or
ii) new release source(s) due to equipment or piping being damaged by the initial explosion.
To avoid new releases, it is important that piping, equipment and their supporting structures are designed to take the loading from the explosion.
As a result of a violent gas explosion, walls or decks may start to move or even break down and fragment. Pipes that are suspended on a moving wall may be sheared off (i.e. guillotine break) as a result of the relative movement of points of suspension. Piping from one module to another module may have to respond to relative structure movements. Cables and control lines may also be damaged due to this type of relative movement.
To illustrate this phenomenon a case history from a gas explosion in an onshore petrochemical process plant is included. The events are shown in Figure 10.17. It started with a violent explosion inside a building. On the roof of this building there was a pipe bridge with a 0.3 m diameter pipe crossing over to a pressurised vessel containing hydrogen. As a result of the explosion the roof was lifted about 1 m and the hydrogen pipe was sheared off.
The result was that large quantities of hydrogen leaked out from the vessel. The hydrogen caught fire and a very intense jet flame burned until the reservoir was empty. The length of the jet flame was about 50 m.
Other phenomena that may cause deformation or damage to the piping system are:
i) drag forces due to the explosion wind and
ii) flying fragments that may cut or break weak connections like instrument lines.
Figure 10.17. A case history: Jet-fire as a consequence of a gas explosion inside a building.
Experimental results (Acton et al., 1990, Bjerketvedt and Bjørkhaug, 1990, Bjørkhaug et al.,1990, van Wingerden et al., 1993, Catlin et al., 1993 ) have shown that ordinary water deluge for fire fighting can have a mitigating effect on gas explosions.
By request from the Department of Energy (D.En.), UK, the Chr. Michelsen Institute (Bjerketvedt and Bjørkhaug 1990) undertook a pilot experimental investigation addressing the effect of water sprays on gas explosions. The objective of this test programme was to identify any beneficial effect of deluge water sprays on overpressures generated by gas explosions.
The 1:5 scale model of an offshore module was used in the tests. The gas mixture was either methane in air (8.5 - 10 vol%) or propane in air (4.25 vol%).
In the tests the deluge system was activated prior to the ignition, i.e. the water spray droplets were inside the gas cloud at the time of ignition. A deluge density of 12 l/min per m2 was used as base case in the tests. This value is recommended in the Department of Energy Guidance Notes on Fire-fighting Equipment on Offshore Installations.
The recorded explosion overpressures ranged from 100 to 700 mbar. The propane-air tests gave about twice as high pressure as the methane-air tests. Some of the tests were performed without the deluge system activated (i.e., dry tests) as reference tests. In tests with central ignition no beneficial effect of the water deluge was observed, actually there was a slight increase in peak pressure. In tests with end ignition and louver walls close to the ignition point, a significant reduction of the explosion pressure was observed when the water deluge was activated. This reduction was as large as up to a factor of three. Figure 10.18 summarises the results.
Figure 10.18. D.En./CMR results
Figure 10.19 shows pressure records from two identical tests with and without water deluge activated. These are tests using end ignition. Here we can clearly see a positive effect of the water deluge.
Figure 10.19. Pressure records from deluge tests
British Gas have also reported results from water spray tests (Acton et al. 1990) (see Figure 10.20). In these experiments the pressure in the dry tests (i.e. no deluge) was in the range of several bar. For the off-shore module Case 2, no pressure for the dry tests was reported. It was only stated that the explosion pressure was several bar. Also for these tests the deluge had a positive effect on the explosion pressure. With the deluge activated, the explosion pressure was about 0.25 bar except for the pipe-rack tests with nozzle type 126.
Figure 10.20. British Gas results.
Recent experiments on waterspray were reported by Catlin et al. (1993). They were performed in two rig geometries, closed on all sides except for one side. On this side either a small or large vent area could be installed. In the experiments with the large vent opening the sprays substantially reduced the overpressures. But the experiments with a small vent opening resulted in higher pressures than those that would have been produced had the sprays not been activated.
From all these experimental programmes it was concluded that the effects of water spray on gas explosions seem to be twofold and competing:
i) The water spray interferes with the low velocity flame in the initial phase of the gas explosion or in situations where the flame cannot accelerate sufficiently, i.e. in compartments with little venting. This causes increased flame acceleration and faster pressure build-up.
ii) It is likely that mist is generated in the unburned gas mixture due to droplet break-up and stripping in the later phase of the gas explosion. The evaporation of the mist in the flame will result in water vapour diluting the mixture and thereby reducing the reaction rate or even stopping the reaction completely. As a result an important reduction of the explosion pressure will occur.
Further the experiments showed that the beneficial effect of waterspray systems increased when:
i) a larger number of nozzles was used.
ii) higher nozzle pressure were used.
iii) a more uniform spray distribution was used
Experiments reported by Wilkins and van Wingerden (1993) showed that the overpressure increasing effect of waterspray systems in more confined geometries and during the initial stages of flame propagation increased when
i) higher nozzle pressure were used.
ii) using nozzles generating higher water velocities .
iii) the water application rate increased.
From simple calculations we know that the droplets from standard deluge nozzles (> 100µm) will not evaporate in the flame front, because they are too large. In order to evaporate in the flame front the droplet diameter has to be 1-50 µm or less. Therefore in order to be effective the large droplets have to break up due to the wind in the unburned mixture ahead of the flame front. This is illustrated in Figure 10.21.
Figure 10.21. One mode of droplet break-up due to wind.
For break-up there must be a velocity difference between the gas flow and the droplet. This critical velocity difference increases with reduction of the droplet size, i.e. big droplets will break up easier than small droplets.
Experiments with very reactive fuels, such as hydrogen and ethylene, have shown that the flame acceleration due to droplets in the initial phase of the gas explosion, can be significant. It is likely that very reactive fuels are more sensitive to this effect. It may therefore be true that the water spray has the most positive effect on the least reactive fuels. No conclusive evidence, however, exists to support this statement.
For a detonation in acetylene-air, Jenssen (1990) has shown that ordinary deluge has no influence on a detonation wave. The time available for break-up within the detonation front is probably too short. Thomas et al. (1990) have shown that small droplets are required to get any quenching effect on detonations.
Unfortunately there are disadvantages related to the use of water deluge also. Since the activation time for an ordinary deluge system is much longer than the duration of the explosion, the deluge system has to be activated on gas detection. Accidents have been reported where the probable ignition source was a discharge in electrical equipment due to moisture from the deluge system. Water sprays and deluge should therefore be activated in compartments with waterproof electrical equipment.
Our general opinion about water sprays is that such systems seem to increase the likelihood of ignition, but they can reduce the pressure build-up, particularly in the high pressure cases.
The tests have given promising results, which indicate that water spray may be a future mitigating device for accidental gas explosions. However, further research is required in order to quantify and relate all the effects of water sprays in an explosion event.
The quality of predictions of gas explosions depends on:
i) the quality and approximations of the physical models and codes that have been used
ii) the representation of the geometry and the scenario parameters by the user
Back in the 60's and the 70's there were several attempts to make simple correlations, formulas or venting guidelines relating maximum explosion pressure and vent area. Figure 10.22 shows a comparison of results of these types of formulas.
Figure 10.22. Comparison of predicted maximum pressure from different formulas (venting guidelines) for prediction of explosion pressures in a compartment. (Solberg 1981)
As we can see there is a wide spread between these formulas. If we compare these results with experimental results we will find order of magnitude differences. The typical weakness with such formulas is that they do not take into account the location of the ignition point and vent areas, and they do not handle the generation of turbulence and flame acceleration. Most of these formulas are also based on small-scale and empty compartment experiments. In the report "Review of the Applicability of Predictive Methods for Gas Explosions in Offshore Modules" (British Gas for the Department of Energy, 1990), it is concluded that the use of these simple formulas with any degree of confidence, must be limited to situations involving empty single-chamber vessels up to volumes of the order of 200-300 m3 and with an aspect ratio of less than 3. Our view is that these simple formulas in almost all industrial situations (accidental releases) are not applicable and should not be used. They should only be used within their stated range of validity.
A much better way to estimate loads from gas explosions inside a compartment is to use numerical fluid dynamic codes such as the FLACS and the µFlacs codes, which are described in detail in Chapters 12 and 13.
Guidelines for equipment location can be given as follows:
- Do not locate large pieces of equipment or rooms near the module vent areas.
- Avoid laydown areas outside explosion vent areas. Containers etc. will block the venting.
- The longest side/dimensions of equipment should normally be parallel with the flow direction during an explosion, i.e. pointing in the direction of the explosion vent areas.
- Locate piping in low drag (i.e. explosion wind) zones.
Guidelines for module shape and explosion venting are:
- Approach a cubical module shape when explosion vents are placed on the two end walls.
- An elongated module needs venting on three walls.
- The two smallest module (confinement) dimensions should be of the same length.
Guidelines for venting:
- Make the vent area as large as possible.
- Distribute the vent area, or make it large close to the potential ignition source.
- Vent into open areas.
- Relief walls should be able to open up quickly and be light-weight.
- Relief walls should not cause any dangerous fragments (projectiles).
- Do not use windows (ordinary window glass) for venting.
To avoid fires as consequences of explosions one should:
- Design the piping and the supporting structure so that deformation of the module(s) will not cause strong deformation of the piping system.
- Locate piping in low drag (i.e. explosion wind) zones.
- Avoid constructions that during an explosion will cause flying fragments.
Guidelines for use of water deluge:
- The use of water spray should be made part of a hazard evaluation. Each case should be viewed individually, and both the positive and possibly negative effects of water spray should be addressed.
- Electrical equipment must be waterproof if water deluge is to be activated based on gas alarm.
- The design of the water spray system will depend on the layout of the compartment.
Guidelines for estimating the loads from a gas explosion in a compartment (see Ch. 12 and 13):
- Simple formulas are not recommended, unless the use is strictly within the applicability range. Almost all scenarios in industrial environments are outside the applicability of the simple formulas.
- The best way to estimate loads from gas explosions is to use advanced numerical codes, like FLACS.
- Use µFlacs as a screening / layout tool. Note that this code is not intended for design purposes, hence main conclusions based on µFlacs use should be verified by additional FLACS simulations.
Applying such tools correctly provides good results but is time-consuming and requires planning and expertise.