Review of expanded aluminum products for explosion suppression in containers holding flammable liquids and gases

Review of expanded aluminum products for explosion suppression in containers holding flammable liquids and gases

ARTICLE IN PRESS Journal of Loss Prevention in the Process Industries 21 (2008) 493– 505 Contents lists available at ScienceDirect Journal of Loss P...

470KB Sizes 3 Downloads 21 Views

ARTICLE IN PRESS Journal of Loss Prevention in the Process Industries 21 (2008) 493– 505

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries journal homepage: www.elsevier.com/locate/jlp

Review of expanded aluminum products for explosion suppression in containers holding flammable liquids and gases A.M. Birk  Department of Mechanical and Materials Engineering, Queen’s University, Kingston, Ont., Canada

a r t i c l e in fo

abstract

Article history: Received 28 January 2008 Received in revised form 7 April 2008 Accepted 9 April 2008

For over three decades, products made from expanded aluminum have been available for explosion suppression in fuel tanks and containers. You can find various products on the internet with names like Explosafe, Deto-Stop, Ex-Co or Explo Control, No-Ex, EM2, and others. It appears that Explosafe was the first product to appear in the late 1970s. The expanded aluminum consists of thin aluminum foil that is sliced, stretched and stacked or rolled to produce a highly porous matrix of low-density expanded aluminum. This matrix can be inserted into a fuel tank and it only takes up 1–3% of the tank volume. The matrix has been shown to dramatically reduce overpressures generated by ignition of fuel–air mixtures in a closed space if the closed space is nearly 100% filled with the matrix. The matrix has also been promoted as having other beneficial properties that reduce slosh pressures in dropped fuel tanks, suppression of algae growth, enhanced discharge of static electricity, corrosion protection for steel fuel tanks, and some state it reduces the risk of a BLEVE (boiling liquid expanding vapor explosion) in pressure liquefied gas tanks and pressure vessels. This paper reviews the history of some of these products and presents a summary of the testing that has been performed to prove some of the benefits. It also includes some limited installation experience that shows some of the penalties associated with the products. & 2008 Elsevier Ltd. All rights reserved.

Keywords: BLEVE Explosion Detonation Flammable liquids and gases Slosh Impact

Contents 1. 2. 3.

4.

5.

6.

7.

Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494 Description of EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 Proven benefits of EA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 3.1. Fuel–air ignition in closed spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 3.2. Slosh loads in fuel tanks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496 Claimed benefits of EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497 4.1. Suppression of BLEVE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497 4.2. Projectile impacts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501 4.3. Static electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501 4.4. Engulfing fire of non-pressurized fuel containers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501 Possibile benefits of EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501 5.1. BLEVEs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501 5.2. Shock release from bursting pressure vessels. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501 Installation issues with EA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502 6.1. Corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502 6.2. Compaction by static loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502 6.3. Degradation by slosh and vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 502 Documented cases of installation studies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503 7.1. Expanded metal in aircraft fuel tanks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503

 Tel.: +1 613 533 2570; fax: +1 613 533 6489.

E-mail address: [email protected] 0950-4230/$ - see front matter & 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jlp.2008.04.001

ARTICLE IN PRESS 494

A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

7.2. Small aircraft fuel systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503 7.3. Expanded metal for rail tank-cars in North America . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503 7.4. US military land combat vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504 EA and the National Fire Protection Association (NFPA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504 Research needs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 505

8. 9. 10.

1. Background Pressure vessels or non-pressure tanks containing fuels or other dangerous goods of various kinds pose a serious threat to life and property if the commodity is released during some kind of accident or malicious act. Threats that can cause accidental release include: (i) (ii) (iii) (iv) (v) (vi) (vii) (viii)

Improvised explosive devices (IED) Jetting fires, engulfing fires, remote fires Impacts and drops Munitions Electrical arcs Static electricity Corrosion Manufacturing flaws

The hazards of interest include: (i) (ii) (iii) (iv) (v) (vi)

Deflagration or detonation in closed spaces Boiling liquid expanding vapor explosion (BLEVE) Burst hazards including blast, projectiles Uncontrolled deflagrations in fireballs and flash fires Detonations and vapor cloud explosions Toxic release

For over three decades, products made from expanded aluminum (EA) have been available for explosion suppression in fuel tanks and containers. You can find various products on the internet with names like Explosafe, Deto-Stop, Ex-Co or Explo Control, No-Ex, EM2 and others. It appears that Explosafe was the first product to appear in the late 1970s. The EA consists of thin aluminum (or other metal) foil that is sliced, stretched and stacked or rolled to produce a highly porous matrix of low-density expanded metal. This matrix can be inserted into a fuel tank and it only takes up 1–3% of the tank volume. The matrix has been shown to dramatically reduce overpressures generated by ignition of fuel–air mixtures in a closed space if the closed space is nearly 100% filled with the matrix. The matrix has also been promoted as having other beneficial properties such as slosh pressure suppression in dropped fuel tanks, suppression of algae growth, enhanced discharge of static electricity, corrosion protection for steel fuel tanks, and some state it reduces the risk of a BLEVE in pressure liquefied gas tanks and pressure vessels. EA products were tested by the US air force by Hogan and Pedriani (1980) and Szego, Premji, and Appleyard (1980) in the late 1970s for use in aircraft fuel tanks. Their testing showed that the matrix suppressed pressure buildup in ignited fuel vapors in the ullage space of fuel tanks (by over 90% if the matrix fills the tank). This could be applied to any fuel tank where the fuel vapor could mix with air to produce an explosive mixture. The potential applications for this technology were wide reaching, including marine, automotive, aviation, fuel storage, portable fuel containers, etc.

Transport Canada sponsored tests of EA in the early 1980s (Appleyard, 1980) to control BLEVE in propane pressure vessels and rail tank-cars exposed to accidental fires. Propane is stored as a pressurized liquid in pressure vessels and there is no air in the vapor space. Therefore, flame propagation in the vapor space is not really an issue in properly purged propane vessels. However, EA manufacturers thought the matrix would protect the vapor space wall from overheating when exposed to external fire and thereby reduce the risk of thermal rupture and BLEVE. Transport Canada sponsored fire tests of one-fifth scale rail tankcars in 1980. In the engulfing fire tests, one of the unprotected pressure vessels suffered a BLEVE and the other unprotected pressure vessel suffered a fish mouth rupture. The EA equipped pressure vessels did not fail at all, and safely vented to empty through the pressure relief valve (PRV). The fire test results suggested that EA suppressed BLEVEs. As a result of this single test series, manufacturers of EA around the world claim that EA stops BLEVEs. However, further analysis suggests that variable fire conditions could also explain the non-BLEVE outcomes in the tests. This question remains today and further research is needed before EA can claim to be a BLEVE suppressor. In the end, Transport Canada did not recommend putting the EA in rail tank-cars to protect against BLEVEs. The EA was too expensive to install and maintain. Today you can buy gasoline containers of various kinds and shapes with EA installed. Home built aircraft enthusiasts have also installed EA in their fuel tanks. The military around the world has considered it for various platforms and has installed it in some. Then the installation problems started to show. There is anecdotal evidence of clogged fuel strainers and filters, blocked intake pipes, and damaged fuel pumps. The promotional literature for the EA was claiming it acted as an anode to suppress corrosion is steel fuel tanks, but this of course means the aluminum corroded and left behind a dark oxide sludge. The matrix was compacting and not fully filling the tanks anymore and this reduced its effectiveness. There were some claims the matrix was causing increased water condensation that fouled tanks. There were claims that engines were failing. It is difficult to find any detailed scientific documentation of this, but there is a lot of hear-say. This was very damaging to the manufacturers of the EA. In the early and mid-1990s, companies in Europe tried to make the product better. They changed the aluminum alloy. One group came up with the idea of forming balls out of the matrix so that it could be easily installed in any container size or shape. The old EA product could be removed easily in a same manner. They claimed they had fixed the problems and put EA in various European military and police vehicles. Some problems came up but these were design issues and installation problems. It is not clear how many of these installations are still in place today. There are rumors that some of these new installations of EA have been removed because of fuel fouling and clogged fuel filters and pumps. The author could not find any reports documenting these rumors. Here we are 30 years later and people are still thinking about EA. There is no question that it does some very good things to

ARTICLE IN PRESS A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

495

Fig. 1. Sketch of EA (slit foil then expanded to matrix) image from reference Szego et al. (1980).

The following have been suggested as the EA properties that play a significant role in explosion suppression in fuel tanks:

Table 1 Summary of EA density and surface area (from Szego et al. (1980)) Expanded width (from 14 in.) (in.)

Density/foil thick (mil ¼ 0.001 in.) (lb/ft3/mil thick)

Surface area (ft2/ft3)

32 35 38 44

1.16 1.09 1.02 0.79

167 152 135 113

suppress explosions and fuel tank ruptures. But it is hard to separate the science from the marketing. The following pages attempt to highlight some of the real data and to identify research needs for this technology.

2. Description of EA The geometry of the expanded metal seems to be the same from all the manufacturers. It is typically manufactured from aluminum foil. The foil is cut as shown in Fig. 1 and then pulled perpendicular to the cuts to expand the sheet as shown. This expanded foil is then stacked or rolled to form batting, coils or balls. The early Explosafe material was 3003 H24 aluminum when it was tested by the US military in the late 1970s. Since then it appears other alloys have been used to improve structural properties and increase corrosion and abrasion properties. In fact, there is also expanded stainless steel foil available as well. The following is a summary of the 3003 H24 alloy characteristics from Szego et al. (1980):

      

Ultimate tensile strength 22,000 psi (151 MPa) Elongation in 2 in. (5.1 cm) 2–6% Melting temperature 1170 1F (632 1C) Foil thickness 0.0015, 0.002 and 0.003 in. (38, 51 and 76 mm) Slit width was 0.04 and 0.055 in. (1.02 and 1.4 mm) Slit length 0.67 in. (17.1 mm) Uncut length 0.11 in. (2.8 mm)

The material started as a foil with a width of 14 in. (356 mm). The Initial 14 in. was then expanded to 30–44 in. (0.76–1.12 m). The final density of the product depends on the foil thickness and the expansion. Table 1 gives a summary of the density and surface area for the expanded material.

(i) (ii) (iii) (iv) (v) (vi) (vii)

Rapid heat absorption High thermal conductivity Resistance to fluid flow Cellular structure High electric conductivity Turbulence generator Boiling nucleation sites

As will be shown, many of these properties can be used to reduce hazards associated with fuel tanks holding liquid and gaseous fuels. The following benefits of EA have been put forward by manufactures of the technology: (i) (ii) (iii) (iv) (v) (vi)

Combustion explosion suppression BLEVE suppression Static electric discharge Slosh pressure suppression during impact or drop Algae growth suppression Fuel tank corrosion protection

The following sections give a brief summary of what is known from the scientific literature.

3. Proven benefits of EA The following are documented experiments that demonstrate the benefits of expanded metal for explosion suppression. 3.1. Fuel– air ignition in closed spaces Of all the tests done on the EA products this is appears to be the one that has been most often repeated and cited in the promotional literature for these products. The first tests were done by the US Airforce in 1980 by Hogan and Pedriani (1980) and Szego et al. (1980). After this tests were done by White (1988). Then other tests were done by the Swiss, Austrians and Germans in the 1990s. They all concluded the same thing. The EA reduces overpressures when fuel–air mixtures are ignited in closed spaces. It acts to absorb heat and slow the combustion wave and this reduces the rate and magnitude of the pressure rise in the closed space. It is believed the matrix takes heat out of the reaction front

ARTICLE IN PRESS A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

and this suppresses the acceleration of the combustion wave. The cellular structure also limits the volume of each reaction zone thereby limiting the local energy release. In the tests, containers of various sizes were filled or partially filled using the EA material. A combustible mixture of gaseous fuel and air was put into the container and a thermal energy source was added to ignite the fuel vapor–air mixture. In all the tests done by the various groups, the matrix dramatically reduced the resulting overpressure after the fuel was ignited. The best suppression occurred with the tank filled 100% with the matrix. As the percent fill was reduced the overpressure increased. In the US Airforce tests by Hogan and Pedriani (1980), they conducted flame tube tests, and ballistics tests. They wanted to compare the EA to poly-urethane foams that were already used in aircraft fuel tanks. They had determined that the EA may have advantages in certain high temperature and high humidity environments. The flame tube was used for studying the performance of EA of different foil thicknesses, orientations, densities and vessel EA fill fractions. The flame tube testing followed the specification for plastic foams USAF (1978) in existence at the time. In that specification the allowable overpressure was 103. kPag (15 psig) for a container initially at 21 kPag (3 psig) with Vc ¼ 20% (i.e. Vc ¼ fraction of tank volume not filled with EA). The flame tube measured 30.5 cm  30.5 cm  228.7 cm. When this vessel was charged with air and 5% (v/v) propane vapor at 21 kPag pressure, overpressures of 790 kPag (114 psig) could be achieved after the flammable mixture was ignited. When this same vessel was filled 100% full with 51 mm thick EA of various densities the overpressures were reduced to 48–90 kPa (7–13 psig) (89–94% reduction). They found that the thicker foil (76 mm) provided little benefit in overpressure suppression but was heavier than the comparable plastic foams. The thinner foil (38 mm) was the lightest but was much more susceptible to damage and settling. They concluded that the 51 mm foil with density of around 32 kg/m3 (2 lb/ft3) was the best overall choice from the EA products tested. It had the best combination of thermal and structural performance. In the ballistic tests, three different tank sizes were used (440, 847 and 1140 l) for testing with 23 mm high energy incendiaries (HEI) and 0.30 caliber armor piercing incendiaries (API). They conducted tests with three thicknesses of EA and one polyurethane foam density (24 kg/m3 or 1.5 lb/ft3). They tested two fill cases of Vc ¼ 0% and 40%. They did a more detailed fill study with the 51 mm foil with the 847 l tank. The test vessels were initially at ambient pressure and temperature. The tests were done with air and propane at 4–4.5% (v/v). They found that the overpressures when Vc ¼ 0% (tank filled 100% with EA or foam) where generally below 69 kPag (10 psig). As the EA fill fraction was reduced the overpressures increased. The HEI damage to the 51 mm EA equipped tanks was slightly worse than that with the foam. The API damage to the 51 mm EA was comparable to the foam. Fauske and Henry (2001a) conducted an analysis of overpressure data from other EA testing and they showed analytically that the EA matrix limits the temperatures reached when the fuel–air mixture burns. The heat from the reaction goes into increasing the temperature of the combustion products and the matrix. They also showed that the cell size in the EA matrix was similar to the critical flame quenching diameter (see for example Glassman, 1977) for laminar flames. They showed that it was possible to estimate the explosion overpressure from the following simple relationship: P ¼ Pmax

 2 Vc . V

From the data shown by Fauske and Henry (2001a), this correlation does not appear to be valid for Vc/Vo0.2, where Pmax is

Overpressure vs % Propane in Air (by Vol)

9

Overpressure Bar

496

8

A = 5.6 m2

7

3.2 1.2

6

0.8

5

0

4 3 2 1 0 -1

0

2

4

6

8

10

12

% propane by volume Fig. 2. Measured internal overpressures in a 20 l sphere as a function of percent propane in air (v/v) for various EA installations (mesh surface area A ¼ 0–5.6 m2) data from CIBA-GEYGY AG (1994).

the overpressure in vessel with no EA, P is the overpressure with EA, V is the volume of vessel, and Vc is the volume not filled with EA where. Fig. 2 shows measured overpressures from a 20 l test sphere including various installations of EA from testing by the CIBAGEYGY AG Technical Group (1994b). The results of Fig. 2 (data from CIBA-GEYGY AG (1994)) show there is still some overpressure event if the vessel is 100% filled with the EA (i.e. case with A ¼ 5.6 m2). This minimum overpressure was typically 5–10% of the maximum overpressure measured with no EA present in the vessel and is consistent with USAF tests. The tests covered propane air concentrations in air from 1.5% to 12% (v/v). The amount of EA in the sphere is indicated by the surface area of the product. The graph clearly shows how the EA suppressed the pressure buildup. The USAF tests showed that varying the thickness and slit width also affected the overpressures developed in the test samples. Thicker foils did better but were a bigger weight penalty which is an important issue for aircraft. The thinnest foil (38 mm) deformed too easily and was not considered practical. They recommended that work be done to improve the structural characteristics of the thin foil, if light weight was a big issue. The clear conclusion from all of these tests was that the matrix works for this kind of accident. 3.2. Slosh loads in fuel tanks When a fuel tank experiences a sudden strong acceleration due to impact or being dropped the fluid can move to impact on one end of the vessel and this can cause significant pressures and forces. In severe cases this could lead to rupture of the container and spilling of the contained liquid. This can then lead to a pool fire. This scenario is very applicable to aircraft such as helicopters. Tests were done by Borghetti, Janszen et al. (2000) to demonstrate how EA affects slosh loads. They conducted tests with a deceleration sled facility. The tests involved a small helicopter fuel tank of unspecified capacity filled with water. The tanks were accelerated to about 5.7 m/s and then brought to rest suddenly. The exact details of this are not clear from the report. Table 2 gives a summary of some of the results. The authors concluded that the EA did the following: (i) (ii) (iii) (iv)

Reduced Reduced Reduced Reduced

acceleration peaks pressure peaks deformations on all sides of the container leakage

ARTICLE IN PRESS A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

Table 2 Impact slosh pressures in a suddenly decelerated helicopter fuel tank partly filled with water (data from Borghetti, Janszen, Morandini, & Rubini, 2000)

Initial speed (m/s) Peak axial (g) Peak sideways (g) Peak vertical (g) Peak end pressure (bar) Peak side pressure (bar)

No expanded aluminum

With expanded aluminum

5.7 260 150 300 1 2.3

5.6 175 100 200 .75 0.75

All of these trends are good and need to be studied further for different vessel scales. We also need to determine how the EA performs compared to other methods of slosh reduction such as baffles, bladders, and plastic foams.

4. Claimed benefits of EA The following are benefits of EA that are claimed by manufactures of EA but there is limited, conflicting or no published scientific evidence to demonstrate them. 4.1. Suppression of BLEVE With the successful demonstration of the explosion suppression benefits of EA for fuel–air mixtures in closed spaces, the EA engineers moved on to the boiling liquid expanding vapor explosion or BLEVE. A BLEVE happens if a pressure vessel holding a pressure-liquefied gas (PLG) fails catastrophically. A PLG is a substance that is normally a gas at ambient temperature and pressure but is stored and transported as an ambient temperature liquid under pressure. Propane is a very common PLG used in our society. If this pressurized liquid is suddenly released from its containment, it will change phase from liquid to vapor violently to produce a BLEVE. A BLEVE can cause blast waves, projectiles and a fireball if flammable (see for example CCPS Guidelines (1994)). The PLG may also be toxic (examples are chlorine or anhydrous ammonia). The failure of the pressure vessel may be caused by severe vessel weakening by corrosion or by a major flaw or by fire impingement. Or a failure can be caused by a severe external impact of some kind (i.e. collision, external explosion, etc.). It can also be caused by an uncontrolled pressure rise in the vessel. Or it can be caused by a combination of all of the above. In the late 1970s, there were a series of train derailments in North America and many of them included fire induced BLEVEs of 33,000 gal (125,000 l) rail tank-cars holding propane. As can be imagined, a BLEVE of such a large pressure vessel would be a catastrophic event producing a fireball almost 200 m in diameter, massive projectiles and blast overpressure. Transport Canada was very open to concepts that could suppress such an event. A fire-induced BLEVE is usually caused by weakening of the pressure vessel by high temperatures in the vapor space wall. Even if the PRV is working properly to control the vessel internal pressure buildup, the pressure vessel can still fail due to reductions in the wall material strength at high temperature. If the pressure vessel internal pressure is limited to about 2 MPa, the pressure vessel wall can still fail in a few minutes by high-temperature stress rupture if the wall achieves temperatures of above 600 1C. The EA engineers thought the high thermal conductivity and large surface area of the aluminum matrix would conduct the heat out of the vapor space wall and convect it into the vapor and thereby reduce the wall temperature in that area. In other words,

497

the EA would act like cooling fins for the vapor space wall, provided the EA was in firm thermal contact with wall. If this could reduce the wall temperature by a small amount (tens of 1C) then this could delay vessel failure in a fire. It only takes a small decrease in wall temperature to change the failure time by several minutes (see for example, Birk and Yoon, 2006). In some cases, even a few minutes delay of failure could lead to the vessel emptying safely through the PRV before the vessel ruptures. One EA manufacturer stepped forward with a proposal to Transport Canada to demonstrate this with engulfing fire tests. This company first did some small-scale testing using a heating plate type of apparatus (Appleyard, 1980). They constructed a thermally insulated vertical (20 cm  20 cm  60 cm long) vessel with a 4.5 kW electric heater at the top. They then filled this vessel with air and 3 l of ambient temperature water and various configurations of the EA material. Test runs lasted 120 min. They heated the top plate of the vessel to 220–240 1C and measured the temperature distribution in the matrix and water heat sink. The temperature readings clearly showed that heat entered the test vessel faster when the EA was present. They determined that in the best EA configuration the heat transfer rate with the EA was double that of the case with no EA. It should be noted that in this case the wall temperature was only 240 1C and therefore the heat transfer was mostly by conduction and convection. In a real fire situation, the wall temperature may be 650 1C where radiation may be the dominant mode of heat transfer. At 650 1C, many aluminum alloys have begun to melt. This test did not simulate the action of a PRV that would also affect the heat transfer in the vapor space. In any case, this was a very promising outcome. From this testing they moved on to large-scale fire testing of propane pressure vessels. To our knowledge the only detailed test done for the suppression of a BLEVE using EA was funded by Transport Canada (Appleyard, 1980). This work involved six one-fifth linear scale models of 33,000 US gal (125 m3) rail tank cars. The test vessels were approximately 260 gal or 980 l. The vessels had a diameter of 0.61 m, and a length to diameter ratio of 6. The vessel was made from SA 285 steel and the wall thickness was 6.4 mm in the cylindrical section. The test pressure vessels were filled to 85% liquid full with propane and exposed to 100% engulfing hydrocarbon pool fires. The fires were fueled with JP4 jet fuel. The fuel was contained in a rectangular fire pan measuring 6.10 m  3.05 m. This fire pan should have been sufficient to give full fire engulfment of the vessel if no cross winds were present. Wind shields consisting of several 1.2  2.4 m steel panels were used on the north and east sides of the fire pan in an attempt to block wind effects. The vessels were instrumented with thermocouples to measure wall and lading (both liquid and vapor) temperatures and pressure transducers to measure tank internal pressure. Fire temperatures were also measured in a limited way using thermocouples. For further details, the reader is directed to reference by Appleyard (1980). The fire tests were conducted outdoors and therefore were susceptible to cross wind effects and this resulted in less than 100% engulfment and fire inconsistencies from test to test. The recorded wind speeds and directions, and air temperatures were as follows: Baseline DRES Special Diablo Howler Nova Enigma

0 1C 22 1C 3 1C 7 1C 2 1C 0 1C

12 km/h from W 10 km/h from W 15 km/h from N 11 km/h from WNW 4 km/h from SSW 3 km/h from S

The cylinder axis of the vessels were oriented EW for the tests.

ARTICLE IN PRESS 498

A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

Table 3 Summary of Transport Canada fire tests (data from Appleyard, 1980) Test

Test duration (min)

Outcome

Peak measured wall temperature (1C)

Peak pressure (bar)

Approx. average fire radiating temperature (1C)

Baseline, no EA DRES special, full EA

10 15.5

BLEVE Empty

No data Poor fire

No data Poor fire

No data Poor fire

Diablo, full EA

15.5

Empty

590 right 400 left

16 bar PRV open 20 bar PRV closed

Start 780 End 630

Howler, full EA

15

Empty severe deformation

710 right 710 left

16.5 bar Peak PRV open 17 bar peak PRV closed

770

Nova wall layer, EA Centre, no EA Enigma, no EA

15

Empty

560 630 570 460

16 18 18 19

650

7.5

Large fish mouth rupture

In these six tests, two of the pressure vessels were unprotected except for standard pop action PRVs. The other four pressure vessels had various configurations of the EA and a relief valve. The relief valves were sized in accordance with appropriate standards for the pressure vessel. The first test was an unprotected pressure vessel and was code named Baseline. The vessel failed violently (BLEVE) after 10 min of fire exposure. Unfortunately, no data was obtained from this test due to instrument failure and therefore it is impossible to define the fire conditions and pressure vessel response. The second test was called DRES Special (DRES stands for Defense Research Establishment Suffield, Alta., Canada, where the tests were conducted). This pressure vessel was equipped with the EA matrix (100% full) and it did not suffer a rupture or BLEVE over the test duration. In other words, the pressure vessel emptied safely through the PRV and the vessel did not fail. However, during the test there were pool fire fuel delivery problems and the fire temperature was decreasing steadily during the majority of the test duration. This test was not considered to be indicative of a fully engulfing fire and the measured results were not used. The next test was code named Diablo and it was a repeat of DRES special. This pressure vessel also survived the fire and emptied in about 15.5 min of fire exposure. This test also suffered from a steadily decreasing flame intensity, and high wind speed and again brought questions about the test validity. The next two tests were Howler and Nova and were equipped with the EA matrix. These two pressure vessels survived the fire without suffering a failure. It was noted that the Howler pressure vessel was severely plastically deformed (bulged) at one end and this indicated the vessel was near rupture at the end of the test. The final test called Enigma was a repeat of the unprotected test Baseline. It suffered a large fish mouth rupture at about 7.5 min into the test. A brief summary of results from these tests are given in Table 3. For full details of the tests, the reader is directed to Appleyard (1980). To summarize then, the two unprotected pressure vessels failed in 7.5 (fish mouth failure) and 10 min (BLEVE). The four vessels equipped with EA did not fail but emptied through the PRV in about 15 min. The simple conclusion might be—the EA saved the pressure vessels from failure and BLEVE. Unfortunately, there just is not enough data to attribute this outcome to the EA alone. An analysis of the data was conducted by Birk (1983) and he concluded that the outcome of the tests could not be clearly attributed to the EA. The pressure vessel response was not just determined by the presence of the aluminum matrix but was also determined by the fire conditions and the behavior of the PRV.

right left right left

bar bar bar bar

peak peak peak peak

PRV PRV PRV PRV

open closed open closed

700

This of course assumes all of the pressure vessels were made of identical steel and with no major manufacturing flaws. Birk has since conducted many fire tests of propane pressure vessels (from a few litres to 1800 l) exposed to various kinds of fire (see for example, Birk & Cunningham, 1994; Birk, Cunningham, Ostic, & Hiscoke, 1997; Birk, Poirier, Davison, & Wakelam, 2005; Birk, VanderSteen, Davison, Cunningham, & Mirzazadeh, 2003). During this research they have witnessed many pressure vessels that survived severe fire conditions without thermal rupture. The exact behavior of the pressure vessel material, the PRV and the exact fire conditions determined whether a pressure vessel would suffer a failure or not. Birk and his coworkers were continuously frustrated when they did engulfing pool fire tests because of very large test to test variability due to fire variations. In fact, they have since given up using open pool fires for engulfing fire testing. They now use an array of liquid propane burners so that test to test variability can be somewhat managed. Wind effects during open pool fire tests are notorious for changing test outcomes. If one looks at the measured wall temperatures from the Diablo test for example, the peak wall temperature on the right side of the pressure vessel was about 590 1C after 15 min in the fire, but it was only 400 1C on the left hand side at the same time. This is due to variations in fire exposure. Most of the tests showed this same trend of higher wall temperatures on the right side of the pressure vessel. This is a bias caused by the prevailing winds during the tests. This means that the pool fire was pushed over to one side of the pressure vessel by the wind. This usually means the top of the pressure vessel is exposed to fire only intermittently. This leads to peak wall temperatures somewhere down the side instead of on the pressure vessel top. Peaks wall temperatures were recorded as follows: Howler Enigma Nova Diablo

Right Right Right Right

side side side side

at 451 from pressure vessel top at 221 from pressure vessel top near pressure vessel top at 221 and then later at 671 from the pressure vessel top

This alone shows the fire test conditions were different for the four tests and makes it difficult if not impossible to make direct comparisons between the tests and the outcomes. Fig. 3 shows the measured fire temperatures from the four tests (Howler, Enigma, Diablo, and Nova) where data was captured and results published. Fire test standards for rail tank car thermal protection systems in North America (Can CGSB 43.147, 2005; CFR 49 Part 179, 2008) require fire temperatures of 871756 1C. None of the above tests come close to meeting this test standard. Of the four fire tests shown, we will discard Diablo and Nova because of

ARTICLE IN PRESS A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

800 750

Fire T (deg C)

700 650 600 550 Diablo

500

Nova

450

Howler Enigma

400 0

5

10

15

time (min) Fig. 3. Mean fire radiating temperatures calculated from transport Canada tests Birk (1983).

21 19

Pressure (atm)

17 15 13 11 9

Enigma Howler

7 5 0

5

10

15

time (min) Fig. 4. Measured pressure vs time for Howler and Enigma tests (data from Appleyard, 1980).

clearly poor fire conditions. We will focus our discussion on Howler and Enigma. Fig. 4 shows the measured pressures in Enigma and Howler. The figure clearly shows that Enigma experienced higher pressures during most of its test. The PRV pop pressure was higher in Enigma. The peak pressure during steady PRV flow was higher in Enigma. This means the average hoop stress was higher for Enigma. Could this difference have been caused by the presence of the EA? Or was it caused by different PRV behavior and different fire conditions? It is well-documented that PRVs of this size and type have highly variable performance (Pierorazio and Birk, 1998). It is also well-documented that different PRV behavior can affect the time to rupture Birk, VanderSteen, et al. (2003). If EA works by conducting heat out of the vapor space wall and into the lading then the EA pressure vessel should see higher pressures—not lower. Did the EA cause more PRV heating? This is a possibility. If the PRV is heated more, the PRV spring will relax and this could reduce the pressure in the tank. This needs further study. Fig. 5 shows the measured peak wall temperatures in the four tests. As can be seen from the Figure, Enigma experienced a more rapid increase in wall temperature than did Howler. Enigma failed just before 8 min and Howler did not fail at all. Did the EA save

499

Howler by keeping the wall cooler early in the test? This is what the small scale hot plate test results showed in reference Appleyard (1980). This is one possibility. But it also could have been variable fire conditions. If we compare Enigma (no EA) and Diablo (100% filled with EA) early in the test, we see they have almost identical peak wall temperatures up until about 4 min. Diablo had EA—why did the EA not cool the wall in that case? The PRV was working for a few minutes by that point so two-phase swell in the liquid should have been working to cool the vapor space wall. However, the full benefits of two-phase swell induced cooling would not take place until the liquid is near isothermal so that full saturated boiling could take place. This saturated state probably was not in place until about 4 min in the test. After 4 min, the wall temperature in Diablo leveled off—maybe because of saturated boiling, but probably due to changes in the fire conditions. If it was swell induced cooling that brought the wall temperature down in Diablo after 4 min, why did this same mechanism not cool the wall in Howler after 4 min? Again, this points to varying fire conditions as the most probable explanation for the lower wall temperatures from test to test. Also note the peak wall temperature in Howler was at a position 451 down the pressure vessel side on the right side. In a fully engulfing fire, we would expect peak wall temperatures nearer to the pressure vessel top where the wall is furthest from the cooling effects of the liquid. If the strongest heating is at 451 down the side, we would expect lower wall temperatures than if the same heating were at 221 down the pressure vessel side. That is probably why Enigma saw higher wall temperatures than Howler. Howler continued on and did not fail even though it had measured peak wall temperatures at 700 1C. It did come very close to failure as indicated by the severe bulging of the pressure vessel at one end. It should be noted that all these observations are based on the peak ‘‘measured’’ temperature. It is very possible the actual peak wall temperature is located on the pressure vessel wall where there is no thermocouple to sense this temperature. In other words, we have no way of knowing if this is the true peak wall temperature on the pressure vessel. With variable fire conditions, the peak wall temperature on the vessel will change location and magnitude from test to test. Enigma failed at 7.5 min with an indicated wall temperature of 570 1C and a pressure of just over 15 atm. Enigma saw a peak pressure of about 18 atm at 4.5 min. Howler did not fail with a peak-indicated temperature of 700 1C at 15 min. Howler achieved its peak pressure of about 16.5 atm at 5.5 min. A pressure vessel in good condition (no corrosion, no major flaws, etc.) will fail due to high-temperature stress rupture Boyer (1988). High-temperature stress rupture is caused by a combination of stress and elevated temperature of the wall. At these pressures, the wall must get very hot to suffer a stress rupture in a few minutes. Let us consider Enigma. If we take the 570 1C wall temperature and the pressure vessel peak pressure of 1.8 MPa, we should be able to estimate the time to failure based on high-temperature stress rupture analysis (see for example, Boyer, 1988). The pressure vessel diameter was 0.62 m and the wall thickness was 6.4 mm. This translates to a hoop stress of 87 MPa. If we use the stress rupture data from Birk and Yoon, this hoop stress and temperature combination would take many, many hours to fail the subject pressure vessel. In other words, this pressure vessel should not have failed at this condition unless something more severe was happening. This suggests that the measured (indicated) peak wall temperature was not the true peak wall temperature on Enigma or there was something else wrong with this pressure vessel. They did not have enough instrumentation on this pressure vessel to measure what was happening in detail.

ARTICLE IN PRESS 500

A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

800 700

wall T (deg C)

600 500 400 300 Howler (right side, at 45 deg from top)

200

Enigma (right side at 22 deg from top) Diablo (right, at 22 deg then 67 deg from top)

100

Nova (right, at top)

0 0

2

4

6

8 time (min)

10

12

14

Fig. 5. Measured peak wall temperatures for Howler and Enigma (data from Appleyard, 1980).

Now let us consider Howler with its 700 1C measured peak wall temperature and peak pressure of around 1.62 MPa. This gives a hoop stress of 79 MPa. At this condition, the data from Birk and Yoon gives a failure time on the order of 1 h (i.e. many minutes). In other words, Howler should have been much closer to failure than Enigma if we use the measured wall temperature and pressure. Howler did come very close to failure in 15 min as shown by the bulged pressure vessel end. It is clear from the Howler pressure data that the pressure vessel was saved in the later minutes of the test by PRV spring softening (see Fig. 4 where pressure drops in final minutes). This is what happens when the PRV spring gets very hot. The spring steel relaxes and the PRV pop and reclose pressure decreases as the spring gets hotter and hotter. This typically happens in propane pressure vessels in fires and is not something that the EA can be credited with. As the liquid level goes down, the PRV gets hotter because the vapor space gets bigger and the vapor space wall temperature gets hotter. If the vessel pressure had stayed up near the PRV set pressure while the pressure vessel wall was at 700 1C, Howler would have failed in a few minutes. Enigma failed with a measured peak wall temperature of 570 1C and analysis suggests that it should not have failed if this was the true peak temperature. What happened in the Enigma test? Was there a major flaw in the pressure vessel, or was the peak wall temperature much higher in a location where there was no thermocouple to measure it? We just do not know. It is possible that the EA has some other beneficial effect to protect the pressure vessel. Fauske and Henry (2001b) suggested there are three mechanisms: (I) Increased number of nucleation sites stabilizes boiling process. (II) EA prevents impingement on the vessel wall due to liquid impact during sudden depressurization and flashing (i.e. after initial vessel failure). (III) Prevents cold liquid from directly contacting the overheated and stressed wall. The first point is easily understandable as the EA would act like boiling chips in a beaker of boiling water. This might enhance the liquid swell that could cool the vapor space wall from inside. This

has nothing to do with the EA thermal conductivity. It is a twophase flow phenomena. The swelling would happen every time a PRV would open. This benefit would only be present when the vessel has a high fill level. If the pressure vessel wall reaches a high enough temperature while the vessel is under pressure then high-temperature stress rupture is possible. Birk and Yoon, (2006) have looked into this process in detail and have used high-temperature stress rupture data to successfully explain the timing of fire-induced failures of propane pressure vessels from the testing of (Birk, Poirier, & Davison, 2006). High-temperature stress rupture leads to local wall thinning and eventually the formation of a fissure or crack. Once the crack forms, a powerful two-phase flow will leave the vessel through the crack. This usually results in liquid impacting the vessel top from inside and this could trigger the crack to grow. The flow also cools the crack tips adding thermal stress, and this may also trigger the crack to grow further. If the crack grows then it may lead to a catastrophic failure and BLEVE. There is a possibility that the EA could mitigate this process and suppress the BLEVE. However, more research is needed. The two-phase swell and splashing of cold liquid on the hot wall may affect the failure process after the pressure vessel wall begins to fail. Once the initial fissure forms in the wall due to stress rupture (i.e. local thinning of the wall in the region of highest accumulated damage due to stress and high temperature leads to a crack or fissure which opens the wall), the resulting two-phase flow through the fissure will lead to new stresses and this could help to make the fissure grow and this could make the difference between a finite rupture (i.e. fish mouth opening with jet release—like Enigma) and a BLEVE (Birk and VanderSteen, 2006). But that is not what we are talking about here—we want the EA to stop the initial failure from happening. Water hammer and cold liquid splashing only happens in a significant way after the initial failure starts. Maybe the EA affects the post failure process and could mitigate the subsequent hazards. In conclusion, there is no data from these tests that show conclusively that the EA saved the pressure vessels from failure and a BLEVE. Yes, the four pressure vessels with EA did not fail and the two without EA did fail. If you look at the measured fire temperatures from these tests you can eliminate three of the four EA equipped tests due to low or decreasing fire temperatures during the tests—and that is probably why they did not fail. Only

ARTICLE IN PRESS A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

the Howler test had steady and severe fire conditions and this pressure vessel came very close to failure. Howler had very high wall temperatures that brought it very near to failure. As for the rail tank-cars in North America, they never did put EA in them. They concluded there were too many installation problems (slosh, corrosion, etc.) and the cost was unacceptable in terms of installation and maintenance. In the end they chose to use external thermal insulation to protect against fire.

501

As noted earlier it is not clear that EA protects against an external fire induced rupture event. If the fuel tanks are plastic, then it is likely a fire will cause them to melt, burn and spill the contents. It is not clear that the EA will have much of an impact on this. As stated earlier, specific research is needed to quantify the benefits, if there are any.

5. Possibile benefits of EA 4.2. Projectile impacts When a projectile impacts a fuel tank, the response of the container wall and the liquid and vapor contents determines whether the projectile will penetrate or just bounce off. Ideally, we would like the projectile to bounce off and not penetrate the wall. If the wall is breached then the fuel can spill and cause a fire. If the projectile enters and leaves the container then there are two holes that fuel can spill from. In some cases, the dynamics of the projectile impact can cause the container to pressurize and burst and this can disperse the liquid fuel as droplets. Tests were conducted by the US military, Copland (1983) to study how the EA matrix affects the hydrodynamic ram effect. Hydrodynamic ram effect takes place when a high-speed projectile hits and penetrates a liquid filled container. It involves three phases: (i) Shock upon entry (ii) Drag from passage through liquid (iii) Cavity generation In the tests, bullets and spheres of different caliber were fired into 20 and 220 l fuel containers holding water and diesel fuel. Some of the cans were equipped with EA. Projectiles were traveling at around 870 m/s (Mach 2.6 in air). The presence of the aluminum matrix made things worse in the 20 l containers. Apparently, the tumbling projectile grabs hold of the EA and carries it along. This produces a piston effect that does more damage on the exit surface. Spherical rounds did not show this same trend but did show more damage on side walls. For the 220 l drums, the presence of the EA did not appear to make things worse or better. Clearly further study is needed here. Maybe the EA works well for application at larger scales. 4.3. Static electricity The buildup of static electricity can cause a spark to ignite an explosive mixture of fuel and air. For this reason it is highly desirable that any fuel tank filler dissipate any static charges safely to ground. The manufacturers of EA claim the high electrical conductivity of the aluminum matrix does this very effectively as long as the matrix is in good electrical contact with the metallic container wall. No detailed reports were found on this subject area for the EA products. 4.4. Engulfing fire of non-pressurized fuel containers The promotional literature for EA products suggest that the matrix will reduce the severity of hazards if fuel containers are exposed to fire. We were not able to find any scientific documentation of this. If a metal container holding a non-pressurized fuel is put in a fire, it may pressurize and burst in a way similar to a BLEVE (note that a BLEVE is usually associated with pressure liquefied gases). When it bursts, the vessel may send out fragments and the liquid fuel may be dispersed as a cloud of burning liquid droplets.

The following are postulated benefits of EA. All of these will need detailed research and testing to demonstrate. 5.1. BLEVEs It is possible that the EA could provide the following benefits to systems that can suffer a BLEVE. (i) Possible cooling effect on vapor space wall by surface area enhancement (i.e. fin cooling effect). (ii) For near full propane pressure vessels (or similar commodities), the expanded metal matrix may act to promote boiling and two phase swell during pressure relief. During the early minutes of the fire, the swelling liquid could wet the pressure vessel upper wall surface to maintain lower temperatures and this may delay failure of the pressure vessel until the vessel empties. (iii) The EA may affect the PRV temperature and this could cause enhanced spring relaxation that could reduce the pressure in the vessel. This too could delay failure because it could result in lower wall stress. (iv) The EA may mitigate liquid impact forces and crack tip cooling during the failure of the vessel and this may suppress total loss of containment. Total loss of containment is needed for a BLEVE to take place. (v) The EA may mitigate the blast overpressure hazards from a BLEVE, if a pressure vessel does fail due to fire engulfment. It is possible that the EA will slow the opening of the vessel and this could suppress the formation of a shock. It is also possible that the EA could make things worse at certain scales and conditions. For example, the low emissivity (reflective) aluminum could act as a reflector to the thermal radiation from the pressure vessel inner wall in the vapor space and this could lead to even higher wall temperatures. The matrix would also provide a huge area for boiling nucleation and this could enhance the pressure transient during failure if the pressure vessel suffers a finite rupture. This could make the difference between a BLEVE and a non-BLEVE finite rupture with jet release. More research is needed to show that EA suppresses BLEVEs by cooling the vapor space wall. If it cools the wall, the mechanism is most likely a fin cooling effect and two-phase wetting. At the wall temperatures that cause pressure vessel failure, the dominant mode of wall cooling is thermal radiation. More research is needed to understand what the expanded metal does in systems that could suffer a BLEVE. It is not clear from the current available data that the expanded metal stops a BLEVE in a fire-engulfed pressure vessel. 5.2. Shock release from bursting pressure vessels When a pressure vessel bursts, there is a possibility a shock wave will be released. If a shock is released we would call this event an explosion. The vessel must open fully and rapidly enough to cause the formation of a shock. This shock then travels out into

ARTICLE IN PRESS 502

A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

the free field and this can produce shock overpressure loading and blast wind drag loading on nearby structures and people. There is a possibility that EA mesh could interfere with the formation of this shock and possibly mitigate it. When the vessel fails and begins to open, the tank contents (vapor and/or liquid) will rapidly flow out of the opening. This flow will pull the EA material with it and this may block the opening and possibly choke off the outward flow. This blockage will also increase the loading on the vessel wall and may affect the vessel failure process (as it did in the case of hydrodynamic ram). If it does block the flow this means the outflow will be reduced and the energy release rate may not be sufficient to produce a shock. If the outflow rate is reduced by the presence of EA, then this could have very good benefits including: (i) For downwind populations, if the commodity is toxic the lower flow rates may dissipate the hazardous material more rapidly and result in lower concentrations down wind. (ii) If the commodity is flammable, the lower flow rate could make the difference between a huge short duration fireball and a much smaller longer duration jet fire. (iii) The lower flow rate will result in lower thrust forces and less likelihood of projectile effects To our knowledge there is no scientific evidence to support these statements. Research is needed to study these possible benefits.

In the 30-day tests, all the EA samples became coated in a gray film of oxide. The mass of the EA material (with the gray film) gained from 4% to 14% with the lower gain taking place with the highest level of FSSI. The samples with tap water had the highest mass gain. Several tests showed precipitate in the bottom of the sample beakers. After sample cleaning to remove the film, there was a net mass loss of the EA samples from 1.1% to 3.6% after the 30-day soak. As before, the mass loss was highest with the tap water and low FSSI level. The 30-day trial samples were then reused for the 6-month trials. With the 6-month trials, the mass change for the EA samples ranged from +2.5% to 7.6%. It was noted that most of the changes happened in the first 30-day elevated temperature trial. The EA material was inspected and it revealed that there was general surface corrosion of the aluminum foil. There were a few isolated areas of severe corrosion. There were no apparent perforations or pitting on the foil. This test was considered to be an accelerated worst-case situation in service. These tests were done using glass beakers and therefore do not account for the EA coming in contact with other metals such as steel or other alloys of aluminum. The report authors noted that special care must be taken in these cases. The suppliers of some EA material claim that it can act as a sacrificial anode in systems with polar molecules such as water or salt water present. Thus, it acts to slow corrosion of steel fuel containers. However, if the EA acts as an anode this means it must degrade and this will result in EA mass loss and an oxide sludge left in the tank. This of course could plug fuel filers and affect engine operation if not properly managed.

6. Installation issues with EA 6.2. Compaction by static loads There are numerous anecdotal references to installation problems caused by EA in fuel tanks. It is not clear whether these problems can be attributed to the product itself or to the installation procedures used. The following is a partial listing of known and anticipated problems that can come up with EA installations: (i) (ii) (iii) (iv)

Corrosion (loss of matrix, deposition of oxide residues) Fuel delivery system restriction or blockage Vibration breakdown including compaction and abrasion General compaction due to normal motion (i.e. slosh, impact, etc.) (v) Added weight and reduction in available fuel (vi) Cost of installation and maintenance

6.1. Corrosion Corrosion of EA materials was studied by the US military (Szego et al., 1980). Reports that consider EA material usually state that there are no corrosion issues as long as the matrix is used with non-polar fluid molecules. However, polar fluids such as water do get into fuel systems. In fuel soak tests of Szego et al. (1980), samples of EA were stored in 2000 ml glass beakers with aircraft fuel for 3 months. They were stored at two different temperatures of 129 and 135 1C. They concluded that the tested EA did not degrade the aircraft fuels in any significant way. They also concluded the fuel did not degrade the EA. They also tested fuel system icing inhibitor (FSII) and this also did not degrade the EA. Other soak tests were done with fuel including tap and salt water (3.5% NaCl by weight) all including FSSI that also acts as a biocide. Soak tests lasted 30 days at 60 1C and 6 months at ambient temperature.

Tests were done to see how the EA matrix deflected under a static load (Szego et al., 1980). In these tests samples with different foil thickness and in various orientations were loaded using a pressure plate. They found the deflection depended on the foil thickness and the sample orientations. They recommended that the material be oriented such that the layers and the long axis of the cells lie in the direction of the highest loads. They concluded that static deflection should be less than 5% for anticipated loads in a fuel tank. 6.3. Degradation by slosh and vibration The US military was concerned that normal operation would result in compaction of the EA, so extensive tests were done to study this issue. Tests were conducted to study large-scale slosh resistance of the EA material (Szego et al., 1980). The tests involved 200 gal external aircraft fuel tanks. The first test was fully packed with coiled and layered EA material. The second test involved the same tank but this time the tank was filled with layered EA only. The tanks were rocked through 301 (151 on each side of horizontal) at 12 cycles per minute. For the first test, the tank was filled to 66% with Type III test fluid. The second test was done at the same fill with aviation fuel. Both tests involved 28,800 slosh cycles (40 h). This 40-h test represents the expected slosh conditions for the expected life of the tank. In the first test, the coiled sections appeared to suffer significant damage and deformation. This is probably due to the fact that coils are packed more loosely than the layered material. They found the material was fatigued and breaking off at the edges. Internal devices rubbed and broke off pieces of the EA material producing loose debris.

ARTICLE IN PRESS A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

In the second test, they observed significant flexing of the EA in regions where the EA was not included due to internal fixtures. This did not result in any damage during the tests. The second test produced much less debris from the EA. Various degrees of wear were observed on the painted internal surface of the tank. There was also a film of aluminum and aluminum oxide. When this was wiped off, it was observed that the surface wear was minor except in one location. A key observation was that the coiled material was oriented such that the foil expansion was in the direction of slosh motion. This opened and closed the cell diamonds and this lead to fatigue and breakdown of the matrix. The layered material did not suffer the same damage because it is generally stronger as a structure because it includes metal stitches to hold it together. As a result, the coiled material was not recommended for this application. Tests were also done with a 90 gal rubber bladder tank. Once again there was some small amount of debris and there was some minor scuffing of the tank inner surface. The wear on the inside of the tanks was mostly done by the exposed cut ends of the layered EA material. However, the degree of wear was considered acceptable as this test also represented the expected life of the subject tank. It should be noted that all the above tests were conducted with rolled or layered EA material. In recent years, the technology has been extended to include small rolled up balls of EA. It is possible that this geometry has advantages over the older configurations. The US Air Force did dynamic motion tests (Szego et al., 1980) which involved a 100 gal auxiliary aircraft fuel tank filled with the EA material and then filled 66% with liquid aviation fuel. The tank was rocked in pitch and roll and vibrated. The vibration was at 2000 rpm with a double amplitude of 0.032 in. (0.81 mm). The rocking took place at about 19 cycles per minute through a full angle of 301 (151 on each side of horizontal). The test involved 12.5 h with pitch roll and 12.5 h in pitch motion. After the test, the fuel was removed and filters were used to collect debris. There was no physical damage or deterioration of the EA. The tank itself was not damaged in any way. The interior surface of the tank was not scoured by the EA material. It is believed that the tank was made of aluminum but this is not confirmed. The filters captured pieces of aluminum up to 0.5 mm in size. The report stated that a very small quantity of material was collected. Some of this material probably came from the manufacture of the tank itself and some came from the EA. They considered the quantity and size of the collected material to be harmless. Other vibration tests were conducted on a 9 in. cube of EA in a container with aviation fuel. They vibrated the sample for 72 h at various amplitudes and frequencies. They found very little damage in the EA. They did find some scratch marks on the container wall due to the EA but this was probably due to installation and removal between tests. There were also smear marks of aluminum oxides on the walls. They concluded the EA and the containers survived the 72 h of testing with no significant damage. It appears from the testing that small-scale installations do not suffer damage but larger scale systems may. Obviously, each installation must be engineered and tested properly to confirm this.

503

these cases in the public domain. Further work is necessary to obtain information on actual EA installation experience. 7.1. Expanded metal in aircraft fuel tanks Exploding fuel tanks on commercial aircraft was very rare and not really an issue until TWA flight 800. After TWA 800 in 1996, the commercial aircraft industry was looking for a way to eliminate the possibility of ignition of flammable mixtures in aircraft fuel tanks. EA was considered for this application. The FAA had a task force (Task Group 4, 1998) that looked into this matter. They noted the benefits from the EA material and made reference to US Military specification MIL-B87162 (USAF,1999) (cancelled in 2004) and the testing done by the US air force (Szego et al., 1980). They also noted the disadvantages of the EA materials including weight increase, temperature increases, fuel volume loss, foreign object damage, and maintenance difficulties. They concluded that the EA would greatly impact removal and replacement of in tank components and this would increase maintenance costs. They estimated installation and maintenance costs and found that annual maintenance costs exceeded the cost of installation of EA for large and medium sized aircraft and were similar for small aircraft. The task force also noted that there are unquantifiable hazards associated with the technology (i.e. weight, FOD, etc.). For example, there have been no overpressure events in US regional airlines and therefore if any negative event could be associated with the technology then the net outcome from the technology would be negative for that sector. In the end, the FAA found a more cost-effective way to inert fuel tanks (Duquette, 2006), which involved nitrogen gas inerting of the ullage space. It should also be noted that the US cancelled its specification MIL-B-87162A in 2004 and no longer uses EA in new aircraft. 7.2. Small aircraft fuel systems If EA is installed in a fuel tank of a land, sea or air vehicle, then it must not interfere with the delivery of clean fuel to the engine. This is most critical for aircraft. There are numerous anecdotal reports of EA interfering with fuel systems. These anecdotal reports also suggest some EA installations have been removed because of this. EA has been used in home built aircraft since the 1980s. Few detailed reports are available on specific problems. One report in Glasair News (1988) documents fuel delivery problems in aerobatic aircraft. EA clogged the intake tube to the fuel pump. When the fuel tank was opened up and the aluminum matrix was removed they found a large quantity of loose particles of the EA foil. These particles were then found all though the fuel system. It was suggested that the loose particles came from the rough edge cuts done to fit the EA batting into the tank. A detailed analysis was not conducted to determine why particles were being released from the product. The report suggested that users make sure that they have periodic inspections of the installation and that fuel strainers and filters be installed in appropriate locations. 7.3. Expanded metal for rail tank-cars in North America

7. Documented cases of installation studies The following are cases where EA was being considered for, or was actually installed in certain platforms. In some cases, problems were identified. It is known that there were other installations, but it was not possible to find documentation of

As noted earlier, Transport Canada funded an in depth study to reduce BLEVEs of railroad tank-cars in the early 1980s. This work began with fire testing of one-fifth scale tank-cars (Appleyard, 1980) and the results looked very promising. This work was followed by a detailed installation design and cost analysis study

ARTICLE IN PRESS 504

A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

and a computer modeling study. In the end, it was decided that the EA was too expensive to install and maintain. It was concluded that the harsh environment of rail tank-cars (slosh, vibration, impact, steam cleaning, inspection, and corrosion) was too much for the EA. In the end, the product was not selected. Instead, the industry selected an external blanket of thermal insulation (Anderson, 1982) and a steel jacket to protect the tanks from fire. This thermal protection system proved to protect the tank from failure for at least 100 min when exposed to an engulfing fire.

matrices for explosion suppression. This new standard gives detailed specifications for metal matrix explosion suppression systems, including: (i) (ii) (iii) (iv) (v) (vi)

Density Surface area Surface area-to-volume ratio Pore or cell size Metal alloy Testing

7.4. US military land combat vehicles The US military conducted numerous studies into using EA for protecting land combat vehicle fuel tanks from various threats. The results were summarized in McCormick, Motzenbacker, and Clauson (1988). The authors of that report stated that test results with one EA product were mixed. They stated:

 EA intensified the effects of hydrodynamic ram in 20 l fuel tanks and had no impact in 220 l tanks.

 No significant benefit when M113 fuel tanks were subjected to

Because there is so little data on actual installations of EA, the NFPA 69 requires detailed testing for each installation to establish performance parameters. The results of this testing are then to be appropriately documented as part of the system design. The NFPA 69 then goes on to include requirements for the following: (i) (ii) (iii) (iv)

Installation details Maintenance and replacement Periodic inspections Disposal

shaped charge attack.

 Mixed results when striking the liquid portion of a fuel    

tank—sometimes increasing the fireball size and sometimes decreasing it. When striking the ullage portion of a fuel tank equipped with the EA, there was no benefit (author note—it is not clear if this impact ignited the mixture in the vapor space). When struck or damaged, the EA leaves particles of aluminum in the fuel and this could damage other parts of the fuel system. The heat sink effect of the EA provided no advantage and may actually increase the temperature of the fuel tank. It did provide significant protection against pressure buildup in fuel tanks, if the ullage space fuel–air mixture was ignited.

The US Marine Corps considered the EA for its amphibious assault vehicles but stopped testing because of three problems: (i) the kit was extremely complicated to install, (ii) matrix interfered with fuel pump operation and (iii) matrix did not reduce the risk of fuel fires when the tanks were punctured. Expanded metal was considered by the US military as being ineffective for ground combat vehicles. As a result, as of the date of that report, expanded metal was not used in any production US Military combat vehicles. These early tests were done with the layered version of the EA. In later years, the product was available in the form of small balls of mesh. This had the potential of significantly simplifying the installations. There is some evidence that the US military was reconsidering the use of EA as late as 2002, but this time in the form of mesh balls. The US Navy is currently doing further studies with possible reporting in mid-2008. Since then it appears several European countries have used the EA in their military vehicles. No detail on this could be found in the public domain. One EA company claims their proprietary EA matrix has been fitted into numerous military and police vehicles in Germany, UK, Holland, Greece, Sweden, Denmark, Spain, Finland and Italy. No publicly available references could be found on this subject.

8. EA and the National Fire Protection Association (NFPA) A recent paper by Zalosh (2007) presented an overview of the newest standard NFPA 69 Rodgers (2007) for Explosion Prevention Systems. This standard includes references to expanded metal

9. Conclusions On the basis of the scientific evidence available in the open literature, it is clear that EA has some very beneficial characteristics for protecting fuel tanks from fire and explosion. Data shows that it suppresses overpressure when combustion takes place in closed spaces, it reduces slosh accelerations and pressures in dropped fuel tanks, and it reduces static electricity buildup. However, it can make hydrodynamic ram effects worse. Some promotional literature for commercially available products seem to be pushing beyond what has been shown scientifically. Specifically, the benefits of EA in terms of protecting pressures vessels from BLEVE type failures. More research is needed before it can be claimed that EA stops BLEVEs from happening. The available literature presents test results that suggest that EA can be installed with minimal impact on the operation of certain vehicles. The promotional literature from EA manufacturers suggests there are no maintenance requirements with EA installations. However, practical, longer-term experience seems to indicate that problems do come up. Specifically, fuel fouling and fuel delivery system problems. However, in this day and age we should be able to overcome fuel filter blockage and other mechanical issues with sound installation design and proper maintenance practices. NFPA 69 presents a standard for such practices. We should also be able to learn from previous installations and therefore users (military, etc.) should be contacted to obtain any documentation that is available. A major obstacle for the EA technology appears to be overall cost. This includes the cost of the material and its installation, its maintenance and the added cost of material removal and refitting if other in-tank systems need maintenance.

10. Research needs On the basis of this review, it is clear more research is needed on how EA can reduce the risks of accidental fuel tank failures, explosions and fires. It is also clear that more work is needed so that EA systems can be properly designed, installed, and maintained. As noted at the beginning of this paper, the following events can lead to the accidental releases and possible ignition of flammable commodities:

ARTICLE IN PRESS A.M. Birk / Journal of Loss Prevention in the Process Industries 21 (2008) 493–505

(i) (ii) (iii) (iv) (v) (vi) (vii) (viii)

Improvised explosive devices (IED) Jetting fires, engulfing fires, remote fires Impacts and drops Munitions Electrical arcs Static electricity Corrosion Manufacturing flaws.

Research is needed to better understand how EA can reduce the hazards associated with the above. The following general research areas are identified: (i) Determine mechanical performance (shock, vibration, drop, impact, etc.) of EA for specific applications (i.e. various shapes and scales of containers). (ii) Controlled fire tests to show benefits of EA with regard to BLEVE type failures and hazards for various scales. (iii) Conduct controlled fire tests to study benefits of EA for bursting non-pressure tanks at various scales. (iv) Conduct IED tests to see if EA mitigates blast, outflow and projectile effects. To support this work, detailed research is needed to better understand the following fundamentals of how EA can mitigate hazards: (i) Effect of EA on heat transfer from a high-temperature wall to a gas filled space. (ii) Effect of surface finish on product performance. (iii) Effect of product on vessel rupture process. (iv) Effect of EA on boiling response and two phase swell during pressure relief. (v) Effect of EA on vessel failure process. (vi) Effect of EA on thermal radiation heat transfer from vessel inside wall. (vii) Effect of EA on PRV behavior and heating. (viii) Effect of EA on passing of a shock or detonation wave.

Acknowledgments This paper was written with financial support from Fusaco SA and the Natural Sciences and Engineering Research Council of Canada. References Anderson, C. E. (1982). Rail tank car safety by fire protection. In Proceedings of the 6th international fire protection seminar. Appleyard, R. D. (1980). Testing and evaluation of the explosafe explosion suppression system as a method of controlling the boiling liquid expanding vapour explosion. TP 2740, Transportation Development Centre, Transport Canada.

505

Birk, A. M. (1983). Development and validation of a mathematical model of a rail tankcar engulfed in fire. Kingston, Ont., Canada: Department of Mechanical Engineering, Queen’s University. Birk, A. M., & Cunningham, M. H. (1994). A medium scale experimental study of the boiling liquid expanding vapour explosion. TP 11995E, Transport Canada. Birk, A. M., Cunningham, M. H., Ostic, P., & Hiscoke, B. (1997). Fire tests of propane tanks to study BLEVEs and other thermal ruptures: Detailed analysis of medium scale test results. TP 12498E, Transport Canada. Birk, A. M., Poirier, D., & Davison, C. (2006). On the response of 500 gal propane tanks to a 25% engulfing fire. Journal of Loss Prevention in the Process Industries, 19, 527–541. Birk, A. M., Poirier, D., Davison, C., & Wakelam, C. (2005). Tank-car thermal protection defect assessment: Fire tests of 500 gal tanks with thermal protection defects. TP 14366E, Transportation Development Centre, Transport Canada. Birk, A. M., & VanderSteen, J. D. J. (2006). On the transition from non-BLEVE to BLEVE failure for a 1.8 m3 propane tank. ASME Journal of Pressure Vessel Technology, 128, 648–655. Birk, A. M., VanderSteen, J. D. J., Davison, C., Cunningham, M. H., & Mirzazadeh, I. (2003). PRV field trials—the effects of fire conditions and PRV blowdown on propane tank survivability in a fire. TP 14045E, Transport Canada. Birk, A. M., & Yoon, K. T. (2006). High temperature stress-rupture data for the analysis of dangerous goods tank-cars exposed to fire. Journal of Loss Prevention in the Process Industries, 19, 442–451. Borghetti, L., Janszen, G., Morandini, M., & Rubini, P. (2000). Fuel tank explosion protection. Hughes Associates Europe SRL. Boyer, H. E. (1988). Atlas of creep and stress rupture curves. Metals Park, OH: ASM International p. 44073. CFR 49 Part 179 Specification for Tank Cars. (2008). US Federal Government, Code of Federal Regulations. Construction, Modification, Qualification, Maintenance and Selection and Use of Containment for Handling. (2005). Offering for transport or transporting dangerous goods by rail. CGSB-43.147-2005, Canadian General Standards Board. Copland, A. (1983). Hydrodynamic ram attenuation. ARBRL-MR-03246. Duquette, A. (2006). Fuel tank safety press release. FAA. Explosafe report. (1988). Newsletter no. 29. Glasair News, pp. 233–234. Fauske, H. K., & Henry, R. E. (2001a). Expanded metal networks: A safety net to thwart gas explosions (pp. 66–71). CEP /www.cepmagazine.orgS. Fauske, H. K., & Henry, R. E. (2001b). The role of expanded metal network in preventing BLEVEs. FAI Process Safety News, Fauske and Associates Inc., pp. 8–10. Glassman, I. (1977). Combustion. London: Academic Press Inc. Guidelines for evaluating the characteristics of vapor cloud explosions, flash fires and BLEVEs. (1994). AIChE Centre for Chemical Process Safety. Hogan, T. A., & Pedriani, C. (1980). Flame tube and ballistic evaluation of explosafe aluminum foil for aircraft fuel tank explosion protection. AFWAL-TR-80-2031, Fire Protection Branch, Fuels and Lubrication Division. McCormick, S., Motzenbacker, P., & Clauson, M. (1988). Passive fuel tank inerting systems for ground combat vehicles. AD-A201 403 no. 13385. US Army Tank Automotive Command. Warren, Michigan: R, D and E Centre. Pierorazio, A. J., & Birk, A. M. (1998). Evaluation of dangerous goods pressure relief valve performance—phase II: Small vessel PRV tests. TP 13259E, Transport Canada. Rodgers, S. A. (2007). Chair, NFPA 69. Standard on explosion prevention systems. 2008 edition. National Fire Prevention Association. Szego, A. K., Premji, K., & Appleyard, R. D. (1980). Evalutation of explosafe explosion suppression system for aircraft fuel tank suppression. AFWAL-TR-80-2043, Explosafe Division, Vulcan Industrial Packaging. Task Group 4, Foam. (1998). Aviation rulemaking advisory committee, fuel tank foam and expanded metal products task group. The effects of security netting material as filling for storage tanks during an explosion of propane gas. (1994). CIBA-GEYGY AG Technical Group of Explosives. USAF. (1978). MIL -B-83054B (USAF) baffle and inerting material. Aircraft fuel tank. USAF. (1999). Baffle material, explosion suppression, expanded aluminum mesh, for aircraft fuel tanks. MIL-B-87162A, US Air Force. White, R. E. (1988). Functional tests for No-Ex(R) explosion suppression. San Antonio, TX: Southwest Research Institute pp. 06-2367. Zalosh, R. (2007). Deflagration suppression using expanded metal mesh and polymer foams. Journal of Loss Prevention in the Process Industries, 20, 659–663.