Failure analysis case study is presented for a two-piece aerosol containing tetrafluoroethane. The bottom exploded off the aerosol container, which hit the man in the eye and blinded him in that eye. The root cause for the explosion was determined to be a lack of strength of the bottom.
Failure analysis case study is presented for a two-piece aerosol containing tetrafluoroethane. The bottom exploded off the aerosol container, which hit the man in the eye and blinded him in that eye. The root cause for the explosion was determined to be a lack of strength of the bottom.
Failure analysis case study is presented for a two-piece aerosol containing tetrafluoroethane. The bottom exploded off the aerosol container, which hit the man in the eye and blinded him in that eye. The root cause for the explosion was determined to be a lack of strength of the bottom.
Case Study of an Aerosol Explosion and a Method to Determine
Explosion Temperatures Michael Fox Richard Hastings Scott Lovald Juan Heinrich Submitted: 24 February 2007 / in revised form: 4 May 2007 / Published online: 26 June 2007 ASM International 2007 Abstract A failure analysis case study is presented for a two-piece aerosol containing tetrauoroethane, commonly referred to as Refrigerant 134a. A gentleman was preparing to recharge the air conditioning system of an automobile when the bottom exploded off the aerosol container, pro- pelling the body of the aerosol container like a rocket, which hit the man in the eye and blinded him in that eye. The aerosol was never connected to the air conditioner, therefore backpressure from the air conditioner (AC) compressor was ruled out as a cause for the explosion. The objective of the study was to determine why the aerosol exploded. Several recently developed test methods were used, including two types of heat-to-burst tests and a puncture chamber to measure the pressure-versus-temper- ature behavior of aerosols. More common test methods were also used, such as water bath pressure tests, hydro pressure burst tests, pneumatic pressure burst tests, hard- ness measurements, weight measurements, metallography, scanning electron microscopy (SEM), energy-dispersive spectroscopy (EDS), and an accident scenario recreation. A semi-empirical correlation between the hardness and weights of the container bottoms was used to determine the explosion temperature and/or pressure. This semi-empirical correlation agrees in principle with an analysis of the explosion pressures using nite-element analysis (FEA). The root cause for the explosion was determined to be a lack of strength of the bottom of the two-piece aerosol coupled with heating the aerosol to temperatures signi- cantly above room temperature. Keywords Aerosol container Burst pressure Failure analysis FEA Predictive method Introduction A middle-aged man purchased a kit to recharge the air conditioning (AC) units on several vehicles he owned. The kit was a do-it-yourself package consisting of three 425- gram, two-piece aerosol containers of Refrigerant 134a (tetrauoroethane) and several adaptors plus a connection hose with a built-in pressure gage. From here forward the aerosol being used will be referred to as R134a. After successfully recharging one vehicle, he moved to a second auto. Using the connectors that came with the R134a kit, he attached one of the kit adaptors to what he believed to be the correct air conditioning tting to determine the ll requirement per the instructions, and then restarted the engine. It turned out to be the wrong air conditioning t- ting, so he stopped the engine at that point after it had run for a period of time. That was the last time the engine ran. The aerosol container of R134a was never connected to the tting but was placed lying on its side on top of the engine valve cover with the hose-gage assembly attached to the R134a container. After stopping the engine, he continued to inspect the engine compartment for several minutes and tried to nd the correct combination of ttings and connections. He was M. Fox (&) Chemical Accident Reconstruction Services, Inc., 9121 E. Tanque Verde Rd. #105, Tucson, AZ 85749, USA e-mail: mikefox@mikefox.com R. Hastings Materials Science Department, University of Arizona, Tucson, AZ 85719, USA S. Lovald J. Heinrich Department of Mechanical Engineering, University of New Mexico, Albuquerque, NM, USA 1 3 J Fail. Anal. and Preven. (2007) 7:165174 DOI 10.1007/s11668-007-9029-1 just about to go get the manual for the vehicle when there was an explosion. The bottom exploded off the two-piece container and the body of the container was propelled like a rocket. The rocketing container hit him with a glancing blow in the eye. He was blinded in that one eye. In Figure 1, A is the hood dent from the hose, B the hood dent from the bottom of the aerosol, C is the location where the hose was found after the explosion, D is where the bottom of the aerosol was found after the explosion, and E is where the aerosol R134a was located just prior to the explosion. The person who came to pick up the pieces and survey the accident after the injured man was taken to the hospital photographed and documented where each piece was found. The body of the aerosol was found approximately 22.3 m (73 ft) away from the drivers side of the vehicle, which is where the injured man was standing next to the engine. It was apparent that after hitting the man in the eye the body of the aerosol continued to be propelled in this same general direction. The bottom of the container and the hose assembly that were attached to the aerosol at the time it exploded were found between the engine and the radia- tor. The engine hood was the type that lifted toward the front of the vehicle, as seen in Figure 1. From the position of the bottom of the aerosol plus a dent visible in the hood (see B in Figure 1), it appeared that the bottom was propelled into the hood, thereby making a small dent in the hood, and then ricocheted downward and landed between the engine and the radiator. There were also bers on the aerosol bottom that were later conrmed via SEM and EDS to be consistent with the hood insulation. This is discussed in more detail later. The hose-gage assembly was also found between the radiator and the engine (see C in Figure 1). There was also hood damage (see A in Figure 1) indicating that the hose-gage assembly hit the hood, thereby being deected downward between the radiator and the engine. If the injured mans description of what happened was true, this meant that the hose-gage assembly would have had to immediately, upon explosion, tear apart from the aerosol to which it was attached. An unconned heat-to-burst test (Pancake Griddle Test), discussed later, conrmed that the hose-gage assembly does instantly separate from the aerosol when the bottom explodes off the aerosol. The objective of the investigation was to determine why the R134a exploded. Was it overheated by sitting on the engine, or was it in some way defective, or did something else happen? Because the R134a was never connected to the AC unit, reverse pressurizing the aerosol container via the AC compressor was not considered possible. The Product The product at issue was a do-it-yourself kit for recharging automobile AC units. Each kit contained three 425-gram aerosols of Refrigerant 134a or tetrauoroethane, referred to in this paper as R134a. The R134a also contained lubricants and sealants and hence was not pure tetra- uoroethane. The aerosol container was a two-piece steel construction. The body and top are one piece formed by impact forging. The bottom is attached to the body via a double seam as seen in Figure 2. The kit also contained various ttings and a hose-gage assembly. The aerosol container was labeled as a Depart- ment of Transportation (DOT) 2Q designation. The al- lowed pressures and burst pressures of aerosol containers are regulated by the DOT [1]. Two DOT requirements of importance in this investigation were the maximum Fig. 1 Schematic preceding the explosion incident as viewed through the windshield Fig. 2 Cross-sectional metallography of a bottom double seam of a two-piece DOT 2Q container. Thickness of metal bottom at double arrows is 0.0356 cm (0.014 in.). This is an exemplar double seam, not the evidence 166 J Fail. Anal. and Preven. (2007) 7:165174 1 3 54.4 C (130 F) vapor pressure of the contents and the minimum burst pressure of the container. Those parameters are presented in Table 1. The Experimental Methodology The rst step after examining the physical evidence from the explosion was to determine the pressure-versus- temperature behavior of the contents of R134a. Following that, a simulation was performed using the automobile that was involved in the explosion incident. The purpose of the simulation was to determine the maximum temperature to which that specic vehicle could heat a container of R134a. Next, various forms of burst testing were performed to determine the burst characteristics of exemplar R134a aerosols. The burst methods included hydrostatic (water), pneumatic (nitrogen gas), and heat-to-burst testing. Scan- ning electron microscopy and EDS were used to examine bers from the hood and from the aerosol bottom along with grease found on the body of the aerosol and the engine belts. After heat-to-burst testing, the aerosol bottoms were cleaned and weighed and hardness measurements were made. Finally, nite element model calculations were made to compare against the experimental and empirical ndings. Each phase of the experimental and theoretical methods is discussed in the following sections of this paper. Pressure vs. Temperature In an investigation of an exploding aerosol it is advisable to measure the temperature-versus-pressure behavior of the contents of the aerosol. This was accomplished by two separate methods. The rst was to submerge the R134a into a water bath at approximately 60 C (140 F) and monitor the pressure as the water bath cooled. This is readily accomplished with R134a because the container is de- signed to accommodate a screw-type puncture valve as seen in Figure 3, which is looking down on a container of R134a submerged in water and taped to two bricks. In turn, that valve can be connected to a hose leading to a cali- brated pressure gage. Two surface thermocouples were attached to the sidewall of the R134a container, one near the top and one near the bottom. The measurement of aerosol pressures (Figure 3) by the water bath method was conducted inside a strong metal cage. The objective of the water bath pressure test is to measure the pressure of the contents at 54.4 C (130 F) and lower. As shown in Fig- ure 3, the 54.4 C (130 F) pressure of the contents was measured to be 1.38 MPa (200 psig), which is 0.14 MPa (20 psig) higher than allowed in a DOT 2Q container (see Table 1). The water bath is maintained at a temperature of 54.4 C (130 F) for a period of time to ensure thermal equilibrium with the R134a contents. For higher temperatures that may approach the bursting pressure of the aerosol container, a device referred to as the Puncture Chamber was developed. A diagram and photo of the Puncture Chamber are seen in Figure 4. The Puncture Chamber is capable of withstanding much higher pressures than aerosol cans. It is designed and built to t tightly around the aerosol container. Once sealed, but prior to testing, the air gaps between the Puncture Chamber walls and the aerosol are ushed with propellant from a sister aerosol. The spring-loaded and self-sealing puncture device on the top of the container is impacted against the bottom of the aerosol inside the chamber. This produces a large hole in the bottom of the aerosol that allows the Table 1 Two DOT Requirements for 2Q Aerosol Containers Requirement Metric Units English Units Maximum 54.4 C (130 F) vapor pressure of contents 1.24 MPa 180 psig Minimum burst pressure 1.86 MPa 270 psig Minimum wall thickness 0.0203 cm 0.008 in. Fig. 3 Water bath method for measuring pressure at 54.4 C (130 F) and below J Fail. Anal. and Preven. (2007) 7:165174 167 1 3 propellant to escape into the small free volume of the chamber. The chamber is then ushed three times briey with the contents of the punctured aerosol. After that, the Puncture Chamber is slowly heated and the pressure and temperature are monitored with a calibrated pressure gage and thermocouple. The pressure and temperature data are continuously recorded during heating and cooling. The Puncture Chamber method has been validated against aerosols containing pure propellants, such as isobutane and pure tetrauoroethane. The pressure-versus-temperature behaviors of those aerosols, as obtained via the Puncture Chamber, were in excellent agreement with the scientic literature for those propellants. Therefore, the Puncture Chamber is considered to be reliable and accurate. Figure 5 presents the pressure-versus-temperature results of R134a using the water bath, which are in good agreement with data from the Puncture Chamber. Also presented are data points from DuPont [2] for pure tetra- uoroethane showing that the results are in good agreement with DuPont, a manufacturer of tetrauoroethane. This agreement is to be expected because the other ingredients in R134a are sealants and lubricants, which are not ex- pected to alter the vapor pressure of pure tetrauoroethane to any signicant degree. Figure 5 also shows that any R134a that explodes below 65.5 C (150 F) would not meet the minimum burst requirement for a DOT 2Q container, which is 1.86 MPa (270 psig). Automobile Simulation Within a few months of the actual incident, an accident recreation simulation was conducted based on an interview with the man who lost his eye. The identical vehicle was used at the same location where the explosion occurred. The goal of the simulation was to measure the maximum temperatures reached when R134a is positioned on a running engine in the same manner as on the day of the explosion. Two thermocouples were attached to the container of R134a on opposing sides of the body of the container, approximately in the center. The R134a was positioned on the engine as it was positioned on the day of the explosion, as shown schematically in Figure 1. Fig. 4 The Puncture Chamber is designed to obtain temperature vs. pressure behavior of aerosol at high temperatures and pressures Fig. 5 Experimentally measured pressure-versus-temperature data of 134a. Data from DuPont [2] are also presented for comparison 168 J Fail. Anal. and Preven. (2007) 7:165174 1 3 One thermocouple was mounted on the R134a sidewall next to the engine and the other thermocouple was mounted on the R134a sidewall facing the sky. Wire mesh was placed over the entire engine compartment for safety, but the hood was kept open as it was on the day of the explosion. The engine (cold) was started and the temper- atures were monitored. The engine was turned off after 40 min. Figure 6 presents the temperature results graphi- cally with time. The maximum temperature reached during the above experiment was 65.9 C (150.7 F) on the underside of the R134a facing the engine, while the topside maximum temperature was approximately 58.9 C (138 F). The average temperature (topside + underside) at that time was 61 C (141.8 F). The maximum temperature happened 14 min after the engine was turned off, which would be expected because water-cooled automobile engines gener- ally get hotter when they are turned off. Also, with no engine fan running, there is no longer any air owing past the R134a keeping it cool. The oscillations of temperature on the thermocouple pointing toward the sky were caused by clouds that would come and go during the simulation. This observation illustrates the variability that can occur in the overall canister temperature during such a simulation. During the simulation, there was no bottom bulging and the exemplar R134a did not explode. An important fact that should not be overlooked is that no point on the R134a used in the simulation got above a surface temperature of 65.9 C (150.7 F), even though in this simulation the engine ran longer than it did on the day of the explosion. The fact that the engine ran for a shorter period of time on the day of the explosion suggests that the R134a was most likely heated to some temperature below 65.9 C (150.7 F) on the day of the explosion. The simulation resulted in the need for more research. On one hand, the temperature did not exceed 65.9 C (150.7 F), thus an R134a should not explode because the minimum DOT burst pressure is 1.86 MPa (270 psig). However, this simulation was made because an R134a exploded and injured the man under conditions similar (temperature wise) to those encountered on the day of the explosion. A plausible explanation is that not all R134a containers are alike and the one that exploded had a lower burst pressure than the exemplar used in the simulation. Examination of the Container Bottom Fibers were found on the inside surface of the bottom of the aerosol container, as seen in Figure 7. Recall that the bottom was found between the radiator and the engine and that there was a dent in the hood suggesting that the bottom rocketed against the hood and was deected downward. Scanning electron microscopy and EDS conrmed that the bers on the bottom of the R134a container were consistent with bers taken directly from the hood. Therefore, the scenario in which the bottom hit the hood and was deected downward appeared consistent with the testimony of the injured man. Black marks were also present on the R134a container body and it was hypothesized (by others) that those were evidence that the R134a had been crushed by moving en- gine belts. All attempts to get a container of R134a into an engine belt such that it could be crushed were unsuccessful. Additionally, it seemed inconceivable that the body of the R134a could be both crushed inside an engine belt and then travel 22.3 m (73 ft) away from the vehicle. It was also observed that the engine belts and pulleys did not exhibit any evidence of physical damage. Nevertheless, SEM/EDS Fig. 6 Simulation testing temperature vs. time data. The hotter curve is for the thermocouple facing down toward the engine Fig. 7 SEM micrograph of bers found on the inside surface of the R134a container bottom J Fail. Anal. and Preven. (2007) 7:165174 169 1 3 of the black marks were consistent with engine grease. Scanning electron microscopy and EDS were also per- formed on the belts removed from the engine of the vehicle, as shown in Figure 8b. The belts contained a high amount of chlorine, whereas the black marks on the R134a contained no chlorine. These results showed that the black marks were not from belt crushing, but were consistent with engine grease that could have been picked up either before or during the explosion. Burst Testing Three different forms of burst testing were performed to determine the burst characteristics of exemplar R134a. The burst methods included hydrostatic (water), pneumatic (nitrogen gas), and heat-to-burst testing. A detailed dis- cussion and comparison of the three methods has already been published [3]. This paper focuses on the heat-to-burst testing, which provides the most conservative estimates of burst pressure and best represents the conditions on the day of the explosion. The testing reported here conrmed the previous published results [3] and showed that hydro testing led to high burst pressures while heat-to-burst testing led to low burst pressures. It should be noted that either test may be used to demonstrate compliance with present day DOT requirements. Due to the energy that can be released in a heat-to-burst test, a device that provides a level of safety during testing is shown in Figure 9. This devise is affectionately called the Rocket Launcher. The heat-to-burst apparatus consists of a 7.6 cm (3 in.) diameter steel tube approximately 46 cm (18 in.) long with one end-cap. Heating tape is wrapped around the steel tube and aluminum bubble insulation is wrapped over the heating tape. A shop rag is placed between the aerosol and the end cap for added insulation. A shop rag is also loosely placed at the opening of the pipe to prevent air currents and thermal uxuations inside the pipe. The rate of heating is regulated by a heating controller that takes its signal from a thermocouple at- tached to the outer or inner surface of the pipe wall. At least two thermocouples are attached to the R134a con- tainer with at least one thermocouple attached to the con- tainer bottom and at least one thermocouple attached to the container body. Temperatures are continuously monitored and recorded. Depending on the rate of heating, there is a Fig. 8 (a) Black residue from R134a container body. (b) Surface SEM/EDS of engine belt 170 J Fail. Anal. and Preven. (2007) 7:165174 1 3 slight temperature difference (5.58.3 C, or 1015 F) between the bottom and the body, with the body being warmer. Heat-to-burst tests on R134a always result in explosions. There were no leaks or partial failures or even step-wise failures. The explosions were all one-step events that oc- curred without warning when the bottom buckles. When the bottom exploded off, the R134a bodies would be pro- pelled approximately 61 m (200 ft) in the direction the apparatus was pointing. The heat-to-burst tests produce the temperatures at which R134as will explode. The temper- ature at which each explosion occurs can be measured accurately and reproducibly using surface thermocouples, heating controllers, and data loggers. The pressure-versus- temperature data (Figure 5) is then used to determine the explosion pressure. The initial motive for the heat-to-burst test was to determine if a thermally exploded R134a could travel at least 22.3 m (73 ft), as it was reported to have done on the day of the accident. In other words, could a thermally ex- ploded R134a with an unobstructed ight path travel the required distance? Because the R134as traveled further than 22.3 m (73 ft), this observation was consistent with the testimony of the injured man and the physical evidence. A travel distance shorter than 61 m (200 ft) can be ex- plained by either the R134a hitting the man in the eye or the arc of ight. The heat-to-burst tests were always con- ducted with a 30 elevation from horizontal and ight paths were always unobstructed. What was surprising and led to further heat-to-burst testing was that the rst R134a exploded at a much lower than expected temperature, approximately 49 to 54 C (120135 F). Further testing revealed a range of explosion temperatures that made the automobile-heating scenario plausible. Of six heat-to-burst tests, at least four were be- low 1.86 MPa (270 psig). The R134a that was used in the automobile simulation (Fig. 1 and 6) exploded in the heat- to-burst test just a few degrees higher than was achieved during the simulation, thus explaining why that specic R134a did not explode during the automobile simulation. Another, unconned form of heat-to-burst test, referred to as a Pancake Griddle Test, was designed to answer questions that arose out of nding both the bottom of the R134a and the hose-gage assembly between the radiator and engine. It made sense that the bottom of the R134a hit the hood and was deected downward between the radiator and the engine, but should the hose-gage assembly also have ended up there, or should it have traveled some distance away from the vehicle with the rocketing body of the R134a? A single R134a with its hose-gage assembly attached was placed on an electric pancake griddle in the middle of a eld. The purpose of the test was to determine what happens when an R134a with its hose-gage assembly at- tached explodes and it is not constrained. To ensure the lack of constraint no thermocouples were attached to the R134a itself. It was a concern that an attachment of any kind, even a thermocouple, might alter the ight path of the pieces once the R134a exploded. The griddle was turned to 100% heat from the very start of the test. In the R134a Pancake Griddle Test, the bottom pointed west and the top east. When the R134a exploded, the top of the container was damaged in a manner very similar to the evidence R134a. The hose-gage assembly ended up 4.9 m (16 ft) directly to the south of the pancake griddle. Hence, the hose-gage assembly was torn away from the R134a instantly upon explosion and did not travel with the R134a body even a few inches. The bottom of the can ended up approximately 45.1 m (148 ft) to the south. The body of the can ended up approximately 74.4 m (244 ft) northeast. Therefore, the result of the Pancake Griddle Test was consistent with the physical evidence from the day of the explosion. Semi-Empirical Correlation As it is in most failure analyses, each new piece of infor- mation led to further questions. The obvious question was: Why, in the heat-to-burst tests, were there such dramatic variations in R134a explosion temperatures (and hence pressures)? In an effort to answer that question, each R134a bottom used in the heat-to-burst testing was thoroughly cleaned in ultrasonic acetone and methanol baths and then dried and weighed. There appeared to be a correlation between the weights of the bottoms and the explosion temperatures (with heavier bottoms exploding at higher temperatures), but the correlation was qualitative and had at least one signicant outlier. The idea that heavier bottoms exploded at higher temperatures and pressures was logical if the Fig. 9 Heat-to-burst test apparatus J Fail. Anal. and Preven. (2007) 7:165174 171 1 3 heavier weight was due to thicker material. Thicker bot- toms would be expected to have greater stiffness and be less prone to buckling than thinner bottoms. Next, the Rockwell 15T supercial hardness of each bottom was measured at multiple locations and a new parameter, which was the multiplication product of the bottom mass in grams times its minimum Rockwell 15T hardness, was created. A fully automated digital readout hardness tester was used. The idea was that hardness is related to strength and weight to stiffness. When this new parameter was plotted against the explosion temperatures, as seen in Figure 10, the correlation factor (calculated via Microsoft Excel) was greater than 0.9. This made sense. The bottom of an aerosol container receives much of its strength from its domelike geometry. The failure mecha- nism of an aerosol bottom is buckling. Once the bottom buckling initiates, the bottom loses almost half of its strength and the bottom can no longer resist the pressure inside the container. At that instant the bottom vigorously buckles outward with enough force and kinetic energy to unravel the bottom double seam. To the unaided eye this appears as a one-step explosion. With the (hardness weight) parameter available, the explosion temperature of the evidence R134a was deter- mined via extrapolation of Figure 10 to be 37.8 C (100 F), which was consistent with the description from the injured man of what happened on the day of the explosion. This 37.8 C (100 F) was determined using an average hardness for the entire evidence bottom. From the automobile simulation data presented in Fig- ure 6, a temperature of 37.8 C (100 F) would require only a relatively short heating time, which was consistent with the description from the injured man of what hap- pened on the day of the explosion. It is worthwhile to note that the evidence bottom was the lightest of all the bottoms (16.5 g) and also had the lowest supercial hardness (HR15T:77.6) compared to all the exemplars. The empirical correlation seen in Figure 10 made it possible to conclude that the R134a not only exploded below 37.8 C (150 F), but also that the specic reason for the explosion was the low weight/stiffness of the bottom. Per DOT Regulations, all aerosols are required to go through a hot water bath such that they are heated to their 55 C (131 F) equilibrium pressure. One might question then, why the evidence R134a did not explode when it went through the manufacturers hot water bath. After all, 55 C (131 F) is greater than 37.8 C (100 F). It is suggested that the residence time and temperature of the manufacturers water bath was insufcient for the contents to be heated to their 55 C (131 F) temperature. Had the water bath been operated at higher temperature and/or the aerosols had a longer residence time in the water bath, this R134a may have been rejected before reaching any con- sumer. Finite-Element Analyses While straightforward stress-strain calculations are appli- cable for simple problems, nite-element analysis (FEA) is a common numerical method for more complicated prob- lems with irregular geometry and loadings. A rst attempt to model the aerosol container bottom behavior was pub- lished earlier [4]. In FEA, a geometrical domain is broken into a nite number of elements, or subdivisions of that domain. The elements are quadrilateral or triangular shaped in the case of a 2D problem or shaped as a brick or tetrahedral in the case of a 3D problem. The elements are connected at a specied number of intersecting points, called nodes. As the mathematical equations are solved for one element, the displacements at that elements nodes are transferred to other elements connected at common nodes. Eventually, as the equilibrium equations for each element are satised, the displacement, strain, and stress elds throughout the entire physical body are known. Understanding these elds al- lows the estimation of the functional failure of a device. The nite element software package ANSYS/LS-DYNA was used to simulate the 2-piece aerosol problem, because the software package has been a leading FEA program for almost thirty years [5]. The software packages are mature and reliable structure analysis tools available with the capabilities to handle highly nonlinear contact problems [6]. For the problem considered here, ANSYS/LS-DYNA with element PLANE162 is used for large-deformations dynamic analysis and the collapse of the container in axi- symmetric geometry. To model contact between surfaces, the 2D automatic single surface (ASS2D) model is used [7]. Fig. 10 Explosion temperature vs. (hardness grams) of R134a bottoms. The linear trendline is drawn by Microsoft Excel 172 J Fail. Anal. and Preven. (2007) 7:165174 1 3 Figure 11 presents a schematic of the dimensions used in the FEA model. Nine separate analyses were completed, varying three levels of two parameters for the R134a bottom, yield strength and thickness. Other than the thickness of the bottom (t B ), all geo- metrical parameters from Figure 11 are held constant as: D = 6.4 cm (2.52 in.), H S = 0.323 cm (0.127 in.), t W = 0.254 cm (0.01 in.), H W = 2.54 cm (1.0 in.), H A = 0.889 cm (0.35 in.), r S = 0.20 cm (0.08 in.), and r A = 6.104 cm (2.403 in.). Each simulation was performed by incrementally increasing the loading pressure inside the R134a by Dp, which is proportional to a pseudo time step, and nding a static solution before advancing the calculation to the next step. The nine analyses performed are listed in Table 2. The failure criterion was selected as the pressure at which there are through-thickness yield stresses in the bottom and the bottom loses its domelike geometry. The reason this failure criterion was chosen is due to the experimental observation that a one-step explosion takes place the instant there is a disturbance of the geometry of the bottom. It is acknowledged that the failure criterion is rather exible at this point in the development of the FEA model and that further research, both experimental and theoretical, is warranted. However, it is a very reasonable criterion based on the information and data available (Figure 12). Table 2 shows that increasing bottom thickness and/or bottom strength will improve the resistance to failure. It also shows that some combinations of thickness and strength predict failures at pressures below 1.86 MPa (270 psig), while other combinations predict failures above 1.86 MPa (270 psig). This is consistent with experimental observations. The failure pressures calculated by FEA can be converted to failure temperatures by using Figure 5. A graphical presentation of Table 2 is presented in Figure 13. Summary and Conclusions A failure analysis case study is presented for an aerosol container. The aerosol container came out of a do-it-yourself Fig. 11 Schematic of dimensions used in FEA model Table 2 Matrix of Yield Strengths and Bottom Thickness and Results of FEA Model YS Ksi Thickness, in. Internal Psig at Failure Deg F at Failure YS MPa Thickness, cm Deg C at Failure MPa cm 50 0.013 209 134 0.34 0.033 56.67 0.011 50 0.014 241 145 0.34 0.036 62.78 0.012 50 0.015 281 157 0.34 0.038 69.44 0.013 60 0.013 232 142 0.41 0.033 61.11 0.014 60 0.014 267 153 0.41 0.036 67.22 0.015 60 0.015 307 162 0.41 0.038 72.22 0.016 70 0.013 245 146 0.48 0.033 63.33 0.016 70 0.014 285 157 0.48 0.036 69.44 0.017 70 0.015 325 165 0.48 0.038 73.89 0.018 Fig. 12 Graphic representation of failure stress in FEA model. Yield strength of bottom illustrated is 34.7 MPa (50 Ksi) and thickness 0.036 cm (0.014 in.) J Fail. Anal. and Preven. (2007) 7:165174 173 1 3 automobile AC recharging kit containing three DOT 2Q, two-piece aerosol containers of tetrauoroethane, com- monly referred to as Refrigerant 134a. A gentleman was preparing to recharge the ACsystemof his automobile when the bottomexploded off the aerosol container, propelling the body of the aerosol container like a rocket, which in turn hit the man in the eye and blinded him in that eye. An empirical correlation using measured hardness and weights of the container bottoms and the experimentally determined burst pressures allowed the calculation of the explosion temperature and pressure on the day of the incident. This empirical correlation agrees qualitatively with a theoretical analysis of the failure pressures using FEA. Several factors were experimentally measured that most likely contributed to the unexpected explosion of the R134a. The 54.4 C (130 F) pressure of the contents was 0.14 MPa (20 psig) higher than allowed by DOT regula- tions for the 2Q container. The strength (hardness) and/or thickness (weight) of the container bottom were insuf- cient to provide the necessary resistance to buckling, thus when the container was heated, the bottom buckled and the aerosol exploded. References 1. Code of Federal Regulations, Transportation, 49 CFR Sections 173.306, 178.33 and 178.33a, October, 2003. 2. DuPont Product Information Bulletin for Suva 134a, February 2003. 3. M. Fox, R. Hastings J. of Failure Anal. Prevent. 6(2), 5056 (2006). 4. M. Fox, P. Zhao and J. C. Heinrich: Computer Stress Analysis of Self-Pressurized Container Bottom, Invited Paper, On-Line Packaging Conference, www.industryids.com, February 2005. 5. S. Moaveni: Finite Element Analysis, Theory and Application with ANSYS, Pearson Education, New Jersey, 2 nd Ed. 2003. 6. ANSYS 7.0: Online Documentation (included with the software). 7. Explicit Dynamics with ANSYS LS-DYNA, Training Manual, Release 7.1, 2 nd edition, 2003. Fig. 13 Graphical presentation of FEA model calculations. The linear trendline is drawn by Microsoft Excel 174 J Fail. Anal. and Preven. (2007) 7:165174 1 3