Explosion Dynamics Laboratory

Explosion Research at Caltech and TWA Flight 800

Why carry out scale-model testing?

The scale-model testing was carried out to examine combustion issues that we could not study in our laboratory testing. These were: a. The motion of the flame between the bays of the CWT. b. Acceleration of the flame by turbulence created by gas motion through the passageways between the bays. c. Dispersal of the liquid Jet fuel into a spray or mist of liquid droplets. d. Effect of the failure of the partitions (FS, SWB3, and manufacturing access panel in SWB2) on the motion of the flame. The principal goal of the testing was to determine if the ignition location would leave a "signature" that could be identified from the wreckage of TWA 800.

What features of CWT were modeled?

All of the features considered essential for understanding how a flame would propagate through the center wing tank were modeled. This included:

1. The 7 bays of the CWT.
2. The partial ribs, beams and spars.
3. The openings (passageways) in the ribs, beams, and spars.
4. The water bottles on the front spar.
5. The vent tubes and vent stringers.
6. Vapor and liquid fuel (thin layer on floor).
7. Structural failure of the spanwise beams, front and mid spar.

What features of the CWT were not modeled?

A number of features were not modeled since these were either considered unimportant (variation in height), modeled in other ways (heating), or else too complex or expensive (failure of tank top, etc).

1. The variation in height from the back to the front.
2. The fuel probes and associated wiring in the tank.
3. The boost and scavenge pumps.
4. The stringers and stiffeners.
5. The variation in elevation of the bottom of the tank.
6. The heating by the air packs.
7. Failure of the top, bottom or sides of the tank.

The top, bottom, rear spar and side of body ribs were modeled as rigid, i.e., nonfailing, structural components. The detailed structure and structural failure mechanism of the tank were not simulated. The only aspect of the structural failure that was simulated was the gross failure of the beams and spars due to pressure differences between the bays.

What type of fuels were used in the 1/4-scale tests?

Two types of fuels were used in the 1/4-scale testing. The first set (1-40) of 1/4-scale tests used simulant vapor fuel rather than hot Jet A. Laboratory experiments with Jet A and other fuels determined that a mixture of hydrogen and propane could be used to simulate Jet A combustion due to Jet A vapor resulting from liquid fuel evaporation at 50 C. This was done to simplify the field testing and enable an unheated tank to be used. The simulant fuel was also selected to compensate for the difference in pressure and temperature between the tests and the conditions in the CWT at the time of the explosion. The second set of tests (41-70) used a heated tank and Jet A. The Jet A was carefully characterized in laboratory combustion tests at Caltech and chemical analyses at UNR. The tank was heated to temperatures between 40 and 60C and evacuated to the pressure altitude of the accident.

How were the tests carried out?

The tank was heated in some cases, partially evacuated, filled with fuel, and mixed. A thermal ignition source consisting of a rapidly-heated, exposed, light bulb filament was used to start the flame. Laboratory tests have shown that the ignition of the flame occurs in an essentially identical fashion for both spark and thermal (filament-type) ignition sources. Video cameras, high-speed (500 frames-per-second) cinematography, pressure, temperature and light measurements are used to record the development of the flame. A special photographic system using parallel light through the transparent sides of the tank was used to follow the motion of the flame through the tank.

What were the results of the tests ?

The laboratory and scale-model experiments with Jet A fuel were carried out under conditions simulating the CWT environment at the event altitude: temperatures of 40 to 50 C (104 F to 122 F) and a pressure of 0.585 bar (8.5 psi). These experiments demonstrated that the fuel vapor-air mixture was flammable over the range of temperatures (40 to 50 C; 104 F to 122 F) and fuel loading (50-100 gallons) present in the CWT at the time of the accident. Furthermore, the elevated temperatures inside the tank, caused by heating from the air cycle machines under the tank, increased the explosion hazard over that posed by a cool tank by substantially increasing the amount of fuel vapor and thus decreasing the ignition energy. Flight and ground testing confirmed this finding.

The measured peak pressure rises recorded during our experiments were between 1.5 and 4 bar (20 to 60 psi), sufficient to cause failure of structural components inside the CWT of a B-747 aircraft. Scale-model experiments, using both Jet A and a simulant fuel, and numerical simulations reveal that the flame front can propagate rapidly between the compartments of the tank once the flame reaches the passageways and vent stringers connecting the compartments. In some cases, the flame will be quenched as it passes through these openings or connections. The motion of the flame through the tank follows a complex path that results in pressure differences between compartments. Scale (1/4) model experiments and numerical simulations have confirmed that the flame path, and consequently, the pressure differentials generated across the fuel tank structures, are dependent upon ignition location.

Our experimental measurements and numerical calculations have shown that combustion-induced pressure differences produce forces on CWT internal structures (span-wise beams and spars) that can cause deformation and failure of these components. The magnitude and direction of these forces determine which components fail, in what direction, and the sequence of failure. The location of the ignition source is reflected in, but not obvious from the examination of, the damage to the various internal structures, since the forces created during the explosion depend on that location, among other factors.