Other Airship Gases

Hydrogen and helium are efficient lift gases for balloons. However, no envelope material can contain them long-term. Hydrogen is easier to contain than helium -- it at least forms a molecule. Methane - Natural Gas - is lighter than air. Methane is a much bigger molecule than H2. Methane will diffuse through the skin more slowly, but note that the higher density will require a larger balloon to lift the same weight.

There are a variety of other gaseous substances of low weight. Some are too heavy to be a balloon gas, such as nitric oxide (NO, 30.006) and formaldehyde (H2CO, 30.026) and ethane (C2H6, 30.069). Methane [CH4 - 16 g/mole] is less dense than air so it floats, and propane [C3H8 - 44 g/mole] is more dense than air so it sinks. A balloon of pure nitrogen would float in air (and even rise slowly), because it has the same volume as a mole of air, but weighs less. It probably would not be very practical to make an actual pure-nitrogen balloon, but it could be done. It just has to be big enough.

By the 1970s considerable success had been achieved in the past in predicting the gross geometric properties of nuclear fireballs as a function of time. However, the amount of physics that is desirable in such calculations is more than that usually contained in hydrodynamic computer codes so that when a discrepancy between the numerical simulations and the experimental data did occur, it was often attributed to phenomena missing from the codes, e.g., radiation, moisture, dust, equations of state of the many constituents of a nuclear burst, etc. This left little opportunity for a good evaluation of the importance of these effects on predictions of large-scale dynamic behavior.

To calibrate the ability to predict the dynamic behavior of nuclear fireballs researchers simulated the balloon detonations conducted under DNA sponsership in November of 1973. These well-instrumented shoes provide excellent data to test the reliability of hydrodynamic models for the rise and expansion of very low yield explosions. To reduce the number of unknown variables often associatet with nuclear effects, in tests conducted in 1973 several large ballo.ms filled with a mixture of methane and oxygen were detonated. These experiments were well ininstrumented and, thus, serve as good benchmark test cases involving primarily hydrodynamic phenomena. These shots consisted of centrally detonated Mylar balloons, originally 9.7 meters in diameter, for which the yields were predicted to be 4 x 10x9 J. The detonation point was 43 meters above ground.

Several important conclusions can be drawn from these results. Most important is that the hydrodynamics does describe the important features of the observed fireball behavior. The code gives rather accurate predictions for the position and size of these fireballs. Since care was taken to keep as many parameters of the code the same as used in nuclear fireball simulations and not to specialize in any way for these smaller shots, the balloon simulations have increased confidence in the ability of codes to predict the hydrodynamic evolution of fireballs from the initial one-dimensional input conditions.

The discrepancy observed with some of the nuclear events, then, must be due either to a lack of accurate Initial data or to some feature of the simulations that cannot be scaled up from the balloon events. In this latter category could be the stratification of the atmosphere over distances involved in nuclear events or the much more complex composition of the fireball which must be modeled with a simplified equation of state. Almost all nuclear fireball calculations have been made with a single equation of state - that for heated air. Although this must be largely correct, it is an approximation, particularly at the edge of the burst region where it can affect radial pressure gradients and significantly modify the growth of the fireball.

For nuclear fireballs simulated previously there had always been some uncertainty concerning the tradeoff between rise rate and radial expansion. One could fit them both for up to, say, 10 torus formation times but then the rise rate would drop below the experimental rate. If one did the calculation with as little numerical diffusion as possible, the result was to achieve too much rise and too little radial expansion. At early times, say 1-3 torus formation times, the fits to all geometric data tended to be good almost independently of the details of the simulation.