The Space Congress® Proceedings 1976 (13th) Technology For The New Horizon
Apr 1st, 8:00 AM
Autoignition -A Liquid Propellant Explosive Potential Limiting Phenomena
Wallace H. Boggs Design Engineering, National Aeronautics and Space Administration, J. F, Kennedy Space Center, Fla.
Follow this and additional works at: https://commons.erau.edu/space-congress-proceedings
Scholarly Commons Citation Boggs, Wallace H., "Autoignition -A Liquid Propellant Explosive Potential Limiting Phenomena" (1976). The Space Congress® Proceedings. 6. https://commons.erau.edu/space-congress-proceedings/proceedings-1976-13th/session-4/6
This Event is brought to you for free and open access by the Conferences at Scholarly Commons. It has been accepted for inclusion in The Space Congress® Proceedings by an authorized administrator of Scholarly Commons. For more information, please contact [email protected]. AUTOIGNITION - A LIQUID PROPELLANT EXPLOSIVE POTENTIAL LIMITING PHENOMENA
Wallace H. Boggs Design Engineering National Aeronautics and Space Administration J. F, Kennedy Space Center, Fla.
ABSTRACT During the design phase of large liquid launch ve quantities to mix, a significant probability of hicles, personnel safety considerations and facili spontaneous detonation exists. The phenomena of ties and equipment design criteria must account for autoignition has been confirmed and quantitatively the unlikely but potentially possible series of described by work performed at the Kennedy Space failures that would lead to unplanned, hazardous Center, the University of Florida and most recently mixing of bulk quantities of propellants. Massive by the Battelle Pacific Nortwest Laboratories. explosion and destruction might be a suspected result. The latest work was completed in May 1975 and no further effort is planned, since sufficient confi Simple theory of chemical energetics would predict dence in predicting TNT equivalence for large quan explosive forces greater than equivalent weights of tities of LOX/LH 2 has been established. A slightly TNT (trinitrotoluene). Judicious use of "buffer" weaker test data base exists for LOX/RP-1 , but for zone land and "facility or equipment "hardening" re this combination safety margins allow sufficient quirements dictates that new projects estimate po confidence for presently anticipated usage. tential explosive yields by precise, realistic analysis. It is appropriate now to summarize the knowledge gained from more than ten years of effort as it has This paper highlights the work done by NASA to become accepted and is no longer a center of re develop and confirm a precise analytical theory search attention. and predictive model for liquid propellant explo sives. It covers a span of almost fifteen years Launch complex layouts are determined by interele- work, most performed under Contract NAS10-1255 with ment distance scales fixed by acoustic hazards and the University of Florida. Dr. Eric A. Farber of explosive hazards. Hazards pertain both to per the University and Mr. J. H. Deese of NASA-Kennedy sonnel and to other facilities and equipment. Static Space Center conceived and conducted tests to es tests were instrumented to give acoustic level pre tablish a theory that autoignition occurs when pro- dictions but explosive levels, generally expressed pellants mix in a certain "Critical Mass" or great as TNT equivalence (percent of equal mass of TNT), er. The author participated in the latter phases were attainable by the more expensive project of of this work and was technical manager of Contract purposefully detonating flight vehicle stages in NAS10-8591. This contract was completed in May statistically significant numbers of ground tests. 1975 by Battelle Laboratories and describes quanti An analytical understanding or a valid empirical tatively the physical phenomena taking place prior model was necessary to avoid the unacceptable cost to autoignition. and waste of large numbers of explosive tests for large mix quantities. The work confirmed that autoignition occurs and prevents the mixing of more than the "Critical During Saturn V planning phases, Mr. J. H. Deese Mass" and therefore limits the explosive yield to of NASA Facilities Design initiated a contract with several thousand pounds (kilograms) which is high Dr. E. A. Farber, of the University of Florida, to ly significant when total propellant loads reach explore a mathematical model approach to TNT equiv hundreds of thousands of pounds (kilograms). alence and other explosive phenomena of liquid rocket propel!ants. It is well known that small The results were useful in establishing explosive quantities of propellants appropriately mixed and safety criteria for Space Shuttle facilities and initiated externally yield TNT equivalence greater operations. than T f OO, which is in accord with the theoretical thermodynamic potential of such "clean" reactions. As apparent mix quantities grow to several hundreds INTRODUCTION of pounds (or kilograms), to tens and even hundreds of thousands of pounds the explosive potential Space vehicle cryogenic fuel and oxidizer pairs do grows progressively less by TNT equivalence. not ignite spontaneously when small quantities are purposefully or accidentally mixed together. How One rather obvious explanation, is that when large ever when some failure mode, such as rupture of a quantities are allowed to mix under the failure separating bulkhead, or bulkheads, causes large
4-33 nodes of interest the percentage actually mixed may aging by the transformation = 1 - X
Upper Bound (95% Confidence Limit)
0.5 J.
a* tux
Propellant Weight, Lb • ~* Figure 1 Estimated Explosive Yield as a Function of Propellant Weight
4-34 Design of Prediction Model Experiments propel!ants would not autoignite. Since large quantities autoignite, it was apparent that the In order to understand the relationships of the transition point or region should be determined by yield function, mixing function, time dependence, experiment and the results then used to better pre chemical and physical reactant properties, and dict large quantity yields. A quantity of pro quantity effects, a "Seven Chart" approach was pel lant mix (at stoichiometric ratio) that would developed. The first chart predicted the maximum certainly (probability of 1) autoignite was postu theoretical energy release as a function of fuel- lated and termed "Critical Mass", an analogy to oxidizer ratio and included a tertiary (LOX/LH2/ nuclear reactions. RP-1) mixture as well as the binary mixtures. The second chart related the yield potential to fuel- Since explosive tests are time and resource con oxidizer ratio, this being different because of suming, the experimental design for obtaining quan- reaction rate differences. Chart 3 related the titive demonstrations requires judicious selection remaining amounts of LOX and Lh^ as a function of of test explosive mix amounts, replications and in time from a relatively low turbulence contact mode, strumentation schemes. Further theoretical work LH2 of course tending to vaporize rapidly. Three and model work was conducted to better estimate thousand pounds (1,350 Kg) of hydrogen was observ this transition region. The dynamics of the great ed to vaporize from a 4400 pound (2000 Kg) terti er mix region were of immediate interest. In tests ary mixture within ten seconds. Chart 4 related with LN2/RP-1, chosen for obvious inertness, and yield potential to time and predicted a maxima at later confirmed with the actual propel 1 ants, the approximately seven seconds for the experiment just mixing was found to be not only non-linear with described. The mixing function-time relationship, time but also not monotonically increasing. The Chart 5, proved the most difficult to analyze. idealized sketch, Figure 2, illustrates the descent High speed photographs, simulation by wax cast at velocity v of a cylindrical plug (cross sec models, vibration mixing for repetitive contact tion) into the surface of the denser propellant, dynamics, and finally a fine wire (low thermal at time t0 and at a later intermediate time, and inertia) thermocouple grid were all used to "map" finally an oscillatory period when density differ progressive mixing dynamics. The three dimension ences and gas bubble phenomena cause the plug to al thermocouple grid proved to be the most power rise above and below an equilibrium position (depth) ful tool but the high speed recording and data re denoted by y and a 0 in the figure. A differen duction were expensive and time consuming. The tial equation can be written and solved by iterative grid was useful in studying the detonation process techniques that will describe the plug motion as itself with rapid propagation and state changes damped sinusoidal. Mixed volume, vapor generated being discernable through skillful data trace ana and other parameters of interest may be related lysis. Combining the yield potential function and to the fluids used, the initial and continuing mixing function led to the expected yield function- time dependent introduction conditions of the less time relationship, Chart 6. dense propellant, and the film heat transfer co efficients. Mixing function predictions were found Before Chart 7, the "Expected Yield" can be devel to be in good agreement with model tests,an example oped, the time from mixing start to detonation must being Figure 3. be determined. With planned, initiated tests this is simple, but where spurious initiation sources Electrostatic charge buildup was measured in the or autoignition takes place this is a much more turbulent mix region during model tests and voltage difficult matter, with some degree of randomness levels observed were on the order of ten thousand inherent. Examiniation of all available data gave volts/cm thus on the order of gas breakdown field a mean of about three seconds and a standard devi strength. ation of about one second. The region of Chart 6 bracketing this time interval then becomes Chart 7, The total Saturn V vehicle was to be analyzed with the relation of primary interest. Better statis the primary objective a total stacked vehicle ex tical estimates were possible with a larger, more plosive equivalence and a secondary objective of controlled sample. estimating a fireball expansion rate for the third stage, the S-IVB. Propellant dispersal systems are When the "Expected Yield" from past tests was activated when flight trajectory error due to a compared to test results, it was noted that the flight system failure threatens damage to any yield (remember that this is defined as a percent) life or property. Figure 4 shows the general ar drops off with increasing propellant quantity rangement of Saturn V, with propellant dispersal available for mixing. Several tests had clearly details shown. These are intended to prevent ex detonated prior to planned initiation. It was plosive potential in flight. The "worst case" of possible in the Project Pyro tests, Reference 2, total explosive release is taken to be the rupture that the process for breaking the fuel and oxidiz- of the tanks in the stacked stages, common bulk er separation wall could have inducted initiation heads on the second and third stage, dual bulkheads sources. The time delay suggested however a self on the first stage. The introduction of propellants ignition termed "autoignition" from that time on. would be vertical under gravity assist plus small The following were considered as possible initia velocity from rupture overpressure. ting causes for autoignition, crystal fracture from mechanical or thermal stress, static electri A series of tests were planned and conducted in 1971 city from internal friction of fluid layers or and 1972 at Kennedy Space Center, to define the pre static electricity from fluid-gas interface fric dictive relationships for LOX/LHo and LOX/RP-1. tion. It was known that small quantities of mixed Replicated tests of six pounds (2.7 Kg), 60 pounds
4-35 V o°o° o o o o 0 O o°o °0 o o \ s Equilibrium o o o O 0 0 \ r \/ Position ° U °o° J
O O O o
Figure 2 The Fluid Plug Model
60 pounds (27 Kg) and 240 pounds (108 Kg) were con Extensions and Use of the Model ducted by dumping stoichiometric mix amounts of one propel 1 ant into another. The mixing took place in Spill mixtures were modeled and estimated as well, ground level dewar after introduction of second but naturally the geometric and time factors are constituent from til table elevated dewar. Figure more complicated mathematically. 5 shows the result of a 240 pound (108 Kg) autoig nition explosion. The tests at the smaller quanti Detonation overpressures and velocities were esti ties did not autoignlte: two tests of twenty of mated from the models and confirmed by instrumenta LOX/LH 2 at the 240 pound (108 Kg) quantity did tion of test explosions. autoignite. The important measurements of electro static charge built up in each test were done with The TNT equivalence estimates of both LOX/RP-1 and wire screen grids. The results are shown (normal LOX/LH2 mixtures were, as a result of examining ized) in Figure 6. The solid lines represent the all test data, reduced from 60 percent to 20 per upper and lower limits established from the data. cent for large stored quantities. The lower limit reaches the charge level at which autoignition occurred in the two tests at approx- It should be noted that some appreciable probabil mately 2300 pounds (1050 Kg) extrapolation. This ity of autoignition does exist for smaller mixes confirmed the estimate made earlier for LQX/LH? down to the order of a few pounds (or Kg) and that that this amount could mix before explosion was a these and smaller quantities can be detonated with certainty. It can be seen from the line slopes high TNT equivalence by an external initiating that scatter is greater for smaller quantities, or cause. said another way that the Central Limit Theorem of statistics tends to make for better prediction For intermediate quantities, both yield and proba confidence at larger quantities. Sufficient repli bility need to be estimated on total analysis of cation could demonstrate autoignition at the lower potential failure modes. or intermediate levels but the coincidence of agreement of both the two in twenty autoignitions at 240 pounds (108 Kg) and the 2300 pound (1050 Kg) Analytical Confirmation of Theory lower limit extrapolation with previous prediction provided enough confirmation of the theory. No In 1974 the author contracted with Battelle Pacific autoignition occurred with LOX/RP-1. Lower limit Northwest Laboratories for a team led by Dr. David extrapolation was 2900 pounds (1320 Kg) to a charge Lester to examine existing data and to construct level equated to a 25,000 pounds (11,400 Kg) test analytical (math) models of phenomena leading to during Project Pyro. Slightly lower confidence autoignition in LOX/LH 2 and LOX/RP-1. The heat exists therefore for LOX/RP-1 but until such time transfer and vapor bubble generation and be- as large stages are planned with these propellants, no further predictive model work is anticipated.
4-36 Payload
Instrument Unit
3' x 22' slot in LH tank S-IV B Stage VI) 47" dia. hole in bottom of LOX tank
LH,
-2' x 31' slot in LH 2 tank
S-II Stage 3' x 18' slot in LOX tank
[. LOX 2' x 41' slot in LOX tank • !
S-I Stage RP "2 1 x 20' slot in RP tank
Figure 4 Schematic Diagram of Saturn V, with Effects of Destruct Initiation Indicated-
4-37 FIG. 5 The 240 Pound (110 Kg) Test Explosion Sequence
4-38 100,000 - AUTO IGNITED EXPLOSION
2300 LB. PREDICTED W/0 AUTOICNITION
10,000 -
, POSSIBLE WEIGHT OF T[, ST *QTEST * ELECTRIC CHARGE ON SCREEN OF TEST w 1,000 - QUANTITY
SCREEN OF SIX LB. QUANITY WITH ONE VOLT APPLIED 100 .
10 10 100 1,000 10,000
MIXED WEIGHT OF LO,, AND LH 2 (5/1 RATIO) IN LB.
Fig. 5 Charge Ratio as a Function of Propellent Weight (L0 9 /LH 2 Mixtures)
4-39 Fluid Plug Height 5 inches Radius 1.5 in.
Experimental
0.10
Time, Seconds Fig. 3 Fluid Plug Mixing Function havior models were extensions of the state of the SUMMARY art. Work was carried to a point of substantiating the "Autoigniti on/Critical Mass 11 theory in large That there is an autoignition process which limits part. It was hoped that the single electrostatic the explosive potential of quantities of the com process which controlled reaction thresholds and/or monly used space booster propel 1 ants has been es rates could be isolated. However the results, de tablished. Even quantities up to millions of pound tailed in Reference 4, showed that several inter- (Kg) can be expected to be limited by the autoigni facial motion phenomena (for example streaming po tion of a mix region of a few thousand pounds (Kg) tential between liquid layers) could produce suf which will disperse the remaining propel 1 ant and ficient possible fields to cause a vapor breakdown. prohibit detonation of the total quantities other The field strengths were of an order higher magni wise suspected. Analytical support has been estab tude, in general, at the 240 pound (108 Kg) quanti lished to the amount consistent with available re ty, and two or more orders at 2300 pounds (1050 Kg). sources and currently planned estimation needs. A discharge from droplet to droplet above the liq uid surface was also near the same order for LOX/ LH2 and was for LOX/RP-1 the only phenomena that REFERENCES could clearly be expected to cause initiation. 1. Prediction of Explosive Yield and Other Charac One of the two coefficient terms of Dr. Farber's teristics of Liquid Rocket Propellant Explosions, Critical Mass equation was confirmed but the high Dr. E. A. Farber, et al ., University of Florida, mix energy term was so configuration dependent that Gainesville, June 30, 1973. (Final Report for confirmation was prohibitive. Contract NAS10-1255.)
4-40 2. NASA/USAF Liquid Propellant Blast Hazards Pro gram - Pyro Quarterly Reports - URS Corp. 3. Final Report: TNT Equivalency Study for Space Shuttle (EOS).Aerospace Corporation Report Number ATR-71 (7233)-4, Vol. 1-3. 4. A Study of Liquid Propellant Autoignition, Dr. David H. Lester, et al., Battelle Pacific North- west Laboratories. Final Report on Contract NAS10- 8591.
4-41