Keeping Disaster at a Distance To safely site potentially explosive materials, engineers use computer models to predict where building debris might fall after a catastrophic explosion. By Patricia Moseley Bowles and PA Cox W enever ammunition is being stored, there exists the slight but daunting prospect of accidental explosions. These internal explosions can occur through impact, thermal changes, friction, static electric discharge, or the instability of aging munitions. But whatever the cause of such catastrophic accidents, the potential for destruction of property, equipment - or more importantly, human life - is sobering. This potentially explosive situation is being addressed by engineers and analysts in the Engineering Dynamics Department of Southwest Research Institute (SwRI), who have been developing and perfecting sophisticated debris prediction tools that can help to minimize the devastation resulting from the accidental explosions of ammunition stored in hardened aircraft shelters or similar magazines. Numerous studies have been conducted over the past 20 years to address the effects of accidental explosions. Many were conducted by SwRI for the Norwegian Defence Construction Service (NDCS) and the Klotz Group*, an informal international explosives safety Tests of concrete walls, such as the sequence shown here, were integral to the development of OISPRE2, a PC-based computer code developed at SwRI for predicting the debris scatter and air blast from accidental explosions in structures. 8 Technolo *The Klotz Group, formerly known as the Klotz Club, is currently composed of delegates from Norway, Germany, The Netherlands, Sweden, Switzerland, and Singapore. Other participating countries have included the United States, the United Kingdom, and France. Toda • Summer 2000 E1 35662 working group . These efforts have led to the development of DISPRE2, an accurate and userfriendly PC-based code that can predict the likely blast field or h azardous debris "throw" outside aircraft shelters or munition sstorage structures. Sw RI and its contractors hope it w ill become the ultimate debris prediction tool for the safe siting of structures containin g explosives. With this critically important p redictive tool, the safe siting of these shelters in relation to surrounding structures, equipment, and p ersonnel can be expedited w ith a minimum of wasted space and a m aximum degree of safety assurance. Software history Patricia Moseley Bowles served as a principal analyst in the Engineering Dynamics Department of SwRl's Mechanical and Materials Engineering Division. She is a specialist in predicting the dispersion of fragments and debris following internal explosions in structures, and has been the DISPRE program manager since 1990. Bowles joined the staff of Applied Research Associates in San Antonio as a senior analyst in July 2000. A 1981 study for NDCS correlated an approximate engineering analysis with the exp erimental results of 1:20 and 1:100 scale model tests of the explosive effects on a third-generation Norwegian aircraft shelter. The objective was to determine a method to predict the blast field characteristics outside the shelter and the m aximum exp ected debris d istances from fragmentation of such a shelter following an accidental explosion of ammunition stored in a chamber beneath the floor. In a follow-up phase, the objective was to use these test results to develop a m odel for better estimating the p robability of lethality for humans from such an explosion . The reproducibility and directionality in these tests provided a reliable database, enabling engineers and others to predict how and where build ing debris might fall after such a catastrophic event. On any given site, this inform ation vastly facilitates the placement of such storage magazines in relation to P.A. Cox is a staff engineer in the Engineering Dynamics Department. A structural analyst with special interests in transient response, limit design, and the finite element method, most of his work has been in applied or basic research related to weapon and vehicle dynamics, ship tank design, and the evaluation and minimization of explosive hazards. rubble berm surrounding the structure. surrounding buildings, equipment, and the personnel who use them. The initial velocities and angles at which concrete debris flew from an aircraft shelThis goal was achieved only throu gh years of testing using h ardened ter following an internal detonation were aircraft shelters, w hich are specifically carefully documented . This project designed to resist blast and fragment included 1:15 scale model tests of three effects. Throu ghou t the 1980s, SwRI p arshelter designs: combined Norwegian / ticipated in several large-scale (1 :4 to 1:3) U.S., third-generation U.S., and thirdto full-scale tests of these shelters, w hich generation German . were subjected to internal detonations. In 1992, SwRI began developing a These programs contributed significode to calculate initial debris p aram eters cantly to the growing database of hazardou s LAUNCH VELOCITY AS A FUNCTION OF SCALED debris throw and the COVER DEPTH FOR VARIOUS LOADING DENSITIES air blast distribution following an explosion. Early in 1991, Sw RI began another test pro• Shallow Underground gram for N DCS to furTunnel Test, 20,000 kg 100 ther study the debris • Fould 2,500,000 kg effects from aircraft 'C A Pre Gondola c: o shelters, this time 20,000 kg u en 80 adding a protective rock • Essex I Q) 10,000 kg - Wall Debris Velocity (Basler & Hofmann) " Small Concrete Bunker The relationship between launch velocity and scaled cover depth (scaled using the cube root of the charge weight) seems to follow a similar trend for a wide range of loading densities. The determination of similar rela tionships for debris launch angle and debris mass distribution was not only considered possible, but became a focused goal for explosives safety. These data could be used to develop, computerize, and refine the prediction of debris velocities, angles, and initial mass distribution to facilitate the calculation of debris effects for a variety of structures and to aid in the subsequent siting of the structures. • Aircraft Shelters 1,500 kg /m 3 (UET untamped) 0.06 kg/m 3 • 1,500 kg /m 3 (tamped) 10.2 10.1 10 Scaled Cover Depth , Kilograms Per Meter1J3 T prhn o l o p'v Tori "V • C;l1mmpr ?nnn Q and throw from such detonations. The development of a computer code based on new and existing test data, as well as first-principle calculations for such parameters as launch velocity, angle distribution, and mass distribution, would eliminate the need to test each new design or modification. Processing .. . S .... D~p' ....... 'lNIX.. Ii I r ..... l ... 20"'21 c..pw",1 1 I 4X 5--TlIkF~N~ S S"""'_5_"''' T_ 1. 1 j rJeN-=P:\tIISPflEz\sC£NAIIIO\NOAU S_1N.S NA AK ........ ~ ...... _ ..... 0.......... ,..._....,,10."" - - . - . . . . , . . . . _ .............. ' - - -... juot ..... UOOIiI .... _ . i o c ....... The OlSPRE2 model is a self-contained software package, including numerous FORTRAN modules and Visual Basic pre- and post-processors designed to run in a Windows ™ environment on an IBM-compatible personal computer. The calculation models are based on the physics of structural breakup and a large database of mostly aircraft shelter response data. The model This predictive model was developed by SwRI several years ago with funding from the U.S. Department of Energy (DOE) and the U.S. Department of Defense Explosives Safety Board (DDESB). The model was called DISPRE, for" dispersion prediction," and was approved as a siting tool for explosives processing and handling facilities in November 1990 by both DOE and DDESB. Various intermediate calculation models and three computer codes developed by the Naval Civil Engineering Laboratoryt comprise DISPRE Version 1.0: SHOCK, short for shock loading; FRANC, short for frangible panel; and MUDEMIMP, the acronym for Multiple Debris Missile Impact Simulation. The DIS PRE model has proven effective in reducing required siting distances for many explosive material processing structures when used within OlSPRE2 includes the prediction of air blast around a storage magazine. Air blast, as well as debris, can cause serious damage to structures and personnel in the vicinity of an explosion and is a significant factor in determining the safe separation of storage magazines. its constraints, based on the limits of the test data used to validate the model. Version 1.0 can be used to predict building debris throw for charge weights up to 120 kilograms (260 pounds) in a rectangular structure. Because of a lack of other accurate predictive models at the time, the model was frequently used outside its validated limits. Under funding from the Klotz Croup, Sw RI modified the DISPRE model, rather than simply extrapolating from it to predict throw. This newest model, known as DISPRE2, compares the differences between the internal loads and breakup of above-ground rectangular structures containing less than 120 kg (260 lb) of TNTequivalent explosives, and arch-shaped magazines and rectangular above-ground magazines storing up to 5,000 kg (11,000 lb) of such exploSives. Because the safe separation of magazines also depends on external air blast, DISPRE2 includes the prediction of air blast around the magazine. The software is based on an analYSis of existing data and fundamental calculations and has been validated to a level of accuracy consistent with the existing test data chiefly aircraft shelter breakup, debris, and air blast data. Version 1.0 predicted hazardous debris density - defined as no more than one hazardous debris piece per 55.7 square meters - and thus, safe siting distances, reasonably well over a fairly wide range of loading densities. More refined versions have expanded the number of structure types that can be analyzed and has improved the accuracy. Recent efforts have resulted in two versions of the software. The first is specifically for aircraft shelters. The second development version covers seven types of magazines, including aircraft shelters. DISPRE2 Version 2.9 contains both versions of the software. Debris Calculations tNCEL is now the Naval Facilities Engineering Service Center of Port Hueneme, California. LETHALITY PREDICTIONS FOR 10,000 kg ACCIDENT IN AIRCRAFT SHELTER 10-1 Zone 1 Zone 5 Hazard Zones 110' . 1 . 2-4 .... 5 110' Zone 6 10-4 Reproducibility of breakup patterns and debris trajectories in 1:100 and 1:20 scale tests suggested that the prototype shelter would fragment in a manner similar to the rupture patterns observed in the tests, with differences in reinforcement considered. The measured external blast field was also similar. The location of concentrated debris after an explosion appeared to be quite directional in both scales tested and in full-scale tests conducted in the U.S. The reproducibility and directionality of debris throw aided engineers in the development of lethality probability curves for personnel in the vicinity of an accident. TOP VIEW Definition of Hazard Zones 10 100 1,000 Distance From Shelter (10,000 kg Charge) Where do we go from here? loading situations and better meet the storage needs of the Klotz Group members. This expansion will require loading realm differentiation. SwRI has collaborated with the Ernst Mach Institut of Freiburg, Germany, to develop several program plans for the Klotz Group that include testing and collecting data, as well as performing analyses, to accomplish this goal. The upper limit of 5,000 kg (11,000 lb) within a storage magazine corresponds to a relatively low loading density for such a structure. At high loading densities - up to 500,000 kg (1.1 million lb) of explosivesthe distance from the explosive charge to each structural component and the components' thicknesses also become important in determining failure modes and subsequent debris throw distances. When the charge References amount is very high, the P.K. Moseley, M.G. Whitney, "Prediction of Blast charge standoff can be and Debris from an Accidental Explosion Inside a quite small, resulting in Norwegian Aircraft Shelter," SwRl Project 02-5881, close-in shock loading of February 1981. the component. For typical magazine wall F. Riis, "Third-Generation Aircraft Shelter. Debris Throw and Air Blast Caused by Accidental thicknesses, this type of Explosion in the Ammunition Cubicle. Report lli. loading causes cataModel Tests, Scales 1:20 and 1:100," Fortifikatorisk strophic failure of the Notat 149/80, November 1980. wall or a portion of the wall into small debris particles. Because typical stores for magazines other than aircraft shelters can well exceed 5,000 kg (11,000 lb) of explosives, the DISPRE2 model should be expanded to cover high A. Jenssen, F. Riis, 'Third-Generation Aircraft Shelter. Debris Throw and Air Blast Caused by Accidental Explosion in the Ammunition Cubicle. Report II. Pressure Measurements within Steel Model, Scale 1:75," Fortifikatorisk Notat 150/80, November 1980. P.K. Moseley, M.G. Whitney, "Prediction of Injury Levels for Humans in the Vicinity of an Accidental Technolo In the high-risk world of explosives storage, out of sight can never be out of mind. Mindful of the life and death importance of this mission, SwRI will continue working to develop predictive tools that are as fail-safe as humanly possible .•:. Comments about this article? Contact Cox at (210) 522-2315 or pcox@swri.org. Explosion Inside a Third Generation Norwegian Aircraft Shelter," SwRl Project 02-6863, September 1982. K.A. Marchand, P.M. Bowles, "Scaled Tests and Debris Analysis of Model Aircraft Shelters," SwRl Project 06-4164, September 1991. P.M. Bowles, M.e. Whitney, C-J. Oswald, "A White Paper on Computer Code for Calculating Initial Debris Parameters for Accidental Detonations in Magazines," SwRl, January 1991. "Prediction of Building Debris for Quantity-Distance Siting," Department of Defense Explosives Safety Board Technical Paper No. 13, April 1991. P.M. Bowles, c.J. Oswald, L.M. Vargas, w.E. Baker, "Building Debris Hazard Prediction Model," February 1991. P.M. Bowles, C-J. Oswald, M.A. Polcyn, "Earth Covered Ammunition Magazines Quantity-Distance Model, DISPRE2," SwRl Project 07-5394, October 1994. Todav • Summer 2000