MP-091-19
Transcription
MP-091-19
NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments D. Desailly – Y. Guengant – B. Briquet – P. Brunet – D. Houdusse Groupe SNPE/Centre de Recherches du Bouchet rue Lavoisier 91710 Vert Le Petit FRANCE ABSTRACT Violence level prediction for Cook-off events is a major concern for Insensitive Munitions, and moreover for solid propellant rocket motor. Before they are accepted as being safe and suitable for introduction into military inventories, systems undergo a wide range of expensive type qualification trials. There is a need, therefore, to develop a greater understanding of the reaction mechanisms that affect the behaviour of confined energetic materials when subjected to external stimuli, and computer models to control the resulting hazard. Many research programs have been investigated at SNPE Propulsion to predict reactions types (I, II, III, IV, V) in respect to IM STANAG. SNPE Propulsion has developed a method for modelling Cook-off response in order to master the design of Insensitive Munitions (IM). The method requires modular tools describing each physical phenomenon. Works have been first aimed at the development of experimental facilities to characterise thermal and chemical properties, and reactivity of thermally damaged energetic material. Consequently, the first part of the model describing self–heating and pyrolysis processes was implemented in an implicit solver (ABAQUS). Recent developments are now concentrated on the numerical modelling of the reaction violence level prediction hence a thermal damage dependent burning model V.R.E.M.E. has been incorporated in an explicit solver (LS-DYNA). First results come from simulations performed on generic vehicles, full scale munitions, warheads and rocket motors. They have shown quite accurate predictions of ignition temperature and time to event that proved the effectiveness of this approach. Concerning the violence of reaction, no validation phase has been performed at this time, but results accuracy appears to be encouraging and attractive for future applications. 1.0 INTRODUCTION Since the Second World War, in peacetime or during military operations, most of munitions accidents have been due to cook off events. These have happened directly in case of unstable aged energetic materials reaction at moderate temperatures, indirectly when conventional fires reached munitions stockpiles, vehicles or combat platform. Everybody has heard comments about major accidents of US Navy ships (Orraksiny, Forrestal, Entreprise, Nimitz) or the Camp Doha accident during Gulf War – Desert Storm. We must not forget two land storage severe accidents during Vietnam War (Qui Nhon for example), and more recently, in India (January 11th, 2002 Bikaner), Nigeria (January 27th, 2002, Lagos) or Thailand (March 28th, 2002, Aranyaprathet). Thus, predict the reaction and master Cook-off events are a major concern for Insensitive Munitions designers, especially for solid propellant rocket motors. After the self-ignition phenomenon and under Paper presented at the RTO AVT Specialists’ Meeting on “Advances in Rocket Performance Life and Disposal”, held in Aalborg, Denmark, 23-26 September 2002, and published in RTO-MP-091. RTO-MP-091 19 - 1 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments some confinement and thermal conditions, burning can accelerate and lead to more or less violent combustions, explosions or even to Deflagration to Detonation Transition events. It was shown that the cook-off reaction level is ruled by the competition between combustion gas flow, the pressure build-up and the stress release resulting from the break-up confinement. Thus, the challenge is to predict reaction types (I, II, III, IV and V) according to Insensitive Munitions codification and to respect IM STANAG requirements. Our objective is to develop and evaluate test methods and computer models for predicting the response of pyrotechnical systems to the perceived threats of Fast and Slow Cook-off. In order to reduce vulnerability while preserving performances, sensitivity of the results to the dimensioning parameters can be considered (composition optimisation, free volumes, confinement …). 2.0 BACKGROUND ON MUNITIONS BEHAVIOUR PREDICTION TO COOK-OFF Predictive methodology developed by SNPE, is based both on experimental characterisations and numerical modelling. The first step is constituted by experiments involving measurements techniques leading to chemical and physical mechanisms. This overall understanding of mechanisms and variables controlling cook-off is essential for establishing predictive tools. For each elementary mechanism, the model development is divided into four stages: 1. Experimental characterisations on Energetic Materials to achieve a database of intrinsic properties, 2. Physics model development, 3. Fit of the numerical parameters from the experimental database, 4. Implementation of this physical model within a finite elements code to take into account multidimensional munitions design. 2.1 Cook-Off Phenomena – Identification of Reaction Mechanisms An Energetic Material subjected to a thermal threat exhibits exothermic reactions due to chemical decompositions. Thus, this self-heating runaway process may induce self-ignition: the event that follows is usually called Cook-off. Typical time to ignition can be range from few seconds to a few tens hours. Due to thermal exchanges, at high temperatures, ignition occurs close to the energetic material grain surface, at moderate temperatures, the self-ignition occurs inside the grain. The chemical degradation reactions generate also some gas by pyrolysis phenomena with cracks and vacuoles as result. Energetic material porosity can increase drastically and when self-ignition occurs, the combustion of this porous media can be very fast and violent. Then, the propagation of this combustion can go from a conductive to a convective form (usually called a deflagration) with regard to influent parameters. 19 - 2 RTO-MP-091 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments Figure 1: Sights of Pristine (on left) and Thermally Damaged Propellant (on right) Porosity in Binder and AP Particles. Energetic materials behaviour subjected to a thermal threat is complex and follows different time scales: self-heating process is counted in minutes or hours whereas pyrotechnic event may require less than a few milliseconds. Thus, we have chosen to develop some modular tools for each phase and phenomenon. 2.2 Main Hypotheses of the Physical Model The methodology developed by the SNPE Research Centre is macroscopic, the ambition is to model the main physical mechanisms. Modelling each chemical degradation (a few tens for propellant) seems very difficult through available resources. In this way, we defined characterisation tests suitable with the main reactions identified previously. More precisely, the objective is to quantify: • the self-heating phenomenon, • the amount of thermally damaged energetic material (quantified by pyrolysis gas flow and mass loss), • the relation between the burning velocity and the reactivity of thermally damaged product. Our methodology is based on correlation between thermal damage, mass loss and reactivity of energetic materials. This assumption is that a one-to-one relation connects thermal damage (which is quantified through mass loss) and the energetic material reactivity. Indeed, it is established that prior ignition, thermal damage is a key parameter to reaction violence level. Thus, we can separately distinguish and treat quasi-static phenomena (chemical and physical damages & self-heating) and dynamical ones (combustion, deflagration, pressure burst, …). 3.0 CHARACTERISATION TESTS Choice of experimental tests is essential to obtain confident data. Samples must be large enough to get all phenomena and thermal conditioning must cover the wide range of full-scale munitions threats. 3.1 Unconfined Thermo-Ignition Test [8] This test has been performed for about thirty years. It is devoted to define critical temperature, data that lead to the self-heating kinetic parameters and potential violence level of each composition. RTO-MP-091 19 - 3 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments Tests are performed for ten constant temperatures with cylindrical samples (50 mm diameter & length) put in an oven where heating is a convective flux type. Time-to-ignition and temperature history curves at the middle of the sample are recorded. Critical temperature is the maximum temperature that will never lead to ignition. oven temperature THERMOCOUPLES 212 193 191 187 184 240 ALUMINIUM CASE SAMPLE CENTER TEMPERATURE °C 230 STEEL OVEN 220 210 200 190 180 170 160 ELECTRICAL RESISTANCE UNITS THE SAMPLE IS PACKED WITH ALUMINUM PAPER 2000 0 4000 6000 8000 10000 12000 14000 TIME IN MINUTES Figure 2: Set-Up of Unconfined Thermo-Ignition Test. 3.2 200 196 Figure 3: Temperature History Curves at Sample Central Point. Under Confinement Pyrolysis Test [8] This test was developed to establish pyrolysis kinetic through increasing pressure. Experimental results enable us to fit kinetic parameters in order to predict the mass loss. PRESSURE POROUS 50 Pression (bars) DISC TRANSDUCER STEEL PART ALUMINUM 40 CASE ELECTRICAL HEATING RESISTANCES 206 °C 216 °C 30 203 °C 20 200 °C 196 °C 10 INSULATION Temps (mn) UNDER CONFINEMENT PYROLYSIS 0 Figure 4: Set-Up of Under Confinement Pyrolysis Test. 3.3 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 Figure 5: Increasing Pressure Curves according to Temperatures. High Pressure Closed Vessel Bomb To establish the correlation between reactivity and thermally damaged product, we perform tests in a High Pressure Closed Vessel Bomb under temperature control. The specific feature of such a system is to perform the tests at high temperature. Thus, we can estimate burning rate of thermally damaged samples without relaxation process. 19 - 4 RTO-MP-091 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments 120 dP/dt (MPa/ms) 100 80 60 40 20 Bouchon capteur + allumage 0 Corps + chambre de combustion montée+ joint 0 50 100 150 P(MPa) 200 250 300 Figure 7: Derived Pressure Curves according to Different Temperatures. Figure 6: “Diodon” High Pressure Closed Vessel Bomb. A combustion law of the Vieille type is deduced from pressure signals and sample geometry. A burning rate versus pressure law is then established with regard to thermal conditioning, and so with regard to mass loss γ: V (γ ) = a(γ ) P n (γ ) This apparatus is adapted to work up to 5 Kbars: we can study a large scope of burning rate according to porosity of thermally damaged energetic materials. Nevertheless, we now focus on experimental devices development quantifying more violent reactions over the burning regimes (Deflagration to Detonation Transition for instance). 4.0 PHYSIC MODEL DEVELOPMENT & IMPLEMENTATION IN NUMERICAL SOLVERS Numerical works was first focused on the time to ignition prediction and thermal damage state assessment, quantified by the mass loss. Recent developments are thus concentrated on the numerical modelling of the reaction violence. 4.1 Thermochemical Model 4.1.1 Thermo-Ignition Kinetic Model A feature of some energetic materials is to exhibit some endothermic or exothermic decomposition processes as a function of temperature. Thus, heat transfer computation is based on the balance equation: λ ∆2 T heat flow λ T ρ S S + = heat source of EM (Endoexothermic processes) thermal conductivity temperature density heat source RTO-MP-091 ρ Cp dT dt heat accumulation ∆2 Cp t Laplacian operator specific heat time 19 - 5 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments Physical parameters ( ρ , C P , λ ) are known as functions of temperature. The heat source parameters are fitted by a three steps Arrhenius law on the unconfined thermo-ignition test results: Reactions: Species: Heat of reactions: 1 2 3 A→ B→ C→ Q1 > 0 Q2 < 0 Q3 > 0 D This scheme originates in experimental observations. Concentration evolution equations enable us to describe reactions rates of progress dε i (i = 1,2,3) : dt dε 1 − E / RT = (1 − ε 1 ) Z 1e a1 dt dε 2 − E / RT = (ε 1 − ε 2 ) Z 2 e a2 dt dε 3 − E / RT = (ε 2 − ε 3 ) Z 3 e a3 dt εi = Progress of reaction i Zi = Frequency factor of reaction i E ai / R = Activation energy of reaction i / Perfect gas constant T Temperature = Volumetric heat flux S generated within the energetic material is written as: S = ρ ∑ Qi i =1, 3 Qi = dε i dt Heat of reaction i The 9 kinetic parameters (Qi , Z i , E ai for i = 1, 2, 3) are fitted on experimental results (Unconfined Thermo-ignition test) through a specific minimisation procedure based on simplex method. Heating rates are assumed to reproduce thermal threats from Slow Cook-off to fast Cook-off. Figures 8 shows, with adapted scale, the calculated temperature curve of the sample central point when a 180 °C ambient air is given. Figures 9 gives reactions progress in the same conditions. Features corresponding to the three reactions can be clearly distinguished. 19 - 6 RTO-MP-091 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments 1 Temperature a coeur - T source = 180 C Avancement 1 Avancement 2 Avancement 3 200 Cinetique tri-etage Reaction 1 Reaction 3 190 0.8 ε1 Temperature en C 180 0.6 170 Reaction 2 ε2 0.4 160 150 ε3 0.2 140 0 10000 20000 30000 40000 Temps en s 50000 60000 70000 80000 0 0 Figure 8: Temperature Curve of the Sample Central Point. 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000 Figure 9: Progress of Reactions. 4.1.2 Pyrolysis Model Pyrolysis kinetic is characterised from the “Under Confinement Pyrolysis test”. The gas production is modelled with an Arrhenius law similar to the self-ignition kinetic. We use a one step kinetic and the gas production is calculated from the formula: G = g (1 - ε) Z e -Ea/RT G gas production Ea activation energy g gas mass produced per EM mass unit R gas constant ε reaction kinetic T temperature Z frequency factor This Arrhenius law is determined analytically from the pyrolysis characterisation test. 4.1.3 Implementation in a Finite Elements Code To take into account bulk effects of systems, the self-heating and pyrolysis models have been implemented in a well-adapted finite elements solver. For designing its grains, SNPE usually operates with the commercial software ABAQUS for thermomechanical analysis of structure under large strains. Once the computation is completed results, can be analysed in terms of: • temperature distribution, location of ignition and time to reaction (self-heating modelling) • thermal degradation and pressure increasing (pyrolysis modelling) All these data will be used for the prediction of the reaction violence. RTO-MP-091 19 - 7 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments 4.2 Numerical Model V.R.E.M.E. ([13], [14]) 4.2.1 Main Ideas – Description This chapter primarily concentrates on the implementation of the combustion model. A general description of the governing processes and constitutive laws is introduced. The simulation of the violence of reaction resulting from the chemical decomposition of the energetic material requires a coupling between the energy issued from the combustion and the energy released by the deformation process of the system. This coupling is ensured by the hydrodynamic model V.R.E.M.E. (Violence of REaction of Energetic Materials). V.R.E.M.E. is a two phases flow model composed of compressible solid and gas phases, combustion inducing a mass transfer between phases. It requires, in addition to the usual conservation equations, an equation describing the kinetics of reaction according to the thermal damage – reactivity relation. The complete equation system requires for its closure an equation of state for each phase and laws of mixture. 4.2.2 Thermodynamics The hydrodynamic pressures of unreacted (s) and reacted (g) products are determined in terms of specific volume and specific internal energy using JWL equations of state: Ps = Ps (e s ,v s ) Pg = Pg (e g ,v g ) es , v s = eg , v g = specific volume and specific internal energy of the solid phase Ps , Pg = pressures of the solid and gaseous phases specific volume and specific internal energy of the gaseous phase ω − R2V ωe ω − R1v e e + + B 1 − P = A 1 − v R v R v 1 2 where A, B, R1 , R2 , ω depend on the composition. The JWL coefficients of the solid phase are deduced from the Energetic material Hugoniot curve. Given the chemical composition, the density and the enthalpy of formation of the explosive material, the thermochemical TIGER code enables us to determine the CJ conditions and isentropic release of the detonation products. The JWL coefficients of the gaseous phase are so established. 4.2.3 Laws of Mixture Laws of mixture are necessary to fix a thermodynamical equilibrium: • The pressure is assumed to be in equilibrium between the two phases. This assumption is justified because at the microscopic scale, during the combustion, the pressure gradient disappears at once after some acoustic reflections between particles. Ps = Pg Ps Pg = pressure of the solid phase = pressure of the gaseous phase 19 - 8 RTO-MP-091 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments • We suppose adiabatic process within the solid phase: des = − Pdv s des dv s P • = internal energy increment of the solid phase = internal volume increment of the solid phase = equilibrium pressure Specific volume and specific internal energy of the two phases are supposed to be additive: v = ∑ xi vi = (1 − λ)v s + λv g i =1,2 e = ∑ xi ei = (1 − λ)e s + λe g i =1,2 v e xi λ vi ei = = = specific volume of the mixture specific internal energy of the mixture mass fraction of the phase i = = mass fraction of gas specific volume of the phase i = specific internal energy of the phase i 4.2.4 Surface Regression Models Two models have been developed in Lagrangian and Eulerian formulations. The second one is more effective in some cases involving large strains and mesh distorsions. • Surface regression in the Lagrangian Formulation A geometrical method is used for propagating the burning front at a celerity according to the porosity γ: Vr (γ ) = a(γ ) P n (γ ) The reacted fraction λ results from the burning propagation model described above: dλ = αVr (γ ) dt (where α is mesh dependant). • Surface regression in the Eulerian formulation Most recently, a surface regression model based on the resolution of Hamilton-Jacobi equation has been developed. We apply a level set formulation to the problem of surface advancement. The level set formulation is based on solving a Hamilton-Jacobi type equation for a propagating level set function, using techniques borrowed from hyperbolic conservation laws. Given a moving closed hypersurface Γ(t ) , we wish to produce an Eulerian formulation for the motion of the hypersurface propagating along its normal direction with celerity Vr (γ ) . The main idea RTO-MP-091 19 - 9 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments is to embed this propagating interface as a level set of a higher dimensional function φ ( φ (→x (t ), t ) = cte ), and thus: → ∂φ ∂x → + ∇φ = 0 ∂t ∂t → If the particle velocity is denoted u , we obtain: → → ∂φ → → + u . ∇ φ + Vr (γ ) n . ∇ φ = 0 ∂t → Since the normal vector is given as n = → ∇φ → , we have the evolution equation for φ : ∇φ → ∂φ → → + u . ∇ φ = Vr (γ ) ∇φ ∂t The main advantage of this formulation is that there are no significant changes to follow fronts from two to three space dimensions. By simply extending the array structures and gradient operators, propagating surfaces are easily handled. 4.2.5 Implementation in a Finite Element Solver V.R.E.M.E. is implemented in the commercial software LS-DYNA that solves dynamic equations with an explicit formulation. This finite elements code is currently used at SNPE Propulsion for solving non linear phenomena: shocks, warhead functioning … Numerical simulations are performed on two and three dimensional systems and lead to estimate pressurisation and expansion velocity of the confinement. These informations are good indicators of the reaction violence. 5.0 APPLICATIONS The purpose of this section is to illustrate our modelling capabilities for predicting both location of ignition, time to event and reaction level of explosive scale model subjected to thermal threats. 5.1 Influence of the Heating Rate on Results The first case deals with the behaviour of a small scale vehicle loaded with PBX explosive (PBHT, RDX). ABAQUS charts are presented below. They point out heating rate influence on ignition location. At slow heating rate (3.9 °C/h), ignition occurs at the centre, at 50 °C/h, it starts near end caps, and at fast heating rate (360 °C/h), it occurs close to the tube. 19 - 10 RTO-MP-091 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments 3.3 °C/h 50 °C/h 360 °C/h Figure 10: Temperature Distributions in a Small Scale Vehicle submitted to Different Heating Rates – Influence of the Heating Rate on the Ignition Locus – ABAQUS Software. 5.2 Reaction Violence Prediction of a Large Vehicle Charged with B2238/SNPE PBX As an illustration of the methodology on large scale experiments, the test case below relates to a Large Vehicle Cook-off Tube (8 litres generic vehicle) subjected to a fire (heating rate of 6 °C/mn). This full-scale test vehicle is made of stainless steel confinement filled with B2238 SNPE PBX. The prediction of the reaction violence of an ammunition subjected to a fire calls upon the methodology developed and exposed previously. The temperature distribution inside the vehicle is shown at two times just before the ignition occurs (2140 and 2150 s): NT11 VALUE +2.06E+01 NT11 VALUE +2.06E+01 +3.89E+01 +3.89E+01 +5.72E+01 +5.72E+01 +7.55E+01 +7.55E+01 +9.37E+01 +9.38E+01 +1.12E+02 +1.12E+02 +1.30E+02 +1.30E+02 +1.49E+02 +1.49E+02 +1.67E+02 +1.67E+02 +1.85E+02 +1.85E+02 +2.03E+02 +2.03E+02 +2.22E+02 +2.22E+02 +2.40E+02 +2.40E+02 +INFINITY +2.42E+02 C B A Figure 11: Temperature Distributions inside the Large Vehicle Cook-Off Test charged with B2238/SNPE – ABAQUS Software. It can be seen a spread in temperatures from 20 °C to 240 °C. Most notably, the simulation predicted the ignition point to be located on the lateral surface near the confinement. Due to symmetrical considerations (heating rate and tube design), the ignition occurs first near the midpoint of the tube. Next figure shows a comparison of the temperature history plots along the (A-C) line. The ignition occurs on the (A-B) line since temperature gradients are insignificant. The pyrotechnical event that follows results from mechanical analysis. RTO-MP-091 19 - 11 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments 280 260 A B C Temperature ( C) 240 220 200 180 160 140 1500 1600 1700 1800 1900 2000 2100 2200 Time (s) Figure 12: Temperature History Plots on (A-C) Line. In this case, the burning front propagates in an undamaged product with a burning rate law determined through the high pressure closed vessel bomb. Calculation performed with LS-DYNA Lagrangian formulation gives the vehicle cook-off tube expansion: EUCLID WP2 : LARGE VEHICLE COOK-OFF TUBE EUCLID WP2 : LARGE VEHICLE COOK-OFF TUBE EUCLID WP2 : LARGE VEHICLE COOK-OFF TUBE time = 0.00000E+00 time = 1.00000E+04 time = 2.22000E+04 Y Z Y X Z Y X Z LS-TAURUS 940.3 Feb99 T=0 X LS-TAURUS 940.3 Feb99 T=10 m LS-TAURUS 940.3 Feb99 T = 22 ms Figure 13: Reaction Level Prediction – Mesh Distortions at Different Times – LS-DYNA Software. The average velocity is about 2 m/s at 10 ms. At the same time, the pressure reached a maximum value of 1.7 Kbars and then shows a slope break. Its decrease is explained through thinning of the tube thickness combined with the zero hardening modulus of the elasto-plastic steel model. That is, beyond yield stress, the curves shapes are ruled by the hardening law of the confinement. 19 - 12 RTO-MP-091 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments EUCLID WP2 : LARGE VEHICLE COOK-OFF TUBE EUCLID WP2 : LARGE VEHICLE COOK-OFF TUBE 1.70 4.25 1.60 4.00 1.50 3.75 3.50 1.40 3.25 1.30 3.00 1.20 2.75 1.10 2.50 1.00 2.25 0.90 2.00 0.80 pressure (Kbars) x velocity (m/s) 1.75 1.50 1.25 1.00 0.75 Hardening law dependant 0.70 0.60 0.50 0.40 0.50 0.30 0.25 0.20 0.00 0.10 -0.25 time (ms) minimum = -4.6843E-01 maximum = 4.3665E+00 time (ms) node# 749 minimum = 1.0000E-03 maximum = 1.7078E+00 Figure 14: Ballistics Prediction – Tube Velocity History. 28.00 26.00 24.00 22.00 20.00 18.00 16.00 14.00 12.00 8.00 10.00 6.00 4.00 2.00 0.00 32.00 30.00 28.00 26.00 24.00 22.00 20.00 18.00 16.00 14.00 12.00 8.00 10.00 6.00 4.00 2.00 0.00 0.00 shell element# 14 Figure 15: Pressure Raise in the Vehicle. The failure is not modelled, but we can notice significant plastic strains after 17 ms. The confinement early fails and the reaction is rapidly quenched, as experiments in the same heating rate conditions. EUCLID WP2 : LARGE VEHICLE COOK-OFF TUBE time = 1.70000E+04 fringes of eff. plastic strain min= 0.000E+00 in element 1154 max= 3.241E-01 in element 673 ref. surface values for shells fringe levels 0.000E+00 > 2.161E-02 > 4.321E-02 > 6.482E-02 > 8.643E-02 > 1.080E-01 > 1.296E-01 > 1.513E-01 > 1.729E-01 > 1.945E-01 > 2.161E-01 > 2.377E-01 > 2.593E-01 > 2.809E-01 > 3.025E-01 > 3.241E-01 > Y Z X LS-TAURUS 940.3 Feb99 Figure 16: Plastic Strains Contours Prediction. 5.3 Figure 17: Pressure Burst Observed during Fast Cook-Off Conditions. Results Accuracy The accuracy of predictions are checked through data issued from trials. From a general point of view, time to reaction and ignition temperature predictions are in a good agreement with the experiments since variations do not exceed 10 %. Because there are no experimental data available at this time, we have no indications how relevant the results concerning the violence level are. Numerical tools may be possibly refined when precise measurement devices recording tube expansion velocity will be available. RTO-MP-091 19 - 13 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments 6.0 CONCLUSIONS AND PROSPECTS Due to the results of research undertaken on the knowledge of the energetic materials reaction mechanisms, test methods and computer models have been developed for predicting the response of generic explosive-loaded munitions to the perceived threats of Cook-off. In this way, SNPE developed a methodology that integrates experimental and numerical tools. The approach is macroscopic since global physical phenomena are characterised. Numerical tools were improved and validated especially for thermochemical processes (self-heating and pyrolysis phenomena). Further work should now concern the validation of the burning model by performing instrumented trials quantifying precisely the violence of reaction. Future plans are to incorporate physical components for simulating more violent events, particularly the Deflagration to Detonation Transition. Military requirements enabled the development of these innovative solutions. The first applications carried out on an explosive load are promising. We are confident on the good behaviour of the method for systems integrating other types of energetic materials. 7.0 REFERENCES [1] R. Kent and M. Rat, “Explosion Thermique des Propergols Solides”, Prop.Expl.Pyro. 7, 129-136 (1982). [2] J. Isler, “Combustion Mechanism of High Explosives and their Relation with DDT process”, Prop.Expl.Pyro.16, 151-155 (1991). [3] J. Brunet and B. Salvetat, “Thermal Threat Behaviour of Solid Propellant Grains”, IM Tech. Symp., 273-288 (June 1992). [4] B. Nouguez and J. Isler, “Insensitive Warhead Concept: SNPE Progress”, IM Tech. SYMP., 99-11 (June 1992). [5] Y. Guengant, J. Isler and D. Houdusse, “Influence of Energetic Material Formulation on the Reaction to Slow Cook-off”, International Symposium on Energetic Materials Technology, ADPA, 1994. [6] Y. Guengant, D. Houdusse and B. Briquet, “Knowledge and Prediction of Munitions Behaviour to Slow Cook-off”, International Workshop on Measurement of Thermophysical and Ballistic Properties of Energetic Materials, Politecnico di Milano, Italy, 1998. [7] Y. Guengant, D. Houdusse and B. Briquet, “Pyrolysis and Self-Heating Characterisations to Predict the Munitions Responses to Slow Cook-Off Test”, Combustion and Detonation 30th International Annual Conference of ICT, Karlsrue, 1999. [8] Y. Guengant, D. Houdusse, B. Briquet, “Pyrolysis and Self-Heating Characterisations to Predict the Munitions Responses to Slow Cook-Off”, IM & EMTS, Tampa, November 1999. [9] W.W. Erikson, R.G. Schmitt, “Modeling and Analysis of Navy Cookoff Experiments”, Sandia National Laboratories, IMEMTS 2001. [10] R.G. Schmitt, T.A. Baer, “Millisecond Burning of Confined Energetic Materials during Cookoff”, Sandia National Laboratories, 2001. 19 - 14 RTO-MP-091 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments [11] B.M. Dobratz, P.C. Crawford, LLNL Explosives Handbook, 1985. [12] T.R. Gibbs, A. Popolato, LASL Explosive Properties Data. [13] D. Desailly – Un modèle de prévision de la violence de réaction d’un matériau soumis à une sollicitation thermique, introduction dans le code LS-DYNA en formulation lagrangienne, V.R.E.M.E. version 1.0 – SNPE Propulsion, juillet 2001 – private communication. [14] D. Desailly – Un modèle de prévision de la violence de réaction d’un matériau soumis à une sollicitation thermique, introduction dans le code LS-DYNA en formulation Euler/Lagrange, V.R.E.M.E. version 2.0 – SNPE Propulsion, juillet 2001 – private communication. 8.0 ACKNOWLEDGEMENT This research program has been sponsored by DGA of the French Ministry of Defence and supported by the EUCLID RTP2 project, a program performed collaboratively by six European countries involved in an industrial consortium. RTO-MP-091 19 - 15 NATO UNCLASSIFIED NATO UNCLASSIFIED Numerical Simulation of Reaction Violence to Cook-Off Experiments SYMPOSIA DISCUSSION – PAPER NO: 19 Discusser’s Name: M.A. Bohn Question: In your presentation you have given a three-step consecutive chemical reaction model to describe the chemical decomposition and the heat output of the chemical reaction. To what extent can such a simplified model describe the real behavior and what different compositions can be described? Author’s Name: David Desailly Author’s Response: To predict the response of pyrotechnical systems submitted to a thermal threat, we developed the following three physical models for each reaction mechanism: • A procedure based on a three steps consecutive chemical reaction to predict self-heating phenomenon. It is macroscopic since the ambition is not to model each chemical reaction that can occur between all the constituents. This model has been tested and validated on PBX explosive and AP based propellant. The results are accurate for these compositions since variations in the time and temperature to ignition between calculation and test do not exceed 10 %. • A procedure to describe pyrolysis phenomenon. • A procedure to predict the reaction violence by implementing a burning model in LS-DYNA FE solver. The response of pyrotechnical systems is assessed with this burning thermal damage model. Discusser’s Name: R. Derr Question: The work you describe here includes phenomenological models that are implemented with empirical data. This approach for cook-off is embraced by the TTCP protocols now maintained by the NATO Insensitive Munitions Information Center (NIMIC). Are you aware of these protocols and have you used them in your work? Author’s Name: David Desailly Author’s Response: SNPE representatives have participated in several NIMIC workshops. NIMIC’s contributions in the field of predictive protocols and associated tests were appreciated (“Cookoff and XDT mechanisms”, 1996, “IM Testing”, 1997, “Small scale testing and modeling”, 2000). The defined protocols give lists of elementary physical & chemical phenomena that happen during vulnerability trials. To establish some munitions predictive assessments, it is necessary to determine energetic material response for each elementary phenomenon. The SNPE paper describes small-scale tests and phenomenological models that have been developed to establish the energetic material responses for each elementary phenomenon during Cook-off tests. The SNPE approach is more physical than chemical, our goal was to develop pragmatic tools for predicting events and reaction types; it is not to list each chemical degradation reaction. Our tools need to be improved, especially for rocket motors, but first results are promising. 19 - 16 RTO-MP-091 NATO UNCLASSIFIED