IM Rocket Motor Design Assessment Toolset – phase 1 RAe_Cullis... · ... (Elastomer Modified Cast...
Transcript of IM Rocket Motor Design Assessment Toolset – phase 1 RAe_Cullis... · ... (Elastomer Modified Cast...
IM Rocket Motor Design & Assessment Toolset – phase 1
Ian Cullis A presentation to: Weapon Systems and Technology ConferenceTrials, Modelling & Validation of Weapon Systems & TechnologiesQinetiQ, Boscombe DownQ Q,
15th November 2012
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/026581
Agenda
1. The Problem
2. Programme Aim2. Programme Aim
3. Strategy & Approach
4. Material
5. Material Algorithm Development
6. Small Scale Testing
7. Energetic Response Model
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
1
The Problem
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
1 The Problem
There is a requirement for the UK to both be able to design and manufacture IM rocket motors and assess foreign rocket motors for in‐service use under a wide range of conditions. Missile systems used in current operations theatre are vulnerable to bullet and fragment impact In many cases thisused in current operations theatre are vulnerable to bullet and fragment impact. In many cases this is due to shock initiation of detonation of damaged propellant spalled across the central bore of the rocket motor (XDT).
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
1 The Problem
The objective of this requirement is to develop, demonstrate and validate a predictive modelling capability for the hazard response of rocket motors, with a particular emphasis on XDT, as part of an improved time and cost effective IM certification process for rocket motorsimproved time and cost effective IM certification process for rocket motors.
Possible responses:
SDT: Shock to Detonation Transition.
The shock generated by the fragment impact is strong enough to generate a prompt detonation.
DDT: Deflagration to Detonation Transition.
The shock generated by the fragment impact initiates burning which transitions into a detonation after the shock has travelled a finite distance into the explosive.
XDT: Unknown Detonation Transition.
The fragment impact fractures the energetic material and projected debris ignites and burns violently/detonates when it strikes a solid surface.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
2
Programme Aim
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
2 Programme Aim
AIMS
• The aim of the programme is to address the development of an ability to design, qualify and certify a rocket motor that is resistant to XDT with an optimised, timely and cost effective limited series of small scale tests, supported where necessary by numerical simulation and material testing.
• The programme elements include:
− Understand Materials ‐ Application of existing material science‐based methods to predict the fracture, ignition and burning behaviour of highly damaged energetic materials.
− Test ‐Model characterisation against small scale instrumented experiments:
− Develop a predictive modelling capability validated against large scale, system level tests.
− Integrate with the related research programmes to understand and define ageing effects, and the wider Smart Certification programme;
− Update and apply the IM assessment protocol
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
3
Strategy & Approach
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
3 Strategy & Approach
Importance of damage
Conditions for ignitionViolence of the burn?
on burning behaviour
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/026589
4
Material
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
4 Material
Propellant (Elastomer Modified Cast Double‐base propellant‐ EMCDB) contains nitrocellulose, nitroglycerine and a nitramine (RDX)nitroglycerine and a nitramine (RDX)
• In service rocket motor propellant.
• Extensive data and experience of its behaviour.
• We need to develop a material model to describe its thermo‐mechanical properties and behaviour.
• Also tested precursor propellants to provide a generalised understanding of the role ofAlso tested precursor propellants to provide a generalised understanding of the role of additives and particulates on thermo‐mechanical properties and behaviour.
• Provides the potential to predict ‘min smoke’ and role and level of nitramines in controlling the violence of any response.violence of any response.
• Provides double base propellant design and assessment capability.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
4 Material
Material Tests performed by Cambridge and QinetiQ
• Tests
− Dynamic Mechanical Thermal Analysis (DMTA)
− Compression TestsCompression Tests
− Damage Tests
− Small Scale Impact Testing
− Fragmentation Testing
− Burn Rate Measurements
h l i− Thermal Properties
− Plate Impact
− Tensile Samples JANNAFTensile Samples JANNAF
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
4 Material
Integrated Modelling‐Experiment methodology.
• Numerical simulation capability based on several integrated semi‐empirical models:
− Experiment measures properties to give the constitutive model for the specific propellant;
− Allows simulation of integrated experiments and prediction of response of the rocket motor;
− Analytic model for mechanical response predicts behaviour;
− Feeds into semi‐empirical capability at various stages including burning.
• Validated by experiments on propellant• Validated by experiments on propellant.
• Exploitation.
− Gives capability to determine constitutive response of other similar compositions with minimal further experimental measurement.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5
Material Algorithm Development
Fully nitrated
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5 Material Algorithm Development
Nitrocellulose (NC)
• Structure and existing data reviewed.
• Basic model developed using Group Interaction Model (GIM).
• Research addressed the additional data required to demonstrate the ability of the model to describe the deformation mechanisms and properties and how they are based on the
Fully nitrated
constituents and microstructure.
• Developing a method to describe the observed statistical variation in nitration level and its importance for mechanical properties and subsequent processing requirements.
• Will provide an ability to assess different sources/batches of nitrocellulose and ultimately reduce the demands for re‐qualification.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5 Material Algorithm Development
NC
• GIM parameters associated with microstructural components that can vary due to processing or molecular architecture.
• Identified 11 checks against measured behaviour/properties to confirm structure (e.g. nitration level) and validate the model.
− Checks include solubility, loss tangent, cross linking, vibration spectra, specific heat, thermo‐mechanical properties.
− Roxel will identify/agree most appropriate from a
Fully nitrated
Roxel will identify/agree most appropriate from a manufacturing perspective.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5 Material Algorithm Development
Nitroglycerine (NG)
• Structure and existing data reviewed.g
• Basic model for solid developed using GIM.
− Potential interactions when nitroglycerine is used to gelatinize
nitrocellulose requires parameters for solid nitroglycerine.
− Requires latent heat of fusion for prediction/validation of liquid properties.
• Allo s st d /prediction of properties of NG NC mi t res ith different le els of NG infiltration• Allows study/prediction of properties of NG:NC mixtures with different levels of NG infiltration.
• Allows interpretation of propellant response following different manufacturing routes as NG sees different environments in separate parts of the composite
− Detectable in DMA
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5 Material Algorithm Development
Prediction of Roxel Propellant A DMTA
i b t d l d d tcomparison between model and data
0.5
0.6This prediction required specific assumptions to be made as to how the NC and NG were
0.2
0.3
0.4
tan δ
modeldata
how the NC and NG were distributed within the propellant
• Subsequently confirmed by Roxel
0
0.1
0.2
150 200 250 300 350 400
Roxel
• Would vary with NC type and content
150 200 250 300 350 400
Temperature (K)• Would vary with age
• Gives the ability to assess propellant components and give quantitative information at an early stage in the production processstage in the production process
• Leads directly to prediction of propellant properties
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5 Material Algorithm Development
Equation of state predicted from composition
• Understanding the microstructural• Understanding the microstructural composition of the propellant allows prediction of the mechanical properties although the propellant behaviour is likely to be complex.
• It will act as a composite material at several levels: by action of the particles in the RDX and by action of the areas with less NG. This complexity can be seen in the Hugoniot data, particularly when compared to the predictions.when compared to the predictions.
• The Stress Hugoniot prediction is very good.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5 Material Algorithm Development
Equation of state predicted from composition
• The shock – particle velocity relationship is not predicted very well on two counts:
− it is a different shape
− it deviates significantly at high particle speedparticle speed.
• The data on shock speed curves upwards – a behaviour not normally seen in materials – and this suggestsseen in materials and this suggests that the material changes somehow when exposed to the more severe conditions. Currently this is thought to be due to the ‘free’ NG interacting with the underlying composition structure.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5 Material Algorithm Development
Constitutive model
• Porter‐Gould format
• Fraction of filler particles not equivalent to previous compositions for which this has been 4.E+07
5.E+07
successful. Assumed Propellant A would act as if it were a particulate composite material.
• Nitramine particles in the propellant would 2.E+07
3.E+07
True
Str
ess
(Pa)
model
stiffen the NC/NG binder.
• Input mechanical work, would debond through a normal damage process, reducing
0.E+00
1.E+07
0 0.1 0.2 0.3 0.4 0.5
T modeldata 243K 1/s
the modulus of the propellant.
• Not the case under most conditions.
• Only one condition suggestive of this
True Strain
• Only one condition suggestive of this behaviour.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
5 Material Algorithm Development
Constitutive model
• Data analysis showed generally the shape of stress‐strain curves different to the Porter‐Gould constitutive model.
• Behaviour typical of a rubbery polymer
3.E+06
• Behaviour typical of a rubbery polymer.
• Volume fraction of particles in the composite is too low to dominate behaviour.
B h i b t d ib d li
2.E+06
2.E+06
Stre
ss (P
a)
• Behaviour best described as linear viscoelastic with strain stiffening as is normally seen in rubbers.
• Peak stress is reached when the pellets start
5.E+05
1.E+06
True
S
modeldata 295K 0.01/s
• Peak stress is reached when the pellets start to debond from one another.
• Different tensile and compression test behaviour
0.E+000 0.1 0.2 0.3 0.4 0.5
True Strain
• Needs better implemented polymer model in numerical simulations.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
6
Small Scale Testing
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
6 Small Scale Testing
Explosively
g
Fragmentation
High speed video
Explosively launched sphere
cameraPropellant sample
Flash array
Diffusing screen
Box PMMA or PolycarbonatePolycarbonate
Foam
Cotton cloth
Simona plastic sheet
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/0265824
6 Small Scale Testing
FragmentationRound 3 Velocity=~750m/s
Impact Crater
Particle Size Distribution by Volume
Fragment Size Distributiony
10%
12%
14%
opor
tion
2%
4%
6%
8%
True
Vol
ume
Siz
e Pr
o
Debris Cloud
0%
2%
0 488
975
1463
1951
2438
2926
3414
3901
4389
4877
5364
5852
6340
6827
7315
7803
8290
8778
9266
9753
10241
10729
11216
11704
12192
12679
13167
13655
14142
14630
15118
15605
16093
16581
17068
17556
18044
18532
19019
19507
19995
20482
20970
21458
Circle Equivalent Diameter [µm]
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
6 Small Scale Testing
Fragmentation predictions based on microstructure of composite propellant and manufacturing route
M d l b d t l th l t hi h• Model based on two lengthscales at which fragmentation can occur
• Size distribution captured reasonably well.
Round 1
• Some obvious mismatches but a consistent argument can account for them.
• Largest number of fragments in the predicted g g pdistribution is from single black pellets.
• If black pellets were found to have fragmented further then they would tend to only fracture into a few
Round 2
y yfragments, increasing the number of fragments in the range 500 – 1000 microns, and reducing the numbers at 1000 – 2000 microns.
Round 3• Fragmentation of the black pellets could occur due to the impact shock as their loss mechanisms should be suppressed in this short time.
Round 3
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
6 Small Scale Testing
Ignition
Round 7: Velocity=~950m/s
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
6 Small Scale Testing
XDT
Round 1714F16: Velocity=~1370m/s
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
7
Energetic Response Model
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
7 Energetic Response Model
Solid
XDT P bl
Solid Material Model
EoSConstitutive CHARM XDT Problem
Rocket Motors
Shaped Charges
ConstitutiveDamageFracture
CHARMEoS
TemperatureHotspots
GRIM
Shaped Charges
Warheads
HotspotsCHEMISTRYProducts
EoSCHEETAH
DYNA
CHEETAH
Ignition Burning
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
7 Energetic Response Model
• Ignition & Burning
− We require complete Equations of State (EoS) for the unreacted propellant and its dreaction products.
− Porter‐Gould Damage model description of hotspot term integrated into CHARM.
− Validation not completed.Validation not completed.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
7 Energetic Response Model
Progress
• Idealised cloud impact to demonstrate functionality.p y
• Reaction delay of ~3µs.
• Retonation reaches main charge by 10µs
• Main charge detonates – over by 15µs.
• Minor developments remain.
• Model m st be able to predict the• Model must be able to predict the
experimental observations.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
8
Model Validation
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
8 Model Validation
Model must be able to reproduce a number of important features in the bulk response of the propellant.
Impact Plume
Debris cloud shapeAnd concave specimen edges
Axial plume ejectionp j
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
8 Model Validation
Progress
• Model implemented in GRIM & DYNA.p
• Damage limiting
− No plume
− Curved edges
− Good debris cloud shape
• No damage limiting• No damage limiting
− Very good plume
− No edge curvatureg
− Wrong cloud shape
Damage Limiting ON OFFV=1009m/s
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
8 Model Validation
Progress
• Resolution study – with damage limiting y g g
V=1009m/s
Baseline Refinement 1 Refinement 2
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
8 Model Validation
Progress
• Resolution study – with damage limiting y g g
• V=1009m/s
Baseline Refinement 1 Refinement 2
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
8 Model Validation
Progress
• Resolution study – comparison with experiment (V=1009m/s)y p p ( / )
Refinement 1 Refinement 2
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
9
Conclusions
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
9 Conclusions
Fundamental Understanding
• Material characterisation has achieved a significant improvement in understanding the• Material characterisation has achieved a significant improvement in understanding the constitutive response of double base propellants.
• Established a methodology for testing other propellants of this type.
D t t d th diff t d f ti f th t i l d t PBX• Demonstrated the very different deformation response of these materials compared to a PBX and role of NG in modifying this response.
• Material science based methods can predict many of the properties of double base propellants, including understanding the role of the interaction of NG with NCincluding understanding the role of the interaction of NG with NC.
• Ability to predict other NG‐NC based propellants represents a significant achievement.
• The semi‐empirical polymer model, however, cannot predict the observed deformation behaviour across the full strain rate regime Needs the implementation of the complete polymerbehaviour across the full strain rate regime. Needs the implementation of the complete polymer model in a new numerical scheme.
• The ability to predict the properties of NC, based on its structure, allows the development of a simple small scale screening test to determine the suitability of a new source of NC. This has p g ysignificant potential in reducing re‐qualification of new sources of NC.
• Cost effective and game changing approach.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
9 Conclusions
Small Scale Testing
S ll l f t ti ft i t h hi d UK fi t d id d• Small scale fragmentation soft recovery experiment has achieved a UK first and provided detailed data on the fragmentation process of the debris cloud in both double base propellants. The measurements of fragment size distribution as a function of impact velocity have validated the damage model based on percolation theory. g p y
• Small scale ignition test successfully observed ignition and growth of reactions in the debris cloud on impact with a transparent surface. The experiments have demonstrated the transition to XDT as the impact velocity increases and the gap distance decreases.p y g p
• The understanding gained of the experimental methodology will allow the development of a robust small scale test to determine the propensity of a propellant formulation to exhibit XDT behaviour together with the importance of rocket motor design parameters and materials to g p g premove/mitigate this response.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
9 Conclusions
Numerical Simulations
N i l i l ti f th f t ti i d t l ti d• Numerical simulations of the fragmentation process, given adequate resolution can reproduce the observed fragmentation behaviour.
• Limiting the rate of damage development in the propellant is an important controlling factor
• The validation of the models in 3D has yet to be completed.
• The integration of the Porter‐Gould damage model with CHARM has been successfully completed but remains to be rigorously validated. The need to fully implement the complete p g y y p ppolymer composite model using a new numerical scheme has been identified as an important requirement for a fully predictive capability.
• The ability of CHARM to generate ignition in a high porosity representation of the debris cloud y g g g p y phas been demonstrated. Validation against the ignition experiments remains to be completed.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658
9 Conclusions
Overall
Th h i t t i l i b d t i l th d h d t t d th b d• The research into materials science based material methods has demonstrated they can be used to predict the properties and behaviour of double base propellants.
• Small scale experiments are capable of understanding and characterising the fragmentation, i iti d b i b h i f th d b i l d i XDT tignition and burning behaviour of the debris cloud in XDT events.
• Numerical simulations, given sufficient resolution, can successfully reproduce the fragmentation process.
• The essential building blocks required to develop a predictive capability for XDT in rocket motors are in place to justify Phase II of the programme.
© Copyright QinetiQ Limited 2012
UNIVERSITY OFCAM BRIDGEUNIVERSITY OFCAM BRIDGE
QinetiQ/12/02658