Small Debris Impact Simulation with MSC.Dytran – Part II
-
Upload
sloane-bryan -
Category
Documents
-
view
43 -
download
2
description
Transcript of Small Debris Impact Simulation with MSC.Dytran – Part II
Small Debris Impact Simulation with Small Debris Impact Simulation with MSC.Dytran – Part IIMSC.Dytran – Part II
Klaus O. Schwarzmeier,Carlos E. Chaves, Franco Olmi
Embraer S/A
André de Jesus, Eduardo Araújo, Paul BuisMSC.Software Corporation
Presentation Contents
• Introduction• Background• Strain Rate• Sphere Flexibility• Failure Model• Plate x Solid Model• General Conclusions
IntroductionEmbraer – Empresa Brasileira de Aeronáutica S.A. • It is one of the world’s leading designer and
manufacturer of aircraft for regional airlines,defense and corporate use
• Founded in 1969• Based in São José dos Campos,
50 miles from São Paulo - Brazil• Plant area: 2.51 million sq.ft• More than 5,300 aircraft produced
and sold in over 30 years
Embraer – Empresa Brasileira de Aeronáutica S.A.
Introduction
• Previous work (MSC Aerospace 1999)• Steel spheres impacting Al-2024-T3 panels with 1.6 and 3.2 mm thickness
panelsphere at v0
cannon
sphere at vf
barrier
impact region
cam 1cam 2
reference scales
• Experimental results compared with numerical results from dynamic analysis with MSC.Dytran
Background
• MSC.Dytran Models Characteristics• Lagrangian shell elements (CQUAD4)
• Impacting sphere initially simulated as a rigid ellipsoid.
• Constitutive model: Johnson-Cook (YLDJC)
• Failure model: element maximum plastic strain failure (FAILMPS)
]1[)]ln(1[)(0
k
rm
rnpy TT
TTCBA
• Material properties: obtained from MSC.Mvision database
• Previous analysis: all parameters in the Johnson-Cook relation as well as the value of FAILMPS assumed as constants
213232221
3
2pppppp
eqpFAILMPS
Background
0
50
100
150
200
250
300
0 50 100 150 200 250 300
Initial Velocity (m/s)
Fin
al V
elo
city
(m
/s)
Test 3.2mm
Dytran 3.2mm
Test 1.6mm
Dytran 1.6mm
• Discrepancy between analytical and experimental results (sphere velocities) from the previous work is shown in the figure
• This work: Influence of parameters used as input for the constitutive equation and failure model, as well as the modeling characteristics, will be addressed
• Parameters assumed as material constants in the previous work (regardless of the dynamic condition) will be analyzed in detail
Background
Strain Rate• Assuming that actual material stress-strain behavior could differ significantly from the data available, a sensitivity analysis for the strain rate parameter C was carried out
0
50
100
150
200
0 0.05 0.1 0.15 0.2Strain Rate Parameter (C)
Fin
al V
elo
cit
y (
m/s
)
Vini = 113.8 m/sVini = 153.4 m/sVini = 183.8 m/sExperimental
]1[)]ln(1[)(0
k
rm
rnpy TT
TTCBA
•Results for 1.6 mm thickness plate shown in the figure
• Value of C available in the MSC.Mvision database is 0.015
Strain Rate• In order to have close agreement with the experimental results for the range of velocities analysed, C value of must change more than one order
• Conversely, small changes in C will imply in small changes in the final velocity
CONCLUSION: strain rate parameter does not exert a major influence in analysis results
• Influence of the sphere flexibility and friction between sphere and plate considered
• Small reduction in the final velocity due to friction observed
• This trend seems to be correct only when initial velocities are smaller
CONCLUSION: sphere flexibility and friction between sphere and plate do not explain the discrepancies between analytical and experimental results
Sphere Flexibility
Failure Model• Present results showed that the failure parameter (FAILMPS) plays a key role in numerical analysis
• Starting with FAILMPS = 0.18 (static result for Al 2024-T3), a series of analyses with varying values for this parameter were carried out
• Results of these analyses for both plates shown in figures below
FAILURE INDEX (Plate t=1.6 mm)
0
50
100
150
200
0 0.1 0.2 0.3 0.4 0.5
Failure Index (Maximum Plastic Strain at Failure)
Fin
al V
elo
city
(m
/s)
Vini = 113.8 m/s
Vini = 153.4 m/sVini = 183.8 m/s
Experimental
FAILURE INDEX (Plate t=3.2 mm)
0
50
100
150
200
250
300
0 0.1 0.2 0.3 0.4 0.5
Failure Index (Maximum Plastic Strain at Failure)
Fin
al V
elo
city
(m
/s)
Vini = 155.8 m/s
Vini = 205.8 m/s
Vini = 257.9 m/s
Experimental
Failure ModelCONCLUSIONS:
1. Value of FAILMPS
• Previous work: FAILMPS assumed as fixed and equals to 0.18
• FAILMPS may change significantly, according to the impact velocity and plate thickness (geometry)
• This work: failure index changes, and can be significantly higher than the one obtained from static tests
• MSC.Mvision database: failure index FAILMPS = 0.5 supplied for Al 2024-T3
• This may correspond to some specific dynamic condition, but will not necessarily cover the impact conditions evaluated in this work
CONCLUSIONS (cont.)
2. Stress Triaxiality
• Sphere penetrating in a plate: stress state is clearly bi-axial
• It is apparent that a failure parameter based on the effective plastic strain may not be suitable for the dynamic conditions analyzed
• Desirable to find a correlation between some triaxiality parameter and the strain rate, such that a more adequate failure model can be implemented (for example, by means of an external subroutine in MSC.Dytran)
3. Failure Modes
• According to the plate thickness, two distinct failure modes observed
• These failure mechanisms are associated to the triaxial strain state due to plate thickness, and can be interpreted only by means of a more appropriate failure criterion
Failure Model
•LEFT: impact energy = 750 J, thickness = 1.6 mmbulge formation and tearing (petaling)
• RIGHT: impact energy = 1500 J, thickness = 3.2 mmprecipitated plug formation, followed by a plug removal
Failure Model
Solid x Plate Model• Possibility to run the analysis with solid elements also investigated.
• CHEXA elements along the thickness direction.
• Plate with thickness 1.6 mm modeled with 12 layers
• Plate with thickness 3.2 mm modeled with 24 layers
• Material properties: same as in the previous work, C=0.015, FAILMPS = 0.18
40
90
140
190
240
290
80 130 180 230 280
Initial Velocity (m/s)
Fina
l Vel
ocity
(m/s
)
Experimental - t = 1.6mm
MSC.Dytran shell - t = 1.6mm
MSC.Dytran solid - t = 1.6 mm
Experimental - t = 3.2mm
MSC.Dytran shell - t = 3.2 mm
MSC.Dytran solid - t = 3.2mm
• In general, plots show decrease in final velocity for model with solid elements, expressing some ability of solid elements to absorb energy during the impact (or to take into account the strain variations along the plate thickness direction)
• This trend is opposite to the experimental results for the 3.2 mm thickness plate
CONCLUSION: solid elements do not to reproduce appropriately experimental results when compared to the shell elements
• Computational effort: in a NT workstation with 512 Mb of RAM memory and one processor, models with shell elements typically run in about 20 minutes, while models with solid elements last about 20 hours
• Obviously 2D model is much more economic and should always be used when there are time or CPU constraints
Solid x Plate Model
General Conclusions• PRESENT STUDY: brings some important concepts that were not taken into account previously
• Influence of the parameters of material constitutive equation is not relevant when compared to the failure index (FAILMPS)
• Sphere flexibility and friction also do not imply in significant losses of energy
• Failure index (FAILMPS): can change considerably according to strain rate, and values higher than the one obtained by static tests can be considered
• Failure mechanism: quite complex, and a parameter based on an equivalent plastic strain will not describe this mechanism completely. There are also stress triaxiality issues that must be considered. The complete understanding of this parameter requires further studies
• Solid elements: do not result in a major improvement in the analysis results, but imply in a significant increase in the CPU time for the analysis