POLITECNICO DI TORINO Engineering Faculty Master s Degree in Aerospace Engineering Final Thesis DETECTION AND ANALYSIS OF BARELY VISIBLE IMPACT DAMAGE AND ITS PROGRESSION ON A COMPOSITE OVERWRAPPED PRESSURE VESSEL Academic Supervisor: Prof. Erasmo Carrera Company Supervisor: Ing. Gerlando Augello Candidate: Valentina Dolci
INTRODUCTION Spacecraft are highly complex systems composed of various structural, hydraulic, propulsion, electronic and avionic elements. Such complex systems require extensive maintenance. The development of structural health monitoring systems relies on different research disciplines and in addition it affects design and manufacture as well as operation and maintenance. A widely recognized definition of Structural Health Monitoring states that Health Monitoring (as subsystem of the network of technologies and processes known as Integrated Vehicle Health Management) is the process of nondestructively identifying the main characteristics related to the fitness of space component (or system) as it operates.
OBJECTIVES With this thesis we will develop an AUTOMATIC METHOD TO DETECT AND MONITOR DAMAGES on a filament-wound COPV (Composite Overwrapped Pressure Vessel)
FIRST VERIFICATION We choose some analytical tests from Abrate s work, Impact on composite structures : Es. 3.20, Es. 3.21. Es. 3.22. All these examples deal with slow speed impacts causing BVID - Barely Visible Impact Damage. These tests will be simulated with the help of the commercial software RADIOSS. Afterward there will be a postprocessing phase during which we will analize displacements and contact forces.
EXAMPLE 3.22 We choose to simulate the analytical example 3.22 from Chen e Sun. The FEA solver RADIOSS will be a really satisfying tool. The error in comparison with the analytical solution will be less than 7.5 %. The test is described in the next slide. We show the inputs the solver will need, starter and engine file and the Hypermesh model.
ES. 3.22-IMPACT ON RECTANGULAR COMPOSITE PLATE ; E 1 E 2 120 GPa 7.9 GPa ν 12 0.3 G 12 G 23 G 13 5.5 GPa 5.5 GPa 5.5 GPa ρ 1580 Kg/m 3
Steel sphere Diameter 12.7 [mm] Initial Velocity 30 [m/s] Mass 8.537 [g]
Running Total Time 2.5 [ms] /TFILE.00001 Time History Writing Frequency: 1/10 5 /ANIM/DT Start for the Writing of the Animation File: 0 [s] 0..00001 Frequency of Writing Animation File: 1/10 5 Outputs: /ANIM/SHELL/TENS/STRESS/ALL Shell Tensor Results for all layers /ANIM/VECT/CONT /ANIM/SHELL/EPSP/ALL Animation of Vectorial Data-Contact Force between plate and sphere Plastic Strain for all shell element layers
RESULTS The example 3.22 is a multiple impacts case. Chen e Sun show that the impacts occurs in the first 800 [s]. On the next slides you can see the analytical results. The impacts are marked with the letters A and B. Between the two impacts there is no contact and the displacement of the plate is greater than the displacement of the bullet. In this area the velocity remains constant and therefore AB is a line segment.
FEM ANALYSIS WITH THE FE SOLVER RADIOSS
FEM ANALYSIS WITH THE FE SOLVER RADIOSS Thanks to the postprocessing software Hyperview, we can see the Contour Plot of the displacement s magnitude. The software is able to indicate the node of the maximum deflection, which is the central node of the plate as we expected.
FEM analysis CPT Analytical Results Blue line: variation with time of the central node displacement of the plate Red line: variation with time of the sphere displacement
POSTPROCESSING CONTACT FORCE FEM Analysis CPT Analytical Results
SUMMARIZING TABLE FEM ANALYSIS CPT * RELATIVE ERROR [%] MAXIMUM DISPLACEMENT CENTRAL NODE [mm] 3.50 3.49 0.29 MAXIMUM CONTACT FORCE [N] 3750 3500 7.14 * Reference Solution: Classical Plate Theory
The FE solver RADIOSS is able to reproduce multiple impacts in a very coherent way. The results in terms of displacement, contact forces and their variations with respect to time are in accordance with the analytical results. Relative errors stay under 7.5 %. This encourage us to go on with the development of the procedure, using it with a simple virtual test.
Plate Dimensions: 200x200x2 [mm 3 ] Constraints: Simply Supported Plate on the four edges Regular Symmetrical Laminate cross-ply : 10 layers with stacking sequence [0/90/0/90/0] S Material Properties: E 11 141.2 [GPa] E 22 9.72 [GPa] E 33 9.72 [GPa] G 12 5.53 [GPa] G 23 5.53 [GPa] G 31 5.53 [GPa] ν 12 0.3
## Material Law No 25. COMPOSITE SHELL /MAT/COMPSH/1 One of the more complex aspects of the numerical simulation is the modelling of the rupture behaviour of the composite material. The best description of our composite material is made by the LAW 25 - CRASURV FORMULATION. This card is based on the TSAI-WU ELASTIC-PLASTIC MODEL and is used with composite shells with at least one orthotropic layer.
- There is a first ELASTIC PHASE. - Then, reached to the YIELD STATE, the material may undergo an elastic-plastic work hardening with anisotropic Tsai-Wu yield criteria. It is possible to take into account the MATERIAL DAMAGE. The FAILURE can occur in the elastic stage or after plasticisation. It is starts by a damage phase then conducted by the formation of a crack. The maximum damage factor will allow these two phases to separate.
The damage and failure behaviour is defined by the introduction of the following SIX INPUT PARAMETERS : ε t1 Tensile failure strain in direction 1 ε m1 Maximum strain in direction 1 ε f1 Total tensile failure in direction 1: when reached the element is deleted ε t2 Tensile failure strain in direction 2 ε m2 Maximum strain in direction 2 ε f2 Total tensile failure in direction 2: when reached the element is deleted
THE SIX PARAMETERS IN THE PLATE CASE ARE : 0.7e-2 0.6e-2 0.5e-2 0.9506 0.8642 0.7202 During the postprocessing phase we choose the stress σ xx as a comparison parameter.
OUTLINE OF THE PLATE CASE - The plate will be subjected to a first PRESSURE ANALYSIS, to obtain a strength criterion of the structure. - The stress state is recorded and we proceed to the second step: the BARELY VISIBEL IMPACT DAMAGE of the plate. - Finally the damaged plate is loaded again with the same pressure of the beginning to verify the REDUCED STRENGTH CAPABILITY of the structure.
STEP 1: PRESSURE LOAD P = 5 [atm] σ xx max = 451 [MPa]
STEP 2: BARELY VISIBLE IMPACT DAMAGE The plate is impacted by a steel sphere of standard dimensions and Initial Velocity of 15 [m/s] DIAMETER OF THE SPHERE: MASS OF THE SPHERE: d=12.7 [mm] m=8.537 [g]
σ xx max = 637 [MPa]
The Medium Tensional State σ xx caused by the impact is 90 [MPa].
We apply the pressure load of 5 [atm] to the damaged plate
As we expected the stress state presents greater values. The Medium Tensional State σ xx in the 1 direction (0 of the laminate) is 629 [MPa] against the preceding 451 [MPa].
After the impact a growth of the stress state is recorded. There is an increasing of 40 % with respect to the undamaged plate. Let s test the procedure on a concrete case...
On the last part of the dissertation the procedure is applied on a real case:. The tank is made of composite material wound on 6061 Aluminium alloy liner. The cylinder is 500 [mm] height, with a radius of 416 [mm]. The impact occurs at 0 to avoid the complication of modelling a filament winding dome.
- COPV of CFRP FILAMENT WINDING MATERIAL + Al 6061 LINER: Cylindrical coordinate system, need to reproduce the adhesion between liner and composite material P11_SHELL_SANDWICH - Composite made of 11 layers, Stacking sequence: [ 90 4 / -12 / +12 / 90 2 / -12/ +12/ 90 ]: No simmetries! M25_COMPSH
PROPERTIES OF THE COMPOSITE MATERIAL M10/T1000 Resin-Fiber composite Thanks to experimental characterizations we provide to the solver the following mechanical properties: E 11 188 [GPa] E 22 9 [GPa] E 33 9 [GPa] G 12 4.3 [GPa] G 23 4.3 [GPa] G 31 4.3 [GPa] ν 12 0.3 ρ 1100 [Kg/m 3 ]
On the next table there are the deformations of the material inserted in the solver through the material law 25_COMPSH. 3e-3 2.7e-3 2,4e-3 2e-3 1.9e-3 1.8e-3
DEFINITIVE FEM MODEL
Caused to the fact that we have to model a composite obtained by filament winding, we have to impose different cylindrical coordinate systems to the two segments of the tank.
STEP 1) PRESSURE ANALYSIS Pressure Load 50 [MPa] The pressure load is imposed by a curve In the postprocessing phase we will have to control the absence of unacceptable oscillations on the variation of the internal energy with time.
RESULTS σ xx max = 387 [MPa] Medium Tensional State = 200 [MPa]
CHECK OF THE CORRECT APPLICATION OF THE PRESSURE LOAD The variation of the internal energy with time is regular, showing that the load is applied in an appropriate lapse of time during the explicit analysis.
STEP 2) BVID Energetic Level: 4.56 [J] Sphere Mass: 7.7 [Kg] Sphere Diameter: 12.7 [mm] Impact Position: 0 Initial Velocity of the Sphere: 1.088 [m/s] TENSIONAL STATE CAUSED BY THE IMPACT
Maximum stress σ xx max = 412 [MPa]
Medium tensional state σ xx med on the damage area = 300 [MPa]
FAILED LAYERS The postprocessing software Hyperview is able to show damage s depth providing the number of the impacted layers.
The delamination depth obtained on the experimental results is 1.5 [mm]. Considering that each composite layer has a thickness of 0.2 [mm], we can say that the damage touches 7 plys and a half. The impact area is 4 [mm]
The numerical simulation overstate depth and area of the damage. In fact we can see that it concern with the 8 th ply. Because each finite element has an edge of 2 [mm], the impact area is overstated too.
STEP 3) PRESSURE ANALYSIS OF THE DAMAGED STRUCTURE A pressure load of 50 [MPa] is applied to the damaged tank. On the following slides we can see the progression of the pressure simulation.
The stregth capability of the tank is reduced and it is no more able to resist to the applied load after the impact.
CONCLUSIONS The FEM simulation is able - To show the damage occurred to the structure after a Barely Visible Impact, - To provide the stress state and an overstate of the geometrical parameters of the damage (dimensions and depth), - To verify the reduction of the mechanical properties of the material. FORESEEN IMPROVEMENTS To avoid the actual overstate of the damage with respect to the experimental results making the mesh thickness more consistent and refining the description of the material. To make a better simulation of the failure behaviour of the composite giving more informations about the amount of plastic work absorbed before the rupture.