Failure Analysis of a Composite Laminate in HyperWorks through Ply Elimination Procedure

Transcription

1 Failure Analysis of a Composite Laminate in HyperWorks through ly Elimination rocedure Ing. Daniele Di anzo Engineering - Rotor ystem Design daniele.disanzo@agustawestland.com

2 Failure Analysis of a Composite Laminate in HyperWorks through ly Elimination rocedure UMMARY Introduction ly Elimination rocedure ubroutine HM_E.tcl resentation of the problem Analysis and Results Future developments and Conclusions 2

3 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure Design assessment of a composite laminate Failure Analysis: a negative M (Margin of afety) does not necessarily imply the failure of the composite laminate. In order to assess the final static ultimate strength, further investigation is needed. ly Elimination is a procedure suitable for this purpose. It allows in-depth investigations into the resistance capabilities of composite laminates. bviously, a real test is needed to have the final validation and approval of the solution. 3

4 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure What is ly Elimination? A negative Margin of afety (M) not always means a catastrophic failure of the laminate. The UD strength is much dominated by the fiber strength. When M negative is caused by transverse stress s 2 overcoming the allowable, the corresponding damage in the matrix can be simulated by degrading the material properties in the transverse and shear directions. Degrading the elastic properties of the areas with M<0, the laminate stiffness changes as well and thus stresses are re-distributed. It is an iterative process, which ends with positive M unless fibers break. In case of fibers break, this actually means the final failure of the laminate 4

5 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure ly Elimination Logical Flow The process can be summarized with the scheme below: Iterative Loop M calculation M < 0 elect elements with M<0 Apply fake material (E 2, G 12 small) M > 0 End of analysis Run analysis End of analysis This is a simplified procedure, which does not consider the type of failure (matrix or fibers failure). Further development will be discussed later. 5

6 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure ly Elimination rocess in HyperWorks HyperView HyperMesh 1. Create/update Actual et (M>0) 2. Create/update Fake et (M<0) M contour et A (Fake) et B (Actual) Assign the proper material Re-run analysis FE oftware: Radioss Linear Iterative Loop 6

7 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure, tartup in HM ly Elimination rocess: tartup in HyperMesh (i.e. tep 0 ) Modify the FE model as follows: reliminary operations 1) create fake material fake_mat 2) create element sets {el_actual_ly# - el_fake_ly#} for each ply 3) create ply sets {actual_ly# - fake_ly#} for each ply 4) define stack sequence (TACK) mat el_actual_ly# el_fake_ly# actual_ly# fake_ly# n plies TACK fake_mat n plies 7

8 tep 0 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure, tartup in HM 1) create degraded material fake_mat E1 nu12 Xt Xc Yt Yc degradation of transverse elastic properties: E2=G12=1 tress allowables 2) create element sets for each ply el_actual_ly# identifies the original ply# at step 0 el_fake_ly# empty set at step 0, gathers the degraded elements 8

9 tep 0 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure, tartup in HM 3) create ply sets - attribute properties and the corresponding element set actual_ly# for each ply (actual) t el_actual_ly# ID fake_ly# for each ply (fake) Empty step 0 t el_fake_ly# 9

10 tep 0 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure, tartup in HM 4) define stack sequence (from HM olver Browser Tree) for the subset of plies, introduce the fake plies in couple with the actual ones and redefine the stack sequence properly, as follows: n plies : : LY_301 LY_302. : LY_313. : n + n plies : : actual_ly_ 301 fake_ly_ 301 actual_ly_302 fake_ly_ 302. : actual_ly_313 fake_ly_ 313. : 10

11 tep 0 Composite Laminate Analysis ( ly Elimination ) in HW Introduction ly Elimination rocedure, tartup in HM Then, ptimization anel is used to define the equation for M, as function of ply stresses (s 1, s 2, s 12 ), for both the actual plies and the fake ones. Failure Criterion: TAI-WU F11 F22 F66 F12 F1 F2 F This allows to have M among the outputs of the analysis in ptitruct. Thus, M pattern is available in HyperView for post-processing. The model is now ready for ply elimination iterative loop, performed by subroutine HM_E.tcl. 11

12 Composite Laminate Analysis ( ly Elimination ) in HW The subroutine HM_E.tcl The subroutine HM_E.tcl, programmed in Tcl-Language, has been developed as an accessory tool to be run in HM (from command window) for an automatic application of the ply elimination loop. ####################################################################### # UBRUTINE HM_E.tcl: MAIN INUT DATA # ####################################################################### set MDEL_ref "MyModel.fem" # original fem model (filename) set actual_matid 20 # ID original material (actual) set fake_matid 21 # ID degraded material (fake) set UBCAE_ID 7 # ID load condition set plyid_list " " # List of plies (ID) for E process set stepid k k = 0:1:K # current step of ply elimination ####################################################################### At each step (k), HM_E.tcl retrieves the analysis results from FE output file MyModel_step(k).out and generates the new FE input file MyModel_step(k+1).fem, with the update of actual plies and fake plies. 12

13 tep 1 tep 0 Composite Laminate Analysis ( ly Elimination ) in HW The subroutine HM_E.tcl The working flow proceeds as following: Warning Message in HM Model_step0.fem Run Analysis Model_step0.out HM_E.tcl Iterate until: o element sets have not been updated anymore o the overall laminate fails Model_step1.fem Run Analysis HM_E.tcl Model_step1.out Model_step2.fem and so on 13

14 Composite Laminate Analysis ( ly Elimination ) in HW roblem Definition roblem: Design optimization of a helicopter MR blade composite tip Composite kin Tip 14

15 Composite Laminate Analysis ( ly Elimination ) in HW roblem Definition Considering a skin tip made up of U/D plies, two different composite materials have been investigated: Glass fibers or Graphite fibers? Which of them is the most reliable solution? The evaluation of M (Tsai-Wu), and the E procedure as well, will help us to find out the best choice. 15

16 Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, M Contour The evaluation has been carried out with the most critical load condition. M CNTUR LT Upper skin tip, ly +45 E tep 0 fake_ly306 M < 0 M 0 actual_ly306 Note that, without stress evaluation, you are not able to say if M<0 is due to an overstress in fiber direction or in transverse/shear direction. 16

17 actual_ly303 Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, Fiberglass plies LY ELIMINATIN: Upper skin tip, Fiberglass plies (ly 0 ) tep 0 tep 1 tep 2 tep 3 tep 4 tep 5 17

18 actual_ly303 Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, Fiberglass plies LY ELIMINATIN: Upper skin tip, Fiberglass plies (ly 0 ) tep 10 tep 15 tep 20 tep 25 tep 30 E convergence has been reached. Updates of the model after step 20 are not significant. 18

19 actual_ly303 Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, Graphite plies LY ELIMINATIN: Upper skin tip, Graphite plies (ly 0 ) tep 0 tep 1 tep 2 tep 3 tep 4 tep 5 19

20 actual_ly303 Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, Graphite plies LY ELIMINATIN: Upper skin tip, Graphite plies (ly 0 ) tep 8 tep 10 tep 13 E convergence has been reached for both configurations. Faster convergence for Graphite skin tip (reached at step 13) The degraded area is smaller than that for the fiberglass solution. Is graphite really to be preferred to fiberglass? 20

21 Upper kin Tip Graphite Fiberglass Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, results actual_ly303 fake_ly303 M > 0 M < 0 Fiber Failure 21

22 Lower kin Tip Graphite Fiberglass Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, results actual_ly303 fake_ly303 M > 0 M < 0 Fiber Failure 22

23 Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, results ly Elimination has shown that graphite skin tip can be subjected to fiber failure. Is it really a local phenomenon or does it propagate? In response to this question, a further degradation of the material can be included in the procedure. A new iterative loop is performed. Actual_mat Actual_LY E2=G12=1 Fake_mat Fake_LY NEW! E1=E2=G12=1 Fail_mat Fail_LY If the new ply set Fail_LY becomes larger and larger as the iteration proceeds, it means that fiber failure propagates: catastrophic failure of the laminate occurs. 23

24 tep 2.2 tep 2.1 tep 2.0 Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, results LY ELIMINATIN (extended): Lower skin tip, Graphite plies actual_ly303 fake_ly303 fail_ly303 24

25 tep 2.5 Composite Laminate Analysis ( ly Elimination ) in HW kin Tip ly Elimination, results LY ELIMINATIN (extended): Lower skin tip, Graphite plies actual_ly303 fake_ly303 fail_ly303 The failed area (M<0, due to s 1 overcoming the allowable) is bound to expand. It is confirmed that catastrophic failure occurs. Thus, the graphite skin tip is not acceptable. Note that: actual_ly + fake_ly + fail_ly = LY (the real ply) 25

26 Composite Laminate Analysis ( ly Elimination ) in HW Conclusions: further developments The current procedure, just as discussed, is conservative. The degradation to fail_mat is not direct, but passes through the intermediate fake_mat (i.e. matrix failure), even though s 1 overcoming the allowable has occurred immediately (that is fiber failure). As development, the routine degrades the areas with M<0 depending on the type of failure. Real LY Fake_mat: E2=G12=1 s 1 >X Fiber failure M<0 actual_ly s 2 >Y t 12 > (s 1 <X) Matrix failure Fail_mat: E1=1 + E2=G12=1 s 1 >X Fiber failure M<0 fail_ly fake_ly Iterative loop 26

27 tep k Composite Laminate Analysis ( ly Elimination ) in HW Conclusions: further developments TAI-WU failure criterion is not capable of distinguishing matrix failure from fiber failure. This limitation can be overcome by including the max stress criterion. The procedure shall be modified as follows: evaluation of the stresses (post-processing of file.h3d, in addition to.out) comparison of the stresses with the allowables: s 1 >X: fiber failure s 2 >Y t 12 > (s1<x): matrix failure degradation of the elements with M<0 according to the type of failure. Model_step(k).fem Run Analysis HM_E.tcl Model_step(k+1).fem and so on M s ij Model_step(k).out Model_step(k).h3d 27

28 Composite Laminate Analysis ( ly Elimination ) in HW Conclusions: further developments The subroutine HM_E.tcl has been modified accordingly. The main benefits are: faster convergence/divergence of the loop: fewer steps are needed distinction of fiber failure from matrix failure, from the very beginning of the procedure if fiber failure occurs, it is found out earlier (thanks to the parallel degradation) Remember that: matrix failure is acceptable fiber failure may be catastrophic In conclusion, the implementation of this subroutine has allowed to automatize ly Elimination in HW. Thanks to this tool, a more in-depth investigation of the structure is possible. It is easier to understand where a reinforcement is needed, thus helping to re-design the structure properly. The results with the modified subroutine are not here presented 28