Basics of Composite Modeling for Engineers

Basics of Composite Modeling for Engineers

Presenter: George Laird, PhD, PE, Principal Mechanical EngineerLocation: Boeing’s Future of Flight, Paine Field, May 19, 2016

FEA, CFD & LS‐DYNA Training, Support and

Consulting Siemens NX CAD, CAM, CAE, Teamcenter and Femap

and NX Nastran Sales


FEMAP Symposium, Boeing’s Future of Flight, May 19, 2016 of 30

SEMINAR OUTLINE Basics of Composite Modeling for Engineers Once you remove a bit of the fear, uncertainty and doubt about composite simulation, building accurate first order FEM’s is pretty basic. We are going to focus on the underlying basis of how composites function and show how this ties to simple and verifiable simulation models that can be easily validated for common engineering applications. This background should allow one to work comfortably with basic structures and also know when to punt to a specialist. Who are Predictive Engineering and Applied CAx? Predictive Engineering is a 20+ year engineering consultancy that specializes in finite element analysis using FEMAP, NX Nastran and LS‐DYNA. We are simulation engineers that have built hundreds of FEA models that have been validated in test and in long term service. Our practice spans from satellites to mining equipment to submarines. We are generalists that can leverage expertize from diverse sectors to provide new insights to our clients. Applied CAx is a Siemens PLM Software value‐added‐reseller that is focused on delivering the industry’s highest level of technical support for the software that they advocate. Predictive Engineering is a co‐owner of Applied and provides first line technical support, training and startup services to Applied’s CAE clients (e.g., FEMAP, NX Nastran, NX Adv CAE).



You Win – YOU ARE HERE



1. INTRODUCTION The idea is to provide a roadmap on how one can idealized a structure into a composite model and have confidence that the final FEA results will validate. We view this as a multi‐step process will a little bit of thinking will save a lot of hard work. Starting with some basics we’ll carry an example analysis forward toward final validation.

1.1 BASIC COMPOSITE PRINCIPLES

1.2 BUILDING COMPOSITE SOLID AND SANDWICH LAMINATES

1.3 IDEALIZATION OF THE STRUCTURE INTO A COMPOSITE MODEL

1.4 VERIFICATION

1.5 VALIDATION



2. BASIC COMPOSITE PRINCIPLES It is all about the fibers since the matrix can be considered just a binder that holds things together while the fibers are loaded. One way to think about it is that the fibers, whether glass or carbon, are at least 20x stiffer than the plastic matrix.

2.1 FIRST MECHANICS ‐ ☼ Pretty much all the load goes through the fiber Carry capacity is determined by the amount of fiber volume fraction aligned to the load Carbon fibers break at ~1% strain while glass fibers break at ~1.5% strain

Although stress views feel comfortable, looking at the strain is often quite useful



2.2 WHAT HAPPENS IN THE PLASTIC? It is the transverse and normal loads that tear apart composites. Take a note that the plastic strain is 4x that of the homogenous plastic. Thus, for your composite laminate of 0‐90 plies, a 1% strain down the 0°fiber direction is a 4% strain for the plastic in the 90°ply.

All Plastic 50% FVF



2.3 IN A REAL COMPOSITE If we keep things simple and say that the fiber fails at 1% and the plastic at 4%, we can see that more or less, FRP’s fail at the fiber breaking point with the matrix fighting for pole position in the race to failure.



2.4 ALTHOUGH THE FIBER CONTROLS, THE MATRIX RULES ‐ ░ Once you start to lose the bond between the fibers and the matrix, failure of the composite will occur at an accelerated rate. With a good bond, load is shared between the fiber and matrix.

Bonded Un‐Bonded Bonded Un‐Bonded



2.5 WHAT ABOUT INTERLAMINAR STRESSES? As shown, the matrix transfers shear and any load carrying capacity has to be aligned with the fiber (viz. 20x stiffer than the matrix). In classical laminate theory (CLT) {another TLA} as used in plate FEA, the theory is valid for thin laminates where bending and in‐plane loading dominate. The interlaminar shear stress or strain is then calculated based on in‐plane strains plus the user defined out‐of‐plane shear modulus. At the end of the day, it is sort of rare to have interlaminar failure unless the plastic is bad or something went amiss in the manufacturing process since it is the fibers that control the failure process.

2.6 COMPOSITE FAILURE THEORIES Our preference is to know your material (next section), do your knock‐down and then look at principal stress or principal strain. We like first principles (structures fail under tensile or buckle if they are slender) and composites fail with a snap, crackle and pop. They are not like metals where we have 20+% plastic strains to failure but more like 1% strain (no plastic) to failure; think brittle carbon or glass fibers. Although composites can have graceful and energy absorbing post failure behavior, their initial failure is noticeable.



2.7 LEARN ORGANICALLY We recommend someone new to composite modeling go through the following steps

2.7.1 UNIDIRECTIONAL COMPOSITE MODEL o Build a simply‐supported beam model using plate elements o Make an isotropic material where the elastic modulus is based on the rule‐of‐mixtures for your fiber and matrix

o Run the model and check against hand‐calc o Change material model to 2‐D orthotropic and input values, align material direction o You should get the same values as with the isotropic material

2.7.2 LAMINATE COMPOSITE MODEL

Build a two ply laminate model with each ply ½ thickness of the previous model. The 2D material model will be the same

Run analysis and the results should be the same Rotate one layer, observe unique behavior Add one additional layer (three plys) and change the thickness of each ply to 1/3 of original model. Align all layers and verify the same results as in unidirectional model

Rotate middle layer 90 degrees and finally rotate middle layer 45 degrees

3. BStraandtypForyou

3

o

BUILDINGaight up cod there aree of talk, tr stuff like tu would ma

3.1 MA

o Rule of M

G COMPOSomposite me a ton of tuhere is no that, we haake a solid

ATERIAL M

Mixtures (C

FEM

SITE SOLImodels, whutorials, wneed to goave resour or sandwi

MODELING

Chamis Mo

Basics of ComMAP Symposium,

D AND SAhen it is juswhitepaperso over settrces at Appich laminat

G OF A CO

odel)

mposite Modeling fBoeing’s Future of

ANDWICHst a solid las, blogs, Yoting up youpliedCAx.cote from ma

OMPOSITE

for Engineers f Flight, May 19, 20

H LAMINAaminate or ouTube infur materialom. But whanufacture

E PLY

016

ATES r just a sanformation l angle or that we caner’s data.

dwich is noavailable. the simple n talk abou

of 30

o big deal In this tutorial. t is how

o

o

o

o Workingo Treat two Uni‐direccrank

from mano plies as octional dat

FEM

nufacturer’one ply ta is comm


’s data

on or hand


d‐


of 30

3Westrumo

3.2 WHe are not neuctures wodeling req

We keepand whecombineWe like tply failurelement criterionPut 10 coroom ancertifyingWorld‐W20% diffeand Test

HAT DO Wecessarily gork but brouirements

p it simple en appropred plies as oto include re and empdeath as o omposite ed you’ll geg a compo

Wide Failureerence bett

FEM

WE DO? good role ought in to are differe

and modeiate modeone ply progressivploy FEA our pass/fa

experts in aet 10 ways site to “pae Exercise tween FEA


models sinverify thatent.

l l

ve

ail

a of ass”is A


nce we are t a design w


often not will make i

016

involved init through

n the desigvalidation

of 30

gn of . Thus our



3.3 MATERIAL KNOCK DOWNS AND HOLES ‐ ☼ This is a good question and depends on your level of risk management. Working from first order material data (i.e., Chamis Rule‐of‐Mixtures) or from manufacture’s data, it is difficult to provide guidance without testing. One suggestion that we have heard is to knock down all first order data by 3x. This has some engineering guidance if one considers that a stress concentration of 3x exists around all holes in plates (isotropic material).

We find it illustrative to just build simple models and do your own exploration –



3.4 SANDWICH LAMINATES There is no trick in modeling a straight‐up sandwich but how you handle the transition can be quite interesting.

3One

3.5 WHe of the joy

HEN IN DOys of simul

FEM

OUBT BUILlation is th


LD SOME at it can be


SIMPLE Fe qualitativ


FEA MODEve rather t

016

ELS ‐ ░ than subjec

ctive.

of 30



4. IDEALIZATION OF THE STRUCTURE INTO A COMPOSITE MODEL

4

4.1 STR

RUCTURED

FEM

D FEA MO


ODEL (CLE


AN MESH


H = CLEAN

016

N STRESSE

S) ‐ ☼

of 30



4.2 3D STRUCTURE IDEALIZATION (SANDWICH AS LAMINATE + CORE) ‐ ☼ Idealization around transitions is not direct but a blend of exploratory models, inspection and “crossed‐fingers”

4Givloadcouawahav

4.3 BOen that onded boltedurse we areay…the bigve a FEA m

LTED CONnce a bolt gd connectioe talking abgger the beodel to pu

FEM

NNECTIONgoes into shons and if wbout solid etter. Thent numbers


NS hear, failurwe have dolaminates.n what abos to your fe


re can’t beoubts abou. What aboout creep, ears.


e fall behinut slippageout sandwiit is “plast

016

d, we strive we recomch composic”? Well,

ve to preloammend inssites? Insethat is wh

of 30

ad highly erts. Of rt hy you



5. VERIFICATION (GOTTA TEST) If you can’t test your laminates, one often is required to use a high knock down factor, say 3x. Although the ASTM test procedures are clear about Four‐Point span to thickness ratios, not all laboratories are up‐to‐date on reading the specifications



Understanding test data sometimes requires a bit sleuthing. At first we couldn’t figure out why our nice hand calc’s weren’t matching the test data.

Four‐Point Bend Failure of Sandwich Composite



To get it right, required modeling more stuff than we liked too. At the end of this effort after fiddling around with 2D shell laminate model, we took the plunge and went full 3D skin laminate with hex element solid core. The correlation was above average (>10%).

Foam Material Model The Joy of Validation



5.1 SOMETIMES ONE HAS TO THINK LIKE THE TEST ‐ ☼ ░ It was all about localized deformation around the anvils and one might say that the test didn’t really measure the strength of the sandwich but how testing can lead to wrong answers.

6. VAlthvalidec

VALIDATIOhough it widation cancompressio

ON would be nin be assureon event.

FEM

ce to say ted, reality i


hat througis often a h


gh expertizharsh task


ze and caremaster. O

016

eful modelur critical t

ing procedtest load w

of 30

dures, was a rapid

6

▫

▫

▫

▫

▫

6.1 TES

First mod

Manufac

Mechani

Lessons L

A Model

ST DATA C

del had iss

cturing Dis

cal Closur

Learned

is a Mode

FEM

AN BE DIF

sues

scoveries

res

el


FFICULT



of 30



6.2 THEN YOU HAVE REAL VALIDATION ‐ ░



7. COMPOSITE SIMULATION CHECKLIST o Do your hand‐calcs and think about fiber‐volume‐fraction alignment to load o Understand that it gets knocked down to handle defects and small holes that a designer might miss – investigate and question the knock down factors

o Make your first model isotropic and work through the load lines and for a sandwich structure, don’t forget you can still use a plate model with some tricks (see Composite Handbook)

o Check your manufacturer’s data and build a FEA test coupon of their material o If the project has value, get independent laboratory testing of the samples o Build your FEA model with simplicity and minimum plies (i.e., merge plies of identical makeup)

o Socialized and talk to other engineers that may have done similar projects o Realize that if you are within 15% of validation that you are a super star



8. REFERENCES FOR FEMAP USERS See Applied CAx for tutorials and reference material

FEM




of 30

Basics of Composite Modeling for Engineers

Documents

Transcript of Basics of Composite Modeling for Engineers