Journal Club Theme of December 2009: Impact Behaviour of Materials with Cellular Structures

Henry Tan's picture

Welcome to the forum! Discussion topics were suggested initially as follow:

Metal foams, cell deformation (bending, buckling, plasticity and fracture), constitutive stress-strain behaviour of cellular materials, energy absorption, hypervelocity impact, shock wave behaviour, 1D shock modelling, shock attenuation, shock enhancement, Material Point Method (MPM) simulation and microscopic tomography experimental observation;

and later were extended to many related:

Natural cellular materials, sunflower stem, negative Poisson’s ratio, nanoscale deformation in nacre, mesh-free methods, eXtended Element Free Method (XEFG), eXtended Finite Element Method (XFEM), simulating discontinuities, hyperelastic model, poroelasticity, loading rate effect, fluid-structure interaction, structural optimization, iMorph software, foam geometry, pattern formation, Kelvin problem, quasicrystal, trabecular bone, gyroid, bulk metallic glass foam, peridynamics.

=================== initial topics ==========================

1. Cellular materials, such as metal foams, are used as impact energy absorbers in crash and blast protection due to their unique constitutive behaviour. Three stages can be identified for the stress-strain curves of the uniaxial compression of metal foams: Stage I: The deformation is, in general, reversible. For closed-cell metal foam this is in the form of bending of the cell walls and edges. At the end of this stage some cells suffer collapse. This may be due to elastic buckling, plastic deformation or fracture. Stage II: The almost constant compressive stress, plateau stress, appears in a wide range of strain. Buckling and plastic collapse occur successively until all cells are collapsed. The deformation in this stage is unrecoverable. Stage III: Cell walls and edges contact each other and are crushed; giving rise to a steeply rising stress.

2. Energy absorbers for crush and blast protection are chosen so that the plateau stress is just below the stress that will cause damage to the packaged object; the best choice is then the one which has the longest plateau, and therefore absorbs the most energy.

3. Impact velocities can vary from a few meters per second to some tens of kilometers per second (hypervelocity). Hypervelocity is relevant to the field of space exploration such as the impact by space debris. Using of metal foams in this field is still under investigation.

4. Analytical solutions of shock waves in cellular materials. For 1D analysis, Reid and Peng (1997) firstly treated cellular materials subject to uniaxial compression using a simplified rigid, perfectly-plastic, locking (RPPL) model. In the RPPL model, the constitutive behaviour in stage-I is simplified as rigid; stage-II is treated as perfect plastic at the yielding plateau stress, and the second stage ends at the locking (densification) strain; stage-III is again idealized as rigid. Radford et al. (2005) used RPPL model to study the shock behaviour in a metal foam projectiles. Harrigan et al. (2009) compared RPPL with other analytical approaches on modelling shock behaviours.

5. Shock waves propagating in cellular materials will, in general, be attenuated by cell collapse at low impact speed. However, shock enhancement will occur during high speed impact; this will affect the structure design. The group led by Professor Han Zhao at the Universit´e Paris VI investigated the shock enhancement both numerically and experimentally.

6. Simulating impact behaviour of cellular materials using the Material Point Method (MPM). MPM was adopted by the U.S. Department of Energy’s ASCI (Accelerated Strategic Computing Initiative) Center for the Simulation of Accidental Fires and Explosions in simulating high speed impact on plastic bonded explosives. The group led by Prof. Hongbing Lu at the Oklahoma State University worked on MPM for many years; his group recently studied the cell-wall buckling, shear-band formation and collapse-wave propagation using MPM simulations (Daphalapurkar et al., 2008).  The microstructure of the foam was determined using μ-CT and was converted to material points. The properties of the cell-walls were determined from nanoindentation on the wall of the foam. Features of the microstructures from simulations were compared qualitatively with the in-situ observations of the foam under compression using μ-CT.

Daphalapurkar, N.P.,  Hanan, J.C., Phelps, N.B., Bale, H., Lu, H., 2008. Tomography and simulation of microstructure evolution of a closed-cell polymer foam in compression. Mech. Adv. Mater. Struct. 15, 594-611.

Harrigan, J.J., Reid, S.R., Yaghoubi, A.S., in press. The correct analysis of shocks in a cellular material. Int. J. Impact Eng..

Pattofatto, S., Elnasri, I., Zhao, H., Tsitsiris, H., Hild, F., Girard, Y., 2007. Shock enhancement of cellular structures under impact loading: Part II analysis . J. Mech. Phys. Solids 55, 2672–2686.

Radford, D.D., Deshpande, V.S., Fleck, N.A., 2005. The use of metal foam projectiles to simulate shock loading . Int. J. Impact Eng. 31, 1152–1171.

Reid, S.R., Peng, C., 1997. Dynamic uniaxial crushing of wood. Int. J. Impact Eng. 19, 531-570.


Henry Tan's picture

reference books

Some reference books that I have read:

J. D. Achenbach. Wave propagation in elastic solids, North Holland, 1973.

L. J. Gibson and M.F. Ashby. Cellular solids: structure and properties, 2nd edn. , Cambridge University Press, 1997.

M.F. Ashby, A.G. Evans, N.A. Fleck, L.J. Gibson, J.W. Hutchinson and H.N.G. Wadley. Metal foams: a design guide , Butterworth-Heinemann, 2000.

H.P. Degischer and B. Kriszt (eds). Handbook of cellular metals: production, processing, applications , Wiley-VCH, 2002.

S. Li and W.K. Liu, Meshfree particle methods, Springer, 2007.

C.R. Ethier and C. Simmons. Introductory biomechanics: from cells to organisms, Cambridge University Press, 2007

H. Zhao and N.A. Fleck (eds). IUTAM symposium on mechanical properties of cellular materials , Springer Science, 2007.

S.J. Hiermaier, Structures under crash and impact: continuum mechanics, discretization and experimental characterization, Springer, 2008


likask's picture

question about geometric non-linearities

Hello

I apologize for my ignorance in that matter.

 
Could someone write how important is analysis involving large strains at cell (micro) level of observation. Do strains in cell walls are large? It is clear to me that geometrical non-linearities like large rotations are important in order to describe buckling and post buckling response.  

Answer to my question probably depend how thick are cell walls. But how important is material, what is in case of metal foams? Can we neglect large strains and other geometric nonlinearities in cell walls made from brittle material? 

Kind regards,
Lukasz


Henry Tan's picture

Cell walls are relatively thin compared with edges

Dear Lukasz,

Cell walls are relatively thin compared with edges in a closed-cell foam. For those foams formed from a liquid phase, the cell formation is determined by surface tension which concentrates material into the edges. Thin walls lead to larger bending strains, and lower critical compression for the occurrence of buckling.

An interesting paper on the average cell wall thicknesses and cell edge lengthes for the closed-cell foams (ALPORAS) of high, medium and low relative densities:
T. Mukaia, T. Miyoshi, S. Nakano, H. Somekawa and K. Higashi, 2006.Compressive response of a closed-cell aluminum foam at high strain rate. Scripta Materialia 54, 533-537.

There are competitions and coupling among buckling, plastic deformation and fracture of the cell walls, which depend on the material properties and geometry (such as thickness). It is interesting to simulate such complex behavious of cellular materials under impact.


Dear Dr.Tan, Recently,

Dear Dr.Tan,

Recently, one group (ICACS) found negative Poisson ratio in sunflower stem (not published now). The sunflower stem has a honeycomb structure, but is more diffcult than the common structure we know, i.e., hexagon in the center and stretched hexagon in the periphery. The special structure results into anisotrophic material,radical modulus is larger than tangential modulus,thus it is understood eaily by mechanisians.


Henry Tan's picture

negative Poisson’s ratio

Dear Kongdong,

I am very interest in this work. Compared with the conventional foam material, the impact energy absorption can be very high for foam material with negative Poisson’s ratio.

Regards,
Henry.

note: ICCAS stands for Institute of Chemistry, Chinese Academy of Sciences 

It seems to me that negative Poisson's ratio comes from two reasons:
(1) microstructure, hexagon in the center and stretched hexagon in the periphery; and
(2) anisotropic properites, radical modulus is larger than tangential modulus.


origin of energy absorption AND negative Poisson ratio


Dear Dr.Tan,

I am new to the impact of honeycomb structures, but interested in this phenomenon. As my understanding, the energy absorption is due to the large deformation or structural collapse which absorbs impact energy. For the classical negative Poisson ratio, in addtition to structural collapse, energy is needed to conquer energy barrier between different conformations, is it this reason for "the impact energy absorption can be very high for foam material with negative Poisson’s ratio"?

However, if one material has negative Poisson ratio, but no conformation transition as the classical negative Poisson ratio material, its energy absorption is restricted to "structural collapse". These days, I calculated mechaical response of sunflower stem structure for that group, and testified the negative Poisson ratio in that material.


Henry Tan's picture

structural collapse

Dear Kongdong,

I am very interested to see how you calculate the “structural collapse” for the sunflower stem structure.

Henry.

Note: What does the "structural collapse" includes? honeycomb rotation, elastic - plastic transition, buckling, fracture, or more?


Henry Tan's picture

biologically inspired cellular materials

Dear Kongdong,

Thanks for the article you email to me. It is very helpful.

The topic on biologically inspired cellular materials for impact protection can be very interesting for discussion in this forum.

Henry.


R. Gabbrielli's picture

Trabecular bone

Here is a sample of a biologically inspired cellular material. The sample has been built using a rapid prototyping equipment. Porosity varies from 0 to 100% throughout the whole sample (0 on the top, fully dense material at the base).

http://people.bath.ac.uk/rg247/swansea/gyroidal_bone.html

-- Ruggero Gabbrielli


Henry Tan's picture

parameters

Very interesting work. Ruggero.

What are the parameters you used for prototyping those foams?


R. Gabbrielli's picture

Nodal surfaces

The surface is an offset of the approximation of the minimal surface called gyroid. For the modelling I used triply periodic trigonometric implicit functions, also called nodal surfaces. Parameters can be chosen as desired, affecting primarily volume fraction and shape factor. The latter needs to be introduced when very low volume fractions are required because pinch-off points are reached before the volume fraction goes down to zero when the first parameter is varied. Here is a 3D model and its mathematical expression. Porosity is 90%.

http://people.bath.ac.uk/rg247/swansea/javaview/gyroid_10percent.html

The parameter is only one in this case, k. Click on the 0.01 volume fraction version to see the 2-parameter function I used.

A brief explanation is contained in this presentation, slides 7-14 (the pdf version unfortunately does not contain the animations):

http://people.bath.ac.uk/rg247/bioceramics20.pps

http://people.bath.ac.uk/rg247/bioceramics20.pdf

For details and references read this article:

http://linkinghub.elsevier.com/retrieve/pii/S0009261400014184

-- Ruggero Gabbrielli


Henry Tan's picture

physical realization of the mathematical foam structures

Dear Ruggero,

Thank you for the information and nice websites!

You mentioned the physical realization of the mathematical foam structures using the Selective Laser Sintering. Can you elaborate the details or point to a reference? Thanks!

Regards,
Henry.


R. Gabbrielli's picture

Details and references

The reference for this specific work can be found here:

http://people.bath.ac.uk/rg247/swansea/development.pdf 

Solid Free Forming techniques are various and many of them can be used to build samples with complex topology, such as the gyroid. In particular, the Selective Laser Sintering process, which builds part layer by layer out of a polymer powder, is able to reproduce overhangs and any feature otherwise impossible on a CNC machine. The first work I am aware of in the bio area is the following:

http://www.springerlink.com/content/0xjmr8xka2wb168j/

Resolution today can go as high as 50 micron. With a different process, called Stereolithography (SLA), which uses a curing resing under a UV light, resolution can be even better, about 10 micron. For cheaper but coarser models, a method called 3D printing is used, where polymer particles are stuck together with a binder. The videos below explain how these techniques work:

selective laser sintering: http://www.youtube.com/watch?v=gLxve3ZOmvc 

stereolithography: http://www.youtube.com/watch?v=eT-OIz-Jt3w

3D printing: http://www.youtube.com/watch?v=R-JOJ91p9Wc

There is no need for support structures as mentioned in the video. Now SLA machines project light from the bottom and the part lies on a glass vat.

Regarding the geometry, if one wishes to check the original references then I would recommend to take a look at the work of the people who first discovered the surfaces as mathematical entities: Schwarz (P and D surfaces) and Schoen (G surface).

1. H.A.Schwarz. Gesammelte Mathematische Abhandlungen. Springer-Verlag, Berlin, 1890.

which is obviously in German. The original document can be found here:

http://openlibrary.org/b/OL23301458M/Gesammelte_mathematische_abhandlungen.

2. A.H.Schoen. Infinite periodic minimal-surfaces without self-intersections. NASA Technical Note, D-5541:1-70, 1970.

which can be downloaded directly from the archive at NASA.

http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19700020472_1970020472.pdf

-- Ruggero Gabbrielli


Henry Tan's picture

Infinite periodic minimal surfaces

The shapes taken by soap films are minimal surfaces, very interesting.

Infinite periodic minimal surfaces can be used to describe materials with complex cellular structures.


Henry Tan's picture

Inverse problem

Dear Ruggero,

Given a foam structure, is it possible (then how) to find a mathematical formula to approximate the surfaces?

Regards,
Henry.


R. Gabbrielli's picture

Which structure?

Dear Henry,

what is a foam structure? This post was about the gyroid, which little has to do with foams (at least, topologically). If you mean a real foam then the answer is maybe yes, but the effort required for such an operation probably overcomes the actual benefit.

Mathematical formulae can be used for many applications and they could potentially approximate every geometrical system we could think of (with Fourier series, for example). All this comes at a cost for the user and for the specific applications it is intended for. I am not aware of any theory used for this scope at the moment, but I wouldn't be surprised if anybody looked at such a thing and developed a theory to deal with them. Please let me know if you know of any.

-- Ruggero Gabbrielli


Henry Tan's picture

PhD thesis

Dear Ruggero,

I think I am getting a better understanding of your PhD thesis now. So you have two parts, one is about foam geometry (Kelvin problem), the other is on porous materials (gyroid).

Regards,
Henry.


R. Gabbrielli's picture

origin of energy absorption AND negative Poisson ratio

Dear kongdong,

the energy absorption in negative Poisson ratio materials seems to be due to the rotation of their internal components. As shown by Bertoldi, Reis et alt., simple squared honeycomb structures can easily give a negative Poisson ratio.

https://bertoldi.seas.harvard.edu/publications/2009/Adv-Mat-2009.pdf

Is this the structural collapse you are referring to? The animation below shows how 2D periodic materials undergo to a process in which their internal components rotate, this way producing high negative Poisson ratios.

http://people.bath.ac.uk/rg247/swansea/javaview/auxetic.html

-- Ruggero Gabbrielli


Henry Tan's picture

auxeticity

Dear Ruggero,

Thanks a lot for the Java demonstrations. The mathematical calculations for the negative Poisson’s ratios of square and hexagonal lattices are quite interesting.

The Java animations show quite similar pictures in the paper by Xiaodong Li et al., 2006. He posted his papers in this forum during the 1st week. His observations under the atomic force microscope showed that nano-grain rotation and deformation, which cause the negative Poisson’s ratio, are the major mechanisms contributing to energy dissipation in nacre.

Using a micromechanics model, such as the Mori-Tanaka method, your mathematical calculations can be incorporated into Xiaodong’s nanoscale experimental observations, and thus give quantitative descriptions of the auxetic behaviour and energy dissipation of nacre. We can develop a quite nice journal article together with Xiaodong, I think.

Regards,
Henry.


R. Gabbrielli's picture

auxetics

Henry,

this sounds like a brilliant idea. I am reading the papers you suggested.

-- Ruggero Gabbrielli


Henry Tan's picture

modelling

Ruggero,

I am considering the following:

(1) To simplify the modelling, let's study a 2D deformation under uniaxial tension. 

(2) For the first step approximation we can treat the grains as rigid, and thus focus more on the grain-rotation.

(3) Then the problem is very simply, it is a 2D web of linked hexagonal stones (suppose we treat each nanograin as a rigid hexagonal stone) that subjects to uniaxial tension!

(4) If assume the uniform distribution (honeycomb lattice) of stones, the Poisson's ratio can be calculated directly.

(5) For the above calculation, the major difficulty is in modelling the cohesion between the hexogonal stones. We can assume it to be linear in the beginning. Later, when we deal with the energy absorbing and failure of the nacre, we consider a nonlinear cohesion law between nanograins.

(6) Further, we consider a non-uniform (random) distribution of hexagonal stones.

(7) For the 2nd step approximation, we may further consider the deformation of the nanograins. There are several ways to model the problem, either through numerical simulation, or analytical solution using Mori-Tanaka method. I think Mori-Tanaka method is especially suitable for this problem of a material with many grains.


R. Gabbrielli's picture

modelling

I'm not sure I understand what the system is exactly. Is it a tessellation of rigid hexagons? If it is, how can this system show a negative Poisson ratio?

-- Ruggero Gabbrielli


Henry Tan's picture

initial shapes should be irregular

You are right! If the system is initially tessellated with hexagonal stones, there will be no fattening after tension (in certain directions).

Therefore, to model the auxetic behaviour, the initial shapes of the stones should be irregular.

Do I interpret correctly?


R. Gabbrielli's picture

Shape regularity

Shape regularity does not affect global material behaviour directly. Auxetic behaviour can be modelled using irregular or regular grains. It's the way they pack that determines their auxetic behaviour.

-- Ruggero Gabbrielli


Henry Tan's picture

definition of the problem

Therefore, the problem for solving can be defined as:

For certain shapes and packing of 2D rigid grains, which subject to uniaxial tension/compression, find the deformation in the transverse direction.

Since grains are assumed to be rigid, the deformation comes from the debonding between the grains at the grain interfaces.


Henry Tan's picture

a micromechanical model for nacre

Henry Tan's picture

did not show negative Poisson’s ratio for nacre

It seems that the model in this paper by Bertoldi et al. (Compos. Sci. Techno., 2008) did not show negative Poisson’s ratio for nacre?


Henry Tan's picture

Unidirectional auxetic behaviour?

Dear Ruggero,

The system studied in the paper by K. Bertoldi et al. (Adv. Mater. 2009, Vol. 21, 1–6) comprises a square lattice of circular holes in an elastomeric matrix which was subjected to uniaxial compression. The system shows unidirectional (compression) negative Poisson’s ratio behaviour.

The nacre studied by X. Li et al. (Nano Lett., 2006, Vol. 6, 2301-2304) shows the negative Poisson’s ratio when the nacre is under tension. Do you think nacre also show negative Poisson’s ratio when subjected to compression?

Regards,
Henry.

 


R. Gabbrielli's picture

directionality

It all depends on the structure of nacre. If the grains are already completely packed, it would be difficult to have a negative Poisson's ratio in compression. From the images in the paper by X. Li et al. (2006) it seems that this is the case. The article by K. Bertoldi also shows positive Poisson's ratio for nacre. A recent work by X. Li and Z. Huang (2009) that has been already posted here shows an interesting mechanism adopted by nature to form the structure of nacre. It seems to me that screw dislocation has something to do with aperiodic arrangement.

Regarding the directionality, it is possible to create a material showing auxetic behaviour in both the loading directions, tension and compression. Simply pierce a thin film with a square lattice of alternatively oriented ellipses.

-- Ruggero Gabbrielli


Henry Tan's picture

piercing a thin film

Why piercing a thin film with a square lattice of alternatively oriented ellipses can create negative Poisson’s ration for both uniaxial tension and compression?


R. Gabbrielli's picture

Alternatively orthogonal ellipses

Because they are very similar to the idealized model for auxetic material I have shown previously. I would invite you to take a closer look at it, if you haven't already. Sometimes images explain a lot better - and quicker - a mechanism of action than a thousand words can do.

http://people.bath.ac.uk/rg247/swansea/javaview/auxetic.html

This time consider the configration half way through the animation as the initial, unloaded state. It is not hard to see that holes are alternatively orthogonal rhombi. If you round their corners you can get the ellipses I mentioned. Now imagine applying uniaxial tension to the specimen. It doesn't matter if this is done through x or y, however it has to be a lattice direction.

Then do the same with a uniaxial compression. In either cases the material has negative Poisson's ratio.

-- Ruggero Gabbrielli


Henry Tan's picture

positive moduli

In your idealized model of a system of thin, rigidly connected squares under uniaxial in-plane loading, the moduli in x and y directions are all zeros. How to modify the model in order to get positive moduli?


R. Gabbrielli's picture

Elastic moduli

Why are the elastic moduli zero? If the squares are rigidly connected the stress is nonzero where they touch each other. Also the strain is not zero, as it can be easily seen by the size variation of the unit cell.

Or maybe you are considering the model a mechanism of hinged rigid squares? In this case simply add a torsional spring at each hinge to get a positive modulus.

--
Ruggero Gabbrielli


Henry Tan's picture

critical load

If so, there will be a critical load Pc.

When the applied load is lower than Pc, the Poisson’s ratio is positive;

When the applied load is higher than Pc, the Poisson’s ratio is negative.


R. Gabbrielli's picture

Critical load?

Why? Poisson's ratio is always negative in this model, as long as the load is applied in one of the lattice (in this case also principal) directions. The rotation of the squares causes the specimen to laterally shrink when subject to compression and to laterally expand when subject to tension.

--
Ruggero Gabbrielli


Henry Tan's picture

I was wrong

Ruggero, you are right, the Poisson's ratio is always negative in this model. I was thinking about a wrong model which does not apply to this case.


Henry Tan's picture

quasi-crystallization and periodic/aperiodic tiling

Dear Ruggero,

Regarding the paper by X. Li and Z. Huang (2009) which shows screw dislocation and amorphous aggregation as two dominant mechanisms during nacre’s biomineralization process, don't you think these two mechanisms can be related to something of quasi-crystallization, and periodic/aperiodic tiling?

Regards,
Henry.


R. Gabbrielli's picture

order in nacre

Yes, it could surely be. Unfortunately it is not easy to say. Do you have additional references about aperiodic order in this (and other) natural materials?

-- Ruggero Gabbrielli


Henry Tan's picture

Nacre has a unique structure for energy absorbing

Very interesting research and nice results!

Nacre has a unique structure for energy absorbing. I will look at this in details.

Any thoughts on the nano-grain rotation when the loading is dynamic under high speed impact, such as the case of struck by an object at speed of hundreds or thousands of meters per second?

Just asking for fun.


Henry Tan's picture

Nature has its own way for impact protection

Nature has its way for impact protection. Snail shells have a unique hard-soft-hard sandwich layered structure, as Dr. Haimin Yao , a MIT postdoc reported.

"The outer hard layer contains small, grain-like particles. When under attack, these granules help to dispel the energy of the blow, spreading it out across the outer region. Any fractures that occur will disperse along jagged lines guided by the granules, forming fissures in the top layer." -- An article on Fox News.

 

snail-shellThe image is from http://seastarboats.us/, a website for sea boating.


Henry Tan's picture

CISM lecture note

I delivered lectures on “Impact of Cellular Materials” for a one-week course (28 Sept - 2 Oct, 2009) “Cellular and Porous Materials in Structures and Processes ” at the International Centre for Mechanical Sciences (CISM ), Udine, Italy. Attached is the lecture note with Professor Shaoxing Qu at the Zhejiang University, written when I was in Manchester and during the summer visit in Hangzhou, China.


Henry Tan's picture

synopsis of book contents

Invited by Professor Andreas Öchsner, Editor-in-chief of the Springer book series on "Advanced Structured Materials", I am writing the book “Impact Behaviour of Materials with Cellular Structures”. Below is a synopsis of the book contents:

1. Cellular materials for impact protection
1.1 Natural and man-made cellular structures
1.2 Kelvin problem
1.3 Constitutive models
1.4 Energy absorbing
1.5 Auxetics

2. Impact dynamics
2.1 Wave dynamics
2.2 Plastic shock waves
2.3 Poroelasticity
2.4 Hypervelocity impact

3. Simulating impact behaviour of cellular materials
3.1 Material Point Method 
   3.1.1 MPM for simulating solids
   3.1.2 MPM for simulating fluids
   3.1.3 MPM for simulating cellular materials 
3.2 Peridynamics
3.3 Smoothed particle hydrodynamics
3.4 Molecular dynamics for simulating hypervelocity impact

4. Behaviour of shock waves in metal foams
4.1 Structures of shock front
4.2 Cell collapse and shock attenuation
4.3 Shock enhancement
4.4 Shock arrest
4.5 Propagating instabilities

References:

CellCell membrane Foam Weaire–Phelan structureHoneycomb, Hyperelastic material, Auxetics 

Impact, Wave, Poromechanics, Hypervelocity

Material Point Method ,  Smoothed particle hydrodynamicsMolecular dynamics, Peridynamics

Shock wave


Henry Tan's picture

Material Point Method and Smoothed Particle Method

Chapter 3 is on simulation of hypervelocity impact of cellular materials. Both Material Point Method and Smoothed Particle Method will be introduced and discussed in step by step details.

A very interesting paper on comparison of MPM with the SPH is by Shang Ma, Professor Xiong Zhang and Professor Xinming Qiu at the Tsinghua University. The title of the paper is “Comparison study of MPM and SPH in modelling hypervelocity impact problems” published on the International Journal of Impact Engineering, 2009, 272-282.


ndaphalapurkar's picture

choice of numerical method

Henry: Thanks for providing link to this paper! Quite excited to hear the conclusion from their results!


Henry Tan's picture

simulating discontinuities using mesh-free methods

Any comments on simulating discontinuities using mesh-free methods?

 

Marginal note:

The future of meshless methods

discontinuities in mesh free methods


CanhLe's picture

XEFG

 Dear Dr Henry,

I have not worked on enrich EFG (XEFG) , but I think the method work well for discontinuities . The idea I think is similar to eXtended Finite Element Method (XFEM) where enrich functions are add to basis function to introduce discontinuities or to improve stress field at crack tip!

 

Regards,

 

Canh Le

 

 


Henry Tan's picture

weak form

Fundamentally, Galerkin method and finite element method are all based on converting a set of continuous differential equations into a weak form.

If the problem is discontinuous itself (like for dynamic cracking, fragmentation, shock wave, etc.), i.e., the differentiation does not exist at the discontinuous surface,  there is no such weak form.

Then what all those methods are trying to do is to find ways to fix this no weak form problem.


Henry Tan's picture

discontinuity from a shock front

Dear Canh,

What do you think if the discontinuity is from a shock front that separates the wave behaviour before and after? Will the eXtend Element Free Galerkin or the eXtended Finite Element Method work well for this problem? What kind of enrich functions will be needed?

Regards,
Henry.


CanhLe's picture

Strong discontinuities!

Dear Henry,

For these strong discontinuities, I think the XEFG still works well. You may find useful information in the following papers:

1. T. Rabczuk and G. Zi, A meshfree method based on the local partition of unity for cohesive cracks , Computational Mechanics 39 (2006), pp. 743–760

2. Development of discrete cracks in concrete loaded by shock waves

Level set techniques is often used to trace the crack paths, see in [1] above!

Regards,

Canh


Henry Tan's picture

hyperbolicity condition

Dear Canh,

Many thanks for the paper. Right now I need to check what the hyperbolicity condition (as the criterion for crack initiation and propagation used in those two papers) is. I guess I may also need a criterion for the initiation and propagation of the shock front.


Henry Tan's picture

discontinuities in mesh free methods

Just noticed a very nice blog on discontinuities in mesh free methods:

http://imechanica.org/node/742 

However, there is no discussion on the discontinuities at the shock wave front.

 

Marginal note

Level set method


Henry Tan's picture

moving meshless methods

Moving meshless methods are numerical methods for unsteady partial differential equations that have shock, high gradient region or high oscillatory region.

I am simulating shock wave generated by high speed impact. Any comments on moving meshless methods?


Henry Tan's picture

Peridynamics

Peridynamics may be useful in simulating the discontinuities at the shock wave front.

In Stewart.A.Silling‘s J-Club forum and other Blogs, it seems to me that most of the example applications of the peridynamics are for the discontinuities in cracks, and some for phase boundaries. No example for discontinuities in shock waves. 


Erkan Oterkus's picture

Peridynamics for shock wave front.

Hello Henry,

Although I am not familiar with shock wave dynamics, these two references might be useful:

1) Viscoplasticity using peridynamics

J. T. Foster , S. A. Silling , W. W. Chen 

 International Journal for Numerical Methods in Engineering, Early View

 2) An approach to modeling extreme loading of structures using peridynamics

Paul N. Demmie & Stewart A. Silling,

Journal of Mechanics of Materials and Structures, Vol. 2 (2007), No. 10, 1921-1945

Regards,

 Erkan.


Henry Tan's picture

interactions between material points in peridynamic theory

Dear Erkan,

Thanks for the information. Just a general question about peridynamic theory:

Considering a linear elastic body with a surface S that contains discontinuities in displacement, how will the material points P1 and P2 interact with each other when they reside
(1)    in the same side of the surface S, or
(2)    in the opposite side of the surface S?

Regards,
Henry.


Erkan Oterkus's picture

Re:interactions between material points in peridynamic theory

Dear Henry,

In peridynamic theory, all material points interact with each other if there is no discontinuity in the body, such as a crack. As in your question, let's assume that there is a surface S in the body which contains a crack as a discontinuity. So,

(1) The material points which are on the same side of the surface S will not be affected from the discontinuity, so they'll continue to interact with each other.

(2) For the material points which are on the opposite sides of the the surface S, depending on the location of the material points with respect to the crack, they may be affected. If the interaction between material points P1 and P2 crosses the crack surface, then the interaction between these material points disappears. So, they'll no longer interact with each other.  If this is not the case, they'll continue to interact with each other although they are on the opposite sides of the surface S.

 Regards,

 Erkan. 


Henry Tan's picture

see each other

Assume that S is the discontinuity surface. Do you mean that if P1 can see P2 (not blocked by S) then the interaction between P1 and P2 is nonzero?  There is no interaction between P1 and P2 if they cannot see each other?


Erkan Oterkus's picture

Re: see each other

Dear Henry,

 Yes, what you said is correct. I tried to upload a descriptive image of this process, but it didn't work. Is it OK if I e-mail the image to you so you can upload it here?

 Regards,

 Erkan.


Henry Tan's picture

Thanks, Erkan. Please

Thanks, Erkan.

Please email me the image, and let me find a way to upload it.


Henry Tan's picture

 Erkan, This is your image

Erkan, This is your image. Thanks!

Please describe it.

pd_interaction.jpg


Erkan Oterkus's picture

Interaction of Material Points If There is a Crack in PD Theory

Henry,

Thank you very much for uploading the image.

Peridynamics is a non-local version of continuum mechanics. As we all know that in continuum mechanics, the body is composed of infinite number of material points.In the figure above, just for illustration purposes, only four of the all material points in the structure are shown explicitly. These material points are named as P1, P2, P3 and P4. As I mentioned before, according to peridynamic theory, all material points are interacting with each other. These interactions are called as "bonds". However, if there is a discontinuity in the structure, such as a crack, some of these interactions will disappear due to the discontinuity. As you can see in the figure, bonds that cross the crack surface S, i.e. red dashed lines between material points P1 and P2, and between material points P3 and P4, are broken which means that there is no longer interaction between these material points due to the presence of the crack. On the other hand, for the bonds which do not cross the crack surface will not be affected from the crack surface and these interactions are shown with green dashed lines in the figure.  

Regards,

Erkan.


Henry Tan's picture

need to keep a record of the discontinuity surface?

Does this mean that during the peridynamic simulation, you need to keep a geometrical record of the discontinuity surface S for its change of location, shape, orientation and etc?


Erkan Oterkus's picture

Re: need to keep a record of the discontinuity surface?

Hello Henry,

No, we do not need to keep the record of the discontinuity surface. Instead, we need to keep the record of the status of the interactions between material points. So, if a bond is broken between two material points, then these two material points are not going to interact with each other again in the future. The definition of the disconitunity surfaces can be done at the beginning of the analysis and based on these definitions, the affected bonds are broken.

Regards,

Erkan.


Henry Tan's picture

bonding condition may change

Erkan, 

How about the peridynamic simulation for the collision of two bodies (A and B), then? Consider body A and B that are separated in the beginning; they move towards each other and make a collision.

Initially, material points in body A will have no interaction with material points in body B. However, during the collision they make contact and there should be interaction between them.

For collision simulation, how are you going to modify the algorithm?

Similar case may be in the contact of the cracked surfaces. Are you going to modify the algorithm: "if a bond is broken between two material points, then these two material points are not going to interact with each other again in the future?"


Erkan Oterkus's picture

Re: bonding condition may change

Dear Henry,

If we bring two bodies closer to each other, they still cannot interact with each other through peridynamic bonds, even if they make a contact .Because peridynamic bonds corresponds to internal forces, not external forces that one body is exerting on another body. In other words, material points in body A can only interact with material points in body A via peridynamic bonds. They cannot interact with material points in body B with peridynamic bonds. But, a material point in body A can interact with a material point in body B by using short range forces.

In peridynamic theory, for collision or contact problems, shortrange forces are utilized in order to prevent unphysical penetration of bodies within each other. So, at a particular time, if two material points are close to each other less than some specified distance, the shortrange forces start to activate between material points. And the task of these shortrange forces is pushing two material points away from each other.

Short range forces are also valid for material points which belongs to the same body.

For more information about short-range forces please see the 3. section of "R. W. Macek and S. A. Silling, "Peridynamics via finite element analysis,"
Finite Elements in Analysis and Design, Vol. 43, Issue 15, (2007) 1169-1178. DOI: 10.1016/j.finel.2007.08.012  "

Regards,

Erkan.

 


Henry Tan's picture

different bodies

Erkan,

When writing a code, things can be quite complicated.

You said that: "when two material points P1 abd P2 are close to each other less than some specified distance d, the short range forces start to activate between those two material points."

However, when P1 and P2 are adjacent two points within a continum body, there interaction can be bond rather than short range separating force.

The code needs to make difference of the two cases.


Erkan Oterkus's picture

Re: different bodies

Henry,

Writing a peridynamic code is not very difficult, but understanding theory might take some time. Peridynamic theory is a continuum theory, so we cannot allow any material points to share the same location because this violates the continuum concept. That's why we need those short range forces between material points even if they are neighbors. These short range forces are in addition to bond forces and activates if the material points are very close to each other.

Regards,

Erkan. 

P.S For an introduction for Peridynamic theory , I recommend these two papers:

S. A. Silling and E. Askari, "A Meshfree Method Based on the Peridynamic Model
of Solid Mechanics," Computers and Structures, Vol. 83 (2005) 1526-1535. DOI:10.1016/j.compstruc.2004.11.026 

R. W. Macek and S. A. Silling, "Peridynamics via finite element analysis,"
Finite Elements in Analysis and Design, Vol. 43, Issue 15, (2007) 1169-1178. DOI: 10.1016/j.finel.2007.08.012 

 


Henry Tan's picture

pairwise force function

I am reading the papers suggested and is getting a much better understanding of the theory.

To my understanding, the pairwise force function between x' and x in the equation

 

pairwise force

 

should depend not only on the vector difference of the displacement u and position x, but also on
the state, such as strain, at both x' and x.


Erkan Oterkus's picture

Re: pairwise force function

Hello Henry,

Can you explain your statement in more detail: "should depend not only on the vector difference of the displacement u and position x, but also on
the state, such as strain, at both x' and x."

Regards,

Erkan.


Henry Tan's picture

local state

Erkan,

Consider the bond between material points A and B.

The bond between A and B with A locally at elastic state is different from that with A locally at plastic state.

Regards,
Henry.


Erkan Oterkus's picture

Re:local state

Hello Henry,

The formulation in the two references that I suggested is based on the original formulation of peridynamics. Although the original formulation is straightforward and eaiser to understand, it has some limitations. For example, Poisson's ratio of the material has to be 0.25 because of the central interactions between material points. In order to overcome these limitations, a more general form of the peridynamic theory, called "state based peridynamic theory", was introduced in 2007.

S. A. Silling, M. Epton, O. Weckner, J. Xu and E. Askari, "Peridynamic States
and Constitutive Modeling," Journal of Elasticity, Vol. 88 (2007) 151-184. DOI: 10.1007/s10659-007-9125-1 

I think what you are mentioning in your statement can be done within the "state based peridynamics" framework.

Regards,

Erkan.


Henry Tan's picture

damage

Erkan,

Thank you for pointing the reference.

The modified peridynamics in the above paper incorporates the feature of damage through the influence function that exclude damaged bonds. In this approach the bond can be in only two states, either damaged or undamaged, not in between.

However, by physical intuition, debonding between points should be a continuous process from 0 (complete debonding) to 1 (complete bonding).

Nevertheless, the modified peridynamic theory can still describe damage evolution since bonds are not necessarily broken at the same time.

The question is, what is the physical meaning of a single bond between two points A and B? Is it an averaged summation of all the inter-atomic forces between the atom-aggregate localized around A and the atom-aggregate localized around B? If so, the bond should not be just in two states, damaged and undamaged, but changes continuously with time evolution.

Regards,

Henry.


Henry Tan's picture

microelastic material

Equation (7) in the paper,

S. A. Silling and E. Askari, "A Meshfree Method Based on the Peridynamic Model
of Solid Mechanics," Computers and Structures, Vol. 83 (2005) 1526-1535

micropotential

is said to be the definition of a microelastic material. I do not agree with this definition of "elasticity".


Erkan Oterkus's picture

Re: microelastic material

Henry,

For the proof of this expression, please see the Section 4 of the following paper:

Silling, S. A. (2000). "Reformulation of Elasticity Theory for Discontinuities and
Long-Range Forces"
. Journal of the Mechanics and Physics of Solids
48: 175–209. doi:10.1016/S0022-5096(99)00029-0. http://www.ingentaconnect.com/content/els/00225096/2000/00000048/00000001/art00029.
  

This paper is the first peridynamics paper appeared in the literature.

Regards,

Erkan.


Henry Tan's picture

continuum version of molecular dynamics

The peridynamic theory may be thought of as a continuum version of molecular dynamics, and the major difference between molecular dynamics and peridynnamics are:

(1) The inter-particle force in molecular dynamics  is not history dependent; while the
inter-particle force in peridynamics is history dependent: once it is broken it is always broken, as described by the mu function in Silling and Askari 's paper (2005):

 

 mu function

(2) In molecular dynamics, the interparticle force depends on the distance between two particles; in peridynamics, this depends on stretch, which I think is a brilliant idea.


Henry Tan's picture

Debonding is a gradual process

The statement that bond is either broken or not broken seems does not reflect the real situlation. Debonding is a gradual process from complete bonding, through partial debonding, to comple debonding (broken).


Erkan Oterkus's picture

Re: Debonding is a gradual process

Hello Henry,

Also in the peridynamic theory, bonds are not broken at the same time, and it will take some time to create a new crack surface. So, I think, peridynamic theory is reflecting the real situation. Yes, I mentioned in my previous e-mail that we are breaking bonds if it crosses a discontinuity, but this is only valid if we have an initial crack before starting our analysis.

Regards,

Erkan.


Henry Tan's picture

bond parameters

What does the bond depend on?


Erkan Oterkus's picture

Re: bond parameters

Bond force depends on  bond constant (material constant) and the strain of the peridynamic bond (for an isotropic material).

 


Henry Tan's picture

definition of strain in peridynamics

What is the definition of strain in peridynamics?


Erkan Oterkus's picture

Re: definition of strain in peridynamics

Hello Henry,

 I am back again. First, I want to apologize for my long silence, but I was really busy. You asked about the strain in peridynamics. Peridynamic theory is not a classical theory. However, terms like stress and strain are used in the classical continuum theory. So, in peridynamic theory (PD), we don't need to define stress or strain of a material point to obtain the solution of the problem.The important thing in PD theory is the interaction of the material points, that we name as peridynamic bond.

But, if we still want to obtain the strain field, I think we can obtain by an approximation technique using the displacement field.

Regards,

Erkan.

P.S. I'll try to answer your other questions related with the PD theory soon. 


Henry Tan's picture

bond definition in peridynamics

Dear Erkan,

I have doubt about the current peridynamic bond definition. See the deformation of a continuum body shown in my figure below: before deformation, material points A and B have no interaction (bond=0); therefore, after deformation material points A and B should still have no interaction according to your definition. This seems awkward for this case.

bond

Regards,
Henry.


Erkan Oterkus's picture

Re: bond definition in peridynamics

Dear Henry,

This is a good question. As I mentioned before, all material points in the structure should interact with each other as long as there is no discontinuity in the structure. And if there is some type of discontinuity, some of the bonds are permanently broken, so associated material points will no longer interact with each other. At this point, we need to think about why these bonds are broken.  Is the reason geometrical or physical? I am sorry that I was not very clear on this issue in my previous messages. Yes, we break the bonds which cross the discontinuity surface and from this definition, the reason looks like something to do with the geometry. And as you showed in your figures, if we follow this logic, these bonds might interact after the deformation. Actually this is not the case. Because we are breaking the bonds due to physical reasons. When we apply an external load to the structure, some of the energy due to external forces will be converted strain energy. So, when a particular bond deforms from its original form, it will gain some strain energy. But, if the bond is broken for some reason, then this strain energy is used to create crack surfaces. So, we already converted the strain energy of the bond in creating a crack surface and we cannot gain this energy back. That's why when a bond is broken, it is broken permanently.

Regards,

Erkan.

 


Henry Tan's picture

does bond breaking need time?

When crack propagates, the newly fractured surfaces may break the visibility of many material points instantly.

In the peridynamic theory, all those bonds that cross the newly formed crack surface are suddenly broken, no matter how far they are away from. I have a question: does bond breaking need time?


Erkan Oterkus's picture

Re: does bond breaking need time?

Hello Henry,

Sorry for my late response. I think I need to correct yor statement. Peridynamic theory is a continuum theory like the classical continuum theory. In peridynamic theory (PD), we do not need to have a discontinuity in the structure. So, PD works very well even if there is no discontinuity in the structure. Its advantage over the classical continuum theory is that it has an integral form instead of the spatial partial derivatives as in the classical continuum mechanics and it works without any problem if there is a discontinuity in the structure or not. So, if there is a discontinuity in the structure, we do not need to use any special criteria to propagate the crack, find the crack direction, orientation etc. And during the analysis we do not need to check if there is a discontinuity in the structure or not. So, we don't break the bonds which cross the discontinuity in the structure during the analysis, because we do not need to check if there is any discontinuity in the structure. Then, how can bonds be broken during the analysis? If the bond strain exceeds some critical value (which depends on the energy release rate of the material), then we are breaking the bonds. And if there is a crack in the body, because the strain will be higher around the crack tips, broken bonds will initiate around that region and the crack will start to propagate. But as I mentioned before, we do not force any criteria on when or how the crack propagates. Yes, in my previous e-mails I said that we need to break the bonds which cross the crack surface, but this check is done only at the beginning of the analysis.  

Regards,

Erkan.


Henry Tan's picture

verification

"Peridynamics theory works very well even if there is no discontinuity in the structure."

Are there any papers published so far that verify the peridynamics theory with analytical solutions exist for comparison?


Matt Lewis's picture

Deafening Silence

Have there been any responses elsewhere to Henry's post asking for verification problems/papers for peridynamics?

 Matt Lewis
Los Alamos, New Mexico


Erkan Oterkus's picture

Re: verification of peridynamics.

Dear Henry and Matt,

If you are asking about verification of peridynamic theory for some basic problems such as a 1-Dimensional bar subjected to tension loading, or a beam subjected to a transverse point load, etc, I myself verified PD theory for many problems like these. Both the analytical solutions and peridynamic solutions agree very well. But, if you are asking about a publication, it looks like there are not many comparisons against the analytical solutions.If you have an access to this PhD thesis, you can find many basic problems solved by PD theory:

 Bahattin Kilic, "Peridynamic Theory for Progressive Failure Prediction in Homogeneous and Heterogeneous Materials", The University of Arizona, 2008.

Another publication of Kilic and Madenci, "Structural stability and failure analysis using peridynamic theory", International Journal of Non-linear Mechanics, 44, 845-854, 2009, demonstrates the application of peridynamic theory for structural stability analysis. Buckling loads for a bar subjected to different types of boundary conditions were successfully captured.

Regards,

Erkan.

 


Zaoyang Guo's picture

General Constitutive Model for anisotropic foam

Dear Henry,

We recently tested a transversely isotropic polymeric closed-cell foam. We found that there is no constitutive model to simulate its 3D mechanical response. Do you have any suggestion on that?

Zaoyang