SPH

© All Rights Reserved

25 views

SPH

© All Rights Reserved

- ANSYS Tutorial Release 9.0
- Computational Methods in Transient Dynamics
- ARTIGO RS3
- Fem Objective Questions
- Dynamic Stability of Thin-walled Structures: A Semi-Analytical
- b 31008012
- Discretization Considerations in Moving Load Finite Element Beam Models
- r05410102 Finite Element Methods in Civil Engineering
- Adaptive Mesh Generation Using a Normal Offsetting Technique
- Ansys Manual Libre
- 277-T763
- Mesh, Loads and BCs
- Stress Simulation
- Willardson_COPV_SAMPE2009
- Me6603 Fea - 2 Marks
- FEA2 the Basic FEA Procedure
- Impact on Water and Soil With Lagrangien
- productFlyer_978-3-540-70697-7
- Cst
- Nishioka 1983

You are on page 1of 7

Tatsuo Sakakibara, Dr.Eng.

Toru Tsuda, Dr.Eng.

Ryo Ohtagaki

ITOCHU Techno-Solutions Corporation

4-1-3,Kyutaro-machi,Chuo-ku, Osaka, Japan

Abstract

The SPH method is well known for the hydrodynamics and recently SPH is applied for the impact and the

penetration behavior into solid materials. SPH is completely mesh free so that it seems to be appropriate to simulate

crash behaviors of concrete and rock materials. However, the applicability and the effects of the parameter of SPH

for quasi-static problems are not clear completely.

In this paper, in order to verify the applicability and the accuracy of SPH on quasi-static problems, a number of

unconfined uni-axial compression simulations are performed by SPH. The specimens of different number of

elements are used and the parameter sensitivities of SPH are examined. The influences of formulation type of SPH

in LS-DYNA are also investigated. The results of SPH are compared with the Laglangian result.

From the results, the effects of the parameter for SPH are made clear. Besides, it is found that the renormalized

formulation is efficient to get the accurate results at boundary. As the result, it is conformed that SPH is applicable

for the quasi-static problems.

Introduction

SPH is recently well-known particle method as Lagrangian mesh free and it is applicable to high

velocity and large deformation analysis like impact and penetration behaviors. SPH has some

advantages, one is that it is easy to built up simulation model even though very complicated

geometry because of completely mesh less method. Another advantage is that SPH is not need to

take into account any distortion so that SPH is applicable to analyze large deformation and

fracture behavior. On the other hand, the accuracy of the results by SPH depends on the particle

density or smoothing length. In specially, it is necessary to pay attention to decrease the accuracy

at boundaries. In case of the problem like penetration behaviors, the affects by the boundaries are

not so large. However, in case of quasi-static problems, the high accuracy in terms of stresses

and strains are required. A large number of simulations on fluid dynamics and impact problems

have been performed using SPH method so far, but the number of applications to the quasi-static

problem and the case study for the investigation of accuracy by SPH is not many because of

mainly cost efficiency compare to continuum method.

In this paper, uni-axial compression simulations to the hard sand stone by SPH were performed

and investigated the applicability on the quasi-static problems by SPH. The results of SPH were

compared with the experiments and FEM result. The version we used is LS-DYNA Ver.971.

Analytical Condition

Figure 1 shows a cylindrical specimen with 10mm in diameter and 10mm in height. We used

four type of the SPH models and a Lagrangian model as shown in Figure 2. FEM model is used

to compare with SPH results. Each SPH model has different particle density. SPH-10 means that

the 10 particles line up in the direction of the diameter. Therefore, 40 particles line up in SPH-40.

The number of particles and the distance between particles of simulation models are summarized

in Table 1. Those of FEM in Table 1 mean the number of elements and minimum element size

16-21

respectively. The constant loading velocity is applied to the particles on the top surface. As

boundary condition, the particle displacements to the Z-direction on the bottom surface are fixed.

The material used in this study is assumed to be elasto-plastic as shown Figure 3 defined by

MAT 3 in LS-DYNA. Material properties are shown in Table 2.

constant velocity

Model Name

SPH-10

SPH-20

SPH-30

SPH-40

FEM

10mm

Z

Y

X

10mm

(1)

number of

particles

880

6636

21480

50560

960(1)

number of elements

(2)

distance between

particles

1.0 mm

0.5 mm

0.33 mm

0.25 mm

1.0 mm(2)

Figure 1 Specimen

FEM

SPH-10

SPH-20

SPH-30

SPH-40

y

E

Elastic modulus

poissons ratio

Density

16-22

MPa

g/cm3

82

0.32

1.5

compressive strength

MPa

4.3

Analytical Results

Effect of loading velocity

During the uni-axial compression test, the loading rate must be slow enough to ensure that the

specimen remains in quasi-static equilibrium throughout the test. In terms of the calculation time,

the loading velocity is required faster. So a number of tests with several loading speeds from 0.1

mm/sec to 1000 mm/sec were examined. Figure 4 shows the relationship between force and

displacement at the top surface. The results with the speed from 0.1 mm/sec to 100 mm/sec are

identical completely. In case of the loading speed of 1000 mm/sec, excessive energy is going

into the kinetic energy. Hereafter the loading velocity of 10 mm/sec was used in all subsequent

tests.

Force (N)

300

200

100 mm/sec

1000 mm/sec

0.1 mm/sec

1 mm/sec

10 mm/sec

100

0

0

0.2

0.4

0.6

0.8

Displacement (mm)

Effect of particle density

Particle density of SPH method is very important factor. Because, an accuracy of the result

depends on number of particles which are used in a unit volume. To clarify the effect of particle

density, four models shown as Figure 2 were examined. SPH-10 is the coarsest and SPH-40 is

the finest. Figure 5 shows the relationship between force and displacement at the top surface.

The results of SPH-30 and SPH-40 are in good agreement with the result of FEM. On the other

hand, the model of SPH-10 and SPH-20 seems to be less stiff than the others. From this result, at

least, number of particles as many as SPH-30 on horizontal surface is necessary to get the

accurate result.

16-23

500

Particle Density

SPH-40

Force (N)

400

FEM

300

200

SPH-30

SPH-20

SPH-10

100

0

0

0.2

0.4

0.6

0.8

Displacement (mm)

Effect of Smoothing length

Smoothing length is a one of specific parameter of SPH. In LS-DYNA, the smoothing length can

be varied by CSLH keyword in *SECTION_SPH card. CSLH is the constant applied to the

smoothing length of the particles. According to the LS-DYNA manual, the default value 1.2 of

CSLH is recommended and values between 1.05 and 1.3 are acceptable. In order to make sure

this recommendation is useful even thought the quasi-static problem, three case of the smoothing

length are applied to each SPH model shown as Figure 2. Figures 6 show the correlation between

force and displacement on the top surface with the different smoothing length. The case of CSLH

1.05 is the most accurate in all models. Even if the particle density is very high, in case of

smoothing length larger than 1.2 makes model weaker. That means a yield stress decrease than

the value specified as input parameter. From these results, the smoothing length should be 1.05

on the quasi-static problem.

500

500

SPH-10

SPH-20

FEM

300

CSLH=1.05

200

CSLH=1.3

100

0

0

0.2

FEM

300

CSLH=1.2

200

CSLH=1.3

100

CSLH=1.2

0.4

0.6

0.8

Displacement (mm)

(a) SPH-10

16-24

CSLH=1.05

400

Force (N)

Force (N)

400

0

0

0.2

0.4

0.6

0.8

Displacement (mm)

(b) SPH-20

500

500

SPH-30

CSLH=1.05

SPH-40

400

FEM

300

CSLH=1.2

200

FEM

CSLH=1.2

300

200

CSLH=1.3

100

0

0

CSLH=1.05

400

Force (N)

Force (N)

CSLH=1.3

100

0.2

0.4

0.6

0.8

Displacement (mm)

0

0

0.2

(c) SPH-30

0.4

0.6

0.8

Displacement (mm)

(d) SPH-40

Figures 6 Effect of Smoothing length

16-25

Effect of Renormalization

As mentioned before, Calculation accuracy declines in specially when particle density is low. An

influence of boundary elements is thought as one reason. The number of particles in the

influence domain at boundary particles is smaller than that at inside particles. Figure 7(a) shows

a relationship between axial stress and axial strain at selected elements on the top surface as

shown Figure 7(b). The element F is located at boundary and stress of this element is lower than

the other elements. This graph shows that particle on the boundary edge has inaccurate stress. To

overcome this inaccuracy, the renormalized formulation technique is implemented in LS-DYNA.

There are seven particle approximation theories which are specified by FORM keyword in

*CONTROL_SPH card. In this study, the results of the renormalized formulation Form-1 are

compared with the results obtained from default formulation Form-0. Figure 8 shows the

comparison between the two formulations using SPH-20 model. The result of renormalized

formulation is good agreement with FEM result. Therefore, the renormalized formulation is very

effective on quasi-static problems.

5.0

top surface

Form-0

4.0

3.0

FEM

A

B

C

D

E

F

2.0

1.0

0.0

0

0.02

0.04

0.06

0.08

Axial Strain

(a) Stress Strain correlation

0.1

500

SPH-20

Force (N)

400

FEM

300

FORM-1

200

FORM-0

100

0

0

0.2

0.4

0.6

0.8

Displacement (mm)

16-26

Figure 9 shows comparison of computation time between SPH and FEM. Abscissa axis means

the number of particles or elements and vertical axis means computing time per one cycle.

Computation time of SPH takes approximately three times as much as FEM.

0.30

0.25

SPH

FEM

0.20

0.15

0.10

0.05

0

0

Number of particles/elements

Conclusions

In this study, the unconfined uni-axial compression tests were simulated by SPH to verify the the

applicability and the accuracy on quasi-static problems. The conclusions obtained from the

simulation results are summarized as follows.

The SPH method can be applied to quasi-static problems by an appropriate modeling.

1. Number of particles

- Almost 30 x 30 particles on horizontal surface are necessary to get the accuracy compare

to the FEM result.

- The computation time increases when the particle increases.

2. Smoothing length

- The case of CSLH 1.05 seems to be the most accurate.

3. Accuracy of boundaries

- Less number of particles at boundary causes inaccurasy.

- Renormalized formulation improves accuracy at boundary

References

1) Y. Sakai, A. Yamashita: Study on Fundamental Characteristics of Structural Analysis by Particle Method on SPH,

Transactions of the Japan Society of Mechanical Engineers. Vol.67, No.659, pp.1093-1102, 2001.

2) Jean Luc Lacome:Smoothed Particle Hydrodynamics Method in LS-DYNA, 3rd LS-DYNA Forum, 2004.

3) Randles P.W. and Libersky L.D.: Smoothed Particle Hydrodynamics: Some recent improvements and

applications, Comput. Meth. Appl. Mech. Engrg, pp375-408, 1996.

16-27

- ANSYS Tutorial Release 9.0Uploaded byRAHUL DHIMAN
- Computational Methods in Transient DynamicsUploaded by문성수
- ARTIGO RS3Uploaded byRegina Maria
- Fem Objective QuestionsUploaded bySree Murthy
- Dynamic Stability of Thin-walled Structures: A Semi-AnalyticalUploaded bynjm001
- b 31008012Uploaded byAnonymous 7VPPkWS8O
- Discretization Considerations in Moving Load Finite Element Beam ModelsUploaded byEugene Commerell
- r05410102 Finite Element Methods in Civil EngineeringUploaded bygeddam06108825
- Adaptive Mesh Generation Using a Normal Offsetting TechniqueUploaded bySARA
- Ansys Manual LibreUploaded byMed Elyoubi
- 277-T763Uploaded byfetni
- Mesh, Loads and BCsUploaded byUber_Konstruktor
- Stress SimulationUploaded byapi-3857201
- Willardson_COPV_SAMPE2009Uploaded byHasrizam86
- Me6603 Fea - 2 MarksUploaded byPrabha Karan
- FEA2 the Basic FEA ProcedureUploaded byOmom Mye
- Impact on Water and Soil With LagrangienUploaded bynguyenvannganpc
- productFlyer_978-3-540-70697-7Uploaded bysuriya
- CstUploaded byRavishanker Baliga
- Nishioka 1983Uploaded byMateus Motta
- IJAIEM-2013-05-24-052Uploaded byKrishna Myakala
- Lecture 3Uploaded bybadboys123
- 4256Uploaded byHilary Watson
- proofUploaded byMouhamad Wehbe
- Proof IndrajitChowdhuryandJitendraPSingh17052012Uploaded bySinNombre
- lec17Uploaded bymahendran
- 1A2a_0030Uploaded bySaif Mohtasib
- Finite Element Method.pdfUploaded byMawri
- XFEM_Presentation18Uploaded bySivakumar Mvn
- Now letUploaded bysuriya

- Fundamental Finite Element Analysis and ApplicationsUploaded bychamanchandel
- Evaluation of Different Material Models to PreUploaded byJitapriya Das
- 40722_Journal of Ocean Engineering and Marine EnergyUploaded byJitapriya Das
- GongyoUploaded byCarlo Damião
- A Study of Quasi Static Problem by Sph MethodUploaded byJitapriya Das
- Word Power Made EasyUploaded byJitapriya Das
- Autobiography of a YogiUploaded byyatri82

- Eco Business ParkUploaded byJavier Herreros
- Book 9Uploaded byRebwar Qaremani
- photogrammetryUploaded bymuhammadrz1
- Illumination Lec 1Uploaded byJohnLaboguin
- Toc Notes LastUploaded byAmit Sharma
- sample lesson plan weeblyUploaded byapi-295143259
- GENERALIZATION OF PARTITION FUNCTION, INTRODUCING SMARANDACHE FACTOR PARTITIONUploaded byDon Hass
- Surveying IIUploaded byPrabava Rajan
- Csform101 AUploaded bynagaer
- Persuasive Communication SkillsUploaded byRohan
- 10 5 20 RBT Workbook FinalUploaded byadamaso09
- DECAHS BOR GuidelinesUploaded byDylan Dong
- MIS Raj EditUploaded byAshlesh Mangrulkar
- Designing Strategy for Cultural Change PPT MBAUploaded byBabasab Patil (Karrisatte)
- MathematicsUploaded byDonghoon
- impact of jet.docxUploaded bySaifullah Mohd
- "Divine of Use" of Swedenborg - George DeCharmsUploaded byD.Elder
- e20s Hap450 ManualUploaded byNhut Vo
- en_epochtimes_The War Is Approaching UsUploaded byapi-3707213
- Applied Mathematics JournalUploaded byHirenPaneliya
- Syllabus CHEM-182 OceanUploaded byaddislibro
- gps project docxUploaded byapi-248437787
- UM-Flint to UM-Ann Arbor Engineering TransferFact SheetUploaded byTimothy Kennedy
- perfume making.docxUploaded byChryzl Dumalay Suobiron
- ch06Uploaded byBayley_Navarro
- HeatKal TechnologiesUploaded byRahul Gawali
- Embedded control system for smart walking.pdfUploaded bySreekanth Pagadapalli
- Direct MethodUploaded byBishnu Pada Roy
- PplUploaded byHarsha Vardhan Reddy
- PIMSUploaded byjoseferrehigieneuru