What’s New in ANSYS

2020 R1 - APDL
Content

• MAPDL
oLinear Dynamics
oElements
oContact
oSolver
oNLAD
MAPDL Linear Dynamics
Component Mode Synthesis (CMS) – New Method
Objective:
Obtain better convergence and improve performances when the master nodes are defined at locations
other than the interfaces.
Example: observation nodes where the displacement solutions are requested without expansion pass.

Feature:
For the free-interface CMS analysis (CMSOPT,FREE), by issuing “SUPPORT = ON” on the M command,
pseudo-constraints are specified on some master nodes to enforce constrains during the mode-
extraction analysis done in generation pass.

Command option:
M,NODE,Lab1,NEND,NINC,Lab2,Lab3,Lab4,Lab5,Lab6,SUPPORT

Note: This CMS method is also called mixed method

Component Mode Synthesis (CMS) - Equations
DOFs partition
• Master DOFs: m = [m1 m2]
• Slave DOFs: s

Component modes are normal modes obtained with:

• FIX method: all master DOFs in m are constrained (fixed)

• FREE and RFFB methods: all master DOFs in m are unconstrained (free)

• FREE method with SUPPORT = ON applied on m1 DOFs

m1 DOFs are constrained, m2 DOFs are unconstrained
Component Mode Synthesis (CMS) - Example

M2
Modal analysis with DOFs of m1 constrained
CMS use pass FULL analysis
M1

M2

M1

M2
Solutions Comparison Tool – RSTMAC Enhancements

• Degree of Freedom (DOF) selection: 1 DOF or a combination of DOFs

o New MACOPT, DOF
• Support non-structural DOFs for coupled-field analyses:
o PRES, VOLT, CONC, MAG, TEMP, and AZ
• Support node matching based on node number
o For morphed mesh or translated/rotated model applications
Solutions Comparison Tool – RSTMAC Example

Based on VM282
• Two different meshes of Piston-Fluid System
using FLUID30
• Comparison of coupled-field modal solutions

****** MATCHED SOLUTIONS ******

Substep in Substep in MAC value
FILE1 FILE2 MACOPT,DOF ,PRES

1 1 0,999
2 2 1,000
3 3 0,998
4 4 0,998
5 5 0,998
6 6 0,998
7 7 0,998
MAPDL Elements
List of New Features
• Current technology 3D 3-Node Cable Element : CABLE280
• Support of incompressible materials with Inverse method
• Pure displacement-based formulation for SOLID285
• General distributed load for SOLID and SHELL elements
• Current technology 2D thermal element PLANE292/293
• User-Defined Material Models for 22x Coupled Field Analyses
3D 3-Node Cable Element CABLE280
Stranding
• Suitable for analyzing Winding
moderate to extremely
slender cable structures
• Computationally efficient
with only translational DOFs
• Mixed Displacement / Force queins.com/en/solutions/stranding/
formulation for superb
solution accuracy and One cable is placed in the center, a second layer
robustness containing six cables is stranded around it.

• Wide range of applications: Pulley and Cable

offshore, civil, and
mechanical
z

1 2
L
4
I
y
3

K
x
J
5
Y USUM
CABLE280 Geometry
Inverse Analysis
• It supports Solid185 B-bar formulation (keyopt(2)=1)
• It supports Solid186, Solid187 and Solid185(B-bar) with mixed u/P (Keyopt(6)=1,2)
• New supports large deformation with incompressibility, such as biomedical applications
• An aortic lumen under 80 mmHg blood pressure is the input geometry
o The calculated “deformed” geometry is the zero-pressure geometry
Results when blood pressure increases to 120mmHg
…Diameter increases significantly

Change in diameter with further loading (120 mmHg)

is shown with given cross-section at 80 mmHg
loading.
Pure Displacement Formulation for Linear TET SOLID285
• A linear tetrahedral element with displacement as DOF only: SOLID285
(KEYOPT(1)=1)
o Efficient and robust for problems without significant bending and incompressibility
General Distributed Load for SOLIDs and SHELLs
• General load options: normal, tangential, Tangential distributed
user defined, tapered, projected, and more load on 3D element
faces
• A single new PREP7 command SFCONTROL
to define general load properties
• Eliminates the need for surface effects
elements for most cases
• All current technology 2D/3D SOLID and
SHELL elements are now supported
o 3D/2D solid elements User defined line load
(185,186,187,190,285,182,183) on 3D element edges
o 3D/2D shell elements (181,281,208,209)
o 3D/2D coupled-field elements (226,227,222,223)
• Enable imaginary distributed loads for the
harmonic analysis
Enhanced 2D Thermal Elements PLANE292/293
PLANE293

Temperature distribution
PLANE292

PLANE293
• Capable of simulating
Radiation +
complex thermal problems convection

• Proper handling of Radiation

Radiation
nonlinear thermal load with to space
to space

consistently linearized

Conduction
element stiffness Thermal
contact PLANE293,
• UPF support ( UserMatth.F ) KYOP1=1

Stove with heat generation, PLANE292

User-Defined Material Models for 22x Coupled Field Analyses
• Coupled-field elements SOLID226 and SOLID227 now support the “UserMat” and
“UserMatTh” subroutines for customizing structural and thermal material behaviors,
respectively.
o To define a custom structural material model (UserMat),
specify user-defined structural properties via TB,USER with 22x Coupled-Field Element
TBOPT = NONLINEAR, LINEAR, or MXUP.
o To define a custom thermal model (UserMatTh), specify user-

defined thermal properties using the new THERM option UserMat  ,

(TBOPT = THERM) with TB,USER. Deformation u
Temperature T
Plastic heat
Thermal strain
Viscoelastic heat
q
• Application example: UserMatTh q,
T
o Manufacturing process simulation
Contact Features
Defining Preload in a Fastener Undergoing Large Rotation
• “PSMESH” can now define “MPC184” joint elements for
applying a preload to a bolt undergoing large rotation or large
deformation
• The joint element supports large deformation and
the bolt axis follows the local coordinate system defined
at the joint node
• You can apply torque and rotation about the bolt axis (FJ and
DJ, respectively)
• Bolt Sleeve Model Undergoing Large Rotation:
 Technique with PRETE179 New Technique Technique with PRETE179
Time =2 Time =2

New Technique
Time =3 Time =3

Stress changes after rigid rotation

20
 New Interference Fit Modeling Technique
• A new method for solving interference fit ramps the normal and tangential contact
stiffness (FKN, FKT) as well as the friction coefficient (MU) from near-zero up to actual
values. The ramping method is active during a load step or a time period that you specify.
The time period can be within one load step or span across several load steps. Unlike the
other interference-fit methods, initial contact engagement is not required
3D Radial Shaft
Plate Springs Seal Ring
2D Axisymmetric Circular Spring
Hooked Spring 2D

21
Exponential Pressure - Penetration Relationship
• The exponential pressure-penetration relationship (KEYOPT(6) = 3 on contact elements) can make contact
behavior smoother. While the default settings of pressure at zero penetration (real constant PZER) and initial
contact clearance (real constant CZER) work well for most contact models, some cases require non-default
values to achieve convergences. You now have the option to define PZER and CZER as scaling factors.
Previously, only absolute values could be input for the real constants
• In addition, the maximum cut-off contact stiffness 𝐹𝐾𝑁𝑚𝑎𝑥 used in the exponential pressure-penetration
relationship has been revised to prevent ill-conditioning of the global stiffness matrix

𝐹𝐾𝑁𝑚𝑎𝑥

𝑐0 𝑝0
 Tangential Fluid Pressure-Penetration Loads
• The 2-D and 3-D surface-to-surface contact elements (CONTA172, CONTA174) and their corresponding target
elements (TARGE169, TARGE170) now support tangential fluid pressure-penetration loads. In prior releases,
only normal fluid pressure-penetration loads were considered
• Example: Viscous shear in thin film of fluid between plates dominates resistance to relative plate movement

Contact Status Tangential Fluid Pressure

23
 Enhanced Force, Moment, and Displacement Convergence
• Predictor, Force & Displacement Convergence Checks Existing
o Linear, quadratic predictor, Predictor off for transient. criterion 1.e-4
Bisection due to large displacement increment
o Current non-linear convergence check Reference values suffer
several drawback
o Reference too low- problem will never converge
o Reference too high- problem converges to wrong solution Automatic Force
o New “Convergence Reference” logic aimed to provide convergence Ref
too low- problem
accurate solution with minimal user intervention will not converge
o The changes improve the robustness and accuracy of the
solution, and nonlinear problems with no external loads (such
as initial penetration resolution for contact and free thermal New
expansion) experience enhanced convergence. criterion 1.0

New Automatic
Force convergence
Ref

24
APDL Solver
Distributed ANSYS Enhancements
• New features
o New –rcopy argument to specify remote file copy command on clusters (defaults to scp)
o Added logic to detect SIGKILL/SIGABRT signals and provide relevant message to user
• Improved scaling
o Significantly reduced memory requirements at higher core counts
o Improved Block Lanczos scaling performance at higher core counts
o Faster performance for fracture parameter calculations
o Faster performance in sparse solver when running in the out-of-core memory mode on
systems, which use the Microsoft Windows operating system
Distributed ANSYS Enhancements
• Significantly reduced memory usage (BGA model)

Memory Usage Comparison

2,000
2019 R1

Total Memory Usage (GB)

Solder
balls 1,600
Mold 2020 R1
PCB 1,200

• 16 MDOF; sparse solver 800

• Nonlinear transient analysis
involving creep and nonlinear 400
geometric effects
• Linux cluster; each compute node 0
contains 2 Intel Xeon Gold 6148 32 cores 64 cores 128 cores 256 cores 512 cores 896 cores
processors (40 cores), 384GB RAM,
SSD, CentOS 7.6
Distributed ANSYS Enhancements
• Significantly reduced memory usage (EPD model)

Memory Usage Comparison

2,500
2019 R1

Total Memory Usage (GB)

2,000
2020 R1

1,500

• 10 MDOF; sparse solver 1,000

• Nonlinear static analysis
involving plasticity
• Linux cluster; each compute 500
node contains 2 Intel Xeon
Gold 6148 processors (40 0
cores), 384GB RAM, SSD, 32 cores 64 cores 128 cores 256 cores 512 cores 896 cores
CentOS 7.6
Distributed ANSYS Enhancements
• Significantly reduced memory usage (Engine model)

Memory Usage Comparison

1,500
2019 R1

Total Memory Usage (GB)

2020 R1
1,000

• 9.1 million DOF; sparse solver

• Nonlinear static analysis
involving contact, plasticity and 500
gasket elements
• Linux cluster; each compute
node contains 2 Intel Xeon Gold
6142 processors, 384GB RAM, 0
SSD, CentOS 7.4 32 cores 64 cores 128 cores 256 cores 512 cores
• Mellanox EDR Infiniband
Distributed ANSYS Enhancements
• Significantly reduced memory usage (ECU model)

Memory Usage Comparison

5,000
4,500 2019 R1

Total Memory Usage (GB)

4,000
2020 R1
3,500
3,000
2,500
• 9.7 MDOF; Block Lanczos
2,000
eigensolver
• Modal analysis requesting 100 1,500
modes; includes expansion step 1,000
• Linux cluster; each compute 500
node contains 2 Intel Xeon Gold 0
6148 processors (40 cores),
32 cores 64 cores 128 cores 256 cores 512 cores 768 cores
384GB RAM, SSD, CentOS 7.6
Distributed ANSYS Enhancements
• Improved scaling for Block Lanczos eigensolver (ECU model)

DMP Scaling Performance

3.5
R19.0
3.0
2019 R1 Increased performance over time
2.5 2020 R1

Solution Speedup
2.0
• 9.7 MDOF; Block Lanczos
1.5
eigensolver
• Modal analysis requesting 100
1.0
modes; includes expansion step
• Linux cluster; each compute 0.5
node contains 2 Intel Xeon Gold
6148 processors (40 cores), 0.0
32 64 128 256 512
384GB RAM, SSD, CentOS 7.6
Number of Cores
Distributed ANSYS Enhancements
• Improved performance for fracture parameter calculations (gear tooth)

DMP Scaling Performance

8.0

7.0 2019 R3
Increased performance over time
2020 R1
6.0

Solution Speedup
5.0

4.0

3.0
• 2.2 MDOF; sparse solver
• Linear static analysis involving 2.0

fracture parameter calculations 1.0

• Linux cluster; each compute
node contains 2 Intel Xeon E5- 0.0
8 16 32 64 128
2690 processors (28 cores),
Number of Cores
128GB RAM, SSD, CentOS 6.7
Distributed ANSYS Enhancements
• Improved sparse solver performance using Windows I/O functions

Model 1
• 4.4 MDOF; sparse solver, out-of-
Solution Performance Comparison
core memory mode 2.0
• Nonlinear static analysis
• Windows workstation containing

Solution Speedup
2019 R3
an Intel Xeon E5-2687W processors 1.5
(12 cores), 64 GB RAM, 10k RPM 2020 R1
hard drive, Windows 10
1.0
Model 2
• 9 MDOF; sparse solver, out-of-core 0.5
memory mode
• Nonlinear static analysis
• Windows workstation containing 0.0
an Intel Xeon E5-2687W processors Model 1 Model 1 Model 2 Model 2
(12 cores), 64 GB RAM, 10k RPM
(2 cores) (8 cores) (2 cores) (12 cores)
hard drive, Windows 10
Miscellaneous Enhancements
• Database file compression (sparsify)
o Activated by default via /FCOMP,DB,SPARSE
o Can be deactivated via /FCOMP,DB,0
o Achieves roughly 20-50% database file size reduction for most models
• Includes .rdb database files used by the multi-frame restart procedure
• Slightly longer times to save database files
Miscellaneous Enhancements
Technology Demonstration Manual
• Database file compression (sparsify) models run on Linux server while
measuring database file size right
before solution
Database File Compression
1.4
2019 R3 (Uncompressed) 45% file size reduction on average
1.2
2020 R1 (Compressed)
Compression Ratio

1.0

0.8

0.6

0.4

0.2

0.0
Variety of simulations from Technology Demonstration Guide
Miscellaneous Enhancements
• Database file compression (sparsify) Customer models run on Linux
server while measuring database
file size right before solution
Database File Compression
1.4
2019 R3 (Uncompressed)
1.2 35% file size reduction on average
2020 R1 (Compressed)
Compression Ratio

1.0

0.8

0.6

0.4

0.2

0.0
Variety of simulations obtained via ANSYS customers
Miscellaneous Enhancements
• Upgraded to the Intel 2019 Update 3 FORTRAN/C/C++ compilers
o Includes similar updates for the Intel MKL and DAAL libraries
NLAD
2020 R1 Developments in NLAD
• Current Developments in 2020 R1:
‐ NLAD with NLGEOM=OFF
‐ Nonlinear Stabilization with NLAD and Rezoning
‐ Mixed Remeshing
‐ “KEEP” option for maintaining and updating element components in NLAD
‐ Support for Tabular input of nodal temperatures and heat generation
NLAD with NLGEOM=OFF

NLAD with NLGEOM=OFF

• Energy criterion used to refine mesh
• Automatic improvement of solution accuracy
• Reduction of structural errors

Without NLAD – Stresses Without NLAD – Errors With NLAD – Stresses With NLAD – Errors
Nonlinear Stabilization with NLAD

NLAD without stabilization:

Analysis fails due to instability

Key Features:
• Hollow cylindrical tube subjected to bending load
• NLAD with MESH and ENERGY criteria
• Activate global stabilization
• Stabilization included during MAPSOLVE in
NLAD with stabilization
NLAD/Rezoning
Mixed Remeshing
• Achieve remeshing due to distortion and refinement-based criteria at the same time!

Plastic cap

Rubber seal

Key Features:
• Initially coarse mesh refined using the mixed
remeshing capability
• Leads to smoother contact conditions during solution

Solution output now shows which method is being tried

“Keep” Option for Maintaining And Updating Element Components in NLAD
• Element components for which NLAD is NOT defined are now updated based
on mesh changes using the “KEEP” option CM, Cname, Entity, KEEP

Initial Components Setup Before Re-mesh After Multiple Re-meshings

 Join the ANSYS
conversation!

Read. Comment.
Join the conversation!

The new and improved

ANSYS blog is live at:
ansys.com/blog

44
 Sign up for an ANSYS Event near you!
ansys.com/about-ansys/events

Filter on “Webinar” under Events to sign up. A Structures 2020 R1

update webinar will be available in late February.

45