Software and Methods for Forming Process Simulation
Lecture 2 · Modelling and Simulation of Processes I – Forming (MSPO2_6I )
Ing. Martin Necpal, PhD.
Winter semester 2026/2027
Faculty of Materials Science and Technology · STU Bratislava
How to control the presentation (bar at the bottom)
← → or the ‹ › buttons – previous / next slide
Dots – jump directly to a chosen slide
▶ or space – play / pause the spoken commentary
Auto on – moves to the next slide when the commentary ends; off – the slide stays
F full screen · Esc slide overview
Alt + click (Linux: Ctrl + click) – zoom into a spot on the slide; click again to zoom out
Underlined abbreviations – hover the mouse over them for an explanation
[0:00–0:02] Welcome. Last week: what modelling and simulation are, physical and mathematical models, the modelling chain. Today: the tools – from "pencil and paper" analytical methods to commercial FEM programs. DEFORM gets the most time because we use it in the lab sessions.
Motivation
Why simulate forming?
From press-shop trials to virtual try-out
Forging dies and stamping tools cost weeks of work and tens of thousands of euros ; every modification adds more
Most of a product's cost is fixed at the design stage , long before production starts
Simulation moves the iterations into the computer : change the preform, die radius or lubricant and run the calculation again
Result: fewer tool modifications, faster production ramp-up, less material (flash, scrap)
[0:02–0:05] Ask the students: how many tool trials would you expect for a new forging without simulation? Answer from practice: several, each taking weeks. Show the dashed loops: red means hours, grey means weeks.
Motivation
What the software tells us
Questions that forming simulation answers
Will the die cavity fill? — underfill, laps, folds, piping
What press do I need? — force and energy versus stroke
Will the part crack or neck? — damage, FLD , thinning
Will it keep its shape? — springback, warpage, distortion
What will the tool life be? — stress, wear, temperature
What is inside? — grain size, phases, hardness
[0:05–0:07] Each question corresponds to a result type we will see in every program today. Remember this list — the review questions at the end come back to it.
Where today's lecture fits
1. Real forming process (forging, rolling, stamping …)
↓
2. Physical model (assumptions, idealisation)
↓
3. Mathematical model → analytical solution (Part I)
↓
4. Numerical model and software
→ FEM / FVM (Parts II–IV)
↓
5. Verification and validation (experiment, lectures 5–6)
Lecture 1: modelling, model types, the modelling chain
Lecture 2 (today): methods and software that solve the model
Lecture 3: material models — flow stress, hardening
Lectures 4–6: preprocessing, sheet and bulk forming examples
Lab sessions: DEFORM — so today's Part III is your toolkit for the whole semester
[0:07–0:08] Short recap of lecture 1. Stress that steps 3 and 4 are today's topic.
Today's agenda (120 min)
Time Block Content
0 – 10 Introduction Motivation, context, agenda
10 – 25 I · Analytical methods Slab method, upper bound, slip lines; solving in MATLAB / Python / GAMS
25 – 67 II · Forming software 1 Ansys Forming · 2 Simufact Forming · 3 Abaqus · 4 Moldflow
67 – 97 III · DEFORM Main part — the software of our lab sessions
97 – 112 IV · Special simulation software Overview · simple examples · use cases
112 – 120 Summary and questions Key takeaways, review questions
[0:08–0:10] Show the agenda. DEFORM is deliberately last and longest: once students have seen four other programs, they can place DEFORM in context better.
Part I — Analytical Methods of Forming
Closed-form and semi-analytical models: fast, transparent and the reference for every FEM result
[0:10] Part I, 15 minutes.
Analytical methods
Overview
Classical analytical methods of plasticity theory
Method Principle Typical output Typical process
Slab method (elementary theory)Force equilibrium on a thin slab; homogeneous deformation Pressure distribution, force Upsetting, forging, rolling, drawing
Slip lines Characteristics of the stress field, rigid-perfectly plastic material, plane strain Exact stress field, force Indentation, extrusion, machining
Upper bound (UBM , UBET )Kinematically admissible velocity field; minimum power Force ≥ actual, die filling Extrusion, closed-die forging
Lower bound Statically admissible stress field Force ≤ actual Limit analysis
Visioplasticity Measured grid deformation → strain rates → stresses Experimental flow field Extrusion and forging research
Common assumptions: rigid-plastic material, constant (or mean) flow stress σf , simple friction law, simple geometry
Source: Hosford & Caddell, Metal Forming; Necpal & Sobota, 2023
[0:10–0:13] Quick map. The upper bound always overestimates the force — which is safe when choosing a press. Slip lines give exact solutions, but only for plane strain and perfect plasticity.
Slab method: upsetting of a cylinder
$$\frac{dp}{dr} = -\frac{2\mu\,p}{h} \;\;\Rightarrow\;\; p(r)=\sigma_f\, e^{\,2\mu (R-r)/h}$$
$$F=\int_0^R p(r)\,2\pi r\,dr, \qquad R = R_0\sqrt{h_0/h}$$
\(p\) contact pressure, \(\sigma_f\) flow stress, \(\mu\) Coulomb friction
\(h, R\) current height and radius (volume constancy)
Pressure peaks in the centre — the friction hill
Larger \(\mu\) or smaller \(h\) ⇒ steeper hill ⇒ higher force
[0:13–0:17] If there is time, derive the equilibrium on the board. Stress that you will simulate exactly this upsetting test in DEFORM in the lab — the analytical force is the first check of the FEM result.
Drawing and extrusion: closed-form results
Sachs (slab method) — wire drawing stress
$$\sigma_d=\sigma_f\,\frac{1+B}{B}\left[1-\left(\frac{A_1}{A_0}\right)^{B}\right],\; B=\mu\cot\alpha$$
\(\alpha\) die semi-angle, \(A_0, A_1\) cross-section before and after
Condition: \(\sigma_d \le \sigma_f\) ⇒ maximum reduction per pass
Upper bound — extrusion pressure
$$\frac{p}{\sigma_f}\approx \underbrace{\ln R_{ext}}_{\text{ideal}}+\underbrace{\mu\cot\alpha\,\ln R_{ext}}_{\text{friction}}+\underbrace{\tfrac{2}{\sqrt3}\!\left(\tfrac{\alpha}{\sin^2\alpha}-\cot\alpha\right)}_{\text{redundant (shear)}}$$
\(R_{ext}=A_0/A_1\) extrusion ratio, \(\alpha\) die semi-angle
The shear term grows with α, the friction term falls ⇒ optimum die angle → optimisation problem
Didactic form — simplified (Avitzur-type upper bound for a conical die); see the full derivation in Hosford & Caddell and Avitzur before using the numbers.
[0:17–0:19] What matters is not the formula itself but its structure: one term grows with the die angle, the other falls — so an optimum exists. This turns the analytical model into an optimisation problem, which leads us to the solvers on the next slide.
Analytical methods
Software
Solving analytical models: MATLAB · Python · GAMS
Problem type Forming example Tool
Closed form Drawing force, Sachs stress NumPy · MATLAB
Nonlinear equation Neutral point in rolling fzero · brentq
Differential equation Slab-method forging, von Kármán ode45 · solve_ivp
Symbolic computation Deriving slab-method equations SymPy · Symbolic TB
Optimisation NLP Upper bound, optimum angle fmincon · GAMS /CONOPT
MINLP Pass schedule, number of passes GAMS /BARON · Pyomo
Framework Groove rolling PyRolL (TU Freiberg)
# Upsetting force – slab method (Python)
import numpy as np
from scipy.integrate import quad
sf, mu = 250.0, 0.2 # MPa, -
R0, h0 = 10.0, 30.0 # mm
for h in np.linspace(30, 10, 5):
R = R0*np.sqrt(h0/h) # volume constancy
p = lambda r: sf*np.exp(2*mu*(R-r)/h)
F = quad(lambda r: p(r)*2*np.pi*r, 0, R)[0]
print(f"h={h:5.1f} mm F={F/1e3:7.1f} kN")
[0:19–0:23] If a notebook is ready, run the script live (it takes a second). Students will see the force rise as the height decreases — the same curve as the force–stroke curve from DEFORM. Only mention GAMS as a tool for optimising pass schedules.
Why analytical methods are not enough
Aspect Analytical methods Numerical software (FEM / FVM )
Geometry Simple (cylinder, strip, cone) Arbitrary 3D from CAD
Material Constant or mean σf σf (φ, φ̇, T), anisotropy, damage
Output Force, mean pressure Full fields: strain, T, stress, defects, microstructure
Time to answer Seconds Minutes to days
Transparency Every term is visible "Black box" — must be checked
Use analytical results for the first estimate and to check FEM results — leave the rest to the software
[0:23–0:25] Transition to Part II. The table explains why industry pays for the software we will see next.
Part II — Forming Simulation Software
1 Ansys Forming · 2 Simufact Forming · 3 Abaqus · 4 Moldflow → 5 DEFORM (Part III)
[0:25] Part II, about 42 minutes, roughly 9–10 minutes per program.
Software
Overview
Two groups of forming software
Dedicated process-oriented
Templates: presses, tools, draw beads, passes
For the process engineer , not the FEM specialist
Sheet: Ansys Forming , AutoForm, PAM-STAMP
Bulk: Simufact , DEFORM , QForm, FORGE
Plastics: Moldflow , Moldex3D
General-purpose nonlinear FEM
The user builds everything: parts, contacts, steps
Maximum flexibility, user subroutines (UMAT )
Abaqus , LS-DYNA , Ansys Mechanical, MSC Marc
Research, teaching, unusual processes
Solver type Principle Where
Implicit (static) Equilibrium in every increment, large steps DEFORM , Abaqus/Standard, springback
Explicit (dynamic) Small stable time steps, robust contact LS-DYNA (Ansys Forming), Abaqus/Explicit
Finite volumes (Euler) Fixed mesh, material flows through it Simufact FV , Moldflow solvers
[0:25–0:28] Give the students two "drawers" to sort every program into. Explicit vs. implicit solver is a common exam question.
Every program follows the same workflow
1.
Geometry — import from
CAD (
STEP ,
IGES ,
STL ), symmetry
↓
2. Material — flow curve, thermal data, damage
↓
3. Mesh — element type, size, remeshing
↓
4. Boundary conditions — friction, heat transfer, tool motion
↓
5. Solution — solver, time steps, stopping criteria
↓
6. Evaluation — fields, curves, defects, report
Steps 1–4 = preprocessing (lecture 4)
Step 6 = postprocessing (lectures 5–6)
Programs differ mainly in how much of steps 1–4 their templates automate
"Garbage in, garbage out": the quality of the answer is set by the flow curve and the friction coefficient
[0:28–0:30] Use this as a lens for every program: which steps does it do for you?
1 · Ansys Forming
Sheet metal stamping
Ansys Forming — stamping with the LS-DYNA solver
Complete application for sheet metal stamping : pre-, solution and postprocessing in one environment
Solver: LS-DYNA explicit (implicit for gravity and springback)
Vendor: Ansys, part of Synopsys since 2025; first version 2022 R1
Project types: cold and hot stamping (press hardening), hemming (beta), One Step (inverse method)
For process engineers and tool makers — no need to be an FEM expert
[0:30–0:33] Ansys Forming replaces older preprocessors for LS-DYNA (eta/DYNAFORM, the LS-PrePost module). Unit system: t, mm, s, N, MPa.
1 · Ansys Forming
Workflow · Results
Workflow and typical results
Process —
CAD , tipping into the press, accuracy
↓
Blank — blank outline, material,
FLD , tailor-welded blanks
↓
Operations — drawing, trimming, flanging, draw beads
↓
Advance — springback compensation, trim line, robustness
↓
Analysis — results,
PPTX report
Result Answers
FLD , formability indexCrack or wrinkle?
Thinning Local thickness loss
Springback Shape after unloading
Stoning, zebra Visible surface defects
Draw-in, skid marks Blank movement
Press force Press size
Rule: 3 elements on a 90° radius
Presets: Fast · Accurate · Very Accurate
Source: Ansys Forming User's Guide 2026 R1
[0:33–0:36] Springback compensation: the software iteratively modifies the tool surface until the part is within tolerance — fewer tool modifications. Time scaling: tool speeds are artificially increased to shorten the explicit calculation.
Video: Ansys Forming — stamping from start to finish
Ansys Forming — stamping simulation demo
[0:36–0:39] Play 2–3 minutes. Students should watch: the blank, the blank holder with draw beads, the coloured FLD plot.
2 · Simufact Forming
Bulk · Sheet · Joining
Simufact Forming — two solvers in one program
Process-oriented software for bulk and sheet metal forming ; Simufact (Hamburg) → MSC → Hexagon
Forging, cold heading, extrusion, ring rolling, open-die forging, fine blanking, self-pierce riveting, clinching , heat treatment
Finite elements (FE ) Finite volumes (FV )
Description Lagrangian — the mesh moves with the material Eulerian — fixed mesh, material flows through the cells
Mesh Hexahedra, tetrahedra, remeshing No remeshing needed
Suited for Cold forming, sheet metal, tool stress, joining 3D hot closed-die forging with flash
Origin MSC Marc solverMSC SuperForge
Switching FE ↔ FV between operations; FV parallelised in 2022 — up to 4× faster
[0:39–0:42] The FE/FV table is the key: Lagrange vs. Euler. We will return to FVM in Part IV.
Simufact Forming — defect prediction
Piping and folds : typical in forging and extrusion — hard to predict without simulation
Damage-based separation: 2D mesh splitting, 3D element deletion — blanking, fine blanking, riveting
[0:42–0:45] Picture b is an animation. Explain piping: material flows away from the end face and leaves a funnel-shaped cavity.
Video: fold formation in forging a fork head
Simulation of folds: material flow during forging of a fork head (Simufact)
[0:45–0:48] The original process design tended to form a fold; the video shows the fold and the modified design. Link to DEFORM: the same defect is found there with point tracking and flow lines.
3 · Abaqus
General-purpose nonlinear FEM
Abaqus — the general-purpose tool
Dassault Systèmes SIMULIA ; Abaqus/CAE and 3DEXPERIENCE; free Learning Edition (≈ 1000 nodes)
No forming templates : the user builds parts, contacts and steps — flexible, ideal for research (UMAT /VUMAT )
Abaqus/Explicit Abaqus/Standard
Use Forming with complex contact, large deformations Springback, smooth quasi-static loading, heat transfer
Strength Robust contact, no convergence problems True equilibrium, large increments
Watch out for Small stable time step → mass scaling Convergence with contact
Quasi-static check in Explicit: \( \dfrac{E_{kin}}{E_{int}} = \dfrac{\text{ALLKE}}{\text{ALLIE}} \lesssim 5\text{–}10\,\% \)
[0:48–0:51] Industry standard: form in Explicit, import into Standard, compute springback. The energy ratio proves that speeding up the process in Explicit did not change the physics.
Abaqus example: axisymmetric deep drawing of a cup
Part — blank: deformable axisymmetric shell; punch, die, blank holder: analytical rigid
Property — steel, E = 200 GPa, ν = 0.3, flow curve
Step — Dynamic, Explicit, 0.007 s
Interaction — penalty friction μ = 0.1 (die, blank holder)
Load — punch 60 mm, Smooth Step amplitude
Mesh — blank only, refined at the radii
Results — PEEQ , thickness STH , punch force
[0:51–0:54] Seven steps = seven Abaqus/CAE modules. Compare with Ansys Forming, where the same model is a template done in a few clicks.
Abaqus: large deformations and damage
Technique Use
ALE adaptive meshingForging, rolling, extrusion (Explicit)
Adaptive remeshing Error-driven (Standard)
Mesh-to-mesh mapping Very large deformations, forging operations
CEL (Euler)Extreme material flow
Damage: ductile, Johnson-Cook , shear, FLD , FLSD , MSFLD , M-K
Springback: ≥ 5–7 integration points through the thickness, kinematic hardening
[0:54–0:57] Remeshing is THE main problem of Lagrangian bulk forming. Abaqus offers ALE; DEFORM does fully automatic remeshing — remember this for Part III.
4 · Moldflow
Plastic injection moulding
Autodesk Moldflow — forming of polymers
Founded 1978 in Melbourne (Colin Austin); Autodesk since 2008; current version 2027
Adviser (designers) · Insight (analysts) · Synergy interface · Communicator viewer
Physics: non-Newtonian viscous melt flow + heat transfer — not solid plasticity
Viscosity: Cross-WLF · PVT : Tait equation
≈ 13,900 thermoplastics in the database
Mould closing
↓
Filling (injection) weld lines, air traps, short shot
↓
Packing shrinkage, sink marks
↓
Cooling largest part of the cycle
↓
Ejection → warpage
[0:57–1:00] Why Moldflow in a metal forming course? Same philosophy — virtual try-out of the mould instead of the die — but different physics. Also relevant for metal injection moulding (MIM).
Moldflow: mesh types and defects
Mesh Principle Use
Midplane 2.5D Shells on the mid-surface; Hele-Shaw Thin walls, fastest
Dual Domain 2.5D Matched meshes on both surfaces Thin simple parts
3D tetrahedraFull 3D flow, no Hele-Shaw Thick, massive parts
1D beam Line elements Runners and cooling channels
Defect Remedy suggested by simulation
Short shot Add/move gate, thicker wall, hotter melt
Weld lines Move the gate to a non-critical area
Air traps Venting, gate change
Sink marks Core out thick sections, longer packing
Warpage Balance cooling, uniform wall
Analysis sequence: gate location → fill → pack → cool → warp (+ fibre orientation → structural FEM )
[1:00–1:03] Hele-Shaw: the flow length is much greater than the thickness — the "4 : 1" rule. Warpage causes split into differential cooling, differential shrinkage and orientation effects.
Video: Autodesk Moldflow overview
Overview – Autodesk Moldflow (Autodesk)
[1:03–1:05] Play ~2 minutes: watch the flow front and the weld lines that form where the fronts meet.
Comparison of the five programs
Aspect Ansys Forming Simufact Abaqus Moldflow DEFORM
Focus Sheet metal stamping Bulk, sheet, joining General-purpose FEM Plastic injection moulding Bulk, HT , microstructure
Solver LS-DYNA explicitFE (Marc) + FV Standard + Explicit Flow (2.5D / 3D) Implicit Lagrangian, ALE
Setup Templates Templates Manual Templates Templates + open
Strength Springback compensation Fast FV hot forging Flexibility, UMAT Fill, pack, cool, warp Remeshing, microstructure, wear
Vendor Synopsys Hexagon Dassault Syst. Autodesk SFTC (USA)
There is no "best" software — only software suited to the process and the question
[1:05–1:07] Bridge to Part III: DEFORM is in the last column.
Part III — DEFORM
Design Environment for FORMing — the software of our lab sessions
[1:07] Main part, 30 minutes.
5 · DEFORM
Overview · History
DEFORM — from ALPID to a complete process simulator
FEM system for forming, heat treatment, machining and joining
Vendor: SFTC , Columbus, Ohio (≈ 30 people)
Roots: ALPID at Battelle (U.S. Air Force) — the first practical FEM tool for production forming
Theory: Kobayashi, Oh & Altan, Metal Forming and the FEM (1989)
Year Milestone
1989 DEFORM-2D
1991 SFTC founded, takes over DEFORM
1993 DEFORM-3D
2024 · 2025 V14.0 · V14.1
2026 V14.1.1; V15.0 announced
[1:07–1:10] Industries: aerospace (superalloy and titanium turbine discs), automotive, fasteners, oil and gas. Widely used in research — one reason universities like it.
The DEFORM product family
Product What it is
DEFORM-2D Axisymmetric and plane problems — our first lab sessions
DEFORM-3D Full 3D deformation, temperature and microstructure
DEFORM PREMIERAll systems and modules in one package
FORMING EXPRESS Simplified system for forging, heading, extrusion, drawing
DEFORM-HT Heat treatment: carburising, quenching, tempering
DEFORM-3D Machining Turning, drilling, milling templates; tool wear (Usui)
Modules Microstructure, DOE /optimisation, ring rolling, open-die forging, shape rolling, extrusion
Processes: forging (hot/warm/cold), rolling, extrusion, drawing, heading, tube piercing, spinning, blanking, riveting, clinching, powder compaction, shot peening
[1:10–1:12] In the lab we start in DEFORM-2D: axisymmetric upsetting and extrusion, then we move to 3D.
The DEFORM solver: rigid-viscoplastic formulation
$$\pi=\int_V \bar{\sigma}\,\dot{\bar{\varepsilon}}\,dV+\frac{K}{2}\int_V \dot{\varepsilon}_v^{\,2}\,dV-\int_{S_F} F_i\,u_i\,dS \;\;\rightarrow\;\; \delta\pi = 0$$
\(\bar\sigma, \dot{\bar\varepsilon}\) effective stress and strain rate · \(K\) penalty constant (incompressibility) · \(F_i\) surface tractions
The unknowns are velocities → Newton-Raphson / direct iteration in every step (implicit, updated Lagrangian)
Elastic strains are neglected — suited to large plastic flow (forging); elasto-plastic option for springback, residual stresses, tools
ALE and steady-state solvers for ring rolling, shape rolling, extrusion, tube piercing
Source: Kobayashi, Oh & Altan (1989); DEFORM-3D brochure
[1:12–1:15] This is the flow formulation from ALPID. Link to Part I: minimising a power functional — the same idea as the upper bound method, but with a discretised velocity field instead of a guessed one.
Working with DEFORM : Pre → Simulation → Post
objects, geometry, mesh, material, movement, inter-object conditions (friction,
HTC ), stopping, steps
↓
generate database (.KEY → .DB )
Simulation engine (
FEM + automatic remeshing)
message and log files — watch convergence and remeshing
↓
Post processor
state variables, force–stroke, point tracking, flow lines, damage
Object type Meaning
Rigid Non-deforming tools (optionally with heat transfer)
Plastic Workpiece, rigid-viscoplastic
Elastic Tool stress analysis
Elasto-plastic Springback, residual stresses
Porous Powder, sintered material
Machine models: hydraulic, mechanical , screw press, hammer (energy)
[1:15–1:18] Walk through the three programs exactly as the students will see them in the lab. Stress: the database file is the whole simulation — back it up.
Key inputs: flow stress, friction, heat transfer
$$\bar\sigma = f(\bar\varepsilon,\,\dot{\bar\varepsilon},\,T)$$
Tabulated data from the material library (steels, Al, Ti, superalloys, Cu …) or from your own compression tests
Neural-network prediction of flow stress from chemical composition (V13+)
$$\tau = m\,k = m\,\frac{\bar\sigma}{\sqrt3} \quad\text{(shear)}\qquad \tau=\mu\,p \quad\text{(Coulomb)}$$
Conditions Friction factor m
Cold, good lubrication 0.08 – 0.12
Warm ≈ 0.2
Hot forging, lubricated 0.3
Hot forging, dry 0.7 – 1.0
Die–workpiece heat transfer coefficient; the Inverse HTC module derives it from thermocouples
Most sensitive inputs : flow stress and friction
m values: guide values for typical categories in DEFORM
[1:18–1:21] Comparison: the slab method used μ (Coulomb); DEFORM mostly uses the friction factor m for bulk forming. Material models in detail — lecture 3.
Automatic remeshing and multiple operations
The Lagrangian mesh deforms with the material → automatic remeshing when an element degenerates or penetrates the tool
Tetrahedra (3D), quadrilaterals (2D), optionally hexahedra; mesh windows refine critical regions
Each remeshing interpolates state variables → keep the number of remeshings reasonable (volume loss )
Multiple Operations (MO ) — one chain, runs automatically
Furnace heating
↓
Transfer (air cooling)
↓
Forging blows 1…n
↓
Flash trimming
↓
Heat treatment → machining
[1:21–1:23] Remeshing is why DEFORM became established in forging. MO: the whole process route in one project, with history (strain, temperature) passed from step to step.
Results: from die filling to tool life
Die filling, laps and folds , flow lines (point tracking, flow net)
Effective strain, strain rate, temperature, triaxiality
Force–stroke curve, energy → press selection
Tool stress (elastic dies, prestressing) and tool wear
≈ two dozen ductile damage models, element deletion
Normalised Cockcroft–Latham criterion
$$D=\int_0^{\bar\varepsilon_f}\frac{\sigma_1}{\bar\sigma}\,d\bar\varepsilon \;\ge\; C$$
Archard tool wear
$$w=\int K\,\frac{p\,v}{H^{\,a}}\,dt$$
σ₁ maximum principal stress · p pressure · v sliding velocity · H tool hardness
[1:23–1:26] Cockcroft–Latham: chevron cracks in extrusion, surface cracks in upsetting. Archard: wear grows with pressure and sliding, falls with tool hardness — hot tools soften, so temperature matters.
Microstructure and heat treatment
JMAK recrystallisation: \( X = 1-\exp\!\left[-\beta_d\left(\dfrac{\varepsilon-\varepsilon_c}{\varepsilon_{0.5}}\right)^{k_d}\right] \)
Dynamic / static recrystallisation, grain growth, phase fractions, hardness, precipitates, texture (crystal plasticity)
[1:26–1:29] Aerospace discs: the customer specifies not only the shape but also the grain size — that is why microstructure prediction exists. DEFORM-HT: distortion and quench cracking.
Video: DEFORM — simulation of metal forming processes
DEFORM – The Premier Process Simulation Solution for Metal Forming (
SFTC )
[1:29–1:32] Show 2–3 minutes. Alternative: the turbine disc microstructure video (next slide).
Video: forged turbine disc — microstructure simulation
Forged Turbine Disk Microstructure Simulation (
SFTC )
Optional if short of time — skip to "What's new".
DEFORM today: V14.1 → V15.0 (User Group Meeting, May 2026)
Area New feature Benefit
Solver Stiffness reuse for elastic objects (MUMPS + ESE ) 1.2 – 2.9× faster with elastic tools
Solver Iterative elasto-plastic solver EPCG , MPI > 50 % faster than CG
Spinning QEM + partial domain14 min instead of 17 – 35 h
AI U-Net surrogate model for multi-pass shape rolling + CMA-ES groove optimisationDesign in seconds
Quenching CFD boiling model (vapour film, nucleate boiling)Realistic HTC
Blanking 3D mesh in the clearance zone and its remeshing Fine mesh only in the clearance
Automation Python DEFORM-API , PPTX report generation Scripting, machine learning
Source: SFTC , DEFORM UGM Spring 2026 (12–13 May 2026)
[1:32–1:34] Main idea: forming simulation is moving towards automation and AI surrogate models trained on FEM results — link to Part IV.
DEFORM in our lab sessions
1. Upsetting a cylinder — DEFORM-2D axisymmetric
↓
2. Compare force–stroke with the slab method (Part I)
↓
3. Forward extrusion — die angle, friction, damage
↓
4. Closed-die forging in DEFORM-3D — die filling, folds
↓
5. Lab assignment (max. 20 points)
Bring: your notes from Part I (the slab method formula)
Check as you go: is the force realistic? was the volume preserved? how many remeshings were there?
Assessment: assignment 20 pts + written exam 80 pts
[1:34–1:37] Practical information. Adjust the order of lab sessions to the actual timetable.
Part IV — Special Simulation Software
Beyond Lagrangian FEM : inverse methods, finite volumes, particles, discrete elements, multiscale and data-driven tools
[1:37] Part IV, 15 minutes: overview, simple examples, use cases.
Special software
Overview
Methods and the software that implements them
Method Principle Software (examples) Use in forming
Inverse one-step Unfolding the finished part into a plane AutoForm OneStep, Ansys Forming One Step , PAM-STAMP Blank shape, quick feasibility
Finite volumes Fixed Eulerian mesh Simufact FV , QForm Extrusion* Hot forging, Al extrusion
SPH (particles)Meshless Lagrangian particles DualSPHysics, LS-DYNA SPH Extreme deformations, MIM , FSW
DEM Contact of discrete grains EDEM, LIGGGHTS, Yade Powder die filling
Multiscale CA , crystal plasticity, phase fieldDAMASK, DEFORM Microstructure Recrystallisation, texture, earing
Data-driven ANN / ROM trained on FEM Python (PyTorch), DEFORM-API Springback, rolling design
Analytical frameworks Coupled analytical models PyRolL Rolling pass design
Physical simulation Reproducing φ, φ̇, T on a specimen Gleeble Flow curves, validation
* QForm uses FEM with coupled die deformation; listed as a dedicated extrusion tool
[1:37–1:41] Also mention other major dedicated programs: AutoForm (sheet metal, the automotive standard), QForm and FORGE NxT (bulk). They exist alongside the five we covered in detail.
A simple example: one extrusion, four descriptions
[1:41–1:44] One picture explains why special software exists: each discretisation removes one limitation of Lagrangian FEM — remeshing (FVM, SPH) or the continuum assumption (DEM).
Simple examples: inverse method and AI surrogate model
Inverse one-step (sheet metal)
Finished part geometry (
CAD )
↓ unfold into a plane
Blank shape + strain estimate
↓
Check in the
FLD : risk of cracking / wrinkling
Minutes instead of hours ; ignores process history
Surrogate model (bulk forming)
Hundreds to thousands of
FEM runs (
DOE )
↓ training
Neural network (
ANN ,
U-Net )
↓
Instant prediction + optimiser
Valid only within the trained design space
[1:44–1:47] Both shortcuts trade accuracy or generality for speed. The FEM model remains the reference in both cases.
Special software
Use cases
Use cases from research and industry
Problem Approach Result
Closed-die forging of a rib-web part UBET + plasticine experimentsForce and flow agree with the physical model
Extrusion of Al profiles, ratio > 25 FVM on an Eulerian mesh (+ ANN /GA )No remeshing; optimised die parameters
Metal powder injection moulding (17-4 PH ) SPH in DualSPHysicsVerified by real injections
Powder die filling (Distaloy AE) DEM → density as input to compaction FEM Filling shoe design before trials
Earing in deep drawing of AA 5754 Crystal plasticity FEM Ears predicted from texture
Grain size in hot forging Cellular automata coupled with DEFORM-2DRecrystallised fraction and grain size
Springback in bending ANN trained on FEM Fast springback for new cases
Sources: literature review for this course (Kazan et al. 2009; Tikhovskiy et al. 2007; Fraunhofer IWM and others)
[1:47–1:50] Pick two examples and explain them in detail — suggestion: FVM extrusion (link to Simufact FV) and CA + DEFORM (link to Part III).
Video: dedicated extrusion simulation (QForm)
[1:50–1:52] Short clip: material flow through a porthole die, profile exit speed. Then the summary.
Summary
Analytical methods (slab method, upper bound, slip lines) give a quick force estimate and are the first check of every FEM result
Dedicated software automates the process: Ansys Forming (stamping), Simufact (bulk + joining), Moldflow (polymers)
General-purpose FEM (Abaqus) is flexible, but the user builds everything
DEFORM : implicit rigid-viscoplastic FEM , automatic remeshing, damage, tool wear, microstructure — our tool for the lab sessions
Special software goes beyond the limits of FEM : FVM , SPH , DEM , inverse and multiscale methods, AI surrogate models
Next time: material models — elasticity, plasticity, hardening, flow curves
[1:52–1:54] Five sentences = five parts of the lecture.
Review questions
Why is the contact pressure in upsetting highest in the centre? What increases this peak? answer
Upper bound force: is it greater or smaller than the actual one — and why is that useful? answer
Explicit vs. implicit solver: which for stamping, which for springback? answer
What is the difference between the Lagrangian (FE ) and the Eulerian (FV ) description? answer
Why is the physics in Moldflow different from that in DEFORM ? answer
Which two inputs in DEFORM affect the result most? What is the friction factor m? answer
Name one damage model and one wear model in DEFORM and what they predict. answer
When would you use the inverse one-step method or an ANN surrogate instead of full FEM ? answer
[1:54–2:00] Let the students answer; all questions are shown at once. Remaining time for their own questions.
References and further reading
Necpal, M.; Sobota, R.: Modelovanie a simulácia procesov tvárnenia. Spektrum STU, 2023 (in Slovak)
Kobayashi, S.; Oh, S.I.; Altan, T.: Metal Forming and the Finite Element Method. Oxford UP, 1989
Hosford, W.F.; Caddell, R.M.: Metal Forming: Mechanics and Metallurgy. Cambridge UP
Dixit, P.M.; Dixit, U.S.: Modeling of Metal Forming and Machining Processes. Springer, 2010
Bílik, J.; Kapustová, M.; Ridzoň, M.: Teória tvárnenia. AlumniPress, 2015 (in Slovak)
SFTC : DEFORM brochures, DEFORM News 2022–2025, UGM Spring 2026 — deform.com
Ansys Forming User's Guide 2026 R1 — ansys.synopsys.com
Hexagon: Simufact Forming 2021.1 and 2022 — nexus.hexagon.com
Abaqus Analysis User's Manual; TECHNIA and CAE Assistant tutorials
Autodesk Moldflow help, What's New 2025–2027 — autodesk.com
Weiner et al.: PyRolL, JOSS 9(93), 2024
Images and videos © their owners, referenced for teaching purposes