MICROMODELLING OF ADDITIVE MANUFACTURING AT ILT/LLT

MICROMODELLING OF ADDITIVE MANUFACTURING AT ILT/LLT Current work, problems and outlook ICME Barcelona 14.04.2016 Jonas Zielinski, Norbert Pirch

MICROMODELLING OF ADDITIVE MANUFACTURING AT ILT/LLT AM: SLM & LMD general principal Simulations: our approach Current work Basic model Selective Laser Melting (SLM) Laser Metal Depositioning (LMD) Comparison with experimental data SLM: cross section LMD: cross section SLM & LMD: Differences outlook Seite 2

SLM & LMD: general principals How does it work? Selective Laser Melting Laser Metal Depositioning Seite 3

Simulations: Our approach What do we want now? Do I need to simulate every aspect of the process chain? Simulate the things that can not (easily) be observed Particle distribution Temperature gradients Microstructure Pore formation Shield gas currents Link the results to thingsthat can be observed Heat emission NO!? Seite 4

Simulations: Our approach What do we want? The Additive Manufacturing Dream Do I need to simulate every aspect of the process chain? NO! Simulate the things that can not (easily) be observed Particle distribution Temperature gradients Microstructure Pore formation Shield gas currents Link the results to events that can be observed Heat emission Seite 5 Distortion CAD Simulation Build it!

Simulating AM: A multiscale problem! In length- and time-scale Meltpool dimension ~100 µm Melt time n x ~µs Microstructure ~µm Partsize ~25cm x 25 cmx 25cm Scanvector length ~km Built-time ~days Seite 6

Simulations: Our approach Classification Microstruture Model: Precipitation, Phases Micro Model: Temperature distribution, Solidification Conditions Macro Model: residual stress, strain, distortion, mechanical properties Seite 7

Basic AM micro model Output / Input Output: 4D temperature distribution Temperature gradients Input: Temperature dependent material properties (c p, λ, ρ, T sol, T liq, T eva, γ l,g, h fus, α, ε) Laser density distribution (LDD) (measured and/or approximated) Process parameters (v scan, P L, h Layer, ρ bulk, d track ) Verified by measured melt pool geometry Solidification conditions important for microstructure evolution

Basic AM micro model Implementation Curvature of meltpool surface (liq., gas.) Young-Laplace equation (γ l,g ) Finite Elemente Method based Heat transfer equation Laser power as surface source: P L t, x, y α Thermal radiation: ε T x, y, z 4 surf Heat of fusion h fus : effective warmth capacity method c p (between (T sol, T liq ) Temperature capped at: T T eva (LDD) Mass balance: powder (from powder-bed or blown-powder processes) is absorbed into the meltpool (v scan, Δd melt, h layer, ρ bulk )

SLM micro model Developement: capillary modell Laser power is deposited on the isothermal surface T = T eva (if it locally exists, else: substrate/meltpool surface) Material does not evaporte at T = T eva but is transparent for the laser radiation v scan Isothermal T = T eva

SLM micro model Developement: evaporation pressure capillary modell Local pressure of evaporating material is respected in the Young-Laplace equation Additional required Material properties: Temperature dependend vapour pressure p eva (T) (in case of IN718: only nickel) Evaporation pressure Isothermal T = T solid v scan

SLM micro model Developement: powder bed modell Strongly reduced heat conductivity in the powder Bulk density ρ bulk : volume shrinking due to total densification after powder melting (assumption: no gas pore formation) Effective absorpion/transmission coefficient for the powder layer for the laser radiation Mesh adaption

LMD micro model Developement: modelling of particle propagation Particle density distribution z = -3 mm z = -7 mm Statistical model Input: Particle velocity Particle-reference-density Linear particle trajectories into the interaction (Laser-particle) zone Output: Spatially resolved particle density Spatially resolved laser transmission Particle temperature (respecting self-attenuation)

LMD micro model Developement: thermalization of optical energy LDD Laser LDD trans LDD particle 14% reduction of laser power (emitted)

SLM: comparison with experimental data Cross section: differences in the melt pool shape Width [µm] Depth [µm] W/D Experiment (EOS) 250 ± 50 187 ± 10 1,33 Evap. pressure model 133 73 1,82 10x10x10 mm³ cubes, last layer parallel to two edges Process parameters: v scan = 960 mm s, h layer = 40 µm, r L = 42 µm, P L = 285 W v scan

SLM: comparison with experimental data Difference analysis and possible causes Width [µm] Depth [µm] W/D Experiment (EOS) 250 ± 50 187 ± 10 1,33 Vapour pressure model 133 73 1,82 Smaller width and depth (absolut) in simulation Ratio between width and depth too small Laser radiation Energy input in material too low Scan direction Effective head conduction underestimated Energy is deposited deeper in the material Powder layer meltpool

LMD: comparison with experimental data Cross section: similarity in melt pool shapes and track geometry 784 µm S1 325 µm v scan 749 µm S3 314 µm Process parameters: P L = 250 W, m p = 1 g, v min scan = 500 mm min, IN718 on CK45 Powder efficiency: 72 76 % Good alignment of simulated and measured track geometry

SLM & LMD micro model What s missing? Why does simulating LMD work better? More measured data is feed into the model! Slower process (cooling rates, scanning speed) No keyhole!

MICROMODELLING OF ADDITIVE MANUFACTURING AT ILT/LLT AM: SLM & LMD general principal Simulations: our approach Current work Basic model Selective Laser Melting (SLM) Laser Metal Depositioning (LMD) Comparison with experimental data SLM: cross section LMD: sideview & cross section SLM & LMD: Differences Outlook Seite 22

Outlook What do we want next? A predictive simulation chain for AM Topology optimization Automatic parameter optimization for given machine Tool: Transfer of process parameters (other machines) Simulation driven machine optimization What is the best material for my process? Chemical composition of powder (grain size)? Seite 23 Thank you for your attention! I am grateful for your ideas!