Design and Optimization of Plants and Components for the Production of Polyurethane Foams using STAR-CCM+

Transcription

1 Design and Optimization of Plants and Components for the Production of Polyurethane Foams using STAR-CCM+ Dr. Carsten Brodbeck, Bettina Landvogt Martin Schamberg Fraunhofer SCAI Hennecke Polyurethane Technology Hennecke Hennecke STAR European Conference 2010, London

2 Project Sponsor and Partners The project is funded by German Federal Ministry for Economy and Technology It is a cooperation project of the research facility Fraunhofer SCAI and the industrial company Hennecke Polyurethane Technology Hennecke is a medium sized company, part of Adcuram Group AG and producer of polyurethane machines and plants with a large variety of products

3 Content Introducing Project Partner Hennecke Introducing Fraunhofer SCAI Description of Slab Stock Foam Plants Description of Mix Head Injectors CFD Approach in STAR-CCM+ Optimization Approach with DesParO Mesh Adaption Approach in STAR-CCM+ First Results Slab Stock Foam First Results Mix Heads

4 Product variety of Hennecke Metering machines Mix Heads Gas loading technology, blowing agent metering units Elastomer lines PU spraying methods Moulding lines Lines for refrigerated appliances Sandwich panel lines Slabstock lines Recycling technologies Tank farms Hennecke

5 Fraunhofer SCAI Multiphysics Software: Optimization Software: Crash Software: Research: MpCCI, SCAIMapper Autonester, PackAssistant, DesParO DiffCrash, DesParO, FEMZIP Simulation Engineering, Numerical Methods, Bioinformatics, Optimization

6 Slab Stock Foam Plant Slab stock foams are produced amongst others as standard, hypersoft, high load bearing or visco-elastic foams Hennecke offers facilities for high and small quantities as continuous and discontinuous lines Hennecke wants to offer facilities with a reduced production rate. Many customers request due to investment costs and logistical problems a smaller piece number produced per time. Question is: How do we have to change the inlet domain geometry and flow parameters to obtain a line with a reduced production rate in order to assure a smooth operation and a high quality foam? The fluid has to pass critical point before blowing starts The fluid has to spread all over the sheet system The fluid may not re-circulate

7 Slab Stock Foam Plant Hennecke Hennecke Hennecke Film Hennecke

8 Mix Heads Mix Chambers Mix heads are the key elements of any PU production line Hennecke offers mix heads types for different applications, small and very large shot weights, mix injection even in challenging positions Question is: How can we modify the mix head geometries and the mix head chambers to obtain a desired mixing quality by spending less energy? Mix heads have to operate for different densities and mass flows There are geometrical restrictions due to production methods and cost reduction To reduce energy expended low-pressure stirrers are preferable, but for maintenance and cleaning high-pressure mixing is better. Is the result a combination of both? Time scale is very small as one shot takes only milliseconds

9 Mix Heads Mix Chambers Hennecke Hennecke

10 Numerical Approach Slab stock foam: VOF, user defined density/viscosity (planned), residence time, java scripting for batch execution, steady and unsteady, mesh adaption Variation of inlet geometry Mix heads: single fluid (filled chamber), power law density, turbulent, with or without cavitation (VOF), stationary and transient, java scripting, mesh adaption Variation of mix head and mixing chamber geometry and viscosity Optimization for both: Using SCAI s software DesParO for multi-objective optimization Define parameters (geometry, etc.), boundary conditions (densities, viscosities, etc.) and criteria (mean age, efficiency).

11 Optimization Approach Robust Design Sensitivity and robustness analysis with an efficient reduction of the design space Meta modeling (response surface modeling with radial basis functions) and advanced design-of-experiment techniques Multi-objective robust design-parameter optimization (target function, sensitivity analysis, Pareto-front determination) Design-of- Experiment Methods Non-linear Metamodeling (Radial Basis Functions ++) Adaptive, hierarchical meta-models: iterative refinement Incorporation of scatter + global and fully local tolerance estimation

12 Optimization Approach Software for interactive, multi-objective robust design-parameter studies and optimization

13 Mesh Adaption in STAR-CCM+ - Workflow Batch automated via Perl scripts writing STARCCM+ java scripts Simulation on coarse mesh Import surfaces as parts and use as volumetric controls for meshing Flag cells for adaption, iterative modification of sensitivity to match selected percentage of cells th md sd md i n n i sd 1 2 Realized by reports and field functions = pressure / velocity / volume fraction gradient md = mean deviation; sd = standard deviation α = sensitivity; th = threshold Split regions by Function, remove small cell groups to decrease noncontiguous regions then further split by Non-Contiguous Extract boundary surface of regions with flagged cells Export extracted surfaces as Nastran Wrap surface if it contains non-manifold edges / vertices

14 Mesh Adaption Process Example Mix Heads initial grid 128,539 cells adapted grid 1,447,400 cells several cycles possible volumes of previous adaptions could be kept stop process if too many cells are flagged

15 Mesh Adaption Process Example Slab Stock initial grid adapted grid adaption to volume fraction and velocity gradient limit adaption to cells in PU fluid limit of non-contiguous splitting to fifty was needed -> otherwise splitting failed multi-regional adaption supported

16 Slab Stock Foam Model Geometry Angular view Side view Dispenser Dispenser Inclination γ Gap width Sp PU Moving upper wall Moving upper wall Gap width s1 Distance L3 Fluid basin, generating back Fluid basin pressure (back pressure) Moving lower wall Outlet height s2 Distance L1

17 Slab Stock Foam Mesh trimmed mesh (size ~4 mil. cells) prism layer mesh imprint mesh of injector volumes shapes are adapted to geometrical parameters

18 Slab Stock Foam Evaluating Results Before starting the optimization process a parameter study was launched with variation of gap sizes, down-grade and inlet length, inclination. The results are evaluated by: printing minimum, maximum, average velocity, PU volume fraction and mean age for closest position and for outlet of channel exporting scene files of several planes x or z = const. exporting streamline animation file automated by Java scripts Closest position Outlet of channel

19 Slab Stock Foam Bad/Better Result Not so good Looks better backflow Fluid not spread Fluid widely spread Film Film

20 Slab Stock Foam Optimization Parameters Dispenser Inclination Gap width Outlet height PU Dispenser position Down-grade height Boundary conditions mass flow density viscosity distinct dispenser Parameters gap width inclination dispenser position down-grade height outlet height Criteria homogeneous residence time distribution (range, standard deviation, maximum) in outlet homogeneous velocity distribution (no backflow, range, mean deviation) in outlet

21 Mix Heads Model Geometry Mix Head Injector Mixing Chamber side view top view Inlets Diffusor Outlet Injector Nozzles Nozzle Angle Nozzle Diameter Nozzle Pin Nozzle Gap Rounded Wall to Ejector

22 Mix Head Mesh trimmed mesh (size ~1 mil. cells) volumes shapes are adapted to geometrical parameters

23 mass-averaged velocity [m/s] Numerical Results - Mix Heads The efficiency of different mix head geometries is measured by mass flow averaged velocities (at different distances from the nozzle) of the fluid for varying gap width (= mass flow rates). D angle = nozzle diameter (var) = nozzle angle (fix) gap width [mm]

24 Numerical Results - Mix Heads A total of about 80 numerical experiments with different parameters (geometrical parameters, material properties, boundary conditions) were simulated and analyzed with DesParO. D angle = nozzle diameter (var) = nozzle angle (fix)

25 Mix head Optimization Parameters Nozzle disk Nozzle disk Nozzle Pin Boundary conditions inlet pressure density viscosity mass flow / nozzle gap width Parameters nozzle disk cone grading nozzle disk widths/heights nozzle pin widths/heights nozzle diffusor Criteria mass-averaged velocity (momentum of fluid) in three planes behind injector Nozzle Diameter Nozzle Gap Nozzle Diffusor

26 Summary Today I presented: Basic features of slab stock foam facilities and high-pressure mixing devices for polyurethane applications Numerical approach for CFD and optimization Automated mesh adaption process for STAR-CCM+ Numerical results for slab stock foam and mixing heads

27 Outlook General: Slab stock foam: Mix heads: make a precise definition of all parameters, boundary conditions and criteria for the optimization process develop further java scripts to enhance automation and flexibility include residence-time dependent density and viscosity in model integrate simulation into optimization process and apply adaption process to all simulation runs integrate mixing chamber and a agitator device to simulation model launch more DoE controlled simulations for mixing chamber and mixing heads applying mesh adaption process expand optimization result visualization

28 Thanks for your attention!