FIXTURE ANALYSER AND OPTIMISER MODULE

Transcription

1 FIXTURE ANALYSER AND OPTIMISER MODULE CASE STUDIES AND SUPPORTING MATERIALS Software version (22)

2 Versioning and Contribution History Version Organization Comment Date 1.0 Warwick Uni Table of content added 25/05/ Warwick Uni Introduction and Software GUI added 26/05/ Warwick Uni Appendix A added 31/05/ Warwick Uni Tutorial case study added 2/06/ Warwick Uni Final version and proof reading 3/06/2015 Name Organization Dr Pasquale Franciosa Warwick Uni p.franciosa@warwick.ac.uk Prof Darek Ceglarek Warwick Uni d.j.ceglarek@warwick.ac.uk Mr Manoj Babu Warwick Uni m.k.babu@warwick.ac.uk Contact information: Dr Pasquale Franciosa - Digital Lifecycle Management, WMG, University of Warwick, UK p.franciosa@warwick.ac.uk Tel: +44(0) Prof Darek Ceglarek - Digital Lifecycle Management, WMG, University of Warwick, UK d.j.ceglarek@warwick.ac.uk Tel: +44(0) web: (22)

3 Nomenclature CAD: Computer Aided Design CAM: Computer Aided Manufacturing GUI: Graphical User Interface KCCs: Key Control Characteristics KPCs: key Product Characteristics KPIs: Key Performance Indicators (22)

4 Contents Nomenclature... 3 Contents INTRODUCTION How to install What is it? Technical background SOFTWARE INTRODUCTION Graphical Interface and Menus The Graphical Renderer Window The Options and Setting Window Input data Output data Simulation workflow Model initialization Assembly simulation Fixture analysis and synthesis BASIC TUTORIAL Model definition Model initialisation Simulate assembly Fixture analysis and synthesis Fixture development ADVANCED TUTORIAL: WING SUB-ASSEMBLY Model definition Model initialisation Fixture analysis and synthesis APPENDIX A A Quick Introduction to Simulation Builder (22)

5 1 INTRODUCTION The document intends to summarise the main features of Fixture Analyser and Optimiser toolbox and to provide general guidelines on how to use the main capabilities of the software. Please refer to the following short bibliography for additional technical information. Franciosa P., Gerbino S., Ceglarek D., Fixture Capability Optimization for Early-stage Design of Assembly System with Compliant Parts Using Nested Polynomial Chaos Expansion, proceeding of CIRP-CMS 2015 Franciosa, P., Ceglarek, D., 2015, Hierarchical Synthesis of Multi-level Design Parameters in Assembly System, CIRP Annals, 64(1) Ceglarek, D., Colledani, M., Vancza, J., Kim, D-Y., Marine, C., Kogel-Hollacher, M., Mistry, A., Bolognese, L., 2015, Rapid Deployment of Remote Laser Welding Processes in Automotive Assembly, Systems, CIRP Annals, Vol. 64/1 Ceglarek, D., Franciosa, P., Váncza, J., Erdos, G., Kovács, A., Kim, D-Y., Colledani, M., Marine, C., Kogel-Hollacher, M., Mistry, A., Bolognese, L., Franchini, F., Gerbino, S., Agyapong-Kodua, K., Stroud, I., Chryssolouris, C., Magnano N., 2011, Remote Laser Welding (RLW) System Navigator for Eco and Resilient Automotive Factories, FoF-ICT , No , URL: How to install Fixture Analyser and Optimiser can be installed following two main steps (installation file are located at: Software\Installation package): 1. copy the setup package on the local computer/pc (for example: "C:\Demo") 2. run the setup package and follow the instructions. If required, download and install MRC (Matlab Runtime Compiler). It can be downloaded for free at: The main software architecture is Matlab-based and integrates MEX-C++ routines, which takes advantages of multi-core CPUs capability and x64 platforms (x32 platforms are no longer supported). The software has been tested on Windows 7 and Windows 8 with no reported bug. The installation package contains the following folders/files: Software\Installation package: it contains the installation package; Tutorial\Basic Tutorial: is collects input and output data to run the basic tutorial; Tutorial\Advanced Tutorial (1): is collects input and output data to run the advanced tutorial. Fixture Analyser and Optimiser can be executed by double clicking ClampConfTool.exe. 1.2 What is it? The tool offers the possibility to evaluate (to calculate output KPIs for given KCCs) and optimise (to allocate input KCCs for given KPIs) product/process performances and to configure assembly fixtures. The tool has the following features: Variation Simulation Analysis (VSA) for deformable parts; assembly process simulation with compliant sheet-metal parts; fixture capability optimisation with stochastic product variations (22)

6 Tool s integration capabilities are: (i) optimised product design loop to generate a feasible assembly process based on user-defined key performance indicators; (ii) optimum locator/clamp layout; (iii) work-station optimisation loop with robot simulation and path planning. Fixture Analyser and Optimiser can be used as interactive/collaborative framework among process and product design engineers. The developed GUI offers interactive tools to facilitate user s data input and visualisation of results. The tool has demonstrated the following benefits when compared against traditional approaches: (i) reduction in engineering changes necessary in today s industrial practice; (ii) fixture optimisation not only for deterministic CAD parts (deterministic simulation), but also for production batches (stochastic simulation). 1.3 Technical background Fixtures are used to accurately locate and securely hold parts during the joining operations to ensure that quality requirements are achieved. Fixture design is one of the most important tasks during the design phase as it involves the definition of locators and restraining blocks which provide support to the parts being assembled, and as a result highly affect process capability and the final product quality. Assembly fixture needs to satisfy the following design constraints: (1) mechanical constraint - the assembly fixture needs to locate and support parts during joining operations; (2) quality constraint - sheet metal parts are usually fabricated with ±0.5 mm tolerances in stamping processes. Variations usually propagate and amplify during assembly operations. Assembly fixture needs to be designed to compensate variation stack-up propagation. Reference clamp Joint clamp Figure 1.1. Schematic illustration of design constraints. (a) part placement on reference clamps (mechanical constraint); (b) part error compensation by joint clamps (quality constraint); Fixture s elements are classified in two different categories: (i) reference clamps: they provide mechanical restraint and control the final KPIs of the assembled product (for example, gap and flushness between door and side frame for body-in-white systems). Reference clamps are defined by the product designer as they reflect the design intent; (ii) joint clamps: they need to be optimised to satisfy both quality and process constraints. The input data is expressed as follows: (1) CAD product data is provided by the product designer and it contains: (i) product geometry and related GD&T/ISO tolerance specifications; (ii) joint layout specification (number of joints, location, type (i.e., linear or circular shape)); (2) CAM process data is provided by the process designer and it contains: (i) reference clamp layout; (ii) joint clamp layout (22)

7 2 SOFTWARE INTRODUCTION 2.1 Graphical Interface and Menus The Graphical Interface of the tool is divided into two main windows as shown in Figure 2.1: (i) Graphical Renderer Window; (ii) Options and Setting Window. The Graphical renderer window shows 3D and 2D representation of inputs and outputs of the model being analysed and is also used for interactive component selection. The Options and Setting Window is where various predefined menus to build the model are situated. Simulation builder Setting panel 3D renderer panel Log panel 2D renderer panel Graphical Renderer Window Options and Setting Window The Graphical Renderer Window Figure 2.1. Graphical Renderer Window and Options and Setting Window. The graphical renderer window as two panels as shown in Figure 2.1: (i) 3D renderer panel; (ii) 2D renderer panel. 3D renderer panel displays 3D visualization of the model and is useful for: (i) model visualization; (ii) interactive selection of inputs; (iii) interactive visualisation of contour plots of user-based defined features. 2D renderer panel displays plots of various KPIs or KCCs at given points or locations selected by the user The Options and Setting Window The options and settings window has three panels: (i) simulation builder (ii); settings panel; (iii) log panel. The simulation builder has a tree structure with options that control the entire simulation (please refer to Appendix A for comprehensive explanation of the main menus). The settings panel displays the parameters for selected options in the simulation builder panel. The log panel displays important information, error, messages and warnings during the operation of the software. 2.2 Input data The software is capable of handling files from a wide range of CAD modellers (product data) such as CATIA, SolidWorks, ProE, etc. Supported mesh formats are: *.inp, *.bdf, *.stl..inp is the native (22)

8 Abaqus mesh file, whereas the ".bdf" file is the standard Nastran 8-character fields. ".stl" is the standard file format native to the stereolithography CAD software. Process data can be handled by MS Excel spreadsheets and formatted ASCII files (example, txt files). 2.3 Output data The software is capable of handling most common file formats. Simulation results can be exported as MS Excel spreadsheets and formatted ASCII files. Moreover, fixture configuration results can be exported using the standard.stl format and imported in any 3D CAD modeller (Figure 2.2), where more detailed fixture development can be accomplished. Simulation and 3D configuration 3D model and fixture development 2.4 Simulation workflow Figure 2.2. Fixture configuration vs. fixture development. The software is capable of performing both fixture analysis and synthesis. The workflow is shown in Figure 2.3 and consists of the following steps: (i) model initialization; (ii) assembly simulation; (iii) fixture analysis or synthesis Model initialization During this process the user imports one of the supported formats of CAD or mesh files into the software, which is followed by specifying various model parameters for the relevant process to be simulated. As an example, for a welding process they could be: (i) the position and length of welds; (ii) location of clamps and locators; (iii) technological parameters; (iv) product variation specifications Assembly simulation Once the model is completely defined the user needs to setup the virtual experiments by specifying the parameters and the range over which they are to be varied. Also the solver to be used for a particular simulation has to be specified. The various options available to the user during this stage of the process will be explained in detail in the following sections via case studies Fixture analysis and synthesis Depending on the needs of the user one can choose either the analysis of a given fixture design or the synthesis to automatically achieve the optimum fixture design. Fixture synthesis is achieved through a combination of regression and optimization of the process variables. The optimization uses the output (22)

9 MANUAL CONFIGURATION Software documentation: Fixture Analyser and Optimiser of the assembly simulation (which leads to the estimation of the assembly response function which links input KCCs to outputs KPIs) and hence has to be performed after the results of assembly simulation are available. (1) MODEL INITIALISATION Load product data Load process data (2) SIMULATE ASSEMBLY Set KCCs parameters Solver settings (4) FIXTURE ANALYSIS (5) FIXTURE SYNTHESIS (4.1) What-if analysis (5.1) Calculate Regression (4.2) Plot results (5.2) Solve optimisation problem (6) EXPORT FIXTURE CONFIGURATION AUTOMATIC CONFIGURATION Figure 2.3. Simulation workflow (22)

10 Key Performance Indicators Software documentation: Fixture Analyser and Optimiser 3 BASIC TUTORIAL 3.1 Model definition In this tutorial the reader will study how to setup the main simulation workflow. A 2-part assembly is used as a case-in-point (see Figure 3.1). The 3D model of the fixture is at: Tutorial\Basic Tutorial\Outputs\fixture-3D (basic tutorial.pdf). Figure 3.1. Basis tutorial: initial fixture configuration. Left 2-part assembly. Right assembly fixture. The objectives are as follows: to identify the position of the 4 clamps in such a way the part-to-part gap at mating flanges is between [0.0, 0.4] mm. to maximise the probability to satisfy the part-to-part gap requirements; to compare deterministic optimum vs. stochastic optimum (see Figure 3.2 for a conceptual visualisation of determinist vs. stochastic optimum). Stochastic part variation is simulated by a Gaussian distribution (mean=0.0 mm; standard deviation=0.3 mm). Deterministic optimum Stochastic optimum Key Control Characteristics Design constraints Assembly response function (deterministic) Assembly response function (stochastic) Figure 3.2. Deterministic optimisation vs. stochastic optimisation (22)

11 3.2 Model initialisation MODEL LIBRARY PATH Input data can be find at: Tutorial\Basic Tutorial\Inputs. To open the database go to file> Open >: Tutorial\Basic Tutorial\Simulation files\case study[1] deterministic Tutorial\Basic Tutorial\Simulation files\case study[1] stochastic PRODUCT/PROCESS DATA MATERIAL Parts are made by mild steel (Young Modulus=200 GPa; Poisson ratio=0.3). GEOMETRY (Nominal) The thickness of both parts is 1.0 mm. GEOMETRY (Variation) Geometry variation is emulated by morphed geometry. User-defined inputs are (see Figure 3.3): Influence hull: it envelops the geometry affected by the deformation. It is defined by the location (Pi), the orientation (Ni) and the extension (Ri). Control point: the geometry within the influence hull is deformed by deforming the control point (Pc). The amount of deformation is defined by the user-defined parameters (Pr[1] and Pr[2]). In case of Gaussian kernel, Pr[1] and Pr[2] are mean and standard deviation of the distribution, respectively. Stochastic distribution Pr[1] Ri[z] Main direction (Ni) Influence hull Ri[x] Ri[y] Control point (Pc) Influence hull point (Pi) Figure 3.3. Parameters definition for geometry variation ( morphed geometry). Left parameters definition. Right example of deformation Pc Pr[1] Pr[2] Pi Ni Ri Table 3.1. Parameters used for morphed geometry. STITCH LAYOUT Type Starting Point Ending point Image Linear Circular (22)

12 Linear Circular Circular Circular Table 3.2. Stitch layout. LOCATORS Type Point Image Pin Pin Slot Slot Table 3.3. Pin/Slot layout. Type Point Image NC Block NC Block Table 3.4. NC block layout. Type Point Parametrisation Foor-print size Image Clamp(M) T=[-40, 40] Clamp(M) T=[-40, 40] Clamp(M) T=[-40, 40] Clamp(M) T=[-40, 40] A: 20 B: 20 A: 20 B: 20 A: 20 B: 20 A: 20 B: 20 Table 3.5. Clamp layout (initial layout) (22)

13 PARAMETRISATION Clamps have been parametrised to be movable along the X axis (parallel to the flanges). The parametrisation is controlled through the definition of a local reference frame attached to the clamp (see Figure 3.4). One parameter ( T ) for each clamp has been defined and the translation range has been set to [-40, 40] mm. N vector V vector T vector 3.3 Simulate assembly SOLVER SETTINGS Figure 3.4. Clamp parametrisation and related user-defined reference frame. Solver settings are set as shown in Figure 3.5. If a parallel cluster is available for the current installation, then the assembly simulation takes advantages of multi-cores and multi-processors capability. SET KCCs PARAMETERS Figure 3.5. Solver settings. Assembly simulation is built by defining 4 groups of parameters (one for each clamp). Figure 3.6 shows the model setup (22)

14 3.4 Fixture analysis and synthesis Figure 3.6. Set KCCs parameters. Fixture analysis and synthesis is carried out for given KPIs requirements, as reported in Table 3.6. Stitch Lower gap limit Upper gap limit Acceptance [%] Stitch(1) [70.0, 100.0] Stitch(2) [70.0, 100.0] Stitch(3) [70.0, 100.0] Stitch(4) [70.0, 100.0] Stitch(5) [70.0, 100.0] Stitch(6) [70.0, 100.0] Table 3.6. Tolerance limits on KPIs (stitches) and related acceptance limits. DETERMINISTIC SOLUTION Fixture Analyser and Optimiser offers specific tools to interactively analyse multiple design intents. For example, Figure 3.7 depicts 2 fixture configurations as related to the part-to-part gap of Stitch(5). The reader is now encouraged to try multiple design solution and identify the best fixture configuration as per your expectation (22)

15 Selected solution [2] Selected solution [1] Solution ClampM(1) ClampM(2) ClampM(3) ClampM(4) Part-to-prt Stitch(5) [1] [2] Figure 3.7. Fixture analysis results. Interactive response based on user selection from regression model. Nevertheless, the software has a built-in tool which automatically calculate the optimum fixture configuration for given KPIs requirements (see Figure 3.8 for optimisation settings). Figure 3.8. Optimisation settings and options (22)

16 DETERMINISTIC SOLUTION Figure 3.9 shows the results of the deterministic (only one single instance of product variation) optimisation. The reader is now encouraged to implement different design requirements and to compare among results. ClampM(1) ClampM(2) ClampM(3) ClampM(4) Part-to-prt Stitch(5) Figure 3.9. Fixture synthesis results. Optimum (deterministic) Stitch(5). STOCHASTIC SOLUTION Figure 3.11 shows the results of the deterministic (only one single instance of product variation) optimisation. The reader is now encouraged to implement different design requirements and to compare among results. ClampM(1) ClampM(2) ClampM(3) ClampM(4) Part-to-prt Stitch(5) - Probability of acceptance Figure Fixture synthesis results. Optimum (stochastic) Stitch(5) (22)

17 3.5 Fixture development Simulation outcomes can be transferred to any 3D CAD modeller, where the complete fixture development is designed. The 3D model of the fixture is at: Tutorial\Basic Tutorial\Outputs\fixture- 3D_optimum_layout_deterministic (basic tutorial.pdf). Figure 3.11 compares the optimised fixture development solutions (both deterministic and stochastic). Figure Optimised (deterministic solution) fixture development (left) vs optimised (stochastic solution) fixture development (right) (22)

18 4 ADVANCED TUTORIAL: WING SUB-ASSEMBLY 4.1 Model definition This tutorial deals with the design of an assembly fixture for a wind sub-assembly (see Figure 4.1). The objectives are as follows: Figure 4.1. Advanced tutorial: initial fixture configuration. To optimise the clamp position in such a way the part-to-part gap between the outer skin and the skin reinforcement is [0.0, 1.0] mm; To compare deterministic and stochastic solution. 4.2 Model initialisation MODEL LIBRARY PATH Input data can be found at: Tutorial\ Advanced Tutorial (1)\Inputs. To open the database go to file> Open >: Tutorial\Advanced Tutorial (1)\Simulation files\case study [2] deterministic Tutorial\ Advanced Tutorial (1)\Simulation files\case study[2] stochastic PRODUCT/PROCESS DATA MATERIAL Parts are made by mild steel (Young Modulus=200 GPa; Poisson ratio=0.3). GEOMETRY (Nominal) The thickness of parts being assembled is 1mm. GEOMETRY (Variation) Geometry variation is emulated by morphed geometry. User-defined inputs are collected in Table 4.1. They simulate a deformation pattern as a combination of bending and twisting of the first skin reinforcement (22)

19 Pc Pr[1] Pr[2] Pi Ni Ri Table 4.1. Parameters used for morphed geometry. STITCH LAYOUT Type Starting Point Ending point Image Linear Linear Linear Linear Linear Rigid Link Table 4.2. Stitch layout. LOCATORS Type Point Image Pin Pin Pin Pin Pin Pin Pin Pin Pin Pin pin Pin Table 4.3. Pin/Slot layout. Type Point Parameter Foot-print Image Clamp(M) Reference D: 20 Clamp(M) Reference D: 20 Clamp(M) T=[-50,50] D: (22)

20 Clamp(M) N=[-2,2] D: 20 Clamp(M) T=[-50,50] D: 20 Clamp(M) Reference D: 20 Clamp(M) Reference D: 20 Clamp(M) Reference D: 20 Table 4.4. Clamp layout. In this case study three clamps have been parametrised, two in tangential and one in normal direction. These can be identified from Table 4.4 where the parameter column has values other than reference. Tangential variation is in the interval [-50, 50] and similarly normal variation is between [-2, 2]. 4.3 Fixture analysis and synthesis Fixture analysis and synthesis is carried out for given KPIs requirements, as reported in Table 4.5. Figure 4.2 compares the optimum clamp locations obtained for the deterministic and the stochastic simulation. Stitch Lower gap limit Upper gap limit Acceptance [%] Stitch(1) [70, 100.0] Stitch(2) [70, 100.0] Table 4.5. Tolerance limits on KPIs (stitches) and related acceptance limits. Solution ClampM(1) ClampM(2) ClampM(3) Part-to-part Stitch(1) Deterministic e-11 mm Stochastic % Figure 4.2. Fixture synthesis results. Left deterministic solution; Right stochastic solution (22)

21 APPENDIX A A Quick Introduction to Simulation Builder This Section briefly presents the main menus of the simulation builder. Simulation Builder Settings panel/output Description NA NA PARAMETER PART (GRAPHIC) PART (MODEL) The parameter menu provides design or related experimental information which can be used for setting up the simulation. Acceptable information is in the form of two dimensional arrays. There can be any number of parameters with each parameter containing a m x n matrix. The Part (graphic) menu is used to import parts which help in visualization but are not used for numerical calculations during the simulation. After selecting add option in the Part (Graphic) dropdown menu a new part is created. The Part (Model) menu is used to import parts that are used for numerical calculations during the simulation. This menu also has options for importing cloud of points, deviation data and then generating variational geometry. With the model imported and saved, the geometry has to be built accessing the Build Geometry option from dropdown menu. STITCH The stitch menu has options to create various joining operations such as linear joint, circular joint and rigid link. With the joint parameters set manually or imported from a file, the model has to be saved. Then the geometry has to be built by accessing the Build Geometry option from dropdown menu. PIN LAYOUT (HOLE) PIN LAYOUT (SLOT) (22)

22 Pins are used to constrain the motion of the assembly in given directions. They act as bilateral constraints. With the model imported and saved, the geometry has to be built accessing the Build Geometry option from dropdown menu. LOCATOR REGRESSION VIEWER There are three types of locators (NCBlock, ClampS and ClampM). NCBlock defines a unilateral constraint used to model NC block acting on an individual part. ClampS defines a bilateral clamp acting on an individual part. ClampM defines bilateral clamp acting on two parts, at the same time. An important detail is the geometry of the locator. The shape of any locator can be chosen to be cylindrical, prismatic or L-shaped. With the model imported and saved, the geometry has to be built accessing the Build Geometry option from dropdown menu. This menu offers a high level of automation to optimise KCCs for given KPIs requirements. A regression model is built in 3 steps: (i) model training; (ii) model validation; (iii) optimisation. Model training: KPIs are firstly defined (i.e., point deviation, part-to-part gap, residual stress). Regression is then trained using the KPIs data set. Model validation: a cross validation is implemented to make the model insensitive to the sample size and to overfitting. Optimisation: Optimisation is performed using GA method. Simulation results are graphical visualised and 3D geometry is automatically updated. Results can be exported toward any 3D CAD modeller (22)