2.

Solid Modelling of Castings

The solid model of a cast product is the backbone for various CAD/CAM programs that help in improving the accuracy and speed of different tasks in casting development. In this chapter, we describe various geometric features of castings and different techniques for creating a solid model. We will also review model representation schemes, storage and exchange formats, model verification techniques and the requirements with respect to casting applications. 2.1 Casting Features Casting process enables economical manufacture of products with intricate geometry to near net shape. Components having complex curved surfaces, including blends between sculptured surfaces, intricately shaped pockets, and meandering holes besides features such as bosses, slots, steps and ribs can be produced. Several geometric elements or features can be economically combined together in a single casting, which otherwise require joining or assembly of multiple parts. A geometric feature is defined as a region of interest considering design or manufacture. Casting features, related to product and tooling, may be classified as follows: Base feature: This defines the overall shape of the casting. It can be expressed in terms of the minimum-volume shape that completely encloses the casting after removing small projections (like bosses) and filling up small depressions (like holes). The shape can be a rectangular block, cylinder, sphere, hemisphere, torus, spiral, L-bracket, wheel, etc. It can be solid or hollow. Local feature: This can be a depression (hole, pocket, slot, etc.) or a protrusion (boss, rib, etc.) on the base feature. The feature cross-section may be simple (circle, rectangle) or complex (set of lines defining a closed loop). It may remain constant or vary along the axis of the feature (ex. ellipse becoming a circle). The axis of the feature may or may not be perpendicular to the base surface. Further, the axis may be straight or curved. Tooling feature: This includes different elements of tooling and their features: pattern, mould, core box, undercut, parting line, core support, ingate, runner, sprue, feeder, feeder neck and various feedaids (chill, insulation and exothermic). Many of these, including pattern, mould and core box can be derived from the part geometry. Connecting feature and modifier: Connecting features are those in between adjacent features or those connecting two or more features, such as taper, chamfer, fillet and wall between two holes. Feature modifier implies a minor modification to the standard shape of a single feature using draft, chamfer, fillet, etc.

2.2 Modelling Techniques The two most common approaches to solid modelling include extrusion of a 2D section and constructive solid geometry. Another approach, called features-based modelling, is also available in most solid modelling systems today. These techniques are useful to define the base feature of a casting as well as its local features. Extrusion or sweep: This technique involves sketching a cross-section on a plane and sweeping it through a straight or curved path. The cross-section can have inner loops, leading to hollow shapes. Depending on the path, we can have straight extruded shapes and solids of revolution. For straight extrusion, the distance has to be provided. Revolution can be along the complete circle or a part of it. It is also possible to have negative extrusion, referred to as cut extrude, which produces depression features in an existing solid. For example, a cut extrude circle produces a hole.

Constructive Solid Geometry: This technique uses Boolean operations: union, intersection and difference on pairs of simpler models to create the desired shape (Fig.2.1). A library of solid primitives such as cube, cylinder, sphere, cone and torus are provided to initiate the modelling. Each of these is instantiated by supplying the values of its dimensions, location and orientation. The model created by extrusion or by combining primitive solids is then combined with other primitives or other solids to eventually obtain the designed shape of the component. The final as well as intermediate models are represented in a CSG tree that shows the primitives and steps used for making the model.

Fig.2.1: Results of Boolean operations on a cube and sphere. Left: union, middle: intersection, right: difference.

Fig.2.2: A modern solid modelling system has graphical icons, context sensitive menus, features tree (can be expanded or collapsed) and shaded display of solid model. Features-based modelling: Standard features include hole, pocket, slot, boss, rib and fillet. These can be modelled by Boolean operations; for example a hole can be produced by subtracting a cylinder from the base solid. Features-based modelling makes this operation more intuitive to engineers by adding a hole feature (instead of subtracting a cylinder), which is essentially a negative cylinder. The user selects a face (on the base feature) or a set of edges/faces (as in the case of a fillet), and specifies the feature parameters. The features tree is preserved, so that the user can easily move, modify or delete a specific feature. Special techniques: This includes analytical surfaces, lofting and shelling. Analytical surfaces include Coon, Bezier, B-spline or non-uniform rational B-spline. These are useful for defining intricately curved surfaces. Lofting involves defining a number of cross-sections normal to a straight or curved path, and joining them using an analytical surface. Shelling involves creating an offset surface, usually inside, and removing the inside part to obtain a shell. Most of the recent systems also support parametric modelling. The user can specify relationships between geometric elements (for example, the distance of a hole from a given rib). The software automatically re-evaluates these relationships when the relevant parameters are modified (for example, hole diameter or rib thickness), and positions the feature appropriately. This also implies that standard shapes of castings (such as gear wheels and crankshafts) as well as tooling features (such as feeders) can be created in

terms of a few selected parameters. The appropriate shape can be automatically created by supplying the values of the parameters involved. The majority of castings require a combination of all techniques. For example, the overall shape of a cylinder block or exhaust manifold is created by sweeping a few sections, then combining them with other primitives using Boolean operations, followed by subtracting the volume corresponding to the internal ducts, which are themselves modelled by complex sweep and CSG techniques. Local features like holes, bosses and ribs are modelled next. Feature modifiers, such as fillets and taper are applied last. The designer requires training and experience in deciding the strategy for modelling, in particular, combinations of primitive solids that will lead to the final shape in the least number of steps. Sometimes it becomes necessary to retrace the steps during modelling and take a different approach to complete the shape. Evan an experienced designer may take several days for creating the model of a complex casting. The benefits of downstream applications however, amply justify the initial modelling effort.

Fig.2.3: Sculpting in virtual space using a robotic hand 2.3 Graphical User Interface Most of the solid modelling software available today use a standard geometric kernel that performs the calculations necessary for generating and displaying the solid model. The main difference lies in the graphical user interface that allows the user to command the software, give necessary inputs and view the results and messages. Typical commands include opening a new or existing model, setting the coordinate axes and sketching planes, modelling or modifying the model using various functions (extrusion, Boolean and features-based), viewing the model in different ways (transparent, shaded, orthographic, isometric, close up, etc.), dimensional measurements, geometric property calculations and finally saving the model file in an appropriate format useful for other applications also.

Depending on the operating system, solid modelling software can have a Unix-based or Windows-based user interface. Earlier solid modelling programs were developed in Unix environment, but the newer ones are developed in Windows environment owing to its user friendliness and wider application. The user can communicate with the program through command lines, pull-down menus or icon buttons. Command lines provided the only interface in earlier modelling programs developed in DOS and Unix operating systems. This required memorizing the necessary commands, which is difficult for new or occasional users. Pull-down menus organize various commands in different categories (such as file management, Boolean operations, features-based modelling and model display). Each category may have its own cascading pull-down menu. Icon-based interface comprises a number of buttons with pictures; clicking the mouse over the button executes the particular function. This is faster than selecting a command from a pull-down menu, but the user has to remember the meaning of various icons. Many modern solid modelling programs provide all three types of interfaces (command line, pull-down menus and icon buttons) to suite different people. During solid modelling, the user can change the display of the model by panning (moving vertically and horizontally), rotating (about different axes) or zooming (close-up or distance view). The model can be shown as a wireframe, flat shaded or illuminated object, which take progressively more computation power or time, especially if the model has many surfaces and features. Advanced modellers include reflections, transparency, shadows, illumination from multiple light sources and texturing effects (depending on part material) to create photo-realistic images of the modelled object. Indeed, the computer-generated image of a virtual part may be difficult to distinguish from a photograph of a real part. The model can also be dissected in different locations and the cross-sections can be viewed, useful for parts with internal features. All these functions facilitate understanding the geometry of even intricate parts (such as an engine block) with little effort compared to that required for building up a mental image of the part from its 2D drawings. The user input as well as program output of the current generation of solid modelling programs are through 2D interfaces: mouse moving over a flat surface and a computer display, respectively. One step forward is to use a 3D input device (3D mouse, joystick or robotic arm). Some of these, such as the FreeForm system (Sensable, Inc.) also provide a haptic sensation that allows the user to touch and feel the virtual model. The final goal is a 3D interface for both input as well as output, which is achieved by virtual reality systems. These comprise a data glove, which is worn by the user to move the hand in real space, and stereoscopic goggles, which project two slightly different images into the two eyes to provide the illusion of 3D space. If a physical part already exists, then its geometry can be captured using reverse engineering techniques. This involves moving a contact probe or a non-contact sensor (optic or laser-based) over the part surface and getting the (x, y, z) coordinates of a large

number of points. Then surfaces are mathematically fitted through the cloud of points and stitched together to obtain a solid model. 2.4 Model Representation A complete and unambiguous representation of the solid model is crucial to its visualization, property computation and other downstream applications. Three major methods include the CSG/features tree, Boundary Representation and Cell Enumeration. CSG or Features Tree: This stores a solid model in terms of nodes representing solids and junctions representing operations. The starting nodes represent the initial primitives, whereas the end node defines the final model. Thus the CSG tree not only describes the solid model but also the way in which the model was created. Each node (representing an intermediate model or feature) also stores the geometric parameters of the model or feature. This is useful for quickly viewing and modifying the feature parameters. By itself however, the CSG tree cannot be used for viewing or analysing the solid model; it has to be converted into a Boundary Representation for these purposes. Boundary Representation: This defines a solid in terms of a set of contiguous faces that completely enclose the part volume. This implies that from a given starting face, any other face (including faces of internal features) can be reached by traversing through intermediate faces. Each face is defined by its bounding edges and each edge is defined by its bounding vertices. Geometric information regarding the coefficients of the plane of each face, direction cosines of each edge and coordinates of each vertex are stored. Topological information such as adjacent faces of each edge and edges emanating from each vertex is also stored to speed up computations. Curved surfaces may be decomposed into several flat facets for facilitating display and analysis. This is referred to as a facetted model. Cell Enumeration: This involves decomposing the solid into a number of cubic cells (called volume elements or voxels) and storing the location of each cell. Inclined and curved surfaces cannot be represented accurately, because of the staircase effect. The error can be reduced by smaller size of voxels, but this increases the requirement of storage space and computation time. A similar approach, in which the solid is progressively decomposed into smaller elements depending on its shape and accuracy required, is called Octree enumeration. This is more efficient in terms of storage space. 2.4 Model Exchange Formats Every CAD/CAM system uses a proprietary format to store the solid models internally, which may be quite different from other systems. The internal format primarily depends on the geometric kernel, which in turn may be proprietary or public (such as ACIS, PARASOLIDS and CASCADE). To facilitate exchange of solid models between two or more systems, the models are exported and imported in widely agreed upon standard formats. Several formats are

available today. One of the earliest de-facto formats was DXF used in AutoCAD program (Autodesk, Inc.). A few other formats have been promoted by industry consortiums or countries (for example, VDAF in Germany). Currently, the most widely used exchange formats include IGES, STL and STEP, described here. IGES: It stands for Initial Graphics Exchange Specification. It was primarily meant for surfaces, though the latest versions of IGES fully support solid models using Boundary Representation. It is the most widely used format for models with intricate contoured surfaces. |An IGES format file essentially contains an analytic representation of surface patches belong to the part. The accurate definition of surface is necessary for numerically-controlled (NC) manufacture of tooling (patterns, core boxes and moulds). However a few major CAD/CAM system developers have extended IGES in different ways, leading to minor losses of data and errors during exchange between systems. However, even a minor loss makes the file useless for NC manufacturing. Patching up or fixing the errors in an IGES file imported from a different system requires considerable effort, leading to loss of productivity. STL: It stands for Stereolithography Tessellated Language, which originated from rapid prototyping systems. It stores the Boundary Representation of a solid model in terms of triangular facets. Thus curved surfaces have to be decomposed into a number of facets by the CAD system before exporting. Higher faceting gives a more accurate surface, but increases the memory and computation requirement. The degree of faceting can be controlled during file export by specifying the extent of deviation from the original surface. The STL file essentially contains the (x, y, z) coordinates of vertices belong to each triangular facet. It is a well defined and easy to handle format, and has therefore quickly gained wide support in CAD/CAM community. It is the preferred format for visualization and analysis programs, since these do not require accurate surface data. It is however, not suitable for NC manufacture, particularly with coarse faceting. STEP: The Standard for Exchange of Product data is proposed and managed by International Standards Organization. The STEP handles not only geometric data related to a product, but also non-geometric information such as bill of materials, physical location of database, conformance testing procedures, display attributes, application protocols and data required by specific applications. A protocol for casting application (ISO 10303-223: Exchange of design and manufacturing product information for cast parts) has been proposed by ISO and is in the process of ratification. The protocol includes design data for cast parts (geometry, materials, tolerances, required physical and mechanical properties, and required tests); process plans for parts made by sand, die and investment casting processes; characterisation of products used to make cast parts (moulds, dies, equipment, materials and consumable items); and specifications for patterns and die assemblies. The proposed standard also includes input to and output from casting process simulation software; data exchange between customer and foundry, within the foundry, and between the foundry and supplier; use of data for foundry automation and shop floor control; and use of data for archival of design and manufacturing data. The STEP is expected to standardize product data management and eliminate the need for developing translators and formatters for different applications.

Most solid modelling systems maintain two or more representation schemes, and support several model formats for importing and exporting. In general, the CSG/features representation is useful for modifying specific features and retracing the steps during modelling, whereas the Boundary Representation is useful for visualization, geometric property computation and analysis. However, the specific format for storing CSG or Boundary information varies between different modelling systems. It is therefore not surprising for translation errors to crop up when a solid model is exchanged between two or more different systems, especially for parts having intricate curved surfaces with fillets. 2.5 Model Verification and Requirements The solid model has to be verified for its conformance to the actual shape being analysed or manufactured. Errors may be introduced during modelling, exporting or importing. The STL format files can have errors such as missing facets and edges, usually in small fillets at internal corners. It is therefore necessary to verify the model after exporting from one program and importing into another program. Model verification methods include visualization, dimensional measurements, property computation, thickness check, simulated radiography and Eulers equation. Model visualisation and measurements: The model can be viewed as an orthographic, isometric, cross-sectional, hidden line or shaded image. It can also be panned, rotated and zoomed to visually verify all surfaces. Newer solid modelling programs have automated dimensioning facility driven by the part geometry, making it easy to check all key dimensions. In addition, check points can be clicked on the part surface and the distance between them can be measured for verification. It is also possible to check the angle between two edges and between two faces. Geometric properties include volume, weight, centre of gravity and moments of inertia. The comparison of computed weight with the actual weight of the part is a widely used check to catch gross errors in modelling. This however, requires accurate modelling of even minute features such as fillets and tapers. The actual part must be free of manufacturing defects and the correct value of average density of the part must be available. A volume comparison is more reliable.

mm

Fig.2.4: Model verification using simulated radiography (left) and thickness map (right). Thickness check: This is very useful for models with intricate surfaces and internal features, which are more susceptible to modelling errors. This involves generating model cross-sections and interactively measuring the distance between two points specified on the opposite ends of any wall. More sophisticated approaches include producing a colourcoded thickness map of the entire model (Fig.2.4). Simulated radiography: This is very useful for identifying errors such as missing facets in STL files, which can lead to problems during feature recognition and process simulation. In this method, parallel rays are passed through the model in a specified direction, and the model thickness along a ray is indicated by a grey shade (Fig.2.4). Any unexpected streaks of lines in the radiography image indicate errors in the model. The distance between the rays influences the size of the error that can be identified. Eulers equation is valid for facetted solid models, and is given by: V+F = E+2, where V, E and F are the number of vertices, edges and faces, respectively. The check must be available in the solid modeller (or the program which imports an STL file). 2.6 Application Requirements Major casting software applications and their requirements with respect to the part model are listed below. Casting design and analysis: A facetted model of the product in STL format is sufficient, and is indeed preferred owing to its simplicity and robustness. This format facilitates the generation of mesh required for numerical simulation of stress analysis, mould filling and casting solidification. If the model contains very small fillets, these may be eliminated before exporting the STL file to reduce its size, which improves the speed of subsequent analysis. This also minimizes errors (such as missing facets and dangling edges) in the solid model. Another way to reduce the file size is to use the STL Binary format, which is typically half the size of STL ASCII (or text) format. NC manufacture of tooling: A complete and accurate description of the casting surface is required for generating the numerically controlled (NC) cutter tool paths using a computer-aided manufacturing (CAM) software. The IGES format file is suitable for this purpose. This enables accurate manufacture of the tooling (pattern, core box, mould) using an NC machine and also subsequent inspection using a coordinate measuring machine (CMM). However, there are several versions of IGES, and also the files

generated by different CAD systems may not be exactly the same, creating compatibility problems. A preferred solution is to use the CAD and CAM programs from the same developer, so that data translation errors are eliminated. Manufacturing resource planning: For materials planning, cost estimation and other planning tasks, non-geometric information related to the casting is also required. This can be best handled by the STEP standard for casting, after it is widely accepted. Web-based collaborative engineering: In near future, castings will be collaboratively developed by product, tooling and foundry engineers (who may be located anywhere in the world) by exchanging relevant information with each other over the Internet. This will require web-friendly standards for handling casting life-cycle information.

10