Chapter Five. Modeling and Design

The process of refining and analyzing the design is iterative. As the refined design is tested, the test results suggest modifications to the design, and once the design is changed, it needs further analysis.

Creating a CAD model is an important step in the refinement process. A huge amount of design information can be stored in CAD models that are easily modified. The CAD model can be used to test and evaluate the design. By making changes easy—and in some cases, automating the update process—CAD modeling can eliminate some barriers to a thorough review and refinement of the design.

CAD models also encourage refinement by making it easier and less costly to get feedback about a design while it is in development. Accurate part descriptions allow you to test how well parts will fit together. Customer acceptance and styling issues can be assessed early in the design process using realistically shaded models. Analysis programs can import 3D CAD files or run as part of the CAD software, saving much time. Some analysis packages suggest optimizations to incorporate directly in the CAD model. Animation and kinematic analysis uses CAD data to determine whether the design will function properly before any parts are actually built.

Many different methods are used to create CAD models, each with its strengths and weaknesses for capturing design information. Each item shown in Figure 5.1 may be modeled differently, possibly requiring different software and techniques.

In this chapter, you will learn how different modeling methods represent and store design information, and the advantages and disadvantages of each. This information will aid you in choosing the most effective method for a particular design problem.

A set of written specifications for a design is a descriptive model. If all the specifications are followed, the system will perform correctly. Sketching is also a type of descriptive model for your design ideas on paper. 2D and 3D CAD drawings are also descriptive models. A physical model or prototype is another type of descriptive model, although sometimes physical models are made to a smaller scale (called a scale model).

The figures at right show a scale model of a design for the Next Generation Space Telescope (NGST) developed by NASA. A model helps you visualize the design. It has the added benefit of being easy for non-engineers to understand while decisions are being made about the next step in the project.

Part of creating an effective analytical model is determining which aspects of the system’s behavior to model. In the circuit model, some information, such as interference between some components, is left out because it is too complex to represent effectively. A finite element analysis (FEA) model, such as that used to generate the stress plot shown in Figure 5.3a, simplifies the CAD model in a similar way. The FEA model breaks the model into smaller elements; reducing a complicated system to a series of smaller systems allows the stresses more easily to be solved. Understanding and using analytical models effectively requires knowing how the model differs from the actual system so the results can be interpreted correctly.

In the design process, use different types of models where they are appropriate. During the ideation phase of the design, sketching is often the best technique to use. As you begin to refine the design, you will use both descriptive and analytical models to represent the design more accurately and to provide insight into its behavior.

A 3D CAD model combines qualities of descriptive and analytical models. Because a 3D CAD model accurately depicts the geometry of a device, it can fully describe its shape, size, and appearance as a physical or scale model would. Additional information about the final product, such as the materials from which it will be made, can also be added to the model description stored in the computer database. Figure 5.4 shows a CAD model of NASA’s Space Infrared Telescope Facility (SIRTF).

A 3D CAD model can also be used for analytical modeling. The 3D model of the device can be used to study its characteristics (see Figure 5.5). Sophisticated software allows you to analyze, animate, and predict the behavior of the design under various physical conditions. The 3D CAD model is a detailed representation of the final object, suited for testing and study.

Multiview drawing techniques were designed to improve the robustness, measurability, and accuracy of paper drawings. These same techniques require some skill to be understood, however, and make multiview drawings harder to understand than a 3D model.

Equipment costs for paper drawings are minimal, but they take as long or longer to create than CAD drawings. Changes involve considerable erasing and redrawing, making them difficult to modify as the design changes. Because paper drawings are difficult to change, the labor costs associated with them usually outweigh the equipment savings.

Paper drawing accuracy is about plus or minus one fortieth the drawing scale. For example, a paper drawing of a map drawn at a scale of 1 inch = 400 feet has good accuracy if you can measure from it to plus or minus 10 feet. This makes paper drawings not particularly measurable or accurate—which is why proper dimensioning technique was developed. Proper dimensioning overcomes this limitation and provides measurements to the desired accuracy on the drawing. It is not good practice to make measurements from paper drawings.

Paper drawings can be very effective for quickly communicating the design of small parts for manufacturing. A properly dimensioned sketch can quickly convey all the information needed to make a part that is needed only one time (see Figure 5.7). Also, manufacturing facilities occasionally are not able to read electronic files and require paper drawings.

2D CAD drawings can quickly be printed to different scales. Different types of information can be separated onto different layers that can either be displayed or turned off, making the model more flexible than a paper drawing.

Using CAD, you can accurately define the locations of lines, arcs, and other geometry. In AutoCAD, for example, you can store these locations to at least fourteen decimal places. You can query the database, and the information will be returned as accurately as you originally created it. Of course, the adage Garbage In, Garbage Out (GIGO) applies. If your endpoints do not connect, or you locate distances by eye instead of entering them precisely, you will not have accurate results.

CAD models represent the object full size, unlike paper drawings, so you can make measurements and calculations from 2D CAD models. Also, you can “snap” to locations on objects to determine sizes and distances that may not be dimensioned on a paper drawing. If there must be a clearance of 10 feet from the center of a tank to the location of another piece of equipment, you can measure from the CAD file to determine whether proper clearance is provided in the design.

Text and other information can be saved in a database and linked to the drawing information for later retrieval. The 2D database is limited, however, in its ability to represent information that depends on a 3D definition (such as the volume inside the object), so it is not generally useful for determining mass and other physical properties of the object.

The strengths and weaknesses of paper drawing are shared by 2D CAD models. To create them and read them you must be able to interpret the 2D views to “see” the 3D object. To measure from a 2D CAD model, you must show distances and angles true length in a view. Orthographic projection and descriptive geometry are needed to create these views. Descriptive geometry is the study of producing views of an object that show true lengths, shape, angles, and other information about an engineering design. Mastering these subjects enables you to use auxiliary views created with a 2D CAD system along with the traditional methods of descriptive geometry to solve many engineering problems.

Flexibility and accuracy make 2D CAD systems a cost-effective tool in a wide variety of businesses. For many civil engineering projects, the difficulty of capturing all the 3D information needed to model irregular surfaces, such as terrain, may not be worth the benefits of working in 3D. For projects such as highway design, civil mapping, electrical distribution, and building systems, 2D CAD models may provide enough information and be created more quickly from the information at hand (see Figure 5.8). Areas and perimeters can be calculated accurately, and drawings can be revised quickly. 2D CAD models are more accurate than paper drawings.

Relationships defined between parts of the 2D model are maintained by the software when you make changes to the drawing. Geometric constraints can be a valuable tool, but to benefit from them, you must apply them with a good understanding of basic drawing geometry and the behavior desired when changes are made to the shape.

The robustness and cost effectiveness of physical models are often linked. A simple model made of clay or cardboard may be enough for some purposes (see Figure 5.9), but determining the fit of many parts in a large-scale assembly, or producing a model realistic enough for market testing, may be cost prohibitive until very late in the design process. A full-size prototype of the BMW 850I cost more than $1 million to produce. The amount of information and detail built into the model adds to its cost.

The accuracy of a physical prototype also drives up its cost. Many objects are designed to be mass produced so that the cost of individual parts is reduced. When each part in the model must be produced one at a time, the cost of the prototype can be many times greater than the manufacturing cost of the final product. The more the model must match the final product in terms of materials used and final appearance, the greater the cost. Rapid prototyping systems offer quick and relatively inexpensive means of generating a physical model of smaller parts without the cost of machining or forming parts one at a time (see Figure 5.10).

Physical models are a very good visual representation of the design, but if they are not made from the materials that will be selected for the design, their weight and other characteristics will not match the final product. Sometimes, owing to the size of the project, the physical model must be made to a smaller scale than the final design. The accuracy of the final part also may not be possible with the materials and processes used for the prototype.

Probably the least attractive feature of a physical prototype is its lack of flexibility. Once a physical prototype has been created, changing it is expensive, difficult, and time-consuming. Consequently, full-size physical models are not usually used until fairly late in the design process—when major design changes are less likely. This can limit the usefulness of the model even though it provides important feedback about items that are not working well. When the information comes late in the design process, it can be too expensive to return to a much earlier stage of the design to pursue a different approach. Only critical problems may be fixed. Solutions to problems found late in the process are generally constrained to cause the least amount of redesign while fixing the problem. When a prototype is created late in the process, there may not be time or resources to create a new prototype for the new solution. New problems introduced by the change may not be seen until actual parts are produced.

Despite the trade-offs, even very costly physical models have been a cost-effective way for companies to avoid much more costly errors in manufacturing.

Some 3D interfaces, such as a 3D mouse or controller, let the user interact with the items in the 3D model in a way that enhances the illusion of reality. Some gloves (similar to those shown in Figure 5.13) even use model data to provide physical feedback to the user when an object is encountered. If a virtual object is squeezed, feedback to the glove provides the feeling of the resistance a solid object would give when squeezed. Systems may even interpret how much force would crush the object and provide this sensation to the user.

The term virtual prototype describes 3D CAD systems that represent the object realistically enough for users, designers, and manufacturers to get the same type of information they get from creating a physical prototype.

As more sophisticated software becomes available to analyze, animate, and predict the behavior of the design under various physical conditions, the 3D CAD model becomes a better representation of the final object—and better suited for testing and study.

3D CAD models offer a high degree of accuracy and measurability and a high degree of flexibility. Each new generation of CAD software automates more common tasks, allowing the designer to focus more on the design and less on the mechanics of changing the model. This flexibility allows the CAD model to become a preliminary work that is refined and tested until it is ready to be used as a plan for the final product.

The ability of 3D CAD models to be used throughout the process, serve in lieu of a physical model, and be reused and modified indefinitely makes them very cost effective, especially in designing mechanical assemblies. However, not all 3D CAD models are the same. Familiarity with different types of 3D modeling systems lets you select the method and software most suitable for your design project.

You create a wireframe model in much the same way that you create a 2D CAD drawing. Each edge where surfaces on the object intersect is drawn in 3D space using simple geometric tools, such as a 3D line or arc. The X-, Y-, and Z-coordinates for the endpoints are stored in the database along with the type of entity.

3D wireframes can quickly display process control, piping, sheet metal parts, mechanical linkages, and layout drawings showing critical dimensions that use relatively simple geometric shapes. Interferences and clearance distances are difficult to visualize in 2D piping drawings. It can be critical that they are not overlooked in the design. For example, to quickly check clearances for the design of an electrical substation, the engineer may create a wireframe model of the high-voltage conductors to ensure that conductors are not too close to other equipment that could produce an electrical arc, causing a short.

A 3D wireframe can be useful for piping layout around other 3D equipment, which can be difficult to visualize in 2D. The centerline of the pipe is easy to represent in a 3D wireframe, as shown in Figure 5.14. This prevents problems with clearances that are acceptable in two dimensions, but not in a third. For a system with moving parts or where complex shapes need to fit, a different 3D modeling method may better allow you to assess interferences.

Wireframe models do not include surfaces that can be shaded, so they are not realistic looking. Because you can see through the model, some shapes cannot be represented unambiguously, and it may be difficult to see from a single view which areas are holes and which are surfaces. The model in Figure 5.15 shows a wireframe model. Without rotating or checking coordinate locations, you may visualize the part as showing the hole from the top or the bottom (Figure 5.16).

Before you can measure oblique edges and surfaces in a 2D drawing, you must first create an auxiliary view showing that feature true size. The true size is already contained in the 3D wireframe and can be measured directly. For example, the length of edge A in Figure 5.17 can be determined from this 2D AutoCAD drawing only in the auxiliary view. The 3D wireframe model of the same part can show the length of the edge directly.

Surface modelers are traditionally used by industrial designers and stylists who create the outside envelope for a device. These designers need modeling tools that convey a realistic image of the final product. Software tools developed for their use focused first on the challenge of modeling the exterior of the product, not the mechanical workings within. Many surface modelers offer specialized tools for lighting and rendering a model, so designers can create photorealistic images that cannot be distinguished from the actual product (see Figure 5.18).

Concurrent engineering and the use of the 3D CAD database as the product description contributed to the development of modeling software that satisfies the aesthetic needs of industrial designers as well as the functional information needed by engineers. Understanding surface modeling techniques will help you assess the kinds of features that are available in your modeler—and whether a dedicated surface modeler is required.

Storing the definitions of the surfaces in the CAD digital database is termed boundary representation (BREP), meaning that the information contained in the database represents the external boundaries of the surfaces making up the 3D model. Remember that these surfaces have no thickness.

The following are basic methods used to create surface models:

• Extrusion and revolution

• Meshes

• Spline approximations

Figure 5.19 shows a surface model created by revolution, and the profile and axis of revolution used to generate it. Regular geometric entities can be revolved or extruded to create surface primitives such as cones, cylinders, and planes. Surface primitives are sometimes built into surface modeling software to be used as building blocks for complex shapes.

Some mesh surfaces are called triangulated irregular networks (TINs), because they connect sets of three vertices with triangular faces to serve as the surface model. A mesh surface can be useful for modeling uneven surfaces, such as terrain, where a completely smooth surface is not necessary. A sufficiently large matrix will result in correspondingly smaller triangles that can better approximate a smooth surface. Refining the mesh will produce a smoother surface, but the size of the CAD file will increase and result in slower software performance. Some modeling packages allow you to define a smoothed representation of the surface instead of the somewhat bumpy appearance of a strictly mesh surface.

Rational curves and surfaces have the advantage that they can be used to generate not only free-form curves but also analytical forms such as arcs, lines, cylinders, and planes. This is an advantage for surface modelers that use NURBS techniques, as the database does not need to accommodate different techniques for surfaces created using a surface primitive, mesh, extrusion, or revolution.

Spline curves can be used as input for revolved and extruded surfaces that can be lofted or swept. Lofting is the term used to describe a surface fit to a series of curves that do not cross each other. Sweeping creates a surface by sweeping a curve or cross section along one or more “paths.” In both cases, the surface blends from the shape of one curve to the next (see Figure 5.21).

NURBS surfaces can also be created by meshing curves that run perpendicular to each other, as illustrated in Figure 5.22.

A Coon’s patch is a simple interpolated surface that is bounded by four curves. Mathematical methods interpolate the points on the four boundary curves to determine the vertices of the resulting patch (see Figure 5.25).

Surface patches are joined by blending the edges of the patches. Areas created by blending, or the surfaces created by fillets, corners, or offsets, are called derived surfaces. They are defined by mathematical methods that combine the edges of the patches to create a smooth joint.

Sometimes, complex surface patches are created by trimming. For example, a circular patch may start with a rectangular patch, then be trimmed to a circle, and finally be blended with other surface patches.

Some surface modeling systems include the use of Boolean operations, whereas others do not. Systems that do not include Boolean operations or good tools for trimming surfaces can be difficult to use to create a feature such as a round hole through a curved surface because the exact shape of the surface, including the hole, must be defined by specifying its edges.

For NURBS surfaces, each control point can be edited to produce local changes in the surface model. (Editing Bezier surfaces, by the same token, produces global changes to the surface.) Grabbing and relocating vertices is a highly intuitive way to edit a surface model that contributes to its usefulness in refinement. The term “tweaking” is used to describe editing a model by adjusting control points individually to see the result. For example, the NURBS surfaces used to define the shape of the Predator’s exterior made it easy to use the surface model to explore design options in several areas (Figure 5.26).

Surface models used for computer-aided manufacturing (CAM) require a high degree of accuracy to produce smooth surfaces. The trade-off between speed and accuracy may be resolved by setting the surface display to a less accurate faceted representation and storing a smoothed or more highly defined surface definition in the database. It is important to distinguish between the screen display of the model and the surface definition stored in the database.

Splines used to create surfaces can also vary in accuracy. Splines can have more or fewer control points that you can use to shape the surface. To model surfaces smoothly, fewer control points may allow more fluid curves. A mesh surface used to calculate area may not be as accurate as one created using a smoothing algorithm such as NURBS or Bezier.

Tessellation lines are used to indicate surfaces in a wireframe view, and they may or may not reflect the accuracy of the surface (see Figure 5.28). It is important to distinguish when tessellation lines represent the accuracy of the view of the model, and when they represent the accuracy of the model stored in the database. If the software accurately represents the actual geometry of the model in the model database, the tessellations simply represent the model in the on-screen view.

Because the relative locations of surfaces from a particular direction of sight can be calculated, surface modeling systems can automatically remove back edges (or represent them as hidden lines). Surface definitions remove the ambiguity inherent in some wireframe models and allow you to see holes and front surfaces by hiding the nonvisible parts of the model.

The complex surfaces defined by a surface model can be exported to numerically controlled machines, making it possible to manufacture irregular shapes that would be difficult to document consistently in 2D views. Both surface and solid models can be used to check for fit and interference before the product is manufactured. Often, surface models can be converted to solid models and vice versa.

Because surface models define surfaces, they can often report the surface area of a part. This information can be useful in calculating heat transfer rates, for example, and can save time, particularly when the surface is complex. The accuracy of the calculations may depend on the method used by the software to store the surface data.

Complex surfaces can be difficult to model. The cost-effectiveness of surface modeling depends on the difficulty of the surface, the accuracy required, and the purpose for which the model will be used. Many modeling software systems offer a combination of wireframe, surface, and solid modeling capabilities, making it possible to weigh the benefits against the difficulties and the time required to create each type of model.

In addition, 3D solid models are particularly suited for use with analysis packages. They contain all the information about an object’s volume, and volume is important to many engineering calculations. Mass properties, centroid, moments of inertia, and weight can be calculated from the solid model as often as needed during the refinement process. The system behavior can be simulated from the wealth of information available in the model.

Solid models provide information needed for other analyses. Finite element analysis (FEA) methods break up a complex object into smaller shapes to make stress, strain, and heat transfer easier to calculate. Because the solid model defines the entire object, FEA software can use this information to generate an FEA mesh automatically for a part. This allows inclusion of this type of analysis earlier in the design process, incorporation of changes, and testing of the new model with FEA analysis again. Some FEA and solid modeling packages are integrated to provide this analysis within the modeling software. Others even allow direct model optimization based on the FEA results, creating a new version of the model that the designer can then revise further.

Neptune Seatech Oy made the decision to switch from a 2D design method to a constraint-based 3D modeler, SolidWorks, so it could get its products to market faster while preserving the iterations needed for complex, accurate designs. Neptune Seatech is a Finnish engineering company that specializes in the design of vehicles for subsea applications—passenger and research submarines. To meet increasingly rigorous design, performance, and safety standards, the company found 2D methods made it hard to complete the desired number of design iterations and still meet product development deadlines. Its first project using the constraint-based 3D modeler, a floating bridge for ferries, took only 18 weeks to produce (see Figure 5.32).

The company enjoyed two other benefits of constraint-based modeling. First, the ease with which constraint-based models can be updated makes it possible to create families of designs. Neptune’s floating bridge allows cars to pass from the ferry to land and vice versa during loading and unloading. Because each landing site is slightly different, the main dimensions of a given bridge vary slightly from one site to another. The bridge manufacturer that hired Neptune had several designs previously created for specific landing sites. To eliminate confusion in manufacturing, the customer wanted a design that could be updated for new bridges. Neptune proposed to design one bridge, then produce the others by changing the length and width factors. The constraint-based model of the bridge made it easy to change the pertinent dimensions and have the software update the related parts to a new size. The investment in the bridge model was used to refine the original bridge design, then used again as a starting point for variations in the product family (see Figure 5.33).

Neptune also enjoyed the software’s ability to analyze mass properties. The weight and volume data for a floating bridge need to be evaluated during design so that the resulting bridge floats at the correct level. Calculating the weight of a floating bridge design would have taken 1 to 2 weeks using Neptune’s previous methods, but the built-in capabilities of SolidWorks made it possible to monitor this data throughout the design process. The ease with which the model could be modified made it possible to analyze the model, make changes, and analyze the model again. The software also made it possible for analysis to occur earlier in the design process, allowing more time to optimize the design.

Constraint-based modeling also improves designs by focusing the modeler on the design intent for the product. If a constraint-based model is to be updated successfully when changes are made, the rules for building the model must capture the intended design solution for the system or device. The extra attention and planning required to capture design intent in the model makes the designer carefully consider the function and purpose of the item being designed, which in turn results in better designs.

• Size constraints are the dimensions that define the model. The choice of dimensions and their placement is an important aspect of capturing the design intent.

• Geometric constraints define and maintain the geometric properties of an object, such as tangency, verticality, and so on. These geometric constraints are equally important in capturing design intent.

In constraint-based modeling, a parameter is a named quantity whose value can change; it is similar to a variable. Like a variable, a parameter can be used to define other parameters. If the length of a part is always twice the width, you can define its size as “2 × Width,” where Width is the name of the dimension.

Unlike a variable, a parameter is never abstract: it always has a value assigned to it, so that the model can be represented. For example, the parameter Width is defined by a value, such as 3.5. The value for the length parameter is calculated based on the Width, so in this case Length is 7. If the width value is changed to 5, the length is automatically updated to the new value of “Width * 2,” or 10 (see Figure 5.34).

Even something as simple as this rectangular block may have design intent built into its model. As it is being updated, do you want the center of the part to stay fixed and the part to elongate in both directions, or do you want it to lengthen all in one direction? If so which end should stay put?

Figure 5.35 shows a part that has been modeled in a 3D constraint-based design package. The parameters driving the geometry of the model are indicated on the drawing. When the parameter value for the length of the part was changed, the part was updated automatically, as shown in part b of the figure. Notice that the threaded length did not increase. If you wanted the thread length to increase as the bolt was lengthened, you would need to define the constraints and parameters differently.

A feature is the basic unit of a constraint-based solid model. Each feature has properties that define it. When you create a feature, you specify the geometric constraints that apply to it, then specify the size parameters. The modeler stores these properties and uses them to generate the feature. If an element of the feature, or a related part of the model, changes, the modeling software regenerates the feature in accordance with the defining properties assigned to it. For example, an edge that is defined to be tangent to an arc will move to preserve the tangency constraint if the size of the arc is changed.

Some features, such as holes and fillets, may use predefined features with additional properties that are maintained as the model is updated. For example, a “through hole” (one that goes completely through the part feature) will be extended if the part thickness increases.

When you look at the fixed-height shaft support shown in Figure 5.37a, can you identify the features that make up its shape? Figure 5.37b shows the major cylinder, hole, plate, end plate, slot, and rounds that form the model of the shaft support.

Being able to update models easily to reflect changes in the design is a key ability of constraint-based modelers. If you are not using a constraint-based solid modeling program, you may have to re-create the feature just to change a size. This can result in considerable time and effort, especially for interrelated features.

In addition to defining relationships between features, constraint-based modeling software also allows you to use constraints and parameters across parts in an assembly. Thus, when a part changes, any related parts in the assembly can also be updated. Because the software regenerates the features and parts from the relationships stored in the database (see Figure 5.38), planning constraint-based relationships that reflect the design intent of the part or product is the key to efficient and useful constraint-based models.

Constraint-based modeling software in many ways parallels the design process. To start creating a part, you often start with a 2D sketch of the key shape of a feature, such as that shown in Figure 5.40. The initial sketch may be rough in appearance because the software will apply constraints to define the relationships between the simple 2D geometric elements of your sketch. This process is similar to the way one of your engineering colleagues interprets your hand-drawn sketches. If lines appear to be perpendicular or nearly so, your colleague assumes that you mean them to be perpendicular, without your having to mark the angular dimension between the lines. In a similar way, constraint-based modelers apply constraints to your rough sketch, so that lines that appear nearly perpendicular, parallel, vertical, horizontal, concentric, or collinear will have a constraint relationship added to them so that they function that way in the model. This process is called “solving” the sketch in some modeling software, because the software is interpreting and assigning dimensions and constraints to generate a 2D “profile” of the feature. Before you generate the actual feature, you can review and change the dimensions and geometric constraints so they are what you intended. This step is key to defining your design intent. The sketch constraints in Figure 5.40 have already been added.

To generate a 3D feature, you then select a command to extrude, revolve, sweep, or blend the profile geometry. Figure 5.41 shows the profile in Figure 5.40 as it would appear after being (a) extruded, (b) swept, (c) revolved, and (d) lofted to create different types of features from the 2D sketch.

Constraint-based modeling programs allow you to review the constraints that are applied to the sketch geometry so you can change them to those needed to reflect your design intent. Table 5.2 describes some possible sketch constraints and shows symbols often used to indicate them on screen. (Different modelers use different symbols and means of displaying the constraints—see Table 5.3 and Figure 5.42). These constraints may create relationships to other lines in the sketch or to geometry in an existing feature.

Not all constraint-based modeling programs apply all these constraints, nor do they apply them in the same fashion. Study the constraints available in the software you use and determine their effect on the sketch. You also may be able to set the threshold values for various constraints. For example, instead of having lines at 1° be automatically constrained as horizontal lines, you may wish to increase this value to 3° (Figure 5.42). If you are having difficulty getting the constraints you desire applied to the sketch, you may want to turn off automatic constraint application and select each specific constraint to apply to your sketch.

Most constraint-based modeling programs let you remove or override a constraint that is applied, or specify that the geometry is exact as drawn. Sometimes, you may have to force the sketch closer to the desired relationship. For example, if two lines are too far off in the sketch to be constrained as perpendicular by the software, you could add an angular dimension of 90° between them to force them to that relationship. With the dimensions driving the geometry to the condition you want, the software may apply the perpendicular constraint. Once the constraint is applied, you often can delete the dimension you added.

It is usually beneficial to let geometric constraints determine much of your sketch geometry, then dimension the sizes and relationships that cannot be defined using constraints. You could use a constraint-based modeler as you would a solid modeler, adding dimension values to define all the features, but this would not build in the intelligence that allows the software to automate updating the model.

The sketch shown in Figure 5.43 is overconstrained. Because the top arc is constrained to be tangent to both vertical lines, and its radius is dimensioned (sd0), the width of the sketch is defined. The horizontal dimension shown at the bottom (sd2) defines this width again. If both the arc radius dimension and the overall width dimension were allowed in the model and one of them changed, it would be difficult or impossible for the part to be updated correctly. If the overall width changed and the tangent constraint or arc radius did not, the sketch geometry would be impossible.

The sketch shown in Figure 5.44 is underconstrained. It lacks a height dimension controlling its vertical size. Without this dimension the sketch is not fully defined. Some software will allow you to create a 3D feature from an underconstrained sketch as long as the sketch follows basic rules—for example, the sketch does not cross itself, and the sketch encloses an area. If the sketch is underconstrained, the software assumes values for unspecified dimensions based on the sizes drawn. This can be helpful when you are trying out ideas, but it is best to plan to create a model that reflects your design intent.

Dimensions can easily be changed by typing a new value, so go ahead and guess the size and change it later if needed; but keep in mind that choosing which dimensions to use to define the sketch is a critical aspect of making sketches and features that will be updated as expected when a dimension changes.

When you are dimensioning a sketch used to create a feature, place the dimensions where you can view them clearly. Later, if you create a drawing from your model, you will be able to clean up the appearance of the dimensions if needed. You can move dimensions, change their placement, and arrange them so they follow the standard practices. Good placement practices such as placing dimensions outside the object outline and keeping the dimensions a reasonable distance from one another will make the dimensions in your model easier to read.

3D models are accepted as final design documentation. When you are going to use the model as the design database and not provide drawings, it is even more important to ensure that dimensions and tolerances shown in the model are clear. You will learn more about practices for documenting inside the 3D model in Chapter 11.

If you do a good job selecting the dimensions that control the drawing geometry, you will have little cleanup work to do later when you are creating drawing views.

The base point is indicated in Figure 5.45a by the × in the lower right corner of the sketch. When the width dimension changes, the sketch is updated to the new size while leaving the base point fixed on the coordinates. The rest of the sketch geometry is reoriented, as shown in Figure 5.45b.

If your software uses a base point, it should become the base point for the entire model. When the model is updated, all dimensions will be updated from this point. A good rule of thumb is to identify a fixed point on your model that you would use as a starting point for measuring the part for inspection purposes. Use this point as the base point, or locate your dimensions from it as if it were the base point.

Locating most of your dimensions from a common edge or major surface can be effective for two reasons. First, it helps you create dimensions that will be useful for inspecting the part after manufacture; the dimension values in your model will measure from the same point that an inspector will use to measure the finished part. Second, it helps you anticipate how the model will be updated. When a value is changed, the features on the model will be updated (move) relative to the edge referenced in their dimensions. When you locate dimensions relative to other features you may have difficulty predicting how the model will be updated because you have to add the effect of changes to the intermediate features.

More important than any rule of thumb governing dimension selection or placement is to consider thoroughly the parts you are designing and to make the drawing geometry, constraints, and dimensions reflect the design intent for the part. In some cases, dimensioning a part from another feature is required to reflect the design intent. Figure 5.46 shows two sketches that have the same geometry but are dimensioned differently. When the 2.00 height dimension is changed to 2.50, how will each of these drawings be updated?

In the first case, shown in Figure 5.46a, the height of the 2.00 “step” will change to 2.50, narrowing the gap between it and the 3.00 step. In Figure 5.46b, the height of the 2.00 step will change to 2.50, and the height of the next step will be updated to 3.50 units so it remains 1.00 unit above it. In this case, the narrow gap will occur between the 3.50 and the 4.00 step. Either is correct to the extent that it reflects the intended design.

Figure 5.47 shows the cylinder that could be created as the base feature for the shaft support and the other features that will relate to it. The major cylinder would make a good base feature because it is a significant feature of this part, and the other features make sense related to it. If the design intent of the part centers on the cylinder, its size and location determine the size and shape of other features. The width of the base plate will depend on the stability needed for the size of the cylinder. The hole for the shaft must be centered inside the cylinder and sized to leave enough remaining material for the wall thickness. Using the cylinder as the base feature also makes sense from the standpoint of the part’s role within a larger assembly, as its size could depend on the shaft being designed in a different part. By building your model around the major cylinder, you make it possible to update all features automatically if the cylinder size changes.

To create the second feature, you can use a dimension to locate it relative to an existing edge or use a geometric constraint to align one of the sketch entities to the existing geometry. For example, you could dimension the center point of a hole to be a certain distance from the end of an existing feature or constrain the hole to be concentric to a cylinder or arc on an existing feature.

The way you dimension, constrain, or align your second feature determines how it will be updated with respect to the first feature. Use the things you know about the design to add features. Where are the fixed points on the model, and how does the feature you are adding relate to them? What are the conditions that the new feature needs to satisfy at all times? If existing features were to change, how would the feature you are adding need to change? In some cases, a simple dimension from an existing edge may be all that is needed to add the feature. In other cases, you will want to build in geometric or size relationships needed in the design. The basic process of constraint-based part modeling is illustrated in Figure 5.48. Starting with the base feature, you build a model by adding features and relating them to existing features. For each feature, the same basic process applies.

The base feature is the parent feature for the next feature created from it. Just as real children depend on their parents, child features are dependent on their parent features. If a parent feature moves, the child feature moves in relation to it. For example, when the hole feature was added to the shaft support part, its location was defined as concentric with the major cylinder. When the cylinder’s location changes, the hole feature remains concentric with the cylinder (Figure 5.49). If the hole is located relative to the baseplate or to some other feature and the cylinder is moved, the hole will not move with it (Figure 5.50).

If you delete a feature, you will also delete its child features unless you redefine them so that they are related to a different parent feature. This is another reason for selecting a good base feature; you do not want the first feature you create to be something that might be deleted later in the design. Not all features will have child features, but all features other than the base feature must be related to some parent feature.

One way to visualize the parent-child relationships as you plan your model is to build a tree diagram. The relationships in the Clamp Bottom model in Figure 5.51 are illustrated in the Parent/Child Relationships window. The panel labeled Children shows that the feature named Loop strap has Handlebar hole, Boss strain fillet, and Outer edge fillet as its children, along with the sketch that formed its cross-sectional shape. These features may or may not have children (or “grandchildren”) of their own. The panel labeled Parent shows that the parent features for Loop strap are Sketch4, the Front datum plane, the Screw boss feature, and the Origin of the coordinate system.

Many constraint-based modelers allow you to “play back” the model to see how it is being generated by the underlying parameters. The sequences shown in Figure 5.52 and Figure 5.53 illustrate how the features used to create the parts are ordered in relationships that reflect the design intent for the parts.

Often, it is very useful to use a set of datum planes (three mutually perpendicular planes) as the base feature in your drawing. In fact, most software provides them as the default starting features. Figure 5.54 shows the shaft support part with a set of datum planes used as the base feature.

One of the benefits of starting with a set of datum planes is that they can provide a set of normal surfaces that you can use to orient sketch geometry. This is particularly helpful when there are no planar surfaces on the part that you can use to build subsequent features easily. With a cylinder as the base feature for the shaft support, for example, there is no flat surface to which the end of the base plate can be constrained to be parallel. The datum plane “Right Plane” in Figure 5.54 provides a reference to which the plate is parallel.

Whether or not you start with a set of datum planes as a base feature, consider the surfaces that will serve as datum surfaces on your part. Parts are inspected to see whether they were manufactured acceptably based on a similar set of three mutually perpendicular, theoretically exact planes. Measurements of the finished part are taken from surfaces coincident with one of the datum planes. As you define your features consider the dimensions that will determine whether the part will fit. How will the part be measured after manufacture? If there is a surface from which measurements will be taken, it can be advantageous to identify it as a datum surface in your part (see Figure 5.55).

Like the base point in a sketch, a datum plane serves as a reference from which the other features build. When you define a set of datum planes as the starting point for your model, you will be well on your way when deciding which datum surface to identify for use in inspecting the finished part.

Datum surfaces can also be helpful when defining tolerances for features. Defining your model using the dimensions that will be used to inspect the finished part will make it easier to determine the impact of variations from the stated dimension that occur during manufacture. (You will learn more about tolerances in Chapter 12.)

Datum planes also provide a common reference point for parts in an assembly. When parts share a set of datum planes, it is easier to align the part with other parts when you bring them together in an assembly model. If you do not use a set of datum planes as a base feature, you can locate new features from existing surfaces on the base feature.

Creating datum planes, lines, and points as base features can be used as a strategy called “skeleton modeling” (see Figures 5.56–5.58). Because the datums have no mass, they do not affect mass property calculations. They are easy to model and often do not change. For example, if you create a datum axis that is the centerline of the part, this is unlikely to change, whereas the shape of the part might. Even when some parts are not fully modeled or are represented only as a centerline, they can still be assembled using the datums as references.

Datum planes and datum surfaces can be added and identified at any time, but considering them as you plan your model will help you use them more effectively in capturing design intent in your model.

Depending on how the model was created, you may have more or less success in updating the model. For example, the baseplate that connects to the cylinder (indicated in the figure) can be created many different ways. Each way may have advantages or disadvantages for how the part will be updated.

Figure 5.60 shows three ways the baseplate could be generated:

a. Sketch the shape of the bottom surface on a sketch plane aligned with the bottom of the cylinder and extrude it upward.

b. Sketch the cross section on the midplane and extrude to both sides.

c. Sketch the rectangular end view on a plane offset parallel to the center datum plane and extrude the plate to meet the cylinder feature.

There are several other ways, too. Each of these methods works about the same when changing the length of the part. They vary if you decide to make the plate wider than the cylinder. Often, when a feature is extruded to an existing feature as in Figure 5.60c, the new feature cannot be updated so that it is wider than the surface to which it is extruded. The boundary, or end, of the feature is limited by the next surface. Where there is no portion of the surface at which the feature may end, it will continue to infinity and therefore be undefined.

Many constraint-based modeling programs use the same terminology for these features as that used in manufacturing and design. Review the standard feature types on the next page.

Many constraint-based modelers provide a number of options for quickly placing holes, often through a “hole wizard” or other dialog box. Properties that can often be set for holes are listed in Table 5.4.

A Hole command may also allow you to choose the specialized constraints for locating holes, such as concentric, placed by edge, and on point, as illustrated in Figure 5.62.

A fillet is a common built-in feature that rounds the edge formed by two surfaces. Typically, a fillet can be created by selecting the edges to be filleted (several can be selected at once) and specifying the radius of the fillet. A round is another type of standard feature created in the same way, usually using the same Fillet command. When the surfaces move or are lengthened, the fillet or round adjusts automatically. Figure 5.64 shows a typical fillet and a typical round with a uniform radius.

Sophisticated constraint-based modelers allow you to create uniform radius fillets and rounds, chained fillets and rounds that blend around a corner of a surface, constant-chord fillets and rounds, and variable-radius fillets and rounds (shown in Figure 5.65) using different methods. A constant-chord fillet is used in situations similar to Figure 5.65b, where a constant-radius fillet would result in thicker material on the uphill and downhill side of the blend. The constant-chord distance creates a fillet with a uniform amount of added material. More sophisticated blends between surfaces are also available in some software.

C0, C1, and C2 are terms used to describe the geometric continuity of surface blends (see Figure 5.66). (You may also see the letter G used as a designation for geometric continuity.)

• C0: Coincident surfaces meet but are not tangent. These are essentially intersecting surfaces, not ones that have a fillet.

• C1: Tangent surfaces meet and are tangent. This is the typical tangency produced by Fillet commands, and is suitable for most rounded edges.

• C2: Curvature continuous surfaces meet and are tangent, and where surfaces blend, they share a common center of curvature. Surface blends with this continuity do not have any abrupt change in radius. These blends are important for styling of surfaces such as automobile hoods and exterior plastic parts.

• C3: The next level above C2, except that the rate of change for the curvature is also continuous.

Because the complex surface can slow down model regeneration, an approximation of the surface may be used as a placeholder until later in the refinement process.

The assembly provides a framework for the model. In the assembly, global parameters can be used to build intelligent assemblies. For example, if several parts assemble onto a shaft, the shaft diameter may be made a global parameter. The hole sizes on the parts that fit onto the shaft may be created by adding or subtracting a value from the global parameter for the shaft diameter. If the shaft diameter is changed, all the parts that fit with it may automatically be updated. Using global parameters can be an effective way to coordinate the design effort for a team. You will learn more about using assemblies to refine your design in Chapter 10.

If the size of a feature changes in the model, the value for the dimension in the drawing is automatically updated to show the correct values. When this is a oneway process—when changing the model causes the dimension value to be updated on the drawing—the dimensions are said to be associative. When this is a one-way process, changing the dimension value on the drawing has no effect on the model geometry. In fact, it may break the link between the dimension value and the model geometry.

Most constraint-based modelers are bidirectionally associative: changing a dimension in the drawing changes the model geometry, and vice versa (see Figures 5.71 and 5.72). Constraint-based modelers store and use the dimensions and constraints to generate the model. When dimensions are shown in a drawing, the values are the dimensions used to create the model. Whether these dimensions are changed in the part or on the drawing, the new values are stored in the database, and the model and all associated drawings are updated to reflect the new dimension values. Bidirectional associativity allows drawing changes to update the model, and model changes to update the drawings.

Key considerations are the time required to model the part and the purposes for which it will be used. A simpler method that provides all the needed information is more cost-effective than one that takes longer to model. However, time invested in a complete digital model of the product can pay for itself if it is used to generate visuals that help shape the design, reduce manufacturing difficulties, foster concurrent engineering—or promote the product later.

The accuracy of the modeling method is another key consideration. What types of analysis must be or could be completed with a model? What level of accuracy will be required to interface with CAM programs?

As you have seen, 3D modeling packages that are strong in surface modeling are needed to model the geometry of smoothly contoured features such as the ergonomic mouse shown earlier in the chapter. Piping systems and other structures of simple geometric shapes can sometimes be modeled very effectively using 3D wireframe modeling. Many inexpensive CAD packages support this type of modeling. However, a solid modeling package is required to create a model that can be used to check parts for fit and interference; calculate the weight of a final assembly from the many individual parts; generate a rendered view of the assembly that can be used in manufacturing, marketing, service and repair; and drive the computer-aided machinery to manufacture parts. The same assembly model created using constraint-based modeling methods would have the added advantage of being updated automatically when key dimensions change. It would also make it easy to create a family of similar parts.

Whether you choose the modeling package or learn to use the tools provided by your employer, you should be aware of the strengths and limitations of the software so you can use it most effectively.

Characteristic	3D Wireframe	Surface	3D Solid
Visual	Can be viewed from any direction and used to create standard 2D views; lack of surfaces may make views ambiguous.	Usually includes lighting and background options that can be used to create photo-realistic images of the model.	Shaded and rendered views present a realistic view of the object; offers automated 2D view generation; with most equipment only 2D view is available on monitor; part must be rotated to picture clearly.
Understandable	Requires viewer to mentally add surfaces to see the 3D shape.	Shaded views are easily understood by the viewer.	Shaded views are easily understood by the viewer.
Flexible	Editable, but changing individual objects more time-consuming than in other methods.	Each vertex is editable, but ease of changes depends on method used to create and store the surface.	CSG or hybrid models offer the most flexibility; can edit at the level of the solid object instead of individual geometric entity.
Cost-effective	Requires less computing power, and software is generally lower priced; good for modeling geometrically simple shapes.	Most cost-effective for modeling irregular surfaces that must be conveyed to manufacture with a high degree of accuracy; photo-realistic displays can offset modeling costs by adding value in marketing.	Completeness of the solid model cost-effective when model can be used for multiple purposes, such as testing, presentation, and CAM.
Measurable	Full-size entities are measurable without auxiliary views; can only measure between wireframe entities, however.	Can be used to calculate surface area.	Can be used to measure not just size but also weight, mass, and other physical properties.
Accurate	Offers a high degree of accuracy for the entities represented.	Depends on method used to store model data; smoothed surfaces generally more accurate than mesh surfaces.	Depends on method used to store model data; high degree of accuracy possible.
Robust	Model stores only vertices and edges; lacks surface or volume information; may be ambiguous as to voids or holes.	Fully defines the surfaces of a part; lacks volume information.	Includes volume as well as surfaces and vertices; can eliminate need for physical model.
Application	Good for applications where relationships in 3D space must be modeled, simple geometric shapes are sufficient for the objects to be modeled, and realistic views are not required.	Good for designs that include irregular or free-flowing surfaces and for those where the appearance of the product is a critical design criterion.	Good for applications where fit between parts and other engineering properties need to be evaluated before manufacture.

Assembly Mode

Associative

Base Feature

Base Point

Bidirectionally Associative

Boundary Representation

Built-in or Placed Features

Chamfer

Constraint-Based or Parametric Modeling

Coon’s Patch

Cosmetic Dimension

Counterbore

Datum

Datum Planes

Derived Surfaces

Descriptive Geometry

Descriptive Models

Design Intent

Drawing Mode

Driven Dimensions

Driving Dimensions

Feature

Feature-Based Modeling

Fillet

Finite Element Analysis

Geometric Constraints

Global Parameters

Interpolated

Layers

Lofting

Model

Parameter

Parent-Child Relationship

Part Mode

Patches

Prototypes

Size Constraints

Subassemblies

Surface Normal Vector

Tessellation Lines

Triangulated Irregular Networks

Virtual Prototype

Virtual Reality

Wireframe Modeling

You should be able to list advantages of constraint-based modeling over solid modeling. You should be able to correctly use the term parent-child relationships as applied to part features. Associativity between drawing views and the 3D model is an key characteristic of sophisticated constraint-based modeling programs. Selecting the base feature of the model is important, as it is the first parent feature. The exercises ask you to demonstrate your understanding of parent-child relationships and base features by capturing design intent in a number of part models. They also provide additional practice with your constraint-based modeling package.

a. Preliminary design review with your immediate engineering (technical) supervisor

b. Discussions of installation details with the campus architectural staff

c. Meeting with a group of nontechnical advocates for accessibility by persons with disabilities

d. Meeting with machine shop technicians who will fabricate your prototype

2. Which modeling method contains the most information about a design? Identify a situation in which this would not be the best model to use.

3. Recent developments in Global Positioning System (GPS) technology have resulted in small, relatively inexpensive, handheld GPS receivers (Figure 5.97).

a. A need exists for a mount to attach the system to mountain bikes in such a way that the display is visible while the rider is mounted. It is not desirable to modify the GPS to provide attachment. Consideration should be given to protecting the electronics and display from damage during rugged use while providing a clear sky view for the built-in antenna. What types of CAD models could be used to develop this mounting attachment?

b. At what stage of the process would each type of model be appropriate?

4. For the following situations, choose a modeling method and explain why you would use it.

a. Designing a prosthetic limb

b. Routing a piping system for a building

c. Designing a one-of-a-kind machined part

d. Verifying clearances for an electrical substation located near an airfield

e. Mapping an oil reservoir

5. Not all machines have moving parts. A wedge is an example of a simple machine. Use each of the modeling methods discussed in the chapter to model a wedge. Use the same size wedge for each model.

6. Constraint-based, or feature-based, modeling uses defined relationships between model elements to control various features. What two basic types of constraints are used to control features?

7. Why are the linear, angular, or other physical dimensions of a constraint-based model feature given a name? Give an example where this practice might be useful.

8. An acronym sometimes used to describe the process of preparing for the creation of a constraint-based model is TAP, representing the steps Think, Analyze, and Plan. Consider what it means to think, analyze, and plan a constraint-based model. Why are these steps especially important prior to beginning the constraint-based model?

9. What is meant by parent-child relationship in constraint-based modeling?

10. Define bidirectional associativity with respect to a constraint-based modeling program’s drawing mode. How does this program feature enhance ease of use of the program?

11. You are considering modifying a model by changing dimensions in a drawing view. What problems might you encounter when using a constraint-based modeler without bidirectional associativity? What problems might you anticipate when using a constraint-based modeler with bidirectional associativity in this task?

Exercise 5.2 For each part shown, list the modeling method you would choose to create it if you were creating the original product definition for mass-producing these items. Give the reasons for your choices.