Machine Design

New ideas reshape solid modeling

Drag-and-drop features, drawings that almost read themselves to you, and translators that turn dumb models into manufacturable geometry are just a few of the capabilities added to modeling systems.

Joseph Dunn
Application Engineer
SolidWorks Corp.
Concord, Mass.

Edited by Paul J. Dvorak

Capable modeling systems such as SolidWorks from SolidWorks Inc., give users a range of features for tasks they are likely to encounter. Sheet-metal functions, for example, allow modeling detailed parts, such as the bracket, and then unfold them for manufacturing.

The mounting plate with a center hole shows how design tables can be applied in SolidWorks. Should a user require a fifth instance, adding the row automatically produces the name in the first column. Typing in values to the other columns tells the software to generate the plate. The lower image shows four configurations. Adding a fourth column would let users make the center hole diameter a variable as well.

Despite the success of other file transfer schemes, IGES is still selected for most of that work. Unfortunately, IGES generates what's called dumb geometry, that without features. To help translate IGES and other formats into features, SolidWorks Inc. has built FeatureWorks into the modeler. IGES models such as the wire frame can then be translated into more easily editable features. The part tree at left shows the recognized features. Menus let users assist in the recognition operation.

The drawing of a machinist's vise is in the electronic format from SolidWorks, called an eDrawing. This one stores in 725 kbytes and downloads over a 56k modem in about 2 min. Simple browserlike commands let users navigate the drawing, zoom in and out, and animate details for easier information finding.

A top view of the vice in an eDrawing shows two section lines. Picking on the A or B takes the viewer to that cross section.

The How-to window pops up when opening an eDrawing. In 5 min, a user can learn most of what there is to know about the format, such as how to animate sections, link the same feature in different views, and arrange different views of interest for easier examination

Selecting and placing fasteners in a large assembly is numbing work that begs for automation. Smart Fasteners in SolidWorks, for example, lets users keep their sanity by placing nuts, bolts, and rivets into a design more quickly than previously possible. Built-in intelligence needs only the bolt to find appropriate nuts and washers. The software lets users customize the feature to industry norms and company preferences.

In the last 24 months, developers of solid-modeling systems have packed more productivity features and technology into their software, making them increasingly powerful design tools. The developments are all the more important because design-department influence extends throughout a company. One study, for example, estimates that for every person generating design information, there are at least 20 users. That makes the timeliness and accuracy of design information critical to the company's survival.

The pace of software development is so rapid that it's a good idea to step back from the whirl of daily activity and take a closer look at the state of modeling software — and the productivity features you might be overlooking.

What's new in modeling
Take sheet metal, for example. Design tasks occasionally call for simple enclosures, brackets, and shelves. Well-designed software should let users quickly shape concepts in sheet metal for sizing studies and elementary strength calculations.

Really good sheet-metal features in solid modelers let users bend and shape metal on screen as if working with sheet-metal tools. Applying each tool should clearly show cutouts, tabs, flanges, rips, slots, louvers, dimples, bend reliefs, and so forth. The better modelers also use drag-and-drop methods so users drag, for example, a dimple from a library to a part.

Complex sheet-metal geometry is difficult to draw in 2D. So the more complex the design, the more 3D software pays for itself.

And when finished with the 3D shape, the software should unfold it into a flat pattern. When compared to drawing in 2D, 3D sheet-metal features trim 90% of the time previously needed.

Design software should also provide a way to apply configurable standards that let companies set, for example, neutral-bend radii, or use bend tables applicable to the intended fabrication machine.

Recognizing features in the geometry from other systems is also useful. This broaches the issue of interoperability — CAD compatibility with other products. Features are important because they carry intelligence, function, and purpose. For example, a through hole should "know" it goes through a part, so if the part is thickened, the hole still goes through. Features turn small parcels of geometry into useful things — more than a collection of surfaces.

CAD programs may have elementary geometry readers that work with limited success. They bring in IGES files which are just dumb geometry — shapes only, no features recognizable by the software. Such geometry comes from many translators including those for STEP, Parasolid, and ACIS.

Sophisticated modelers, on the other hand, have translators that can read cylindrical voids and interpret them as holes with alterable lengths and diameters. These modelers can read geometry from widely used systems such as Pro/Engineer and translate every feature so the holes, walls, bosses, and ribs in a Pro/E model are the same features in, for example, Solid-Works.

Recognizing features makes changes simpler. In a plate with holes, for instance, better feature translators recognize the hole pattern so it can be changed in one step. Without the capability, each hole has to be treated individually.

A limitation to the capability is that it works only on machined and sheet-metal parts. If someone were to let the software recognize features on an automobile's body-in-white, the translation's success would be limited. Lofts, for example, would be unrecognized.

Assigning fasteners also gets a boost from automation. Manually placing nuts and bolts is grunt work and should be automated. Such hardware is so well standardized that if a user identifies a hole diameter, the CAD system should respond with correct fasteners. A CAD system should be smart enough to know that specifying a screw determines the hole size along with a particular nut and washer. Designers can then focus on more creative tasks.

Configuration management, another area improved by modelers, includes everything from generating families of parts to managing assemblies. Any difference, variation, or change in a design assembly alters the configuration and should be tracked by the CAD software.

Take a socket-head cap screw, for example. There are thousands of variations. The more efficient modelers store all variations in a compact file, but they only store the rules for configuring different heads, diameters, and threads.

Assemblies are treated similarly — one collection of parts with many variations. Parts can be turned on or off, or simplified or not. A computer mouse, for example, could have one, two, or three buttons, be left or right handed, and come in different colors. Parts have to change to accommodate the design, but they would be stored in the same assembly file. More-efficient modelers also store the intelligence for building different variations in one file and build specific variations after users make selections from a simple menu or spreadsheet.

Where the Web fits in
The Web is turning into an indispensable engineering tool. A few of the inventions it has inspired include electronic drawings, improved model translations, and instant Web sites.

Electronic drawings are a sort of PDF file for the CAD industry. In the same way that Adobe PDF files have made brochures and magazine articles easy to transmit through e-mail, so too have electronic drawings made it easier to send design information.

And it's about time. Despite the Internet's widespread use, the most often-used delivery service remains the overnight express. The DWG format had been a standard in 2D, but it carries many problems. For one, DWG files can be modified, and they are huge. Despite DWG's seeming pervasiveness, it is not always compatible because of so many CAD add-ons.

E-mail, on the other hand, is the intended delivery mechanism for electronic drawings. The format is so compact even older modems can handle them. Electronic drawings have a self-contained viewer so they can be opened and seen on all Windows-based computers.

Instant Web sites also improve communications by letting CAD users easily publish models and documents on the Web. Software developers such as SolidWorks let users publish a model, for example, as a Web site. Pushing a single button generates a Web page for whatever is on screen along with a URL. It's more than an HTML document. It's a separate Web site viewable by those who know its URL. And they can view the model, rotate it, zoom into features, and so forth. They cannot, however, change the model.

Several CAD programs create HTML documents, but they are not automatically published. Most CAD users are unfamiliar with Web details and would have to give the model or document to their MIS department to publish.

The simpler and more useful software makes it nearly a one-button operation. What users do with the information depends on the viewer sent with the URL. An assembly or 3D part could be rotated for examination. Or a detail drawing could be published as an electronic drawing with zoom and measure features.

How solid modelers encourage e-commerce

Operations such as from SolidWorks Inc., lets companies parameterize models so interested parties can size needed components over the Internet. In this case, the user is resizing a bearing for a pneumatic system. Users have the three options at screen bottom after applying size values.

Online catalogs have the potential to let companies toss their paper versions — which are probably out of date — and replace them with an inventory that is more easily searched. But few companies take advantage of this capability because it's expensive and time consuming to turn every product into a 3D model.

A better way provides the software and services to transform paper catalogs into Web versions. The technology for doing so stores the rules for shaping a collection of similar parts and builds instances as they are needed.

The buzzword here is 3D commerce. Recent thinking is that all transactions on the Web will involve 3D parts because they are more useful. That is, when designers download parts, the exact items are likely to be purchased. The file formats are not the problem they once were because about 20 are readily available. Companies are making the transition to so-called e-catalogs because table-driven configurations make it easier to do so.


An Excel spreadsheet lets users do more number crunching than design tables. For instance, the dimensions on the scuba tank have been linked to a spreadsheet. Now changing only the required volume, 480 in. 3 , lets the spreadsheet recalculate the values on the tank.

Most companies design a few particular products in many variations. Design work gets tiring when every new variation calls for starting from square one. A better tactic might be to take advantage of several little-publicized functions in modeling systems that build simple or complex parts from a few users-supplied inputs.

For one, users could store frequently generated models or features in simple record-and-playback files. Design tables, another method, provide builtin spreadsheets that store dimensions describing a family of parts. An external Excel spreadsheet could be added to handle more complex number crunching, and Visual Basic delivers a relatively simple, yet powerful, programming capability. In addition to speeding up design work and removing the drudgery, the following methods preserve corporate knowledge.

Record and Playback may be the easiest way to save frequently generated models and features. The technique works much the same as Record and Playback functions in a word processor: hit Record, build a generic model, hit Stop Recording, and name the file. Next time a job calls for a similar part, use the playback feature, and start modifications from there.

Design tables also provide a way to quickly build families of parts. They can also build a family of assemblies. Tables are actually spreadsheets built into the modeler.

They work this way: Suppose it is necessary to build a family of bolts characterized by diameter and length. First, build a generic part with the essential dimensions assigned as parameters, D and L, for example. To design a new bolt, pull up the design table and assign two values in a new row. The first column usually shows part or dash numbers, and these can be automatically assigned by the system. The second column has diameter dimensions, and the third carries lengths. After filling out the design table, insert it into the generic part model. The CAD modeler interprets each row as a configuration.

The system works with inputs from the model or the table. For example, pick the diameter dimension in column C and then fill in the needed diameter. There is no programming involved from here. Hitting Enter tells the system to build the new part. The system is smart enough to know that entering a dimension on an unused row initiates a new configuration, so it will assign the next dash number in the sequence.

Spreadsheets are the next step in system complexity for driving solid modelers. Users manually link columns to dimensions in the model. A row of cells describes a configuration and each cell holds a dimension. In most cases, the spreadsheet is Microsoft's Excel whose language is Visual Basic (VB).

The advantage of Excel over a design table is that the spreadsheet can do more number crunching to drive a model. It can also use sophisticated logic and calculations with if-then statements. The spreadsheet can query models and additional lookup tables to find required values, run through a few iterations with its number-crunching capability, and feed calculated values back to the model.

Goal seeking is another advantage of Excel. Users need not resort to separate programs for weight optimization. Goal seeking works like this: Suppose a scuba tank must be optimized for least weight and maximum pressure. Equations from Roark's stress and strain text relate pressure to tensile strength and volume. After typing the equations into the spreadsheet, users might enter a maximum pressure and volume. The spreadsheet and modeler would then calculate the thinnest tank-wall thickness and size.

Visual Basic and C++ are more complex methods for driving solid models. Examine a file generated during a record-and-playback session and you'll see VB statements.

More often, users are recording product design knowledge with the program. One company, for example, builds hydraulic valves for helicopters. They start with a master manifold that can accommodate many variations. Their customers typically have five or six different requirements (pressure, flow rate, and so on) that nail down a valve's design. The engineering knowledge in the program now lets the sales force design valves in the field.

Think of VB as a more capable macro language. VB and CAD commands use identical names such as Insert Feature, Extrude, along with X, Y, and Z values. To start using the language, identify some simple, repetitive tasks, such as inserting a part with the intention of modifying a dimension or two. Then, move on to more complex tasks such as applying company standards.

C++, on the other hand, is a bit more complex, but it's the preferred programming language for professional code writers because it executes so quickly. Most add-on programs to solid-modeling systems are written in C++. The bottom line is that it and VB can control a modeler and quickly generate models based on accumulated knowledge stored in equations and tables.


Is durable plastic an oxymoron?

I don't think I'm hard on my possessions. But am I the only one who notices that in any mechanism with moving parts, the one seemingly most likely to break is made of plastic? The experience of three recent failures started me wondering about techniques for designing with polymers. I took a close look at the three, expecting to see flaws that were more or less obvious. It didn't turn out that way. But the effort brought insight nevertheless. First, the anecdotes.

The first problem was on plastic latches attached to rear, swing-out car windows. They work like a holding clamp, so that pushing a plastic lever past a center point locks it closed. But I found it necessary to put my palm on the lever to generate enough force. That didn't feel right. It just seemed like too much work. My impression was that every closing was straining the material. I was right, because latches on both rear windows snapped the same summer, right where the plastic was thinnest.

Another failure involved a plastic threaded connection to a kitchen-sink spray nozzle. A flexible steel hose threads onto the plastic spray head, which also doubles as the faucet spout. The unit came with a 3-lb weight for the hose. When you're done spraying, the

weight pulls the hose into the spout.

It was quite natural to pull the nozzle out and immediately twist it 90° to spray something. This little maneuver applies a moment load to the threads at the base of the spray head. Every time I'd pull out the nozzle to rinse a dish, I could feel the weight pulling back and wondered about the bending moment on the threads. Of course, the load was highest where the plastic was thinnest. No surprise. The threaded portion snapped clean off in less than a year.

Most recently, the handle on a toaster broke after less than two years. Intrigued, I measured the break area, guessed a few loads, and calculated a stress value. The handle is shaped like an inverted L — pushing on the top lowers bread into the toaster. The pressing also applies a moment load to the right angle. A person's fingers touch the handle about 1.25 in. from the vertical column, and the area of the break measures

Senior Editor

1 /16

3 /16 in. Surely, I reasoned, a few simple calculations could reveal high stresses in this area.

Estimating loads turned out to be more complicated than I first imagined. What is the average load on a toaster handle? And what happens to loads at the bottom of the stroke? There is a slight impact as the mechanism suddenly latches, and

the natural inclination is to continue pushing until one hears the connection. Push hard enough and this could boost loads three to fivefold. If one assumes a 2-lb load, = Mc/I calculates a stress value of 474 psi. Stress should be half this figure, or about 237 psi because the handle attaches in two locations, assuming even loading.

Of course, I don't exactly know what plastic makes up the handle. But scanning a table of plastic properties shows that 10,000 psi is an average ultimate strength, and that provides a whopping factor of safety of about 42. Despite the apparent margin, the part still broke after about 600 to 1,000 load cycles. On paper, at least, it's difficult to label it as a poor design.

So what are the lessons from all this? Loads are a big question mark, impact is equally difficult to calculate, and the fatigue life of plastic is next to unknown. A few handbooks list values for the toughness of plastics, but they don't tell how to use them in stress calculations.

Regardless, I can still offer one piece of wisdom: If a plastic part looks flimsy, too thin, or underdesigned, you can be reasonably assured that it is, and it will break long before the rest of the product wears out.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.