Sponsored By

Q+A: Intelligent machining comes of ageQ+A: Intelligent machining comes of age

October 7, 1998

7 Min Read
Plastics Today logo in a gray background | Plastics Today

Today's toolmakers apparently bear two paradoxical crosses. First, they must satisfy customer demands for increasingly shorter lead times. The second challenge, however, makes it more difficult to meet the first. Namely, there are fewer experienced toolmakers available in the labor pool than there were even a decade ago.

Using machining software to bridge the gap is one answer. But will the assorted CAM and CAD/ CAM packages ever really be able to capture an experienced toolmaker's knowledge?

To learn what to expect from machining software today and in the near future, IMM spoke with Dan Marinac, president and CEO of Cimatron Technologies Inc. (Burlington, ON). As an industry spokesperson, he believes capturing the tooling engineer's knowledge in a database is the next step in the software industry's pursuit of the CAD/CAM "holy grail."

IMM: CAD packages now create parametric, feature-based parts that capture an engineer's design intent. Together with design knowledge databases, these technologies let designers download their experience to novice engineers. Where is the corollary capability in CAM?

Marinac: Software development for CAM has not kept pace with its CAD counterpart. We have seen such advances as multiple surface toolpaths and automatic NC, but most attempts to automate CAM began with 2.5-axis programs that focused on machining design features. Toolmakers are not allowed to redesign parts; they must work with electronic data supplied, which is often in the form of an IGES file. Most IGES files for toolmakers are sculpted surface models with thousands of surfaces. Tool designers need full 3-axis automation because mold geometry is complex. There are literally dozens of strategies available for this kind of machining.

Although these two mold cores appear geometrically different, a machinist recognizes that the two parts use similar machining techniques. They have technologically similar attributes identified in yellow by intelligent NC software. These areas have been adapted by the software to be machined the same way.

Several CAM developers are now focusing on intelligent NC, which integrates a full complement of capabilities-manufacturing feature recognition (MFR), automatic NC (ANC), and knowledge-based machining (KBM)-with knowledge of stock remaining (KSR). (For more complete definitions, see sidebar, below) Intelligent NC holds the key to completely automating manufacturing by capturing the user's knowledge of the process and recognizing the appropriate machining strategies. This new process plan automatically adapts to and applies new geometry to produce safe, efficient, high-quality toolpaths.

IMM: How does the software "know" which machining strategies to use? One of the problems with automatic NC in the past has been that it did not take an experienced user's knowledge into account.

Marinac: In the past, NC software contained feature-based machining, which is actually a CAD approach to CAM that restricts machinists to cutting geometrical features. But manufacturing engineers don't make toolpaths that way. They machine technologically similar attributes of the mold the same way in every mold they do. They don't refer to features such as bosses and then say that boss is going to be cut into the mold.

Moldmakers cut the core and cavity only once or the electrode once. For this reason, the CAD approach, based on high volume parts, doesn't serve moldmakers either. The toolmaking industry needs a batch approach to machining a mold. Rather than saying, "Here's a pocket, and we'll cut it," CAM systems must deal with thousands of surfaces. You would not cut each surface one at a time, so the software must take a more global approach, trying to involve the entire job.

With intelligent NC, moldmakers actually "teach" the software the kinds of strategies they use for machining features with technologically similar attributes using strategies that detect shallow or steep regions, large or small curvature, and bounded or unbounded volumes. Based on the way previous jobs were machined, the system can decide which machining strategy to apply. This is not feature-based machining, which restricts the software to geometrical features the designer used to model the part. Those geometrical features have nothing to do with the optimal method for machining a core and cavity.

Combining MFR, ANC, KBM, and KSR means one-button NC using shop practices, not a software developer's algorithm. And when I say similar parts, I don't mean geometrically or parametrically similar. It can be a completely different part with shallow areas, horizontal areas, and steep areas common to other molds. For example, to teach the system how to machine shallow regions, from horizontal to 30°, the company's strategy may be that all cavity surfaces are stored in levels 40 to 60 with the parting line in level 63. The operator then fills in a table of relational parameters to determine machining criteria. For the next job, all shallow regions can be machined using the same technique.

IMM: How similar must molds be to take advantage of the teachability factor of next-generation CAM?

Marinac: Most shops will have dozens of mold styles but only a few machining styles. Some may cut only one type of mold. In general, most mold shops have focused on a segment of the industry and continue to get business in that direction, so their molds are similar from a machining perspective. These are the shops that benefit the most because many of their molds have technologically similar attributes. On the other hand, for operations that produce hundreds of unique mold types, intelligent NC may not significantly save time or improve productivity.

Many mold types use similar machining strategies. With an intelligent system, operators decide where the cutter starts, at what depth, where it finishes, and what size steps it takes. They don't enter raw numbers but rather the first height as a variable maximum and minimum. Each downstep, then, can be a calculation of a percentage of tool diameter. This system is based on relational parameters, just like parametric CAD.

IMM: Why will intelligent NC save time and increase productivity for these types of moldmakers?

Marinac: To answer that question, it helps to take a look at the way NC machining currently operates. In general, originating data comes to moldmakers via an IGES file. If you're running Pro/E or I-deas, the part has to be remade. It cannot be imported as a solid from the monolithic (without parametric or variational features) IGES file. Once the data is in an appropriate format, it is separated into two streams for mold design. The first stream involves passive components such as mold bases and ejection and cooling systems. These are combined in a large assembly using a solids modeler. Active components-slides, lifters, moving pieces, and parting lines-are the second data stream. Here, the design requires complex surface work, such as splitting core and cavity, developing parting surface lines, and extracting and designing slides and lifters. Both active and passive streams need detailed drawings.

Final designs are then sent to separate NC programs. Passive components are typically drafted using software such as AutoCAD; then they often go to a 2.5-axis package, such as MasterCAM. Active component toolpaths are generated in parallel with a 3 to 5-axis CAM package such as Cimatron. Remember, you now have three "current" databases.

Along comes an engineering change, most commonly an IGES file update. For a nonassociative system, changes must be incorporated into each database, then physical changes must be made to each drawing to keep the databases current. This process can take anywhere from several days to weeks.

In an integrated system, a change made once is made everywhere by applying parametric, bi-directional, and associative features to CAM. Now, there is only one database to be updated, and all associated drawings are changed automatically.

Sign up for PlasticsToday newsletter

You May Also Like