FEA has been useful in evaluating designs. Now it's finding more ways to improve manufacturing.
Cosmos Analysis Div.
Los Angeles, Calif.
Mention finite-element analysis to most engineers and they think of stresses and structures in traditional design work. The future of the technology, however, lies more with nonconventional structures such as the cutting edge of tools, predicting the behavior of unfamiliar materials, or providing a look inside complex tooling assemblies.
Manufacturers using FEA say the combination of 3D CAD with an analysis package lets them quickly change models and see results shortly thereafter. What's more, some say they devise several ways to fabricate a product and then analyze them. In effect, simulation programs let them develop virtual stamping, milling, and extrusion machines that don't tie up production equipment. It is, they say, the future of finite-element analysis.
SIMULATE TWICE, CUT ONCE
A manufacturer of deep-drawn metal-stamping equipment simulates manufacturing to identify and troubleshoot tooling problems before building production tooling. "FEA provides insight into how dies, punches, and stamping presses will work by simulating tooling operations before production," says Peter Mayer, advanced development engineer at Jagemann Stamping Co., Manitowoc, Wis. "It also reveals how machines will behave. For example, while troubleshooting a problem, analysis showed that the cast-iron frame of a large stamping press could deflect up to 0.003 in. in some cases. Though deflection was not the source of that particular problem, FEA told us more about the machine."
Jagemann works mostly with tool steel, which is quite notch sensitive. "That means it's easy to generate stresses that crack the punch legs off. The best way to find a working punch for a new part is to model about three different designs analyze them, pick one with lowest stress, and then try it," says Mayer.
SEEING WHAT NO EYE CAN SEE
FEA let engineers at Castool Tooling Solutions in Scarborough, Ontario, Canada, see what was happening inside a dummy block, a multipart piston that works in the head of an extrusion press. The company die casts and extrudes aluminum. Die casters force molten aluminum alloys into molds, while extrusion equipment pushes a heated aluminum billet through a die to create long parts with complex cross sections, such as window frames. Simulating such operations helps the company develop die-casting and extrusion components that resist high pressures and temperatures.
Unfortunately, a key component in the dummy block was wearing too fast at high temperatures. It occasionally cracked and needed replacing. "One of the first simulations we ran was a simple static-stress and thermal analyses on the entire block," says Dan Dunn, a Castool application engineer. FEA indicated the location where the component would crack. "The software also showed how components interacted inside the block. The assembly-analysis capabilities in the FEA software lets us understand the impact of distortions and compressions. Now when normal wear necessitates part replacement, instead of replacing one large part, maintenance swaps out the replaceable ring, a relatively new part and wear item designed into the block — a far less-costly operation," he says.
Castool also uses analysis software to develop extrusion dies that work longer than conventional tooling. "One design trend toward smaller parts means dies must be made of harder alloys that can handle the high pressure needed to form smaller details," says Dunn. "But for designs that do not require higher than normal pressures, our tooling performs longer."
TESTING NEW MATERIALS
There is plenty of experience to draw on when working with common materials. But a new material with unfamiliar characteristics can give a company fits if it has few resources for the trial-and-error tests that would reveal work-arounds for the material's idiosyncrasies. Analysis software, however, and a material's basic properties are often all it takes to give manufacturers insights on how to work with unfamiliarities.
For example, Guill Tool & Engineering Co. Inc., West Warwick, R.I., makes dies for pipes, hoses, and other plastic, composite, and rubber parts. "Frequent new shapes and materials mean first simulating how they will perform or flow in the die. Fluid-flow software lets us find areas where material flow is low or stagnant, or the pressure uneven. Results from the software guide our equipment and die adjustments that can alleviate the problems," says Richard Guillemette, vice president of engineering.
Conducting a flow simulation along with thermal analysis lets engineers study temperatures in the die and, when necessary, inside the manifold. Thermal FEA lets Guillemette's team see how temperature changes alter the viscosity of extruding materials. "Ideal extrusion conditions call for uniform die temperatures," says Guillemette.
To some extent, the software allows simulating oscillating and rotary dies. These dies improve cross-linking of molecular chains in the plastic. (Material suppliers can provide viscosity values at elevated temperatures needed for the simulations.)
FEWER PROTOTYPE MOLDS
Some firms use FEA to develop molds. Bryco Tech LLC, Carrollton, Ga., designs and builds rotational-molded plastic parts using FEA software. Rotational molding spins a fluid plastic inside spherical molds to create relatively large hollow vessels and tanks. "Instead of going through lengthy prototyping cycles, analysis shows us what we could not have predicted," says company President Bryan Dunne.
For example, the software showed that two particular designs would have failed without additional ribs. This let the company eliminate at least two design iterations.
Dunne says the software provides useful data about products before they exist. For example, given the needed internal pressures and final material properties, the software estimates minimum vessel-wall thicknesses and product prices.
Of course, the software allows testing many designs. Seeing where loads are tells where to place ribs or if wider flanges would work better than narrower ones. What's more, adds Dunne, simulation results can be presented in AVI files that can be shared with clients to show why a design must be changed.
PUTTING PARTS ON A DIET
Before analysis was readily available, many parts were excessively beefy to account for uncertainty. But parts can be made with adequate safety margins and still cut material use by 5 to 10%. "Those are sizable figures when you consider we turn 100 tons of cast iron into manhole covers each day," says Steve Clinch, president of LeBaron Foundry, West Warwick, R.I. The weight translates to an annual saving of $500,000.
"Manhole covers have to support at least 10 tons, and that includes an impact safety factor," he adds. "To see if the covers could be trimmed of some material without reducing safety, we created a family of templates using FEA and CAD to provide cover designs with different safety factors. Analysis then pointed to the best geometry that uses less material."
In production, the company uses a 3D CAD model of a manhole cover to generate a plastic pattern. LeBaron then presses a series of one-shot, sand-and-clay molds for castings from that pattern.
Bryco Tech LLC, GoBryco.com, (850) 723-1264
Castool Tooling Solutions, castool.com, (416) 297-1521
Guill Tool & Engineering Co. guilltool.com, (401) 828-7600
Jagemann Stamping Co., jagemann.com, (920) 682-4633
LeBaron Foundry, lebaronfoundry.com, (508) 586-3130
SolidWorks Corp., cosmosm.com
Here's how I see FEA changing in the near term and over a longer period.
In two years, analysis will become more mainstream and a natural part of design. CAD programs will include motion and flow simulations just as SolidWorks now includes FEA.
Flow simulations in CAD will, for example, let valve designers or electronics-packaging engineers simulate flows with water or air. Simulations will be limited, but easy to use. Motion simulations will let users apply a motor to mechanisms and watch them operate. These will be kinematic studies at first and dynamic later.
Designers will find less need to apply rules of thumb for tasks such as how many bolts or ribs an assembly needs. They will examine models for how they carry loads and then build the necessary structure.
In five years, look for more computing power (of course) to change the way designers work. The possibilities can be imagined from the dual-core machines coming to market and in the quad-core processors that are due shortly. Eight-core computers may not be far behind.
The 64-bit OS will, just now making inroads to computer infrastructures, be more prevalent and make it easier to work with enormous models. This OS accesses more memory than 32-bit systems, and that will make a big difference for analysis.
Software as a service will be more widespread. For instance, why maintain 50 licenses of expensive analysis software when a hosting facility will maintain the software on high-speed computers for you? Engineering services will be similar to Google spreadsheets.
Today if you make model changes in CAD, you must hit at least one button to initiate an update to the FE model and then wait about a minute before the solver starts working. Then wait for an additional period for the solver to finish. In five years, FE model updates will be automatic and new simulation results will be available almost instantly.
Besides hardware, software speed will be credited to multithreaded solvers that make use of the available cores. The dual-core computers currently don't help much with simulations. But in five years, software will take advantage of however many cores it finds. The OS will likely assign some processors to immediate foreground tasks and analysis will be done in the background.
More complete and accessible material properties will change analysis. If you select plastic or rubber, today's solver selects the material model. What's missing is accurate material data. It is expensive, especially for new materials. For example, it's not easy to analyze a simple gasket because the material might be proprietary. Simulation companies will partner more with organizations such as Matweb and other Internet material databases.
In 10 years, elements might disappear. Simulation software will become meshless. Software will place nodes throughout a model making it look like a cloud. Nodes will flow into high stress areas with each iteration. Impacts and short duration events such as explosions, crash analysis, and air-bag inflations will be easier to examine.
Nanomaterials will have emerged from labs and components such as carbon nanotubes will be accounted for in simulations.