No. 19 - Injection Molding Simulation

Until the late 1970s, the two tools molders employed to optimize mold filling were trial and error. Although molders had some idea of a resin’s processing characteristics, there was little understanding of its behavior inside the mold.

Many basic parameters of mold filling—such as the pressure required to optimize filling and packing, gate placement, gate and runner sizes, cooling behavior, and how mold design influenced part density, warpage, and surface appearance—were determined more by experimentation than certainty. The result was long mold commissioning trials, lots of wasted resin, and frequent reworking of the tool.

A faster and more efficient approach, which has become indispensable to many molders and OEMs in automotive and other industries, is computer modeling or software simulation. This technology had its beginnings in the mid-1970s. Prof. Musa Kamal at McGill University in Montreal worked on theoretical models of flow behavior, while G. Williams and H.A. Lord at AT&T Bell Laboratories focused on physical flow visualization. Around the same time, plastic flow and cooling behavior was being analyzed at the IKV (Institute for Plastics Processing) in Aachen, Germany, and at Cornell University in Ithaca, N.Y., where Prof. K.K. Wang directed the Cornell Injection Molding Program.

In 1978, Colin Austin, an Australian plastics design and processing consultant and lecturer in plastics technology at the Royal Melbourne Institute of Technology, drew on the work being conducted in these different research programs, especially those at Cornell and AT&T Bell Labs. He combined this theoretical base with the emerging developments in computers. In the days before personal computers, global computing was advanced by time-sharing networks, which gave users remote access to high-level computing power by dialing into it.

Through these means, Austin developed the first flow-analysis software that modeled in-mold pressure and cooling behavior by using observable parameters like gate and runner sizing and part thickness as controlling variables for analysis. Austin founded Moldflow Corp. in 1978 to market the first commercial injection molding simulation software.

The original software created a 2D “layflat” simulation of the mold-filling process. The user had to predict the filling pattern in advance and specify an injection pressure and total filling time. The computer divided the fill time into equal segments and provided a numerical table of pressure drop, temperature, and fill time for each segment.

In 1982 the company developed a 2.5D method, called midplane analysis, which yielded a more accurate reflection of the filling and cooling process. This method was based on a CAD representation of the part and it predicted the filling pattern, rather than modeling flow based on the user’s “best guess” at the filling pattern. This method used finite-element meshing of the part and calculated the temperature, pressure, velocity, and shear within each tiny element as it filled. From those calculations, the software determined how neighboring elements would fill. In this way, the software deduced the filling pattern on its own.

Other companies active in flow simulation by the early ’80s included several CAD software vendors and Plastics & Computer of Milan, Italy, founded by Giorgio Bertacchi from DuPont Europe. Also, AEC Inc., a maker of mold chillers and temperature controllers, created MoldCool, the first commercial mold-cooling simulation (now owned by C.A.E. Services Corp., Batavia, Ill., started by two former AEC employees). In 1986, the Cornell program spawned AC Technology, later called C-Mold. It became Moldflow’s strongest domestic competitor until it merged with Moldflow in 2000. Recently, simulation technologies from Germany (Sigmasoft) and Taiwan (Moldex3D) have made a push into the U.S. market.

In the last 15 years, simulation capabilities have expanded to encompass shrinkage, warpage, gas-assist, coinjection, multi-shot injection, injection-compression, and thermoset molding. Another significant development in the late 1990s was “true 3D” simulation, which for the first time made it possible to model accurately thick, solid parts that did not lend themselves to 2.5D midplane analysis.