Flow analysis finds counter-intuitive solutions.
What's especially interesting about this project is that it illustrates how computerized flow analysis can help find solutions that run counter to the intuition of an experienced engineer. It also shows how close communication between the person doing the flow analysis and the design and manufacturing engineers helps arrive at a solution that is both functional and manufacturable in the least time.
The project involved an injection molded PVC winged inserter used in a new catheter product. The inserter consists of a tube about 0.75 in. long with wings on either side. The inserter serves to connect a catheter that goes into the patient's vein to tubing that runs to a container of I.V. solution. The wings are used as a tapedown platform and are pinched together like butterfly wings in order to grip the needle and aid insertion.
These wings are about 1.5 in. wide and 0.055 in. thick, with a tapered rib along each side that pinches the tube when the wings are flexed. The wings thin down abruptly to 0.035 in. at the point where they attach to the central tube, in order to allow them to flex. However, the upper half of the wings attach to the tube through an additional narrow strip of webbing 0.010 in. thick.
Early prototypes of the part were single-gated on top of one wing. The problem with this arrangement arose from the fact that the wings are considerably thicker than the tube for functional reasons. During molding, injected melt would first fill the gated wing, then flow around the tube into the other wing, and finally, when enough backpressure built up, the central tube would be the last to fill. By the time the two flow fronts met each other on the far side of the tube, the melt was too cold to blend, and a knit line formed. During prototype testing, leakage of blood or I.V. solution was experienced at the knit line. This had to be corrected quickly in order to meet the product development schedule.
CAD/CAE TO THE RESCUE
Computerized flow analysis had not been performed prior to the prototype stage. When he was notified of the problem, Dr. Twfiq Gangjee, project leader in computer-aided engineering for injection molding at the research center, began by requesting drawings of the part and tool and literature on how the part functions, existing processing conditions, and material specifications. He then generated a CAD solid model and finite-element model of the part using IDEAS software from Structural Dynamics Research Corp. (SDRC), Milford, Ohio. Using a conversion program written at Becton Dickinson, the finite-element model was converted to an input file for the Moldflow flow-analysis software package from Moldflow, Inc., Trumbull, Conn.
Gangjee first performed a flow analysis of the part with the original gate location on one of the wings. The graphic 3-D image of the cavity resulting from this analysis depicted the knit line in virtually the exact position in which it actually appeared on the prototype. Examining a temperature-profile "map" of the part, Gangjee could see that the melt that entered the mold at 392 F cooled to 280 F at the point where the two flow paths finally met in the tube 0.75 sec later. This fairly "cold" melt resulted in a knit line rather than blending. Thus, the goal of Gangjee's design iterations was to find either a configuration in which the knit line would be moved to a less critical location or one in which the flow path would be reduced to increase the melt temperature at the knit line.
To minimize tooling costs, Gangjee first tried to retain a single gate, exploring different locations on both the wing and central tube. Simultaneously, different wall thicknesses were also tried in an effort to get rid of the knit line. Gangjee did find that by increasing the wall thickness of the central tube, a more even fill pattern could be achieved, and this would eliminate the knit line.
However, division engineers present while the analyses were being performed informed him that the required tube thickness would impair functionality of the part. With a thicker tube, it would have been impossible to pinch the wings together to grip the needle tightly enough. This much was determined in about 4 hr using computerized flow analysis. Gangjee estimates that it would have taken several weeks and thousands of dollars in tooling costs to obtain the same information by conventional build-and-test methods.
Realizing that he would have to live with the formation of the knit line on the central tube, Gangjee next considered increasing the temperature of the polymer flow fronts at the knit line in order to make it stronger. Division engineers pointed out that the melt temperature of the polymer could not be increased, due to PVC heat-stability constraints and the need for additional cooling time, which would add to cycle time. Gangjee then tried a number of different dual-gating schemes with gates at various points on the upper and lower edges of the wings and at different distances from the central tube.
Intuitively, it seemed to all parties involved that bringing the gates in as close as possible to the central tube would provide the minimum flow path and maximum melt temperature at the knit line. Yet intuition is frequently wrong in plastic molding. The analysis showed that with the gates very close to the tube, the polymer would flow the short distance to the tube and stop while the rest of the wing filled. After running a number of iterations. Gangjee determined that the optimum gate locations that minimized cooling of the melt were actually toward the outer edges of the wings. Under these conditions, the analysis predicted that the temperature where the flows met would be 322 F, which Gangjee felt would allow the polymer to blend without forming a knit line.
So, by the end of one day, the division engineers had a new gating scheme and molding parameters that had been verified by computer analysis. However, this was not quit ethe end of the story. Because of frequent tears during field trials, division engineers decided to double the thickness of the 0.010-in. webs where the upper half of the wings connected to the tube.
When the tool was modified and tested, the engineers found that a pinhole was forming in the central tube. They went back to Gangjee, who performed an analysis of the modified design. This showed that an air pocket was formed at the precise location of the pinhole by several converaging flow fronts.
Gangjee then performed a series of analyses with web thicknesses varying from 0.010 to 0.035 in. in increments of 0.005 in. These showed that at 0.010-in. thickness, air vented naturally from both ends of the tube. But at web thicknesses from 0.015 to 0.25 in., the chance of air entrapment by converging flow fronts and consequent pinhole formation increased. Below 0.015 in. and above 0.025 in. thickness, air was able to escape from the vents and no pinhole formed. This was another counter-intuitive flow-analysis result. Without the analysis, Gangjee would have assumed that further increasing the web thickness past 0.020 in. would not have removed the pinhole. Division engineers informed Gangjee by telephone that a 0.035-in. web was not acceptable from a functional standpoint. Since the web tearing was much less of a problem than the pinhole formation, the engineers decide to return to the original 0.010-in. thickness.
This approach worked perfectly and the part is now in production. Gangjee estimates that the company saved tens of thousands of dollars and several months compared with what would have been required to perform the same work through conventional trial-and-error methods.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||CAD/CAE; computerized mold flow analysis determines exact changes in gate location|
|Author:||Naitove, Matthew H.|
|Date:||Oct 1, 1991|
|Previous Article:||Retrofit package to enhance injection molding performance.|
|Next Article:||New spray-coating processes for plastic powder.|