Supermercados y grandes superficies en el Comercio Justo: visiones

The cracks in the corners of the second iteration component were likely to be a result of biaxial stretching, such as is seen in the components in Bagudanch, Garcia-Romeu, et al. (2015). Plane strain is seen in flat surfaces formed by SPIF, but biaxial strain generated in edges and corners can increase the thickness strain and subsequently the likelihood of failure due to excessive thinning. Single and multi-pass toolpaths should be designed to ensure the material in corner features does not undergo thinning to the point of failure.

A major change which directly contributed to the success of the final part was improving the orientation of the model within the forming space. Part orientation was noted as an important consideration in Jeswiet, Micari, et al. (2005), and must be carefully evaluated for any component formed with SPIF. The resulting wall angles of the part surface in a particular orientation should be taken into account, as well as facilitating support walls with optimal design to support the forming process.

A particular type of error seen in the first iteration was wrinkling in a convex section of the geometry, shown in the photograph in Figure 4.13. The wrinkle was perpendicular to the toolpath direction and was attributed to undesirable flexing of

Figure 4.12: Photograph of the third attempt of the seat base component, successfully formed without cracks using bottom-up multi-pass SPIF

the entire part, rather than localised deformation only under the tool. The whole convex section was seen to flex each time the tool moved over the area. This shows the importance of stiffness in a part design, to support plastic forming rather than only elastic springback.

The second major change to the forming surface in the third iteration involved the support walls. Primarily, it was seen as important to ensure that the wall angles were in a mid range (40° - 60°) so they would form effectively and therefore be stiff enough to support forming the remainder of the part. The reasoning is as follows:

• Ensure the wall angles are well below the forming limit to eliminate any chance of failure by thinning.

• Ensure the wall angles are not too low, to minimise springback and encourage plastic deformation.

• Evaluate the trade off between wall angle and final thickness, because a lower wall angle means a greater final thickness, and a stronger support wall.

When designing the toolpaths, it was necessary to use a constant Z-level type for the first pass, as the tool must form the material at an even rate. However, bottom-up multi-pass toolpaths do not have this restriction, so an Equidistant Finishing type was selected. Equidistant Finishing generates a toolpath where each loop of the tool is a constant distance (equidistant) to the previous loop, resulting in a superior surface finish compared to Z-level. This also means the tool does not necessarily maintain a

Figure 4.13: Detail of wrinkling resulting from the multi-pass forming of the first attempt for the seat base component.

consistent increase or decrease in depth, leading to toolpath features such as the peak shown in Figure 4.14.

An interesting material flow pattern was observed in the completed part as a consequence of the multiple bottom-up re-forming passes. A lip developed around the part at the transition between the original model and the support walls, shown in the photograph in Figure 4.15. The previously discussed toolpath shape (from Figure 4.14) may have also exacerbated the situation.

In any case, the lip of material did not prohibit the usefulness of the component.

4.1.2.1 Geometric accuracy

For this case study, access was provided to a stamped version of the original compo- nent shown in Figure 4.1. To determine a benchmark level of geometric accuracy, or a maximum allowable deviation, the stamped component was scanned with a Faro Arm and laser attachment. The analysis suggested a benchmark maximum deviation of 1.5 mm, as more than 95% of the scanned points were within this margin of error. Geometric accuracy of the final part from Figure 4.12 was measured using a Faro Arm with a laser scanning attachment. The results, shown in Figure 4.16 are in the form of a colour scale on the surface with areas of positive and negative

Figure 4.14: CAD model of the second-last forming pass, with overlaid toolpath in yellow showing the peak in the trajectory of the tool.

Figure 4.15: Detail of the final formed seat base component, showing the folded-over lip created as a result of the multiple bottom-up re-forming passes.

geometric error. Red indicates+5 mm error and blue represents−5 mm, with the scale distributed in between. This final component did not fulfil the benchmark error of 1.5 mm, with much of the surface showing±2 mm deviation.

Comparatively large deviations are seen in areas of the flange, which would have occurred after unclamping from the forming rig (Figure 4.16 (a)). This is a known issue in SPIF, mentioned in Micari, Ambrogio, and Filice (2007) and Behera, B. Lu, and Ou (2016). It indicates that there is some residual stress in the flange which may translate to inaccuracies in the component if it is cut away from the flange.

The bosses on the large central wall (Figure 4.16 (a)) are not accurately defined. Acute and obtuse edges over the part also have issues with accuracy (Figure 4.16 (b)). Furthermore, the shallow mid-section bulged downwards compared to the model, presenting a type of error to be monitored in future components.