Optimizing Wear Resistance in Composite Materials: Surface Response Methodology
In This research aims to use Response Surface Methodology (RSM) in determining the settings that give the highest wear resistance of the developed (Cu-W) composites by varying reinforcement percentage at 20-40%, temperature at 160-200°C and mechanical load at 80-100N. A regression analysis was carried out to determine the parameters that affect wear rate and indicated that temperature and Cu-W percentage are significant factors affecting the wear rate with p total values 0.017 is significant. Optimization result suggested that at 30% Cu, 200°C and 80N load has the highest desirability of 1.00 with the best wear rate is 3.498 mm3/Nm. Contour and surface plots were used to further elaborate on synergy between the factors. The results of this study will be valuable for creating the adequate Cu-W composites composition that has decent durability in the applications that require the use of electric contacts and tools for high temperatures, in which wear is of great importance. This, therefore, shows how RSM is effective in optimizing the material with minimum experimental runs.
Institutions
- Government Medical College (IN)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2025-05-29
- DOI
- https://doi.org/10.5281/zenodo.15574061
- Primary Topic
- Aluminum Alloys Composites Properties
- Type
- article
- Field-Weighted Citation Impact
- 0.00