Nonreplicated factorial experiments assessing print quality in additive manufacturing

Additive manufacturing techniques such as fused deposition modeling (FDM) are commonly used within engineering for rapid prototyping of designed parts. The Endeavor Lab at Oklahoma State University (OSU) takes advantage of this technology and provides 3D printing services to engineering students. To determine which variables of the Bambu P1P printing process most significantly affect the final print quality, a 25 factorial experiment was designed. Five different process variables were tested at a low and high level. The experimental design included only one replication to simulate time and budget constraints, which are very common in small-scale engineering experiments. The printed specimens were tested according to ASTM D638-14, and the ultimate tensile strength (UTS) for each specimen was recorded. The data were analyzed according to methods presented by Milliken and Johnson (1989). Because the use of a single experimental replication results in a lack of true experimental error, multiple analysis methods were examined to compare results. Techniques including Analysis of Variance (ANOVA) and half-normal plot of effects were employed. For this experiment, traditional ANOVA methods concluded that an interaction between print direction and location on the build platform contributed to a marginally significant (p = 0.1025) effect on UTS for the printed specimens. The alternative half-normal plot approach to the analysis also indicated that the same interaction was marginally significant. Additionally, no obvious drift effects were observed for the printing process. Further experimentation into the effect of print direction and location should be pursued to more confidently conclude the effects of these variables on the UTS of the printed specimens.