Relationship Between Students’ Spatial Visualization Ability and their Ability to Create 3-D Constraint-based Models from Various Types of Drawings

During the fall of 2013, a study was conducted where students completed the PSVT:R and the MCT and then were asked to complete three different modeling tasks. These tasks included modeling a part when given the object in the context of an assembly drawing, modeling a part when given an isometric pictorial of the object, and finally modeling a part when given a detail drawing of the object. Research questions for this study include the following: Does a simpler rubric provide similar scores as the previous rubric? Is there a relationship between a student’s spatial visualization ability and their ability to model a part from a pictorial, assembly drawing or detail drawing? Is the MCT still a better predictor of a student’s modeling success than the PSVT:R? Scores evaluated based on the simplified rubric were significantly higher than the scores evaluated based on the original rubric for the part modeled from the assembly drawing. This study revealed significant positive correlations between each modeling task and the two spatial visualization tests. Review of Literature Engineering graphics educators have been studying the computer-aided design modeling strategies of students for approximately 20 years. Studies involving constraint-based modeling have been more recent. These studies include students’ modeling strategies, conceptual framework research, and methods of evaluating models. CAD Modeling Strategies In a study of 34 technology education students, Chester found that strategic CAD instruction was more effective on novice CAD users than on CAD users with previously reported experience . The experienced CAD users had formed strategies based on previously learned software that impeded their learning efficient strategies in the new software. He also makes some observations about research related to CAD . Initial instruction and experience in the field does not necessarily lead to expertise. Experienced users still exhibit modeling strategies that are not optimal. Also, expertise is more about being able to apply strategic knowledge than just differentiating between command knowledge. True experts have knowledge of modeling strategies, and they know when to apply them appropriately. New CAD users tend to take very erratic approaches to modeling when they lack knowledge in design intent. Even when young students have a background in technical graphics and descriptive geometry, they have difficulty applying this knowledge to create CAD models with good design intent . Conceptual Framework Research Several conceptual framework models of CAD expertise have been developed over the last 15 years. Hartman developed a model of CAD expertise after studying practicing professionals in their native work environments . He reports core themes, subordinate themes, and transitional themes demonstrated by experts. Core themes include strategy for tool use, problem definition and solution, design considerations, domain knowledge, and professional and academic experiences. The subordinate themes are software usage techniques, downstream uses of the CAD model, technical communication, social communication, requisite CAD model characteristics, and problem solving techniques. Finally, his transitional themes include the design environment, the way the expert worked, support structures, artifacts, personal characteristics, typical domain activities, conceptions of expertise, and factors related to CAD usage. Rynne & Gaughran 20 present a framework for parametric modeling and discuss attributes that experts demonstrate that are often missing from the models of novices. These attributes include the following: correct sketch plane selection for the base feature sketch; optimum model origin; correct base feature; correct part orientation; appropriate use of symmetry planes; simple sketch geometry; correct sketch relations; fully defined sketch geometry; correct feature sequence; parent-child feature relations; correct feature terminations; correct feature duplication; correct part design intent; and part accommodates planned and unforeseen design modification without feature failure. They further define the components of CAD expertise to include the general categories of the part modeling task, procedural 3D CAD knowledge, strategic 3D CAD knowledge, declarative 3D CAD knowledge, graphical and visualization capability, modeling deconstruction capability, and metacognitive processes . Evaluating CAD Models Studies involving the evaluation of CAD models have been quite diverse. In a study of the correlation between parametric modeling ability and performance on the Mental Cutting Test, Steinhauer used the general categories of approach, structure, accuracy, robustness, and creativity to assess students’ models . In a comparison of manual and online grading of solid models, Ault & Fraser evaluated models based on the following characteristics: correct geometry, appropriate choice and order of features, proper location of origin, proper view orientation, use diameter and radius dimensions correctly, correct hole placement, use of reference geometry for dependent features, and general modeling strategy . Baxter and Guerci developed an automated grader for solid models. Their system allowed the instructor to determine exactly which constraint-based attributes were present in the student’s file, but they do not offer a specific rubric to follow 2, . More recently, studies have been conducted to examine students’ ability to build specific design intent into models. Devine & Laingen outline a procedure used in their course that students can use to self-assess their models . Students are given two self-check opportunities where they must measure a distance, one face area, and the total face area of the part. For the second self-check, students are required to change several dimensions on the part before measuring. This allows the instructor to determine if the initial dimensions captured the correct design intent. Peng et al. advocate for a similar approach . Finally, Company, Contero & Salvado-Herranz summarize how they define the quality of a CAD model with the following five dimensions (p. 2) : 1. Models are valid if they can be opened by suitable applications, and do not contain errors or warnings. 2. Models are complete if they include all product aspects relevant for design purposes. 3. Consistent models should not crash as a result of editing tasks or design exploration. 4. Conciseness pursuits models that do not include irrelevant information or procedures. 5. Effective CAD models convey design intent. Engineering Graphics Literacy A recent series of studies investigated students’ ability to model parts when given assembly drawing information 4-7, . These investigations revealed that the developed modeling test, as measured by the original rubric , had mixed results when examining relationships with measures in the course (e.g., final project and final exam). Some of these studies also examined whether students’ modeling ability was related to their spatial visualization ability 4, 5, . There were positive correlations between the PSVT:R and the modeling test (not all were significant) and significant positive correlations between the MCT and the modeling test. Recommendations included repeating the study using a shorter modeling activity, examining a more efficient way of evaluating the models, and using qualitative methods for evaluating modeling strategies.