612 Training Engineers in the Use of Constraints to Create Quality 2D Proles for 3D Models Raquel Plumed 1 , Carmen González-Lluch 2 , Jerey M. Otey 3 , Victoria Pérez-Belis 4 1 Universitat Jaume I, plumed@uji.es 2 Universitat Jaume I, mlluch@uji.es 3 College of the Ozarks, jotey@cofo.edu 3 Universitat Jaume I, belis@uji.es Corresponding author: Raquel Plumed, plumed@uji.es Abstract. Historically, dierent strategies utilized in engineering and technical graphics re- ect growing concern about improvement in CAD instruction and the introduction of product quality data in curricular activities. Our vision was that early introduction of quality criteria through "good practices" is feasible and increases the quality, robustness, and reliability of CAD models. This paper describes a strategy applied to improve the knowledge of novice CAD users on the use of geometric constraints in 2D parametric proles. This approach consists of supplementing student training with activities in order to provide rapid and eective feedback. A Chi-Squared Test was performed to assess the eectiveness of this strategy, indicating that trainees need continuous and additional autonomous learning to create quality 2D parametric proles. Future work will include developments to promote student awareness of the need for quality in 3D models using an online checker that acts as a lter of semantic quality errors while providing feedback. Keywords: CAD model quality, Sketching, Constraints DOI: https://doi.org/10.14733/cadaps.2021.612-623 1 INTRODUCTION The development of computer-aided three-dimensional design applications (CAD 3D) has transformed the product development process and has introduced a new paradigm of Model-Based Enterprise (MBE). This principle draws on the use of annotated CAD models as primary elements to support the design, analysis, and manufacture of industrial products. These annotated CAD models contain data and additional information necessary for production and support. For this reason, CAD model quality is essential, since the quality of manufactured products depends on the quality of their data [22]. Consequently, poor data quality compromises CAD model reuse, which is a primary benet of history-based parametric modeling software. Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 613 Three quality levels were identied [13] to classify CAD models: morphological, syntactic, and semantic/pragmatic. The morphological level is related to the geometric and topological correction of the CAD model. The syntactic level measures the proper use of modeling conventions. The semantic/pragmatic level is associated with quality that accounts for the capacity of the CAD model for subsequent modications and reuse. A model is reusable if it allows modications in other situations while maintaining its design intent [16]. Therefore, this study considers as a good model the one that is reusable and is at the same time robust and exible [6], [7]. Reusability and the interoperability of a model are the most common functions in MBE. Previous studies revealed that nearly 50% of CAD models fail after making alterations [20]. In our current work, we focus on feature and history-based variational parametric CAD modeling. These CAD applications enhance the creativity of designers, since they allow exploring various alternatives and solutions during the design process of a product. In addition, this modeling strategy is commonly used to create annotated models in the MBE paradigm, shortening design time and increasing business productivity since the reuse is its primary benet. Specically, we are interested in analyzing models created using SolidWorks®since this application is used to introduce students of graphic engineering courses to constructive geometry. At present, commercial tools exist for Model Quality Testing (MQT) or Quality Testers to detect and correct failures. However, a recent study reports that these tools are limited to analyzing the most elementary aspects of CAD model geometry [15], while quality aspects at the semantic/pragmatic level appear to be absent (ex. the use of x constraints), which compromises model reusability [17]. In this study, GonzálezLluch et al. [15] analyzed the capability of the SolidWorks Design Checker®(Model Quality Testing embedded in SolidWorks®) showing that high semantic level quality criteria are not considered during the process and that the checker does not provide much more help and feedback to users of what they receive while modeling. Constraints are commonly used to acquire robust and exible proles that allow for redesign while prevent undesirable geometric changes. Robust proles must be completely or fully constrained [9]. Prole exibility does not depend on the quantity of constraints, but on their semantic level. The proper selection and introduction of geometric constraints in 2D proles determines their applicability for reuse. Various authors have proposed dierent classications constraints [24], [1], [11]. We classify constraints as: ˆ Dimensional. These constraints dene the size and dimensions of the prole. ˆ Geometric. These constraints dene the geometric relationships between the elements of the prole. ˆ Position and orientation. These constraints relate the prole to the coordinate system. We strongly believe that over-constrained proles with redundant relationships are more dicult to edit than those that avoid redundancies. Many experienced CAD instructors have observed that engineers often use redundant relations when creating 2D proles and that this practice prevents the creation of reusable CAD models. Our goal is to train novice CAD users to create quality parametric 2D proles based on robust and exible 3D models for future reuse. When incompatible constraints are introduced during the creation of 2D proles, the user is alerted by the system. In a previous step, the authors compared the behavior of some representative commercial 3D CAD systems including SolidWorks®when incompatible constraints were added to a prole. In all cases, the systems showed a warning and/or stopped execution until the user changed the strategy. However, the behavior is not homogeneous regarding the detection of redundant constraints. In this case, some applications (such as SolidWorks®) do not issue any warning to the user. In the literature, a previous work [18] conducted an experiment examining if engineers were able to: (1) identify fully-constrained proles and (2) detect the types of constraints that were used. Results indicated that more than half of trainees failed to identify which proles were properly constrained and they also could not identify the type of constraints used in a given example. We concluded that improvements were necessary in training engineers in these skills. A survey of past eorts to improve training of novice CAD users reveals that the following strategies have been successfully implemented: Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 614 ˆ Development of rubrics for assessment of CAD models [9], [2], [12], [8]. ˆ Creation of activities to increase student awareness of the methodological aspect of CAD model construction [5], [4], [14], [23]. ˆ Use of automated electronic tools to homogenize and improve CAD grading [21], [3], [25], [19]. ˆ Prompt feedback used to improve modeling strategies [21]. With this approach and from our teaching experience, we rmly believe that users must be trained from the outset to create 3D models considering the quality aspects, in order to design eective and reusable models. A suitable model must incorporate reusability, robustness and exibility capabilities. This paper is one of the rst steps in this direction. In this paper, we present a strategy to reinforce student training through simple exercises paired with quick and eective feedback. Currently, students enrolled in graphic engineering courses at the Jaume I University, are trained with SolidWorks®through a student license which remains active for an entire year. SolidWorks®is a CAD program widely used by companies, has an intuitive interface which makes it easy to use for beginners and experienced users, and the educational version provides resources for teaching mechanical CAD, design validation and data management. This approach facilitates student performance in identifying and avoiding redundant 2D constraints when drawing 2D proles. The following section describes the technique that is applied to reinforce this training. Then, this methodology is experimentally evaluated, comparing the results obtained from students who followed the reinforcement activity against a control group comprised of those who did not. Finally, conclusions are drawn from this experiment and plans for future work are elucidated. 2 METHODOLOGY In a previous study by González-Lluch and Plumed [18], a pilot experiment was performed with students enrolled in a Graphics Engineering course (third sequential course in an undergraduate Mechanical Engineering curriculum). The results indicated that novice CAD users failed to identify constraints in sample CAD les even when the examples were at low levels of diculty. Students were trained to create robust and exible 2D proles following the rst chapter of an instructor-authored text [10]. This book is aimed for basic 3D CAD courses in Mechanical Engineering and Industrial Design and Product Development degrees grades and in the rst chapter, three-dimensional geometric modeling and parametric design of proles are covered. Moreover, they also received instruction on constraints in additional theoretical and practical-based classes (using SolidWorks®). In a continuation of this research focus, a new strategy has been introduced to students in a subject entitled, Computer-aided Design II, which is the third sequential course in an undergraduate Industrial Design and Product Development Engineering curriculum. This subject, along with the previous one, use the same text [10] and both groups of students have similar backgrounds in technical drawing. The didactic proposal in the subject structures the teaching-learning process into two training activities: lectures and practical teaching (lab sessions). geometrical and dimensional constraints. During the theoretical classes, students are introduced to A common method to identify the quality of 2D proles is to quantify their degrees of freedom. Each geometrical element is dened by xing a certain amount of degrees of freedom (DOF). Simultaneously, each constraint restricts a certain number of degrees of freedom (valency). To obtain a robust and quality 2D prole, it must be completely constrained. In other words, the total number of DOF of the prole must be equal to the total valence of all its constraints. Rules to eciently restrict the proles are provided to the students: the rst step should be restricting the shape and size of the prole (prior to the position) until a rigid gure is reached. This step helps to avoid extrinsic restrictions that relate elements of the gure with external datums in order to make the prole more exible in terms of location and movement. Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 615 Figure 1: Example of a question with incorrect answer and the feedback obtained In practical classes, during the creation of 2D proles in SolidWorks®, students apply the theoretical training while a prole is created. The system provides feedback about this process that has been previously explained by the teachers. In the rst eighteen exercises, students create dierent 2D proles and during the process, students can check their prole [10]. If the prole is under-constrained, a minus sign (-) precedes the prole name (tree model). Additionally, the congurations' menu allows the user to congure the program to assist in detecting fully constrained sketches by line color. On the other hand, the system warns of incompatible constraints which are included in the name of the prole, but the system does not provide warnings about redundant constraints. A supplemental activity was designed to improve novice skills with respect to understanding 2D constraints and consists of an eight-question survey, with corresponding gures. The online questionnaire was delivered through a virtual classroom environment as an optional assignment. Each isolated question is displayed full screen, to capture a student's full attention, in multiple-choice format. Questions were delivered in random order. If a student supplies an incorrect answer, the questionnaire provides comments with explanations of the correct answer, as shown in Figure 1. The online platform collects a register of student responses. Student were unable to alter their answers once they were submitted and there was no time limit to respond. Table 1 shows the survey questions and the bolded correct answers. Forty responses were collected from the virtual classroom. Frequency of student answers is also included in the right column (Freq). Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 616 Questions Figures Answers Freq. a. True 20% b. False 80% Is the following 1 sketch completely constrained? 2 Select the response that explains your answer to the previous question a. The sketch is over- 20% constrained; the parallelism constraints (A) are redundant since both Edges A are horizontal b. The sketch is under- 30% constrained c. The sketch is completely 37.5% constrained d. DK/NA a. 12.5% Edge A will remain ver- 10% tical and Edge B will remain 3 How will Edges A horizontal and B of the b. following sketch strained, so no changes are behave when the permitted angle of 45 ◦ is The sketch is fully con- 25% c. Edges A and B will 65% maintain the constraint of perpendicularity ◦ altered to 60 ? d. Edge B will remain  horizontal and Edge A will change a. An equal length constraint What constraints should be applied to all Edges should be applied so A that all edges A 4 b. have the same Collinearity constraints 60%  should be applied to Edges B dimension (height) dimension is c. Both answers a) and b) 40% are correct modied to 20 mm? d. Each edge should be indi- when the 30 mm  vidually constrained Continues on next page Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 617 Questions Figures between Answers Freq. a. The equal constraint 85% should be applied to both circles Circumferences A so b. that when the Ø20 have dimension is straint What constraint should be applied 5 modied, the two The Ø20 circle should a concentricity c. Equal and horizontal con- circumferences have 5% con- 10% straints should be applied the same diameter d. Each circumference should (equal dimension)?  be individually constrained a. No constraints are needed 15% since the sketch is already fully constrained b. The collinearity constraint In the following tices of the hexagon with re- hexagon, what 6 2.5% should be added to all verspect constraints should to the circumscribed circle be added to fully c. constrain the gure? The diameter of the cir- 15% cumscribed circle is missing d. All vertices of the 67.5% hexagon need the coincidence constraint with respect to the circumscribed circle. The Ødimension is also required a. The height of 35 mm will  increase, but the sketch will not deform b. The values of the outer 5% arc and the circle of Ø25 will increase, but the sketch will What will happen if 7 not deform R25 is modied to c. R30? The sketch will be de- formed because there 2.5% are missing constraints d. The sketch will main- 92.5% tain its shape, the dimension of Ø25 will remain unchanged, and the width of the gure will increase Continues on next page Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 618 Questions 8 Figures Answers How would the fol- Freq. a. Fully constrained 15% b. Under-constrained 55% lowing sketch be dened in Solidworks? c. Over-constrained (the 30% angular constraint is redundant) Table 1: Summary of the questionnaire and alternate answers. Correct answers are in bold. In the light of these results, we notice that when determining if a sketch is completely constrained, students appear to focus only on dimensional constraints, while completely overlooking geometric constraints. For this reason, questions about dening a prole as fully, under, or over-constrained have low ratios of student success. Nevertheless, students understand the function of geometric constraints (i.e. how the constraints inuence the prole's shape when any dimensional constraint is modied). In order to gain more insight about student impressions of the activity, the questionnaire also queried their opinions of the eectiveness of these exercises (Table 2). The most frequent answers are bolded. Questions 9 10 11 12 Has this questionnaire aided you in an increased understanding of how to Answers YES 92.50% use constraints in sketches to build 3D models? NO 7.5% YES 30% Has this questionnaire helped you to distinguish between NO YES 70% 77.5% under-constrained, over-constrained, and fully constrained sketches? NO 22.5% YES 95% NO 5% Have you used or been made aware of any new geometric constraints? Do you think it will be useful for sketches you create in the future? Table 2: Student opinion about the eectiveness of the training. According to the answers obtained, we notice that students showed positive opinions about the utility of the questionnaire to dierentiate between fully, under-, and over-constrained sketches. It is also notable that a majority of participants recognized all symbols and geometric constraints used in the questionnaire. 3 METHODOLOGY ASSESSMENT The eectiveness of the training activity was assessed during the midterm exam. The students were required to solve two questions related to constraints, using the same questions that were proposed by González-Lluch and Plumed [18]. The rst question queried students about whether the prole shown in Figure 2a was fully-constrained, over-constrained, or under-constrained. The correct response is that this 2D prole is fullyconstrained, and this answer is easily veried when using a 3D CAD application (ex. Solidworks®), as shown in Figure 2b. Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 619 Figure 2: Left: The rst question of midterm exam; Right: Verifying the same sketch using Solidworks® The second question referred to the same sketch used previously. Students were required to identify and locate each type of constraint (listed below): ˆ Dimensional or geometric (F) ˆ Position and orientation (P) ˆ Others (B) The students also were required to explain their answer. Considering the possibility of multiple correct answers, we consider that the 2D prole includes the following constraints (Figure 3): Figure 3: Correct answer for second questions 3.1 Assessment Results Fifty-eight answers were collected and considered for the study. Forty students followed the reinforcement activity, while eighteen did not. All students answered both questions, and we classied the answers according to whether students completed the training activity. With reference to the rst question, Table 3 summarizes Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 620 Training No Training Fully-Constrained Total Under-Constrained Over-Constrained Table 3: 27 11 Frequency 67.5% 61.1% Total 3 2 Frequency 7.5% 11.1% Total 10 5 Frequency 25% 27.8% Answers for the rst question. the total answers given by the students in each group, along with their corresponding frequencies. The correct response is that the sketch shown in Figure 2a depicts a fully-constrained prole. Results reveal that more than half of the students correctly answered the rst question. In fact, the percentage of correct answers was slightly higher in the trained group (67.5%), than in the group of students that did not perform the reinforcement activity (61.1%). Students who answered incorrectly considered the prole as over-constrained in both groups (trained group, 25%; without training, 27.8%). To determine whether there is a dierence in success rates between groups, answers were grouped into correct (fully-constrained) and incorrect (under- and over-constrained) categories. A contingency table was then constructed in order to compare the rate of correct answers given by each of the two groups. The observed count and expected frequency results for each group are displayed in Table 4. Training No Training Total Correct Incorrect Count 27 11 38 Expected Frequency 26.2% 11.8% 38 Count 13 7 20 Expected Frequency 13.8% 6.2% 20 40 18 58 Total Table 4: Comparison of the success rates for the rst question. We contrast the null hypothesis (H0 : There is no dierence in the success rate between groups with dierent training) using a Chi-Square Independence Test. No signicant relationship exists between these 2 variables [X (1, N=58) = 0.2248, p< .05], thus we can conclude that there are no signicant dierences in the success rates of the groups. The training reinforcement through the questionnaire has no impact on student ability to recognize fully-constrained proles. A similar procedure was performed on the results from the second question (identifying constraint types as either geometric/dimensional, position/orientation, or others). Table 5 summarizes student answers, with incorrect answers categorized and described. Results reect that although the percentage of correct answers of the trained group (37.5%) is slightly higher than the No-trained group (33.3%), students generally are decient in their ability in classifying the types of constraints. To determine whether there is a dierence in success rates between groups for the second question, answers were grouped into correct and incorrect answers. Table 6 shows the observed count and expected frequency results for each group. Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 621 Training No Training Correct answer Total Ill-dened equality constraints Ill-dened perpendicular constraint Ill-dened angular constraint Ill-dened linear dimensional constraint The origin of the reference system is not xed Table 5: 15 6 Frequency 37.5% 33.3% Total 7 4 Frequency 17.5% 22.2% Total 10 5 Frequency 25% 27.8% Total 10 8 Frequency 25% 44.4% Total 3 1 Frequency 7.5% 5.5% Total 7 4 Frequency 17.5% 22.2% Answers for the second question. Training No Training Total Correct Incorrect Count 15 6 21 Expected Frequency 14.5% 6.5% 21 Count 25 12 37 Expected Frequency 25.5% 11.5% 37 40 18 58 Total Table 6: Comparison of the success rates for the rst question. Using a Chi-Square Independence Test, the results reveal that there is no signicant dierence between 2 the answers [X (1, N=58) = .0933, p< .05]. Therefore, reinforcement training does not appear to improve student outcomes in identifying constraint types. One-time training does not improve student performance in the use of constraints, so we conclude that students need continuous and autonomous learning. As a response, future work on this ambit should include both the creation of exercises necessary to create awareness about the importance of reusability in 3D MCAD models, and the design of an online checker. The latter could provide model quality information to the students and lter errors at the semantic level in 2D parametric prole drawings. Future training should primarily address the construction of 2D parametric proles without redundant constraints. 4 CONCLUSIONS Our driving idea is that the eective use of constraints during the creation of 2D proles to build 3D models improves CAD model quality since constraints help to convey design intent and permit subsequent reusability of 3D CAD models. According to previous studies, many designers, engineers, and students oftentimes apply redundant constraints during the creation of parametric proles when using 3D MCAD applications. In this paper, we describe a tentative methodology used to reinforce the training of future engineers and Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 622 designers through a collection of online exercises with self-correction. The goal is to improve the ability of new users of 3D MCAD applications to create quality parametric proles, while avoiding the use of redundant constraints. Results reect that the ratios of correct answers are slightly higher in students who followed the training, but the dierences are not statistically signicant. The lesson learned from this experiment can be summarized by saying that students seem to forget geometric constraints, considering only dimensional constraints. We conclude that students need continuous and autonomous learning. Furthermore, results of the questionnaire reected positive opinions about its utility. The next step in this process consists of building exercises designed to create awareness in trainees about the importance of reusability in 3D MCAD models. As a future development, we suggest another useful strategy, which is to design an online checker which would act as a lter of quality errors at the semantic level in the 2D parametric prole drawing. In this way, students would have the tools necessary to train themselves by performing the exercises and receiving automatic feedback. This strategy is expected to provide two advantages: reducing the teaching workload and developing independent learning capacity in the students. ACKNOWLEDGEMENTS The research work reported here was made possible by the Universitat Jaume I through project UJI-A2017-15 and DPI2017-84526-R (MINECO/AEI/FEDER, UE), project CAL-MBE. ORCID Raquel Plumed, http://orcid.org/0000-0001-8018-8039 Carmen González-LLuch, http://orcid.org/0000-0002-9839-5615 Jerey M. Otey, http://orcid.org/0000-0002-3763-8759 Victoria Pérez-Belis, http://orcid.org/0000-0002-7545-917X REFERENCES [1] Ault, H.: Using geometric constraints to capture design intent. Journal for Geometry and Graphics, 3(1), 3945, 1999. [2] Ault, H.; Bu, L.; Liu, K.: Solid modeling strategiesanalyzing student choices. In 121st ASEE Annual Conference and Exposition, 2014. [3] Ault, H.; Fraser, A.: A comparison of manual vs. online grading for solid models. In 2013 ASEE Annual Conference & Exposition, 2326, 2013. [4] Brano, T.: Constraint-based modeling in the engineering graphics curriculum: Laboratory activities and evaluation strategies. In In Proc. Midyear Conf. Eng. Design Graphics Division of the Am. Soc. for Eng. Education, 132138. Citeseer, 2004. [5] Brano, T.; Wiebe, E.; Hartman, N.: Integrating constraint-based cad into an introductory engineering graphics course: Activities and grading strategies. In 2003 American Society for Engineering Education Annual Conference & Exposition, 2003. [6] Camba, J.; Contero, M.; Company, P.: Parametric cad modeling: An analysis of strategies for design reusability. Computer-Aided Design, 74, 1831, 2016. http://doi.org/10.1016/j.cad.2016.01.003. [7] Cheng, Y.; Ma, Z.: The design reuse benchmark report. Advanced Engineering Informatics, 33, 115, 2017. http://doi.org/10.1016/j.aei.2017.04.003. [8] Company, P.; Contero, M.; Otey, J.; Camba, J.; Agost, M.; Pérez-López, D.: Web-based system for adaptable rubrics: Case study on cad assessment. Journal of Educational Technology & Society, 20(3), 2441, 2017. Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net 623 [9] Company, P.; Contero, M.; Otey, J.; Plumed, R.: Approach for developing coordinated rubrics to convey quality criteria in mcad training. Computer-Aided Design, 63, 101117, 2015. 1016/j.cad.2014.10.001. http://doi.org/10. [10] Company, P.; González-Lluch, C.: CAD 3D con SolidWorks Tomo I: Diseño básico. Publicacions Universitat Jaume I, 2013. http://doi.org/10.6035/Sapientia86. [11] Company, P.; Naya, F.; Contero, M.; Camba, J.: On the role of geometric constraints to support design intent communication and model reusability. Computer-Aided Design & Applications, 17(1), 6176, 2020. http://doi.org/10.14733/cadaps.2020.61-76. [12] Company, P.; Otey, J.; Contero, M.; Agost, M.; Almiñana, A.: Implementation of adaptable rubrics for cad model quality formative assessment. The International journal of engineering education, 32(2A), 749761, 2016. [13] Contero, M.; Company, P.; Vila, C.; Aleixos, N.: Product data quality and collaborative engineering. IEEE Computer Graphics and Applications, 224(3), 3242, 2002. 2002.999786. http://doi.org/10.1109/MCG. [14] Devine, K.; Laingen, M.: Assessing design intent in an introductory-level engineering graphics course. In 68th EDGD Midyear Conference Proceedings, 5963, 2013. [15] González-Lluch, C.; Company, P.; Contero, M.; Camba, J.; Colom, J.: A case study on the use of model quality testing tools for the assessment of mcad models and drawings. The International journal of engineering education, 33(5), 16431653, 2017. [16] González-Lluch, C.; Company, P.; Contero, M.; Camba, J.; Plumed, R.: A survey on 3d cad model quality assurance and testing tools. Computer-Aided Design, 83, 6479, 2017. cad.2016.10.003. http://doi.org/10.1016/j. [17] González-Lluch, C.; Company, P.; Contero, M.; Pérez-López, D.; Camba, J.: On the eects of the x geometric constraint in 2d proles on the reusability of parametric 3d cad models. Journal of Technology and Design Education, 29(4), 821841, 2019. s10798-018-9458-z. International http://doi.org/10.1007/ [18] González-Lluch, C.; Plumed, R.: Are we training our novices towards quality 2d proles for 3d models? Advances on Mechanics, Design Engineering and Manufacturing II. Lecture Notes in Mechanical Engineering, 6176, 2019. http://doi.org/10.1007/978-3-030-12346-8_69. [19] Hekman, K.; Gordon, M.: Automated grading of rst year student cad work. In 2013 ASEE Annual Conference & Exposition, 2629. ASEE Conferences, 2013. Https://peer.asee.org/19249. [20] Jackson, C.; Buxton, M.: The Design Reuse Benchmark Report: Seizing the Opportunity to Shorten Product Development. Aberdeen Group, Boston, 2007. http://doi.org/10.1137/1.9780898717761. [21] Kirstukas, S.: Development and evaluation of a computer program to assess student cad models. In 2016 ASEE Annual Conference & Exposition, 2629, 2016. http://doi.org/10.18260/p.26781. [22] McKenney, D.: Model quality: the key to cad/cam/cae interoperability. In 1998 MSC software Americas users conference, 1998. [23] Ramos-Barbero, B.; Melgosa-Pedrosa, C.; Zamora-Samperio, R.: summaries of the rules of design intent. 27(3), 481498, 2017. Learning cad at university through International Journal of Technology and Design Education, http://doi.org/10.1007/s10798-016-9358-z. [24] Shah, J.; Mäntylä, M.: Parametric and feature-based CAD/CAM: concepts, techniques, and applications. Wiley, Chichester, 1995. [25] Tshibalo, A.: The potential impact of computer-aided assessment technology in higher education. South African Journal of Higher Education, 21(6), 686695, 2007. 25738. http://doi.org/10.4314/sajhe.v21i6. Computer-Aided Design & Applications, 18(3), 2021, 612-623 © 2021 CAD Solutions, LLC, http://www.cad-journal.net