Civil Engineering Pilot Links Digital Structural Modelling With Calculation Traceability Limits

A classroom intervention in civil engineering education has highlighted a persistent divergence between computational model execution and the technical traceability of underlying structural calculations. Researchers at the Universidad Politécnica Salesiana in Quito, Ecuador, conducted a descriptive educational case study involving an intact cohort of twenty-five seventh-semester students enrolled in a 160-hour steel structures module during the 2024–2025 academic year. The semester-long project evaluated the consolidation of collaborative design work into a single collective deliverable, examining how students transitioned from reference architectural data to computational simulation and comprehensive design reports. The cohort achieved an overall weighted score of 85 percent on an eight-criterion assessment rubric administered by the module instructor. While the deliverable attained top marks for structural layout, regulatory code provisions, and visual presentation, the accompanying structural calculation report scored only 60 percent. This contrast revealed that students could assemble a sophisticated digital model without fully recording the governing equations, assumptions, and step-by-step reasoning that justified their engineering decisions.

The pedagogical structure was organised around project-based learning and cooperative learning frameworks to simulate authentic engineering challenges. Working across approximately 16 weeks, the twenty-five undergraduates organised themselves into five distinct working groups of five students, with each group assuming responsibility for a sequential phase of a residential building redesign. Rather than solving isolated, abstract textbook problems at the end of a module, the students were tasked with converting an existing reinforced concrete reference structure into a cold-formed steel framing alternative. The project sequence moved through three clearly delineated stages: reviewing the base concrete project documentation, developing the steel-framing redesign, and conducting a technical comparative evaluation between the original and alternative structural systems. Group responsibilities encompassed analysing the base documentation, reviewing baseline drawings and calculations, establishing the framing layout and gravitational loads, completing three-dimensional computational modeling alongside seismic verification, and drafting construction drawings and the comprehensive calculation report.

To guide the redesign, the students utilized an academic reference configuration from prior structural investigations, specifically the M5 residential model documented in previous academic literature. This configuration featured a floor plan measuring 26.5 by 28.8 metres, covering an area of 763.2 square metres with an overall building height of 8.0 metres beneath a multi-pitched roof. As part of the technical background retained for the classroom exercise, permanent floor loads were set at 4.77 kilonewtons per square metre for the intermediate floor and 2.80 kilonewtons per square metre for the roof, alongside corresponding design live loads of 1.96 and 0.69 kilonewtons per square metre, respectively. Load combinations and gravitational definitions were established in accordance with the Ecuadorian construction standard NEC-SE-CG, while seismic spectra and verification protocols followed the NEC-SE-DS regulatory framework. Rather than performing a component-by-component material take-off, the cohort applied these established operational assumptions to focus directly on the synthesis of structural configuration, three-dimensional modeling, and regulatory compliance.

Computational analysis was conducted using SAP2000 version 23.3.1, wherein the students constructed a three-dimensional global framework using one-dimensional frame elements to represent cold-formed steel studs, tracks, floor joists, roof trusses, and lateral bracing. The analysis protocol incorporated hinged supports at the column bases, linear static procedures, modal analysis utilizing 100 vibration modes with 5 percent damping, and response-spectrum dynamic analyses. Dynamic modal components were combined using the Complete Quadratic Combination method, and spatial directional components were aggregated using the Square Root of the Sum of the Squares rule. The computational model was explicitly scoped as an educational exercise examining global structural response, omitting micro-level phenomena such as screw shear deformation, local plate buckling, sheathing-to-frame composite interaction, contact mechanics, and non-linear hysteretic behavior. Accidental eccentricity of 5 percent was applied to static lateral load patterns, whereas dynamic response spectrum load cases omitted additional accidental eccentricity, representing a specific modeling discrepancy identified during instructor review.

Formative instructional guidance played a critical role throughout the semester to help students address modeling nuances before final grading. The course instructor conducted between eight and ten feedback sessions focused on resolving complex technical junctures, including roof framing geometry, load application paths, spatial orientation of member local axes, seismic verification routines, drawing coordination, and calculation reporting. An intermediate computational model preserved in the study archives confirmed that instructor intervention prompted necessary corrections to local-axis alignments and member moment-of-inertia orientations across specific frame components. Furthermore, the baseline documentation revealed that while the original architectural project featured a multi-sloped roof, the students simplified these geometries in their final structural model. This geometric simplification contributed to the integrated project receiving a score of 80 percent on the criterion evaluating the interpretation of base project information, as the deliverable omitted specific pitch details present in the initial architectural plans.

Evaluation of the consolidated project deliverable was executed using an internal weighted rubric comprising eight specific performance criteria, designed to reflect constructive alignment between instructional goals and practical engineering performance. Structural layout and preliminary element design received a full 100 percent mark, contributing 15 weighted points, as the submission demonstrated a logical hierarchy of load transfer and global stability. The application of code provisions also earned 100 percent, contributing 15 weighted points, reflecting accurate incorporation of assigned regulatory parameters for load combinations and seismic checks. Report presentation achieved 100 percent, adding 5 weighted points for clear visual communication and technical formatting. Conversely, structural modeling received 80 percent, providing 20 weighted points out of a possible 25, while seismic optimization, base information interpretation, and comparative analysis each scored 80 percent, yielding 12, 4, and 8 weighted points, respectively. The calculation report received the lowest mark at 60 percent, yielding only 6 weighted points out of a possible 10.

The structural calculation report’s lower score of 60 percent exposed critical deficiencies in calculation traceability, defined by the authors as the capacity of an independent technical reader to reconstruct the complete reasoning chain linking stated assumptions, governing equations, applied loads, regulatory criteria, modeling commands, and result interpretations. While the deliverable contained extensive automated data tables and final stress checks exported directly from SAP2000, it failed to fully document the underlying mathematical formulations, physical definitions, and formal verification methodologies required to justify those numerical outputs. The instructor observed that computational software can readily produce polished, visually complete models and stress diagrams that mask incomplete analytical understanding. Because the rubric allocated 25 percent of the total mark to computational modeling and only 10 percent to the calculation report, the overall course mark of 85 percent remained high, highlighting how rubric weighting distributions can inadvertently reward software execution over the transparent written documentation of engineering rationale.

Dynamic and static structural response metrics recorded from the students’ finalized SAP2000 model illustrated both functional global behavior and notable computational limitations. The model achieved cumulative participating mass ratios of 93.46 percent in UX and 91.98 percent in UY. However, the first-mode fundamental vibration period was calculated at 0.52 seconds, notably longer than the approximate empirical period of 0.28 seconds used as a reference benchmark. Static base shear was calculated at approximately 848 kilonewtons, while dynamic base shears obtained from response spectrum analysis were 698 kilonewtons in the X direction and 854 kilonewtons in the Y direction. These dynamic results yielded dynamic-to-static shear ratios of 82.29 percent in X and 100.71 percent in Y. While both ratios satisfied the 80 percent minimum prescribed by NEC-SE-DS for regular structures, the 82.29 percent value in the X direction fell below the 85 percent threshold mandatory for irregular structural configurations, a critical distinction given that the student report lacked a formal verification of plan and elevation regularity.

Detailed member-level evaluations within the collective deliverable further demonstrated the boundary between academic modeling and professional design certification. An automated cold-formed design extraction across the SAP2000 database yielded 13,396 individual frame element evaluation records. Among these records, the software identified 66 elements classified as overstressed, exhibiting structural interaction ratios exceeding the allowable 1.00 threshold, with the maximum observed interaction ratio reaching 3.55. Furthermore, the students did not configure the diagonal bracing elements as tension-only members, meaning the simulation failed to capture the unilateral tensile mechanics characteristic of conventional steel strap bracing. A rigid diaphragm constraint was not explicitly assigned to the intermediate floor, with corresponding dead loads applied directly to the joists, meaning that the model did not explicitly reproduce the action in the diaphragm plane. These unresolved overstressed members and modeling simplifications were preserved in the academic deliverable without retrospective correction, reflecting authentic instructional artifacts rather than verified structural engineering certifications.

The technical comparison between structural systems included in the deliverable provided an educational lesson in distinguishing physical quantities from structural performance. The project utilized structural weights reported in prior literature by Chavez Amaguaña and colleagues, which documented 11,883 kilonewtons for the baseline reinforced concrete building compared to 576 kilonewtons for the cold-formed steel alternative, indicating a 95.15 percent reduction in structural material weight. However, the study emphasised that this material reduction cannot be interpreted as an equivalent reduction in seismic demand, base shear, or environmental impact. When calculated from the reported static base shear of 848 kilonewtons under a reduced seismic coefficient of 0.476, the effective seismic weight of the steel building was approximately 1,782 kilonewtons—over three times the structural framing weight alone. The final report lacked the detailed floor-by-floor drift checks and component inventories necessary to reconstruct equivalent seismic masses across both systems, underscoring the necessity for students to interpret technical comparative metrics with rigorous physical distinction.

The authors contextualized the study within explicit methodological and environmental boundaries, noting that the findings represent a single intact cohort evaluated by a single instructor without an independent control group, external peer review, or validated psychometric testing. The rubric scores served as a descriptive evaluation of a collective student output rather than an individualized measurement of student learning gains or a comparative trial of project-based learning versus traditional lecture instruction. To address the identified weaknesses in future course iterations, the researchers proposed introducing staged, step-by-step calculation logs throughout the semester to enforce transparent documentation of governing formulations before computational models are finalized.

Subject of Research: Civil engineering education and calculation traceability in project-based steel design

Article Title: A pilot classroom intervention using project-based learning for steel framing design in a civil engineering course

Article References: Arias-Rueda, J. H., Guerrero Cuasapaz, D. P., & Romero Romero, C. A. (2026). A pilot classroom intervention using project-based learning for steel framing design in a civil engineering course. Frontiers in Education, 11, Article 2003344. https://doi.org/10.3389/feduc.2026.2003344

Image Credits: AI Generated

DOI: 10.3389/feduc.2026.2003344

Keywords: Civil Engineering Education, Project-Based Learning, Structural Modelling, Calculation Traceability, Steel Framing Design, Cooperative Learning, Authentic Assessment, SAP2000, Cold-Formed Steel, Seismic Verification, Higher Education Pedagogy, Curriculum Design, Engineering Design Thinking, Formative Feedback, Base Shear Analysis, Structural Rubrics