PhD Oral Exam - Araham Jesus Martinez Lagunas, Building Engineering

When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.

Abstract

Construction is one of the world’s largest economic sectors and directly influences national productivity and societal well-being. Despite its scale and importance, industry productivity has stagnated for decades, with growth rates three to five times lower than those of other sectors, while nearly 75% of construction projects worldwide experience schedule delays and cost overruns. A major contributor to this underperformance is the lack of systematic, data-driven methods that effectively integrate and operationalize digital technologies to support operational performance improvement. Many Architecture, Engineering, Construction, and Facility Management (AEC/FM) organizations remain predominantly project-oriented and often underestimate the role of systematic process performance monitoring and management in achieving consistent project delivery. As a result, the industry continues to rely largely on manual and subjective methods (e.g., workshops, interviews, focus groups) to capture and manage construction processes, which fail to reflect their actual “as-happened” execution, dynamic evolution, and operational performance. In the absence of such insights, process improvement decisions remain intuitive, judgment-based, and disconnected from empirical evidence. Moreover, AEC/FM organizations increasingly seek to harness project operations data from siloed source systems to enable automated extraction of dynamic process performance insights; however, most Common Data Environments (e.g., Procore) remain largely static and lack such dynamic process monitoring capabilities. Enabling this capability requires the integration and transformation of raw data-oriented system structures into well-defined, process-aware structures.

To address these challenges, this research proposes an automated, data-driven method for quantitatively assessing the performance of inter-organizational construction business processes by integrating process mining techniques with Lean-based quantitative metrics, including cycle time, work in progress (WIP), and lead time. This hybrid Lean-based Process Mining and Management (LPMM) methodology comprises five levels and seven interrelated, process-oriented modules that follow a stepwise progression, transforming data-oriented architectural layers into process-aware structures. Module 1 establishes prerequisites through automated data extraction, integration, and preparation from heterogeneous project information systems, supported by exploratory data analysis, task mining, and expert consultation. Module 2 translates research objectives into analytical questions by defining performance dimensions, key process performance indicators (PPIs), abstraction levels, and stakeholder perspectives, informed by systematic literature review and domain expertise. Module 3 enables automated event log construction by transforming integrated data into standardized, process-aware representations through automated extractors, schema transformers (e.g., XES), loggers, and enrichment mechanisms. Building on this foundation, Module 4 applies process mining techniques, including Fuzzy Miner and Inductive Miner, to automatically discover “as-happened” end-to-end (E2E) process models. Module 5 derives process efficiency formulations grounded in process mining theory and applies quantitative performance metrics across real-world construction projects to evaluate efficiency, detect deviations, analyze process variants, and assess the impact of Request for Information (RFI) content using natural language processing (NLP). Finally, Modules 6 and 7 translate findings into actionable recommendations and enable continuous monitoring through custom process boards.

The framework was applied to the RFI process across 71 real-world projects comprising 5,564 RFIs, including a cross-case analysis of two large commercial projects. The application involved automated data extraction from the Common Data Environment (CDE), integration and transformation into enriched event logs, portfolio-level efficiency assessment, and detailed process-level analysis of time, cost, control-flow, and team dynamics. Results show that process efficiency among similar projects executed by the same general contractor can vary substantially due to RFI flow length, team dynamics, and staff availability. Improved flow control could have tripled efficiency gains and generated savings of approximately $114,000 (USD) on a single commercial project with an estimated value of $50 million (USD). Scaled across 71 projects, the findings reveal significant opportunities for cost reduction, productivity improvement, and evidence-based decision-making. In total, 42 targeted process improvement recommendations were derived. The extracted as-happened model achieved a fitness of 0.958, a precision of 0.761, and an F-score of 0.845, confirming its representational suitability. Overall, LPMM enables AEC/FM organizations to transform project operations data into process knowledge assets for automated process health monitoring and continuous performance improvement.