📙論文名稱：

Driving ESG Performance in High-tech Infrastructure - A Case Study of Sustainable Construction Governance and the SSD Framework

📙刊登處：2026 IFKAD國際學術研討會

https://www.ifkad.org

📙作者姓名： Yu-Chun Wang, Min-Ren Yan

論文摘要：

This study investigates sustainable governance transition pathways within the EPCM framework for high-tech facilities, such as semiconductors, data centers, and pharmaceutical plants. Driven by intensifying global ESG regulations, traditional management faces acute onsite labor shortages and an operational disconnect between digital platforms and specialized cleanroom cultures. Utilizing a Sustainable System Development (SSD) framework and System Dynamics (SD) modeling, this project quantitatively analyzes the dynamic impacts of traditional project management optimization versus AI-driven visual analytics supervisory technology on construction safety, monitoring efficiency, and project cash flow. Benchmarked against Exyte’s "Sustainable Fab" initiative, this project validates how digital transformation translates into concrete ESG governance outcomes, providing a financially resilient sustainable governance framework for the high-tech construction industry.
 
Macro-Environment and Operational Pain Points 
A PESTEL macro-environmental analysis reveals complex external pressures and sector-specific operational pain points within the semiconductor construction industry:

• Political: Geopolitical tensions drive supply chain restructuring, increasing local content requirements and transnational compliance pressures.
• Economic: Inflation triggers severe material cost volatility, while high interest rates elevate liquidity management to a critical priority.
• Social: Soaring market salaries cause critical shortages of cleanroom supervisors in core hubs (e.g., Taiwan), leaving non-core regions with an absolute scarcity of experienced personnel.
• Technological: Strict security protocols prohibit camera-enabled devices inside facilities, forcing reliance on manual inspections and paper records, which creates a severe digital divide.
• Environmental: Ultra-low energy cleanrooms are the standard; optimizing construction waste reduction and energy efficiency is a mandatory prerequisite for global supply chain inclusion.
• Legal: Occupational safety regulations enforce zero tolerance toward accidents alongside tightening ESG factory compliance codes.

Operationally, cleanroom construction requires extreme precision and possesses a high degree of irreversibility, where undetected minor defects cause an exponential surge in rework costs. The current management dilemma stems from supervisors being consumed by tedious manual inspections, preventing focus on high-value engineering decisions. This triggers a systemic vicious cycle where inefficient supervision increases defect rates, driving up rework costs and worsening financial pressure.
 
Strategic Architecture and Management Decisions 
To ensure comprehensive management, this study maps an SSD strategic roadmap (2023–2028) structured around the four dimensions of the Balanced Scorecard (BSC):

• Financial: Optimize liquidity via schedule optimization and AI supervision to ensure a 10% quarterly cash flow growth by 2028.
• Customer: Enhance trust and brand value by reducing onsite construction defects, aiming to increase supervisors' deep work time by 50%.
• Internal Process: Minimize rework costs by increasing manager-level inspection frequency by 400% (from once to four times daily) to maximize construction precision.
• Learning and Growth: Refine talent cultivation by increasing daily professional training time to 1 hour, ensuring supervisor skills dynamically match rigorous cleanroom construction requirements.

System Dynamics Simulation and Scenario Design 
This study developed Stock-and-Flow Diagrams to quantify the causal evolution between labor gaps and AI alternative solutions under fab expansion pressures. The model tracks three core stocks: cash flow, supervisor skill level, and supervisor workforce headcount.
 
The dynamic mechanism demonstrates that a surge in project volume drives up the overall supervisory workload. Increasing manual inspection frequencies during peak construction periods compresses supervisors' deep work time. When the workload reaches a critical tipping point, competitor poaching accelerates supervisor turnover rates, resulting in a structural labor gap. Conversely, an increase in supervisor skills lowers defect generation; otherwise, backlogged defects accumulate, triggering severe rework costs.
 
This study simulates a baseline scenario alongside five strategic scenarios from 2023 Q1 to 2028 Q4, utilizing a quarterly time step:

1. Scenario 1: Training time increases by 12.5% to upgrade skills and reduce defect rates.
2. Scenario 2: Adaptive scheduling staggers construction phases, mitigating peak labor density and smoothing cash flow.
3. Scenario 3: Inspection frequency increases by 400% to forcefully maintain construction quality.
4. Scenario 4: Integrates Scenarios 1, 2, and 3 to evaluate the synergistic effects of concurrent traditional management measures.
5. Scenario 5: Deploys 100 AI monitoring nodes for 24-hour continuous behavioral recognition, establishing a data closed-loop (detection, notification, improvement, acceptance) as a technological substitution stream filling the labor vacuum. 

Simulation Findings and Managerial Reflections 
The simulation shows that Scenario 4 exhibits strong initial synergistic effects, increasing supervisors' deep work time by 545%. However, all human-centric traditional strategies (S1–S4) suffered a significant performance reversal during the 2024 Q1 peak construction period, with Scenario 2's efficiency plummeting to -70%. Under extreme pressure, traditional manual interventions exponentially increase communication and management costs, causing cognitive overload and supervisor fatigue, triggering a negative feedback loop. Traditional strategies also drove peak financial achievement rates into negative territory (-27%).
Conversely, Scenario 5 (AI Vision System) demonstrated exceptional financial and governance resilience. It maintained a stable 6% cash flow growth during peak periods and reached 9% growth by project completion (2028 Q4), nearing the company’s 10% target. This indicates that the AI system, acting as a technological substitution stream, successfully decouples project performance from labor turnover. It functions as a risk circuit breaker against management collapse and reputational damage, preventing severe liquidated damages from defective rework or delayed delivery.
 
Conclusion and Suggestions 
This study confirms that integrating AI visual analytics and systematic training programs into core EPCM strategies can effectively stabilize project quality and enhance financial resilience in high-tech facility construction. Technology must serve as a risk circuit breaker when human management reaches its psychological and physical limits.
To achieve a successful transition to the "Second Curve" by 2030, future corporate action plans should focus on: refining Digital Twin architectures to establish a decision-support cockpit for real-time forecasting; expanding the data closed-loop to complex construction sequence validation; and cultivating a technology-centric corporate culture rooted in data-driven governance. Through these initiatives, precision governance and sustainable transformation in the high-tech construction industry can be fully realized.