Abstract: For almost half a century, Process Systems Engineering (PSE) has been a strong research pillar in the Chemical Engineering Department at Carnegie Mellon University. The Center for Advanced Process Decision-making (CAPD) serves as the vanguard of this effort, bridging theoretical optimization with industrial practice to solve complex decision-making problems across the chemical and energy sectors.
The Challenge of Industrial Optimization
Modern chemical and petroleum industries operate on a massive scale where even a 1% efficiency gain can translate into millions of dollars in savings. However, the mathematical "heavy lifting" required to optimize these systems is immense. Engineers face a fragmented landscape: they must design new processes, schedule batch operations, control real-time fluctuations, and manage global supply chains—often using disparate tools for each task.
The CAPD addresses this fragmentation by developing unified methodologies for Process Systems Engineering (PSE). The challenge lies in creating robust algorithms that can handle "mixed-integer" problems (decisions such as whether to build a specific reactor) and "nonlinear" physics, all while accounting for the inherent uncertainty of global markets and energy prices.
Research Pillars and Methodologies
The CAPD does not merely apply existing mathematics; it invents new computational frameworks. Its work is categorized into several core domains that span the entire lifecycle of a chemical enterprise.
Key methodological pillars include:
- Mathematical Programming: Advanced algorithms for Large-scale Nonlinear Programming (NLP) and Mixed-Integer Nonlinear Programming (MINLP), including the widely used BARON software.
- Process Synthesis: Systematic strategies for designing heat exchanger networks and integrated process flowsheets to minimize waste.
- Enterprise-wide Optimization (EWO): A holistic approach that integrates planning, scheduling, and real-time control into a single computational framework.
- Multi-scale Modeling: Expanding PSE techniques from the plant level down to molecular design and bioinformatics.
The CAPD Ecosystem: Industry Meets Academia
The center’s success stems from a unique feedback loop. Through a consortium of over 20 member companies—including Dow, ExxonMobil, and Shell—the CAPD gains access to "messy" real-world problems. These industrial case studies serve as benchmarks for testing new algorithms against actual constraints.
Beyond research, the center views education as a core product. Its annual seven-day short course translates cutting-edge research into practical skills for industrial practitioners, ensuring that breakthroughs reach the factory floor. This commitment to "lifelong learning" has created a global network of alumni who now lead PSE efforts in both academia and industry.
Impact and Scale
The economic and academic impact of the CAPD is substantial. Success stories, such as those from The Dow Chemical Company, highlight multimillion-dollar annual savings resulting from novel real-time optimization algorithms.
Currently, the center directs the work of over 70 graduate students and researchers. Its alumni hold faculty positions at top-tier institutions like Purdue, Georgia Tech, and Oxford, and are embedded in major global corporations ranging from tech giants like Google to energy leaders like Saudi Aramco.
Future Outlook
As industrial systems grow in complexity, the "curse of dimensionality" remains a primary challenge—solving massive models fast enough for real-time decision-making is an ongoing battle. Future research at the CAPD is pivoting toward:
- Energy Systems Initiative (ESI): Tackling climate-critical technologies like carbon capture and sequestration.
- Molecular and Materials Design: Applying PSE tools to create environmentally benign molecules.
- Data-Driven Integration: Combining traditional physics-based modeling with machine learning to create more resilient, adaptive industrial systems.
Summary updated