Research Interests

Research in my group is on systems engineering and process control; it brings together modeling, mathematical analysis, control theory, optimization and computation, in order to understand the structure and improve the operation of chemical, biological and energy systems.  Our current research focuses on three themes: complexity (networks), data (learning), and energy (sustainability).

Network approaches to control of complex process networks

Our group has introduced a modern network science perspective on the control of complex integrated plants. These large-scale dynamical systems are complex but are also characterized by structure and sparsity. Distributed control, in which control decisions are made for smaller subsystems, with coordination and communication among them, is a natural framework to account for this structure. Effective distributed control requires a decomposition of the system into subsystems that strikes an optimal balance between localized and cross-subsystem computations. Our research has identified an analogy between this problem and that of community detection in networks which seeks communities (subnetworks) with strong internal connections and weak cross-community ones. We have employed such community detection methods for the decomposition of dynamical systems represented as graphs that capture the structure and strength of dynamic interactions. The resulting decompositions significantly accelerate the computations in distributed control of large-scale plants, while ensuring minimal performance degradation compared to the computationally prohibitive, centralized control designs.  Implementations of these methods on industrial control platforms are currently underway. 

On a separate direction we are pursuing the development of data-driven control methods that use data to “learn” fundamental control properties rather than a dynamic system itself. An example is the property of dissipativity, a concept inspired from thermodynamics that is central in ensuring stability but is also very difficult to establish analytically via a thermodynamic analysis. We are developing a framework for learning dissipativity from input/output trajectories of a dynamical system and using it to design data-driven controllers with well-characterized closed-loop stability and performance properties.

Exploiting structure in the solution of complex optimization problems
Large-scale, complex optimization problems arise naturally in process systems engineering tasks (scheduling, planning, control, and their integration). Monolithic solutions to such problems are difficult to obtain. On the other hand, these problems are also characterized by structure which can be exploited in a variety of existing
decomposition-based solution approaches (e.g., Lagrange or Benders). Yet, no framework existed to systematically generate the decompositions of the optimization problem. Our group has developed such a framework based on community detection (more generally detection of latent block structures, including core-periphery and hybrid ones) in suitable graph representations of optimization problems. This research has led to DeCODe, the first software tool that generates automatically high-quality decompositions of optimization problems and can be coupled with existing decomposition-based solution methods and software, to enable fast solution of complex optimization problems.

We are also currently applying modern network science methods such as those highlighted above to metabolic networks and brain networks. The overarching goal is to identify topological features that impact function. For example, we are collaborating with a team of neuroscientists and computer scientists to elucidate how the topology of brain connectomes relates to behavioral attributes and what internal control mechanisms govern cognitive function.

Sustainable production of power, fuels, and chemicals
Our group’s research has traced several new directions in sustainability research, including biomass conversion and microgrids (autonomous power systems with distributed generation and storage units).  On the first theme, we have developed RING, an automated software for elucidating the complex chemistry involved in biomass (and other non-traditional feedstocks) conversion. RING uses powerful cheminformatics and graph theoretic algorithms to automatically generate reaction networks for large classes of catalytic chemistries, along with on-the-fly thermochemistry calculations and kinetic models. It is already used by experimental and industrial research groups for the discovery and analysis of novel chemical and biochemical reaction pathways.  On the second theme, our research has developed a comprehensive framework for (i) the optimal design of microgrids, (ii) their scheduling and supervisory control based on deterministic or stochastic weather and load forecasts, and (iii) a new market structure governing the interaction between microgrids and the grid. This research has brought forth powerful analogies between process systems and power systems and has traced exciting research avenues in this new application domain.

Green ammonia engineering and economics
Green ammonia refers to ammonia produced using renewable energy (from wind or solar) to produce hydrogen (from water) and nitrogen (from air), as well to power the ammonia synthesis itself. This subject is currently attracting world-wide attention due to its potential to curb the significant CO2 emissions in current ammonia production. Ammonia can also find multiple uses: as a fertilizer, a fuel for transportation or power generation, and as a chemical medium for storing and transporting renewable energy (much more efficiently than hydrogen). It therefore can play a critical role in decarbonizing agriculture and facilitating the adoption of renewable energy resources in general. Our group has led the development of decision-support tools for assessing and optimizing the economic feasibility of green ammonia production and its synergistic utilization in the agricultural and energy sectors. These tools allow for the optimal design, scheduling, and real-time operation of ammonia-based agricultural and energy systems at different scales (farm, community, and regional). This work involves collaborations with farming communities, fertilizer producers, utility cooperatives, power companies, as well as researchers in ammonia combustion and public policy.

Selected Publications

  • Mitrai, I., W. Tang and P. Daoutidis, “Efficient Solution of Enterprise-Wide Optimization Problems Using Nested Stochastic Blockmodeling”, I&EC Res., 60, 14476 (2021).
  • Tang, W. and P. Daoutidis, “Coordinating Distributed MPC Efficiently on a Plantwide Scale: The Lyapunov Envelope Algorithm”, Comp. Chem. Engng., 10.1016/j.compchemeng.2021.107532 (2021).
  • Mitrai, I., W. Tang and P. Daoutidis, “Stochastic Blockmodeling for Learning the Structure of Optimization Problems”, AIChE J., 10.1002/aic.17415 (2021).
  • Tang, W. and P. Daoutidis, “Nonlinear State and Parameter Estimation Using Derivative Information: A Lie-Sobolev Approach”, Comp. Chem. Engng., 10.1016/j.compchemeng.2021.10736 (2021).
  • Wang, H., M. Palys, P. Daoutidis and Q. Zhang, “Optimal Design of Sustainable Ammonia-Based Food-Energy-Water Systems with Nitrogen Management”, ACS Sustainable Chemistry & Engineering, 9, 2816 (2021).
  • Liu, J., B. McCool, J. R. Johnson, N. Rangnekar, P. Daoutidis and M. Tsapatsis, “Mathematical modeling and parameter estimation of MFI membranes for para/ortho-xylene separation,” AIChE J.67, e17232 (2021).
  • O’Brien, C., P. Daoutidis, Q. Zhang and W.S. Hu, “A Hybrid Mechanistic-Empirical Model for In-silico Mammalian Cell Bioprocess Simulation”, Metabolic Engineering66, 31 (2021).
  • Palys, M., H. Wang, Q. Zhang and P. Daoutidis, “Renewable Ammonia for Sustainable Energy and Agriculture: Vision and Systems Engineering Opportunities,” Current Opinion in Chemical Engineering, 31, 100667 (2021).
  • Khatib, S. and P. Daoutidis, “Application of Graph Theory and Filter Based Variable Selection Methods in the Design of a Distributed Data-Driven Monitoring System”, Comp. Chem. Engng.143, 107098 (2020).
  • Tang, W. and P. Daoutidis, “Fast and Stable Nonconvex Constrained Distributed Optimization: The ELLADA Algorithm”, Optim. Eng., 10.1007/s11081-020-09585-w (2021).
  • Tang, W. and P. Daoutidis, “Dissipativity Learning Control (DLC): Theoretical Foundations of Input--Output Data-Driven Model-Free Control”, Syst. Contr. Lett.147, 104831 (2021).
Prodromos Daoutidis - Headshot


Phone: 612/625-8818

Office: 233 Amundson Hall

Research Group

Support Prodromos Daoutidis' Research

  • Diploma, Chemical Engineering, Aristotelian University of Thessaloniki, 1987
  • M.S.E., Chemical Engineering, University of Michigan, 1988
  • M.S.E., Electrical Engineering: Systems, University of Michigan, 1991
  • Ph.D., Chemical Engineering, University of Michigan, 1991