Štrbské Pleso, Slovak Republic, June 6-9, 2017


Plenary lectures

S. Skogestad (NTNU Trondheim): Economic Plantwide Control: Control structure design for complete processing plants

Abstract: A chemical plant may have thousands of measurements and control loops. By the term plantwide control it is not meant the tuning and behavior of each of these loops, but rather the control philosophy of the overall plant with emphasis on the structural decisions. In practice, the control system is usually divided into several layers, separated by time scale: scheduling (weeks) , site-wide optimization (day), local optimization (hour), supervisory and economic control (minutes) and regulatory control (seconds). Such a hierchical (cascade) decomposition with layers operating on different time scale is used in the control of all real (complex) systems including  biological systems and airplanes, so the issues in this section are not limited to process control. In the talk the most important issues are discussed, especially related to the choice of ”self-optimizing” variables that provide the link the control layers.  Examples are given for optimal operation of a runner and distillation columns.

M. Mönnigmann (RU Bochum): Constructive Nonlinear Dynamics in Optimisation and Process Systems

Abstract: Model based optimization is common practice in process systems and control engineering. Nonlinear programming, for example, can be applied to find economically optimal steady states, if a system model (such as a set of nonlinear ODE) is available. However, optimization naturally drives dynamical systems to their limits. This may result in modes of operation that are optimal economically, but unstable, or that are optimal but have other undesirable dynamical properties.

An approach is presented that integrates stability boundaries and related boundaries into nonlinear programming. These boundaries cannot be treated as simple constraints in nonlinear programs, because they are hidden in the model and therefore no explicit characterization exists for them. Essentially, the critical boundaries of interest are manifolds of bifurcation points, and the distance of any candidate optimal point to these manifolds can locally be described with normal vectors. The approach has successfully been applied to the steady state and periodic mode optimization of nonlinear ODE, discrete time systems and delay differential equations with uncertain parameters. It is illustrated with examples from energy systems, chemical and biochemical engineering.

Workshop - free of charge, available to all participants

B. Houska (ShanghaiTech): Distributed Optimization and Control with ALADIN

Video from the workshop: youtube

Abstract: This workshop consists of two interactive tutorial presentations. The first presentation gives an overview about distributed convex and nonconvex optimization algorithms of recent interest such as dual decomposition, the alternating direction method of multipliers (ADMM), and the augmented Lagrangian based alternating direction inexact Newton method (ALADIN). We discuss how these methods perform for different types of problems and share experience on which algorithms and settings are suited for which types of optimization problems.

The second presentation focuses on applications of ALADIN with a particular focus on large-scale applications in model- based process control. In particular we discuss how to detect and exploit structure in different types of control applications ranging from distributed traffic control at intersections, nominal and stochastic optimization of AC power flows in large electrical networks, to plant-wide chemical process optimization and control.