Process integration is a holistic approach to process design and operation which emphasizes the unity of the process. Any process is characterized by two commodities; mass and energy. Therefore, process integration can be categorized into mass integration and energy integration.  This short course will provide a step-by-step coverage of the fundamentals and industrial applications of process integration with focus on mass integration. Systematic techniques will be presented to guide in developing cost effective solutions to process design and operation problems.  In particular, the mini-course will address the following: a) How to identify best achievable performance targets for a process without detailed calculations; b) How to understand the big picture of a process and use it to optimize any plant?; c) How to determine optimal direct-recycle strategies; and d) How to determine the tasks of retrofitting the process to meet the targets. Rigorous methods and applications will be presented.

This lecture will first provide an overview on the different approaches to process synthesis, highlighting the one based on superstructure optimization for which a classification of network representations is presented as well as underlying MILP, MINLP and generalized disjunctive programming formulations. A brief review of discrete-continuous optimization methods will be first presented, including global optimization methods to handle nonconvexities arising in process systems. An overview of several solution strategies (e.g. multilevel models, simultaneous optimization, decomposition, handling of rigorous models) is also presented. Applications presented include synthesis of heat exchanger networks, steam and power plants, integrated process water networks, and simple and complex distillation column configurations. Simultaneous optimization, energy and water integration models for the optimal synthesis of process flowsheets are also presented with applications in the synthesis of flowsheets for chemicals and biofuels.

The primary objective of the lecture is to provide an introduction to the theory and application of model predictive control (MPC).  A tutorial review of chemical process dynamics and control will first be provided, including both continuous and discrete-time models and methods. Model predictive control is the class of advanced control techniques most widely applied in the process industries. A primary advantage to the approach is the explicit handling of constraints. In addition, the formulation for multivariable systems with time-delays is straightforward. MPC was developed in the process industries in the 1960’s and 70’s, based primarily on heuristic ideas and input-output step and impulse response models. The basic principle is to solve an open-loop optimal control problem at each time step. The decision variables are a set of future manipulated variable moves and the objective function is to minimize deviations from a desired trajectory; constraints on manipulated, state and output variables are naturally handled in this formulation. Feedback is handled by providing a model update at each time step (often the “additive disturbance correction”), and performing the optimization again. A historical perspective of various MPC approaches is presented, then we derive the analytical solution for the unconstrained problem. The quadratic programming solution for linear models and constraints and a quadratic objective function is then presented. The use of state estimation, with a focus on improving the rejection of unmeasured disturbances will also be covered. Finally, extensions to nonlinear systems using different types of nonlinear models will be studied.

This lecture will discuss the usage of advanced equation-oriented (EO) CAPE tools for process analysis and dynamic simulation. First, the basic concepts of differential-algebraic equation (DAE) systems will be addressed and the corresponding numerical methods reviewed. Afterward, a methodology using dynamic models described by DAEs for real-time state and parameter estimation with constraints will be presented. Finally, the advantages of using dynamic simulator integrated with path-following tool for analyzing complex system behaviors will be demonstrated. Examples of bioprocesses will be used for illustrative purpose and hands-on exercises using EMSO dynamic simulator will be used during the lecture.

A survey is presented of advances in nonlinear programming (NLP) algorithms for challenging, large-scale applications of process optimization. These applications include topics in process modeling, design and control, for both steady-state and dynamic systems. The lecture outlines concepts that are embedded within state-of the-art NLP methods and provides a descriptive comparison of widely used NLP solvers, especially those that are incorporated within GAMS. In addition, applications in process design, operations and control will be developed for a number of challenging large-scale case studies. Finally, a number of hands-on GAMS examples will illustrate these concepts.

This lecture will provide an overview of mixed-integer programming (MIP) theory, modeling methods and solution approaches for applications in the broad area of supply chain management. Applications include short-term scheduling, medium-term production planning, and some well know strategic planning problems. A review of basic concepts on linear and mixed-integer programming will be presented followed by a discussion on some widely studied combinatorial optimization (CO) and network problems. The lecture will outline the connections between combinatorial optimization problems and integer programming formulations and cover how combinatorial insights can be used to advance MIP solution methods and how such results can be generalized to address wide range of problems.

Computational models have been used in metabolic engineering to find non-obvious genetic alterations that lead to improved production phenotypes. Computational simulations allow the best strategies to be identified for experimental implementation. An overview of current genome-scale metabolic and regulatory modeling methods will be presented, including examples of how these models can be used to analyze experimental data and predict cellular phenotypes. Recent optimization approaches that account for metabolism and regulation in the design of microbial strains for improved biofuel production will be presented.

This presentation will discuss mathematical programming approaches for modeling, optimization and control of biological and ecological systems. First, we address the development of first principles-based biogeochemical models for water quality to determine restoration policies. Optimal control problems are formulated within a simultaneous dynamic optimization framework. The objective is phytoplankton concentration reduction through biomanipulation and its associated costs. A submodel for toxin production by cyanobacteria is included. In the second part, we discuss global sensitivity analysis and parameter estimation in kinetic models of metabolic networks. The design of metabolic networks is formulated as a parameter optimization problem, in which targets for genetic engineering are maximum reaction rates and kinetic constants.

Synthetic transportation fuels come into place to address the problem of energy shortage.  In the context of environmental sustainability, alternative fuels often implies an ecologically benign renewable fuel, representing a move away from processes that emit greenhouse gases and toward the proper utilization of carbon. In the present approach, the change is into CO2 reuse from emissions to produce renewable energy (liquid synthetic fuels) with value added carbon co-products that may represent a negative carbon impact. Sustainable process design requires analysis of the environmental impacts of the process/product, incorporating elements from risk assessment. The applied methodology uses data from process simulation in the design phase, and accounts for uncertainties using Monte Carlo approach. Process design on this ground requires integrated and optimized flowsheets, with respect to materials and energy use, ecologic performance and inherent safety. Environmental protection implies that ecological issues are included in the conceptual design methodology. Besides economic efficiency, metrics for ecological and sustainability performance must be included in assessing the quality of a design solution, under uncertainty, and increased process safety. Equally, a process modification, which reduces one hazard, will always have some impact on the risk resulting from another hazard. The advantages and disadvantages of each option must be compared for a particular case and the choice made based on the specific details of the process and materials.  A method to estimate the inherent safety of different design alternatives is needed. Safety metrics are required to solve the safety evaluation problems in the conceptual engineering phase.

Mass integration strategies have been applied to minimize the use of fresh resources for the treatment of polluted process streams. A recent approach has based the design of mass exchange networks on the concept of properties, in which constraints from environmental regulations on pH, viscosity, toxicity, and COD, among others, are considered. Several structures for mass exchange networks can be conceptually designed. We present here the use of mathematical programming techniques as a tool for the optimal design of property-based mass exchange networks. The structures considered include the use of a final centralized treatment system for waste streams, or the implementation of in-plant process interceptors. The implications on the nature of the mathematical models and their solutions are addressed.

In this lecture, we will discuss some key issues associated with the application of advanced control system in industrial plants. Specifically, we will focus on regulatory and advanced control strategies. There are many advantages for the process units in using these advanced systems as more stability, respect to the restrictions and an increase in the profitability, energy efficiency and safety. However, there are still many open issues related to the acceleration of new implementations and the maintenance of good performance along the years. Therefore, this module will discuss the problems and challenges of advanced control in petroleum industries nowadays.

This seminar will present a review of methods for process synthesis with applications in mining operations. The characteristics and challenges of the minerals industry, especially those aspects related to sustainability will be analyzed. Procedures based on superstructures and mathematical programming will be emphasized, although other procedures based in heuristic or hybrid will also be introduced. The applications include, among other, fractional crystallization, mineral flotation, and hydrometallurgy.

In the first part of this seminar knowledge about reactive scheduling is systematized and relevant contributions are analyzed. Concepts about rescheduling strategies, policies and methods are presented. Regarding performance measures, different appraisals of schedule change are evaluated, besides the regular ones associated with schedule efficiency and cost. In the second part, advances in the development of an event-driven, reactive support environment are presented. The framework explicitly models the status of the in-progress schedule, and typifies the unexpected event to characterize its context. This allows making a proper specification of the problem to be faced. The resulting specification is translated into a Constraint Programming (CP) model and, finally, the resulting formulation is solved by means of a CP approach.

The mathematical modeling in human physiology has had a significant growth in recent decades. In particular the level of detail achieved in modeling the cardiovascular system and its interactions with organs and tissues, can now provide practical computational tools for scientists and physicians. This talk will review this exciting branch of modern science from the perspective of process systems engineering and the contributions made with lumped parameter models.

Problems of dynamic optimization with inequality path constraints are common in industrial plants. These constraints describe conditions of the process when it operates with extreme values of the variables, based on safety and/or economics restraints. Normally, during the optimal trajectory, some of the inequality constraints are activated, and those remain active during a certain period. This behavior can produce a change in the differential index of the DAE system, leading to the so-called floating index phenomena. In this seminar, we present a new method that unifies the advantages of special regularization functions with numerical codes, which integrate higher index DAE systems, avoiding the reinitialization and index reduction steps. All the inequality constraints are described by appropriate continuous functions and the resulting DAE system can be numerically integrated directly. The main advantage of the new method is that the DAE system can be integrated continuously; preventing the restart of the numerical integration every time an inequality constraint is violated. Several illustrative examples are presented.

All manufacturing optimization problems can be distilled down into a canonical set of objects with associated attributes from which we can abstract the complexity of these problems. These objects are enumerated by what can be called the unit-operation-port-state (UOPS) superstructure. The attributes related to each object in the proposed modeling approach refer to the phenomena of quantity, logic and quality (QLQ). Quantities include flow, rate, holdup and yield as well as duration for time; logic includes setup, startup, switchover, shutdown, standby including start-times and end-times as well as switchover-to-others for sequence-dependent changeovers, and quality includes phase (solid, liquid, gas, etc.), density, component, property and condition information. Once a superstructure has been defined for the manufacturing problem as well as all of its phenomena of how raw materials are processed into finished products, a mathematical model is generated by creating different types of variables and constraints. Given the scale, scope and size of these models, the usual formulation results in a mixed-integer nonlinear program (MINLP). Due to tractability issues, these problems are solved by using decomposition methods i.e., hierarchical, spatial, temporal and/or phenomenological decompositions. Phenomenological decomposition is the practical technique of combining the quantity and logic attributes into a “logistics” subproblem and the quantity and quality attributes into a "quality" subproblem. The logistics subproblem is solved using MILP. After the logic variables have been fixed to one of the MILP subproblem solutions, the quality subproblem is solved using NLP to assess the global feasibility of the overall problem, and then, iterations between the logistics and quality problems may be performed. Details of how these UOPS superstructures are configured including some of the QLQ phenomena will be discussed, and several industrial examples will be highlighted.

Bioreactors engineering is a well-established field. Nevertheless, the difficulties that biological systems - or even enzymatic reactors - impose are still a challenge. Taken in broader sense, bioreactors encompass enzymatic reactors and the so-called fermenters. Modeling such systems implies dealing with some (or all) of the following problems: complex reaction networks; nonlinear dynamics; feedback-feedforward regulation; great parametric variability; on-line measurement of primary variables often not available in practice; structural model mismatches. Far from intending to overview the abundant literature on these subjects, this communication presents some solutions that have been implemented by our group for this problem. Four processes will be used as proofs of concept for different approaches: tailor-made enzymatic hydrolysis of cheese whey proteins; enzymatic synthesis of beta-lactam antibiotics; high density cell cultivations for production of heterologous proteins; enzymatic hydrolysis of lignocellulosic substrates. In all cases, the proposed solutions were validated against lab-scale experiments. Different techniques had been applied to these processes: model-based control with online re-parameterization; heuristic algorithms; fault detection based on PCA; multivariate calibration for online monitoring of state variables. The use of computational intelligence tools was often necessary: neural networks, fuzzy logic, among others.

Biomass, as a renewable and locally available resource, has great potential for weaning our chemical process industry (CPI) from fossil based feedstocks. There are many different routes to transform biomass into feedstock for the bulk chemicals production. The switch from our current fossil fuel based CPI to a future CPI that utilizes biomass feedstock will require substantial amounts of research & development (R&D) and capital investments for technology development both by the CPI and the government. As such there is great opportunity for investigating how these investments will impact the evolution of the biomass feedstock system. This problem has two unique characteristics: uncertain decision-dependent endogenous technology evolutions and game theoretic issues, which limit the applicability of the existing representations and methods. This talk will explain the problem characteristics in detail, and give an overview of our research activities on developing novel frameworks and large-scale solution capabilities for this problem.

Different strategies to solve bi objective optimization problems including environmental life cycle assessment and operating cost will be discussed. The power plants of the Argentinean national electricity network will be selected minimizing the Environmental Life Cycle Assessment. The benefits of reducing Life Cycle Carbon Dioxide emissions could be included in the objective function, considering the price in the Carbon market. The Environmental Life Cycle Assessment of the energy generation including thermoelectric, hydroelectric and nuclear power plants will be shown. Other applications considering energy consumption in the operation of steam and power plants and the design of hybrid separation system including distillation columns and pervaporation membranes will be discussed.

Distribution is concerned with the shipment and storage of multiple products downstream from the supplier side to the customer side in the supply chain. A distribution network generally includes factories, distribution centers (DCs), retailers and end users at different levels, with product flows going from the highest to the lowest echelon. Typically, final product outputs from multi-site factories are shipping to and stocking up at DCs closer to the markets before delivering them to retailers or end consumers, using vehicle fleets and/or pipeline systems. This is the case for the downstream segment of the oil supply chain. This Seminar will present general optimization frameworks for the short-term operational planning of multiproduct multi-echelon transportation networks involving vehicle fleets or pipelines.

There is little argument over whether the North American forest industry is in crisis.  Merger activity in recent years has undoubtedly been critical, however it is clear that alone, mergers are not enough – a new business model is essential. The North American forestry industry has begun to discuss the need for “transformative changes”, where one possibility for this change is the so-called forest biorefinery.  This presentation overviews the role of supply chain considerations for forestry companies setting their biorefinery strategy – it focuses on the multidisciplinary industrial context, and demonstrates how systems analysis tools can be used to address complex issues.  What is the best supply chain policy for the biorefinery?  What are the manufacturing flexibility requirements needed to mitigate risk from product price volatility?  For the case of a partnership between forestry companies and other manufacturers, how can the supply chain be assessed in such a way as to maximize synergies at the strategic as well as tactical/operational levels?  These strategic issues will be addressed, with reference to industrial case studies underway in our Design Engineering Chair.

Brazil is developing technologies that will allow the use of the whole sugarcane to produce ethanol or other renewable fuels, or even, through the Biorefinery concept, to aggregate value to the sugarcane chain, through the production of new chemical products.

Simultaneously with the development of the research agenda, different scenarios are being constructed, including conventional 1st generation plants (producing sugar, ethanol and bioelectricity); 2nd generation ethanol production (through hydrolysis of bagasse and trash) or 2nd generation liquid fuels production (BTL process) integrated to the 1st generation plants; biorefineries producing different products, besides ethanol and sugar (ethanol-chemistry, sugar-chemistry, others); biorefineries incorporating new agricultural technologies to produce sugarcane, as well as new technologies applied to ethanol and other products usage. The mathematical modeling and simulation of the different processes and operations included in the selected scenarios – that will constitute the so called Virtual Sugarcane Biorefinery, will allow to estimate, to compare, and to optimize the economic, social and environmental impacts obtained with the new technologies under development. These results will allow assessing the stage of development of the new technologies, as well as the interest in accelerating the implementation process, orienting the laboratories participating in the development, about possible optima operating conditions, looking for their experimental confirmation.