Course Outline

← ChE 477: Process Optimization

  • ChE 477 Process Optimization Course Outline Fall Semester 2002/2003 1999-2001 Catalog Data: ChE 477: Process Optimization. Credit 3+0+0 = 3. Fundamentals of analytical optimization. Survey of one dimensional line-search methods, and multi-dimensional unconstrained and constrained numerical optimization algorithms. Applications of linear programming, nonlinear programming, mixed integer linear/ nonlinear programming, and parameter estimation in chemical engineering. Feasible-path and infeasible-path techniques for chemical process flowsheet optimization. Prerequisites: ChE 334, ChE 342, ChE 386 Goals: This course is designed to give seniors in chemical engineering an ability to formulate optimization problems, to select and use optimization software packages, and to compare various optimization algorithms. Instructor: Prof. Dr. Uğur Akman Reference Textbooks: Edgar T.F., Himmelblau D.M., and Lasdon L.S., Optimization of Chemical Processes, 2nd Edition, McGraw-Hill Book Co., 2001 Edgar T.F. and Himmelblau D.M., Optimization of Chemical Processes, McGraw-Hill Book Co., 1988 Venkataraman P., Applied Optimization with MATLAB Programming, Wiley, 2001 Winston W.L., Operations Research - Applications and Algorithms, Brooks Cole, 2004 Ronald L. Rardin, Optimization in Operations Research, Prentice Hall, 1998 Bazaraa M.S. and Shetty C.M., Nonlinear Programming: Theory and Algorithms, John Wiley & Sons, Inc., 1979 Humphreys K.K. Jelen's Cost and Optimization Engineering, 3rd Ed., McGraw-Hill Book Co., 1991 Grossmann I.E., Edt., Chemical Engineering Optimization Models with GAMS, CACHE Design Case Studies Series, Case Study No:6, 1991 Prerequisites: matrix algebra chemical engineering systems, engineering economics computer programming ChE 386 Topics: 1. INTRODUCTION: definitions; terminology; mathematical representation; formulation of objective function; static versus dynamic optimization; unimodality; convexity and concavity; characterization of stationary points; contour plots; equality and inequality constraints; problems encountered in optimization; review of matrices and matrix algebra 2. ANALYTICAL TECHNIQUES: classical optimization theory; necessary and sufficient conditions; Hessian matrix; determinant and eigenvalue analyses of Hessian matrix 3. UNCONSTRAINED ONE-DIMENSIONAL OPTIMIZATION: numerical methods; interval of uncertainty; Line-Search Without Using Derivatives: region-elimination methods; sequential search; Dichotomous search; Golden-Section search; Fibonacci search; polynomial approximation methods, Line-Search Using Derivatives: bisecting search; Newton, Quasi-Newton, and Secant methods 4. UNCONSTRAINED MULTIVARIABLE OPTIMIZATION: use of line search in multidimensional search; Multidimensional Search Without Using Derivatives: cyclic coordinate method; method of Hooke-Jeeves; method of Rosenbrock; flexible polyhedron (simplex) search; Multidimensional Search Using Derivatives: steepest descent; Newton, Quasi-Newton, Marquardt-Levenberg, and Broyden methods; Methods Using Conjugate Directions: method of Fletcher and Reeves 5. CONSTRAINED MULTIVARIABLE OPTIMIZATION: Linear Programming (LP): graphical solution; slack and artificial variables; simplex method; sensitivity analysis; duality in LP; Penalty Function Techniques; Lagrange Multiplier Method; Kuhn-Tucker conditions 6. OPTIMIZATION OF STAGED AND DISCRETE PROCESSES: Dynamic Programming; Integer (IP) and Mixed Integer Programming (MIP: MILP, MINLP) 7. PARAMETER ESTIMATION: Linear/Nonlinear Regression 8. SPECIAL TOPICS: Genetic Algorithm, Simulated Annealing Computer Usage: Students are encouraged and required to use computer in homework assignments and projects. Student version of the GAMS software is distributed and used in several homework assignments and in final GAMS-based project. EXCEL's Solver (optimizer) is also used and compared with GAMS in some homework assignments. Grading: Approximate breakdown is as follows (may vary by e.g., ± 5%): Exam I ........ 20% Exam II ....... 20% Final Exam .... 20% Homework ...... 20% Geometric Average ! Project ....... 20% ___________________ 100%