210 likes | 285 Views
Explore various equipment and process alternatives above the base case to determine their technical and economic impact. Learn about Pareto analysis, optimization procedures, objective functions, constraints, and calculus methods for optimization. Discover LaGrange multipliers and other optimization methods including search methods like Monte Carlo for multi-dimensional systems.
E N D
EQUIPMENT & PROCESS ALTERNATIVES ABOVE THE BASE CASE
PROCESS OPTIONS • NEED TO BE INVESTIGATED TO HAVE A COMPLETE SURVEY • TECHNICAL EVALUATION TO DETERMINE RISK • ECONOMIC EVALUATION TO DETERMINE IMPACT http://www.astm.org/BIZLINK/BusLinkB01/images/line.jpg
ECONOMIC EVALUATION • CAPITAL COSTS ARE BASED ON A ±20% COST FOR PURCHASED EQUIPMENT. • OPERATING COSTS ARE BASED ON A ±5% ACCURACY • CASH FLOW RANGES SHOULD BE DEFINED FOR THE HIGH AND LOW END VALUE FOR THESE ESTIMATES • RESULTING SELLING PRICE RANGE INDICATES THE LIMITS FOR CONSIDERING PROCESS REVISIONS. • NOTE THAT THE EXTREMES OF THE PRICES BASED ON THE CASH FLOW ANALYSES ARE WHAT DETERMINES THE BASIS FOR RECOMMENDATIONS • ALSO INCLUDE THE RISK ASSOCIATED WITH THE ALTERNATE TECHNOLOGY.
GENERAL OPTIMIZATION METHODS • PROJECTS CAN BE OPTIMIZED ON A UNIT BY UNIT BASIS OR ON LARGER SYSTEMS • DETERMINATION OF THE MOST SIGNIFICANT COMPONENTS IS CALLED A PARETO ANALYSIS (http://www-personal.umich.edu/~westj/files/cds/individual/Disk2/lectures/08/08t-pareto.pdf) • THE BASIC IDEA IS TO PUT THE EFFORT IN THE AREA THAT HAS THE HIGHEST TOTAL RETURN POTENTIAL
1Pareto Analysis(http://www-personal.umich.edu/~westj/files/cds/individual/Disk2/lectures/08/08t-pareto.pdf) • Pareto* Principle provides the foundation for the concept of the “vital few” and a “trivial many” • Examples: • Quality – a small percentage of defect categories • (causes) will constitute a high % of the total # defects. • Cost – a small percentage of components will constitute • a high % of total product cost. • Others: Inventory, absenteeism, downtime • *Note: Wilfredo Pareto – 19th Century Italian economist studying wealth who observed that a large proportion of wealth is owned by a small percentage of the people. Pareto principle was later applied to quality by J.M. Juran
Pareto Analysis(http://www-personal.umich.edu/~westj/files/cds/individual/Disk2/lectures/08/08t-pareto.pdf) • 80/20 Rule • In quality, this rule suggests that ~20% of defect categories will account for ~80% of the total number of defects. Example for Bid Preparations
Pareto Analysis(http://www-personal.umich.edu/~westj/files/cds/individual/Disk2/lectures/08/08t-pareto.pdf) • Pareto Chart
Pareto Analysis(http://www-personal.umich.edu/~westj/files/cds/individual/Disk2/lectures/08/08t-pareto.pdf) • Pareto Analysis may be performed using: • Frequency of occurrence (expressed as a frequency count or relative frequency %), • Or Total cost, • Or Severity, adverse outcome, or avoidability • Note: the most frequently occurring item may not be the most important item to address first
OPTIMIZATION PROCEDURE • DEVELOP OBJECTIVE FUNCTION • DEVELOP CONSTRAINTS • MATHEMATICALLY OPTIMIZE http://www.ltponline.com/services_images/lp_graph_lg.gif
OBJECTIVE FUNCTIONS • CAN BE LINEAR OR NON-LINEAR AND INCLUDE MANY VARIABLES. Y = Y (x1, x2, ..., xn) = Y () http://virtual.clemson.edu/groups/mathsci/graduate/seminar/02_3.jpg
CONSTRAINTS • INCLUDE SAME VARIABLES AS THE OBJECTIVE FUNCTION • CAN BE EQUALITIES • Φ1 = Φ1 (x1, x2, ..., xn) = Φ1 (x) • Φ2 = Φ2 (x1, x2, ..., xn) = Φ2 (x) . • Φj = Φj (x1, x2, ..., xn) = Φj (x) • OR INEQUALITIES : • Ψ1 = Ψ1 (x1, x2, ..., xn) = Ψ1 (x) ≤ L1 • Ψ2 = Ψ2 (x1, x2, ..., xn) = Ψ2 (x) ≤ L2 • Ψk = Ψk (x1, x2, ..., xn) = Ψk (x) ≤ Lk www.britannica.com/eb/art-3028?articleTypeId=1
CALCULUS METHODS • BASED ON Y’ = 0 AT OPTIMUM • USING TOTAL DERIVATIVE • AT THE OPTIMUM, ALL PARTIALS EQUAL ZERO http://www.math.lsu.edu/~verrill/teaching/calculus1550/optimize.gif
LA GRANGE MULTIPLIERS • SIMULTANEOUS SOLUTION OF n EQUATIONS FOR n UNKNOWNS. • THE FUNCTION TO BE OPTIMIZED HAS THE FORM: • WHERE THE GRADIENT IS • AND λi IS THE LAGRANGIAN MULTIPLIER • SET UP AN EQUATION FOR EACH VARIABLE BASED ON THE GRADIANT VECTOR FOR THE SCALAR OF THE OPTIMIZATION FUNCTION: • WHERE THE UNIT VECTOR IS
OTHER OPTIMIZATION METHODS • FOR MULTI-DIMENSIONAL SYSTEMS • SEARCH METHODS - EVALUATE Y AT VARIOUS POINTS TO LOCATE THE OPTIMA (MONTE CARLO METHOD) • TYPES - DICHOTOMOUS SEARCH, FIBONACCI SEARCH, GOLDEN SECTION, MONTE CARLO METHOD • MAY BE THE ONLY OPTION FOR COMPLEX SYSTEMS