INDE 2331 DATA ANALYSIS USING EXCEL LINEAR OPTIMIZATION University of Houston Dept. of Industrial Engineering Houston, TX 77204-4812 (713) 743-4195 AGENDA • Linear Optimization Linear Optimization • Some IE problems involve deterministic (nonprobabilistic) data • Example – How many parts of different types can be produced given limited resources or production facilities – How do I load vehicles for best utilization – How many people do I need to finish a project on time • Linear Optimization – INDE 3381 BASIC CONCEPTS • Linear objective function… • Decision variables… • Set of constraints… LINEAR OBJECTIVE FUNCTION • Equation consisting of – Decision variables – Coefficients for each decision variable • Result of the equation is what you want to minimize or maximize • Result is commonly known as “Z” • Examples – Maximize profit – Minimize cost DECISION VARIABLES • Independent variables • Identify the values for these variables so that the objective function is either – Minimized – Maximized • Example – How many of each type of product to produce SET OF CONSTRAINTS • Conditions which must be satisfied – Limited resources – Limited time – Market demand • Example – You only have a total amount of raw material to make the products – Must make a minimum quantity of eqch specific type of product SIMPLE EXAMPLE • To augument his professor’s salary, Dr. Chung makes • • • toy soldiers and toy trains in his spare time. He makes $3 in profit for every toy soldier and $2 in profit for every toy train. A toy soldier takes 1 hour of carpentry work and 2 hours of finishing time. A toy train takes 1 hour of carpentry work and 1 hour of finishing time. Dr. Chung has a total of 80 hours of carpentry time and 100 hours of finishing time available each month. Only 40 toy soldiers can be sold a month, but as many toy trains that can be made can be sold OBJECTIVE FUNCTION • Maximize profit Z • X1 is the number of toy soldiers to make • X2 is the number of toy trains to make • Make 3 dollars for each toy soldier • Make 2 dollars for each toy train CONSTRAINTS • Total carpentry time is 80 hours • Total finishing time is 100 hours • Can’t sell more than 40 toy soldiers • Must make 0 or more toy soldiers • Must make 0 or more toy trains STANDARD FORM • Maximize Z = 3x1 + 2x2 • Subject to: X1 + x2 <= 80 (carpentry) 2x1+x2<=100 (finishing) X1 <= 40 (soldier sale limitation) X1 >= 0 (non-negativity constraint) X2 >=0 (non-negativity constraint) SOLVING LP • Graphical solution • Simplex algorithm by hand • Solve it in Excel • Solve it in some other software package GRAPHICAL SOLUTION • Plot lines corresponding to the constraints • The resulting polygon is the set of feasible solutions • Plot the objective function for different levels of profit • The highest isoprofit line to intersect with the polygon is the maximum profit • The intersection of the isoprofit line and the polygon is the optimal solution • What does this look like? CONSTRAINT LINES 120 100 80 carpentry 60 finishing sale 40 20 0 0 20 40 60 80 100 FEASIBLE SOLUTION POLYGON 120 100 80 carpentry finishing 60 sale 40 20 0 0 20 40 60 80 100 ISOPROFIT LINE 120 100 80 carpentry finishing 60 sale isoprofit 40 20 0 0 20 40 60 80 100 SIMPLEX ALGORITHM • Algorithmic procedure which utilizes matrix operations • Solves for solutions by calculating the values at the polygon vertices • Determines which is the highest • Entire undergraduate class on this INDE 3381 • Excel uses this type of algorithm HOW TO DO THIS IN EXCEL • Name decision variable cells • Create objective function cell • Create cells with constraint formulas • Tools-solver • Designate – – – – Target objective function cell Max or min By changing cells (decision variable cells) Add constraints with cell references, inequalities, rhs values • Options – assume linear • Solve