IE230 Operations Research I ::::: Homework #5 due Wednesday, 22 March 2006 1. Linear Programming Model Formulation: Brady Corporation produces cabinets. Each week, they require 90,000 cu ft of processed lumber. They may obtain processed lumber in two ways. First, they may purchase lumber from an outside supplier and then dry it at their kiln. Second, they may chop down logs on their own land, cut them into lumber at their sawmill, and finally dry the lumber at their kiln. Brady can purchase grade 1 or grade 2 lumber. Grade 1 lumber costs $3 per cu ft (cubic foot) and when dried yields 0.7 cu ft of useful lumber. Grade 2 lumber costs $7 per cu ft and when dried yields 0.9 cu ft of useful lumber. It costs the company $3 per cu ft to chop down a log. After being cut and dried, one cubic foot of log yields 0.8 cu ft of lumber. Brady incurs costs of $4 per cu ft of lumber that is dried in their kiln. It costs $2.50 per cu ft of logs sent through their sawmill. Each week, the sawmill can process up to 35,000 cu ft of lumber. Each week, up to 40,000 cu ft of grade 1 lumber and up to 60,000 cu ft of grade 2 lumber can be purchased. Each week, 40 hours of time are available for drying lumber. the time it takes to dry 1 cubic foot of grade 1 lumber, grade 2 lumber, or logs is as follows: grade 1: 2 seconds grade 2: 0.8 seconds log: 1.3 seconds a. Formulate an LP to help Brady minimize the weekly cost of meeting the demand for processed lumber. Be sure to state precisely the definitions of your decision variables, and briefly (e.g., “sawmill capacity”) explain the purpose of each type of constraint. Grade 1 Grade 2 Logs from their Capacity own land Cost 3$ 7$ 0 0 Useful lumber 0.7 cu. ft. 0.9 cu. ft 0.8 cu ft 90000 cu. Ft Hours 2 sec 0.8 sec 1.3 sec 144000 sec Chop down 0 0 3$ 0 Dry 4$ 4$ 4$ 0 Sawmill 0 0 2.5$ 35000 cu. ft Capacity 40000 cu ft 60000 cu ft 0 X1 – The quantity (cu.ft) of grade 1 X2 – The quantity (cu.ft) of grade 2 X3 – The quantity (cu.ft) of logs from their own land Min z = 7*x1 + 11*x2 + 9.5*x3 0.7*x1 + 0.9*x2 + 0.8*x3 >= 90000 2*x1 + 0.8*x2 + 1.3*x3 <= 144000 X1 <=40000 X2 <=60000 X3<=35000 xi >=0, (i = 1, 2, 3) IE230 HW#5 2/17/2016 page 1 of 4 Rows= 6 Vars= 3 No. integer vars= 0 ( all are linear) Nonzeros= 17 Constraint nonz= 9( 3 are +- 1) Density=0.708 Smallest and largest elements in abs value= 0.700000 144000. No. < : 4 No. =: 0 No. > : 1, Obj=MIN, GUBs <= 3 Single cols= 0 Optimal solution found at step: Objective value: Variable X1 X2 X3 Row 1 2 3 4 5 6 1 1033585. Value 40000.00 55471.70 15094.34 Slack or Surplus 1033585. 0.0000000 0.0000000 0.0000000 4528.302 19905.66 Reduced Cost 0.0000000 0.0000000 0.0000000 Dual Price 1.000000 -12.64151 0.4716981 0.9056604 0.0000000 0.0000000 b. Solve the problem, using LINGO, LINDO or other LP solver. c. State (in words) the optimal solution in such a way that the manager of this company could understand it. 2. Consider the following LP: M inimize z = 3x1 + x2 s.t. x1 - 2x2 2 x1 + x2 3 x1 0, x2 0 a. Convert the constraints (excluding the nonnegativity constraints) into equations, by the use of "surplus" variables. Minimize z = 3x1 +x2 s.t. x1 – 2x2 – e1 +a1 = 2 x1 + x2 – e2 +a2 = 3 x1>=0, x2 >=0, e1 >=0, e2>=0 b. How many basic variables will the set of two equality constraints have? The set of two equality constraints have two basic variables: a1, a2. c. How many basic solutions (both feasible and infeasible) will the two equality constraint have, i.e., how many ways may these basic variables be chosen from the set of four variables in the equations? Because we have 2 constraints and 6 variables (2 basic and 4 non-basic) in this case 6 we will have =15 basic solutions (both feasible and infeasible). 2 IE230 HW#5 2/17/2016 page 2 of 4 d. Prepare the tableau to use the two-phase method to find an initial basic feasible solution of the problem above, using artificial variables. (You need not solve it). Minimize w = a1+ a2 s.t. x1– 2x2 – e1 + a1 =2 x1+ x2 – e2 + a2 = 3 -w 1 0 0 0 -z 0 1 0 0 X1 0 3 1 1 X2 0 1 -2 1 E1 0 0 -1 0 E2 0 0 0 -1 A1 1 0 1 0 A2 1 0 0 1 rhs 0 0 2 3 3. Revised Simplex Method We wish to solve the LP problem M aximize z=10X1 + 6X2 + 4X3 subject to: X1 + X2 + X3 100 10X1 + 4X2 + 5X3 600 2X1 + 2X2 + 6X3 300 Xj 0, j=1,2,3 a. Complete the original tableau for solving this problem using the simplex method: -z 1 0 0 0 X1 10 1 10 2 X2 6 1 4 2 X3 4 1 5 6 X4 0 1 0 0 X5 0 0 1 0 X6 0 0 0 1 RHS 0 100 600 300 Suppose that, after several iterations, we obtain the tableau below (in which some values have been omitted): -z X1 X2 X3 X4 X5 X6 RHS 10/3 5/6 5/3 -1/6 0 200/3 1 0 1/6 1/6 0 2/3 0 1 100 2 b. What is the current basis for the constraint rows? B = { 2___ , ___ , ___ } c. What is the “substitution rate” of X4 for X1? ___________ d. If X4 increases by 1 unit, X1 will ( circle one: increase or decrease) by ________ units. ___ B e. What is the basis matrix? A ___ ___ IE230 HW#5 ___ ___ ___ 2/17/2016 ___ ___ ___ page 3 of 4 _____ _____ _____ f. What is the inverse of the basis matrix? A _____ _____ g. What are the values of the simplex multipliers for this basis? _____, _____, _____ B 1 _____ _____ _____ _____ h. What is the “relative profit” of X3, i.e., the number in the objective row of the X3 column in the current tableau? _______ i. Compute all of the missing portions of the tableau above, using the basis inverse and the original tableau. IE230 HW#5 2/17/2016 page 4 of 4