Upload
amelia-singleton
View
222
Download
2
Tags:
Embed Size (px)
Citation preview
QMB 4701 MANAGERIAL OPERATIONS
ANALYSIS
Jay Marshall TeetsDecision and Information Sciences
University of Florida
Chapter 3Linear Programming: Graphical Solution
Procedure & Spreadsheet Solution Procedure
Agenda• Graphical linear programming
– Example: Par, Inc. Problem – Graphical solution procedure– Example: Dorian problem
• Additional considerations– Degeneracy– Inconsistent problem – Multiple optimal solutions– Unbounded problem
• Spreadsheet solution basics
Example: Par, Inc. Problem
Par, Inc: Manufacturer of golf supplies
New products: Standard golf bags and Deluxe golf bags
Manufacturing operations:
Objective: Maximize total profit
Production Time (Hrs)Standard Bag Deluxe Bag
ResourceConstraints (Hrs)
Cutting &Dyeing 7/10 1 630Sewing 1/2 5/6 600
Finishing 1 2/3 708Inspection
& Packaging 1/10 1/4 135
ProfitContribution ($)
10 9
Cutting & Dyeing
Sewing
Finishing
Inspection & Packaging
Par Inc. Model Formulation
1. What is the objective in words?
2. What are the constraints in words?
Maximize the total profit
Cutting & Dyeing Cutting & Dyeing Hours Allocated Hours Available
Sewing Hours Allocated Sewing Hours Available
Finishing Hours Allocated Finishing Hours Available
Inspection & Packaging Inspection & Packaging Hours Allocated Hours Available
Par Inc. Model Formulation
3. What are the decision variables?
4. Formulate the objective function:
5. Formulate the constraints:
6. Do we need non-negative constraints?
x1 = Number of Standards Producedx2 = Number of Deluxes Produced
Profit Contribution Maximize 10 x1 + 9 x2
Subject to:Cutting & Dyeing: 7/10 x1 + x2 630Sewing 1/2 x1 + 5/6 x2 600Finishing x1 + 2/3 x2 708Inspection & Packaging 1/10 x1 + 1/4 x2 135
Logic x1 0 , x2 0
How to Solve It?Profit Contribution Maximize 10 x1 + 9 x2
Subject to:Cutting & Dyeing: 7/10 x1 + x2 630Sewing 1/2 x1 + 5/6 x2 600Finishing x1 + 2/3 x2 708Inspection & Packaging 1/10 x1 + 1/4 x2 135
Logic x1 0 , x2 0
Trial and Error?
Feasible Solution
Infeasible Solution
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
x1 = 600,x2 = 100?
Profit = 6900
520383.33666.67
85
x1 = 700,x2 = 100?
Profit =7900
590433.33766.67
95
Graphical Approach
Solutions
Solution: any production plan
Feasible solution:solution that satisfies all constraintsFeasible solutions
A. Determine feasible region(draw constraint lines)
Optimal solution:feasible solution that achievesthe maximal objective value
B. Construct isoprofit lines
C. Locate the optimal solutionOptimal solutions
Constraints Determine Feasible Region
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200
x2
Approach: 1. Draw lines representing the constraints solved as equalities2. Show the direction of feasibility3. The feasible region is the set of values which simultaneously satisfies all the constraints
Cutting & Dyeing 7/10 x1 + x2 630Sewing 1/2 x1 + 5/6 x2 600Finishing x1 + 2/3 x2 708Inspection & Packaging 1/10 x1 + 1/4 x2 135 x1 0, x2 0,
A. Constraints Determine Feasible Region
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2 Finishing
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
Inspection& Packaging
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2 Cutting & Dyeing
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2 Sewing
A. Constraints Determine Feasible Region
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
Feasible Region
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2 Feasible Region
Redundantconstraint
Construct Isoprofit Lines Approach:1. Select a constant C2. Draw the “profit line” 10 x1 + 9 x2 = C3. Construct parallel profit lines
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200
x2 Isoprofit Lines
4500
9000
Slope of the objective function: x2= -10/9 x1+C/9
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200
x2
4500
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200
x2
Locate The Optimal Solution - Observations
Optimal Solution
Observations:1. The optimal solution cannot fall in the interior of the feasible region2. The optimal solution must occur at a “corner point” of the feasible region3. In searching for the optimal solution we only have to examine the corner points
200 400 600 800 1000 1200 1400
x1
200
400
600
800
1000
1200x2
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
1 2
3
45
F
C&D
I &P
Locate The Optimal Solution Find extreme points (corner points) and calculate objective valuesExtreme Points: the vertices or “corner points” of the feasible region. With two variables, extreme points are determined by the intersection of the constraint lines.
Point Constraint Constraint x1 x2 Obj. 1 Non-negative Non-negative 0 0 0 2 Non-negative Finishing 708 0 77080 3 Cutting & Dyeing Finishing 540 252 7668 4 Cutting & Dyeing Inspection
& Packaging 300 420 6780
5 Non-negative Inspection & Packaging
0 540 4860
Locate The Optimal Solution - Find Extreme Points
Example: Find extreme point #3:
Intersection of the cutting & dyeing constraint and the finishing constraint:
Cutting & Dyeing: 7/10 x1 + x2 = 630
Finishing x1 + 2/3 x2 = 708
Solve the two equations for x1 and x2:
7/10 x1 + x2 = 630 7/10 x1 = 630 x2 x1 = 900 10/7 x2
x1 + 2/3 x2 = 708 (900 10/7 x2) +2/3 x2 = 708
900 - 30/21 x2 + 14/21 x2 = 708 16/21 x2 = 192 x2 = 21/16 192 = 252
x1 = 900 - 10/7 x2 x1 = 900 10/7 (252) x1 = 900 360 = 540
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
F
C&D
I &P
Slack and Binding Constraints
At optimal solution (540, 252) Hrs used Hrs availableC&D: 630 = 630S: 480 < 600F: 708 = 708I&P: 117 < 135
BindingSlack = 600 480 = 120BindingSlack = 135 117 = 18
Binding constraints
Slack constraint
Summary of Graphical Approach
A. Graph Feasible Region– Draw each constraint line
• For each constraint, plot points on the x1 axis (x2=0) and x2 axis (x2=0) and connect the points with a line.
– Indicate the feasible area with arrow• Check if (0,0) is include, if so, draw arrow towards (0,0). If
not, draw arrow away from (0,0). If (0,0) is on the constraint line, you must choose another point to use as a check
– Hatch the area in common for all constraints and their respective arrows
B. Find Extreme Points– Set the intersecting constraints equal to one another - basic algebra
Summary of Graphical Approach
C. Find optimal solution
Option #1
– List all the extreme points and calculate objective function values for each extreme point.
– Choose the point with the highest (profit) or lowest (cost) value
Option #2
– Find the slope of the objective function
– Choose which extreme point is last touched by the isoprofit/isocost line as it is increased/decreased. This point is the optimal point
– Calculate the objective value for the optimal point.
Note: Simplex Method
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
1. Start with a feasible corner point solution2. Check to see if a feasible neighboring corner point is better3. If not, stop; otherwise move to that better neighbor and return to step 2.
Graphing Example: Minimization Problem
MIN 50,000 x1 + 100,000 x2STHIW: 7x1 + 2 x2 28HIM: 2x1 + 12 x2 24
x1 0, x2 0
Dorian Problem
Extreme Points x1 x2 OBJ1 0 14 14002 3.6 1.4 3203 12 0 600
2 4 6 8 10 12 14 16x1
2
4
6
8
10
12
14
16x2
0,14
3.6,1.4 12,0
2 4 6 8 10 12 14 16x1
2
4
6
8
10
12
14
16x2
HIW
HIM
200 400 600 800 1000 1200 1400
x1
200
400
600
800
1000
1200
x2 Feasible Region
Additional Consideration: DegeneracyDegeneracy: Assume Inspection & Packaging time is 117 hours (instead of 135)
Profit Contribution Maximize 10 x1 + 9 x2
STCutting & Dyeing: 7/10 x1 + x2 630Sewing 1/2 x1 + 5/6 x2 600Finishing x1 + 2/3 x2 708Inspection & Packaging 1/10 x1 + 1/4 x2 117Logic x1 0 , x2 0
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200
x2 Feasible Region
Additional Consideration: Inconsistent Problem
Assume there is an additional constraint: need to produce at least 725 standard bags, i.e., x1 725 Profit Contribution Maximize 10 x1 + 9 x2
STCutting & Dyeing: 7/10 x1 + x2 630Sewing 1/2 x1 + 5/6 x2 600Finishing x1 + 2/3 x2 708Inspection & Packaging 1/10 x1 + 1/4 x2 135
x1 725Logic x1 0 , x2 0
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
Additional Consideration: Multiple Optimal Solutions
Assume the objective function is 12 x1 + 8 x2 (instead of 10 x1 + 9 x2) Profit Contribution Maximize 12 x1 + 8 x2
STCutting & Dyeing: 7/10 x1 + x2 630Sewing 1/2 x1 + 5/6 x2 600Finishing x1 + 2/3 x2 708Inspection & Packaging 1/10 x1 + 1/4 x2 135Logic x1 0 , x2 0
Slope of the new objective function:x2 =-12/8 x1 +C/8= =-3/2 x1 +C/8
Additional Consideration: Unbounded Problem
Profit Contribution Maximize 10 x1 + 9 x2
STDemand: x1 + x2 1000
x1 725Logic x1 0 , x2 0
Assume we have the following problem:
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
200 400 600 800 1000 1200 1400x1
200
400
600
800
1000
1200x2
Excel Solver Solution Procedure
• Excel can do the following:– Solve linear programs Note: Activate the “Assume Linear Model”
option if your problem is a LINEAR program, otherwise the Solver will treat it as a NONLINEAR program.
– Solve nonlinear programs– Solve integer programs– Perform sensitivity analysis
Implementing an LP Model in a Spreadsheet
• Organize the data for the model on the spreadsheet including:
– Objective Coefficients
– Constraint Coefficients
– Right-Hand-Side (RHS) values
• clearly label data and information
• visualize a logical layout
• row and column structures of the data are a good start
Implementing an LP Model in a Spreadsheet
• Reserve separate cells to represent each decision variable – arrange them such that the structure parallels the input
data – keep them together in the same place – Right-Hand-Side (RHS) values
• Create a formula in a cell that corresponds to the Objective Function
• For each constraint, create a formula that corresponds to the left-hand-side (LHS) of the constraint
• NOTE: Once you’ve formulated a spreadsheet model for a certain type of application, it’s useful to save it as a template for future use!
Solver Terminology• Target cell
– Represents the objective function– Must indicate max or min
• Changing cells– Represents the decision variables
• Constraint cells– the cells in the spreadsheet that represent the LHS formulas of the
constraints in the model – the cells in the spreadsheet that represent the RHS values of the
constraints in the model – TIP: If you scale the constraints such that they are all "" or all
"", then it's easier to input the constraints as a block. Then, you avoid having to enter each constraint as a separate line.
Solver Example: Par, Inc. ModelA B C D E F
12 Operation Standard Bag Deluxe Bag Hrs Used Hrs Available3 Cutting & Dyeing 0.7 1 0 <= 6304 Sewing 0.5 0.83333333 0 <= 6005 Finishing 1 0.66666667 0 <= 7086 Inspection & Packaging 0.1 0.25 0 <= 13578 Total Profit9 Profit per Unit 10 9 0
1011 No. of Units Produced 0 0
Production Time (hrs/unit)
A B C D E F12 Operation Standard Bag Deluxe Bag Hrs Used Hrs Available3 Cutting & Dyeing 0.7 1 =B3*$B$11+C3*$C$11 <= 6304 Sewing 0.5 =5/6 =B4*$B$11+C4*$C$11 <= 6005 Finishing 1 =2/3 =B5*$B$11+C5*$C$11 <= 7086 Inspection & Packaging 0.1 0.25 =B6*$B$11+C6*$C$11 <= 13578 Total Profit9 Profit per Unit 10 9 =B9*B11+C9*C111011 No. of Units Produced 0 0
Production Time (hrs/unit)
Par, Inc. Model - Answer Report
Solver Parameters: Objective: MAX F9 Variables: B11:C11 Constraints: D3:D6 <= F3:F6 Options: Assume Linear Model Assume Non-Negative
Worksheet: [LP.xls]ParReport Created: 1/27/00 11:09:15 PM
Target Cell (Max)Cell Name Original Value Final Value
$F$9 Profit per Unit Total Profit 0 7668
Adjustable CellsCell Name Original Value Final Value
$B$11 No. of Untis Produced Standard Bag 0 540$C$11 No. of Untis Produced Deluxe Bag 0 252
ConstraintsCell Name Cell Value Formula Status Slack
$D$3 Cutting & Dyeing Hrs Used 630 $D$3<=$F$3 Binding 0$D$4 Sewing Hrs Used 480 $D$4<=$F$4 Not Binding 120$D$5 Finishing Hrs Used 708 $D$5<=$F$5 Binding 0$D$6 Inspection & Packaging Hrs Used 117 $D$6<=$F$6 Not Binding 18
Optimal objective value
Optimal solution
Constraints Binding & Slack
Information
Summary of Excel Solver Procedure
LHS Inequality RHS=>=<=
Objective Function
Right Hand Side Coefficients
Formulas
Decision Variables
Coefficients of Decision Variables
Constraints Left Hand Side Coefficients
Solver Example: Minimization ExampleMIN 50,000 x1 + 100,000 x2
ST7x1 + 2 x2 282x1 + 12 x2 24x1 0, x2 0
Dorian Problem
A B C D E F12 Comedy Ads Football Ads LHS Requirement3 High-income women 7 2 0 >= 284 High-income men 2 12 0 >= 2456 Total Cost7 Cost per minute($M) 50 100 089 # of Ads paid 0 0
Population (M) reached by Ads
Solver Example: Minimization ExampleA B C D E F
12 Commedy Ads Football Ads LHS Requirement3 High-income women 7 2 =SUMPRODUCT(B3:C3,$B$9:$C$9) >= 284 High-income men 2 12 =SUMPRODUCT(B4:C4,$B$9:$C$9) >= 2456 Total Cost7 Cost per minute($M) 50 100 =SUMPRODUCT(B7:C7,B9:C9)89 # of Ads paid 0 0
Population (M) reached by Ads
Summary of Additional Considerations
ADDITIONAL CONSIDERATIONS
GRAPHICAL EXCEL SOLVER
MANAGERIAL IMPLICATION
Degeneracy More than two constraint lines intersect at one
point
Inconsistent Problem Feasible region does not exist
Could not find a feasible solution
Too many restrictions
Multiple Optimal Solutions
Slope of objective
function is same as that of one
constraint
Output shows only a single
optimal solution
Unbounded Problem Optimal value is infinite
The set of cell values do not
converge
Problem is improperly formulated
Summary
• Graphical solution procedure– Feasible region– Extreme points– Optimal solution– Binding and Slack Constraints
• Additional considerations– Degeneracy– Inconsistent problem– Multiple optimal solutions– Unbounded problem
• Spreadsheet solution basics