Topic: What-if Analysis (Sensitivity Analysis) for Linear Programming
Instructor: Rim Jaber
Course: ADM2302
Assumption: Model parameters are known with certainty.
Reality: Model parameters are estimates and can change.
Example: Referring to the RMC Problem for context.
Company Overview: RMC, Inc. produces chemical-based products using three raw materials for two products.
Material Usage per Product:
Fuel Additive:
Material 1: 2/5 tons per ton
Material 2: 0 tons
Material 3: 3/5 tons per ton
Solvent Base:
Material 1: 1/2 tons per ton
Material 2: 1/5 tons per ton
Material 3: 3/10 tons per ton
Available Material for Production:
Material 1: 20 tons
Material 2: 5 tons
Material 3: 21 tons
Profit Contribution:
Fuel Additive Profit: $40/ton
Solvent Base Profit: $30/ton
Goal: Determine tons of each product to maximize total profit.
Solution:
it will remain the same because between $40 and $50 there is a change of 10 which is allowable because it is within the allowable increase of 20 (0.4X1+0.5X2<=20)
The new optimal solution would now be (35,0) because the profit contribution of solvent based has dropped (Q2)
if we change binding constraints then the corner points would change
Objective Function:
Maximize Z = $40 * x1 + $30 * x2
Constraints:
(2/5)x1 + (1/2)x2 ≤ 20 (Material 1)
(1/5)x2 ≤ 5 (Material 2)
(3/5)x1 + (3/10)x2 ≤ 21 (Material 3)
x1 ≥ 0, x2 ≥ 0
Variables:
x1: tons of fuel additive
x2: tons of solvent base
Graph:
x1 on horizontal axis
x2 on vertical axis
Constraints represented as lines:
2/5x1 + 1/2x2 = 20
1/5x2 = 5
3/5x1 + 3/10x2 = 21
Corner Points Identified:
(25, 20), (18.75, 25), (0, 25), (0, 0), (35, 0)
Optimal Solution using Corner Point Method:
Calculation of Z for each point
Calculated Values of Z:
Z(25, 20) = $750
Z(18.75, 25) = $1500
Z(0, 25) = $0
Z(35, 0) = $1400
Z(25, 20) = $1600
Optimal Solution Found: at (25, 20)
Maximum Isoprofit Line:
Represented as Z = $1600
Optimal Solution Point identified at (25, 20)
Spreadsheet Reference: RMC_Problem.xlsx
Formula Examples:
fx =SUMPRODUCT(B5:C5,$B$3:$C$3)
Variables in Spreadsheet:
D5: Solution
D8 to D10: Material constraints analyzed
Profit Totals:
Original: $0
Final Value: $1,600
Variable Cells:
x1 (Fuel Additive): 0 to 25
x2 (Solvent Base): 0 to 20
Binding Constraints Identified:
Material 1: Binding
Material 2: Not Binding
Material 3: Binding
Variable Cells Summary:
Final Value, Reduced Cost, Objective Coefficient
Allowable increase/decrease for x1 (fuel additive) and x2 (solvent base)
Constraints Analysis: Shadow price evaluations
Definition and Purpose: Sensitivity analysis questions the effect of parameter changes on optimal solutions
Changes in Objective Functions Coefficients
Changes in Right-Hand Sides (RHS) Values
Shadow Prices
Range of Feasibility
Simultaneous Changes
Reduced Costs
Pricing out New Variables
Understanding Coefficients (both are found from the objective function):
c1 for x1 (c1=40)
c2 for x2 (c2=30)
Goal of Sensitivity Analysis:
Determine range of c that keeps the current solution optimal.
Implication: Objective function coefficients do not affect feasible region size (the only thing that happens when changing the coefficients is a change in the slope of the objective function)
Range of Optimality:
x1: 24 ≤ c1 ≤ 60
x2: 20 ≤ c2 ≤ 50
Zero value decision variable: Non-use does not change Z
Non-zero value decision variable: Affects Z positively or negatively
Question and Answer: What if profit estimates are inaccurate? A wide range of values ensures likely optimal solution integrity
Impact of Changes in Constraints: Affect feasible region and optimal solution value
RHS Values: e.g., Q1 = 20, Q2 = 5, Q3 = 21
Optimal Points: Evaluating through graphical representation
Understanding Shadow Prices:
Increase profits per additional resource unit
How to Determine: Increase RHS in spreadsheet, solve anew
Definitions:
Marginal profit & max price willing for resources
Profit decreases by 1 unit reduction
Objective: Examine small changes in RHS and their impact
Value Ranges:
Q1: 14 ≤ Q1 ≤ 21.5
Q2: 4 ≤ Q2
Q3: 18.75 ≤ Q3 ≤ 30
Impact: Changes in binding constraints change optimal solution; non-binding do not affect profit unless within range
Purpose: Determine potential changes in optimal solution under multiple coefficient changes simultaneously
Application: Total allowable changes must not exceed 100% to maintain optimal solution
Illustration: c1 and c2 adjustments confirmed the stability of the current optimal solution
Outcome: Re-evaluate optimal solution needs resolving due to rule breach
Sum of changes: Assess RHS adjustments under feasibility ranges
Topic Validity and Examples discussed
Concept: Explanation of reduced costs for unused activities; relevant for analysis and strategy
Criteria: Identifying zero optimal values and implications for solution stability
Sensitivity Report uses: Assessing the impact of variable changes on current production
Example Evaluation of Ultra Base with material consumption and profit contribution
Checking shadow price validity; guiding production decisions based on opportunity cost assessment
Summary of solutions discussed in-depth during sessions
Final Values: Review usage and changes in objective coefficients
Allowable Changes: Define operational ranges for coefficients
Final Values and Shadow Prices: Implications for resource adjustments and their operational impact.