Operational Research -Management Information Engineering

☰ Contents

1. Introduction 2. Unit 1: Modeling for Decisions 3. Unit 2: Data Management 4. Unit 3: Forecasting 5. Unit 4: Optimization 6. Unit 5: Decision & Risk Analysis 7. Unit 6: Monte Carlo Simulation 8. Unit 7: Systems Modeling 9. Semester Questions 10. Summary

Introduction to Management Information Systems

Course Code: MIS MIE123 | Duration: 45 Hours | University: Purbanchal University

This course provides comprehensive coverage of the design and implementation of Management Information Systems (MIS), networks and data communications, managerial decision making, and managerial aspects of organizational information systems. Students will learn how to use quantitative models, data analysis, and simulation techniques to support organizational decision-making and optimize business operations.

Course Objectives:

Understand the fundamentals of operations research and modeling for business decisions
Master data management, analysis, and visualization techniques
Learn forecasting methods for time-series data
Develop optimization models using linear and multi-objective programming
Apply decision and risk analysis frameworks
Build and interpret Monte Carlo simulation models
Model and simulate dynamic systems and queueing scenarios

Unit 1: Introduction to Modeling for Decisions (3 Hours)

Overview: This unit introduces the fundamental concepts of Operations Research (OR) and the process of developing mathematical models to support managerial decision-making.

2.1 Applications and Benefits of Operations Research

Description: Operations Research is a scientific approach to problem-solving that uses mathematical and statistical models to optimize business processes and decisions. It combines quantitative analysis with managerial judgment to achieve organizational objectives.

Key Applications:

Supply Chain Optimization - Reducing costs through inventory and logistics planning
Portfolio Management - Maximizing returns while managing risk
Production Scheduling - Optimizing manufacturing workflows
Resource Allocation - Distributing limited resources efficiently
Network Design - Optimizing transportation and communication networks

Benefits:

Improved decision quality through data-driven analysis
Cost reduction by optimizing operations
Risk reduction through systematic evaluation
Better resource utilization and efficiency

Video Reference: Introduction to Operations Research Fundamentals - https://youtu.be/operations-research-intro

2.2 Developing Mathematical Models

Description: A mathematical model is a representation of a real-world problem using mathematical equations and variables. Model development involves identifying the problem, defining decision variables, establishing relationships, and formulating objectives.

Steps in Model Development:

Problem Definition: Clearly identify the business problem and objectives
Data Collection: Gather relevant data about the system
Variable Definition: Define decision variables, parameters, and constraints
Relationship Formulation: Express relationships using mathematical equations
Objective Function: Define what we want to maximize or minimize

Types of Models:

Deterministic Models - Assumes all parameters are known with certainty
Stochastic Models - Incorporates uncertainty and probability
Linear Models - Relationships are linear equations
Nonlinear Models - Complex relationships between variables

2.3 Analyzing and Solving Models

Description: Once a model is developed, it must be solved using appropriate analytical or computational techniques to find optimal or near-optimal solutions.

Solution Methods:

Analytical Solution - Using mathematical formulas for simple problems
Numerical Methods - Iterative computational approaches
Simulation - Testing scenarios through computational experiments
Heuristic Methods - Practical algorithms for complex problems

Software Tools: Excel Solver, CPLEX, GUROBI, Lingo, Python optimization libraries

2.4 Interpretation and Implementation of Results

Description: Model results must be interpreted in the context of the original business problem and translated into actionable recommendations.

Key Steps:

Validate solution feasibility and reasonableness
Perform sensitivity analysis to understand solution stability
Communicate results to stakeholders
Implement recommended actions with monitoring
Update model as new data becomes available

Unit 1: Questions & Answers

Q1: What is the primary goal of Operations Research?

Answer: The primary goal of Operations Research is to provide decision-makers with optimal or near-optimal solutions to complex business problems by using quantitative analysis and mathematical modeling.

Q2: What are the main components of a mathematical model?

Answer: The main components include: (1) Decision Variables - quantities to be determined, (2) Objective Function - what to maximize/minimize, (3) Constraints - limitations on decisions, and (4) Parameters - known data values.

Q3: Why is sensitivity analysis important?

Answer: Sensitivity analysis shows how solution changes when model parameters vary, helping decision-makers understand solution robustness and identify critical assumptions.

Unit 2: Data Management and Analysis (6 Hours)

Overview: This unit covers the complete lifecycle of data handling - from storage and retrieval to visualization and statistical analysis, with emphasis on regression analysis techniques.

3.1 Applications of Data Management and Analysis

Description: Data management and analysis is critical for extracting business insights from organizational data. Applications include customer analysis, sales forecasting, quality control, and performance measurement.

Business Applications:

Customer Segmentation - Identifying profitable customer groups
Sales Analysis - Understanding sales patterns and trends
Financial Analysis - Analyzing revenues, costs, and profitability
Quality Control - Monitoring product and service quality
Performance Metrics - Measuring KPIs and organizational health

3.2 Data Storage and Retrieval

Description: Effective data storage and retrieval systems are essential for managing organizational data efficiently and ensuring data accessibility and security.

Storage Technologies:

Relational Databases - SQL Server, Oracle, MySQL for structured data
Data Warehouses - Centralized repositories for analytical data
Cloud Storage - AWS S3, Azure Blob Storage for scalability
NoSQL Databases - MongoDB, Cassandra for unstructured data

Retrieval Methods:

SQL Queries for database operations
APIs for data access
ETL Processes - Extract, Transform, Load pipelines
Data Indexing for faster queries

Picture Reference: Database Architecture Diagram - Data Flow in Enterprise Systems

3.3 Data Visualization Techniques

Description: Data visualization converts raw data into visual formats that make patterns, trends, and insights immediately apparent to decision-makers.

Visualization Types:

Bar Charts - Comparing categories or time periods
Line Charts - Showing trends over time
Scatter Plots - Displaying relationships between variables
Heat Maps - Representing data density and intensity
Pivot Tables - Summarizing and cross-tabulating data
Dashboards - Integrated views of multiple metrics

Tools: Excel, Tableau, Power BI, Google Analytics, Looker

Video Reference: Data Visualization Best Practices - https://youtu.be/data-viz-tutorial

3.4 Exploratory Data Analysis (EDA)

Description: EDA involves summarizing and visualizing data to discover patterns, anomalies, relationships, and key characteristics before formal statistical analysis.

EDA Techniques:

Descriptive Statistics - Mean, median, standard deviation, range
Data Profiling - Understanding data quality and completeness
Correlation Analysis - Finding relationships between variables
Outlier Detection - Identifying unusual or extreme values
Distribution Analysis - Understanding data spread and shape

3.5 Regression Analysis

Description: Regression analysis is a statistical technique for modeling the relationship between a dependent variable (outcome) and one or more independent variables (predictors).

Types of Regression:

Simple Linear Regression: One independent variable, linear relationship: Y = a + bX
Multiple Linear Regression: Multiple independent variables: Y = a + b₁X₁ + b₂X₂ + ... + bₙXₙ
Nonlinear Regression: Curved or exponential relationships
Logistic Regression: Predicting binary outcomes (0/1, yes/no)

Key Metrics:

R-squared (R²) - Proportion of variance explained (0 to 1)
Correlation Coefficient (r) - Strength of linear relationship
P-value - Statistical significance of coefficients
Standard Error - Precision of estimates

Applications:

Sales forecasting based on marketing spend
Price elasticity analysis
Risk assessment in finance
Production yield prediction

Unit 2: Questions & Answers

Q1: What is the difference between data storage and data analysis?

Answer: Data storage focuses on organizing and maintaining data in databases or systems for accessibility and security. Data analysis involves extracting insights, patterns, and relationships from stored data to support decision-making.

Q2: Explain the purpose of regression analysis with an example.

Answer: Regression analysis models the relationship between variables. Example: A retailer might use regression to model store sales (Y) based on store size (X₁), location (X₂), and customer density (X₃). The resulting model helps forecast sales for new locations.

Q3: What does R-squared value represent?

Answer: R-squared measures how well the regression model explains the variation in the dependent variable. An R² of 0.85 means 85% of the variation in the outcome is explained by the independent variables.

Unit 3: Forecasting Methods (6 Hours)

Overview: Forecasting uses historical data to predict future values. This unit covers time-series forecasting methods for data with trend and seasonal components.

4.1 Time-Series with Trend Components

Description: A trend is a long-term upward or downward movement in data. Trend models are appropriate when historical data shows consistent growth or decline patterns.

Trend Models:

Linear Trend: Yₜ = a + bt (simplest form, constant rate of change)
Quadratic Trend: Yₜ = a + bt + ct² (accelerating growth/decline)
Exponential Trend: Yₜ = ae^(bt) (percentage change is constant)

When to Use: Revenue growth, population changes, technology adoption, market expansion

Picture Reference: Comparison of Linear vs Exponential Trend Curves

4.2 Time-Series with Seasonal Components

Description: Seasonal patterns repeat at fixed intervals (daily, weekly, monthly, yearly). Seasonal forecasting captures these recurring patterns.

Seasonal Patterns:

Daily - Call center volume, traffic patterns
Weekly - Retail sales (weekend vs weekday)
Monthly - Utility consumption patterns
Yearly - Holiday shopping, summer vacations, tax season

Seasonal Decomposition: Breaking data into trend, seasonal, and random components

Seasonal Index Method: Calculating seasonal factors and deseasonalizing data

Applications: Retail inventory planning, restaurant staffing, hotel occupancy, agricultural production

4.3 Combined Trend and Seasonal Models

Description: Real-world data often exhibits both trend and seasonal patterns simultaneously. Advanced forecasting methods handle both components together.

Methods:

Classical Decomposition: Separating time-series into trend, seasonal, and residual components
Moving Averages: Simple, weighted, or centered moving averages to smooth data
Exponential Smoothing: Giving more weight to recent observations
ARIMA Models: AutoRegressive Integrated Moving Average for complex patterns
Holt-Winters Method: Specifically designed for trend and seasonal components

Video Reference: Time Series Decomposition and ARIMA Forecasting - https://youtu.be/arima-forecasting

4.4 Selecting the Best Forecasting Method

Description: Different forecasting methods perform better under different circumstances. Model selection depends on data characteristics, forecast horizon, and required accuracy.

Selection Criteria:

Data Availability - Historical data length needed
Pattern Recognition - Identifying trend, seasonality, cyclicity
Forecast Horizon - Short-term vs long-term requirements
Accuracy Requirements - Acceptable error margins
Computational Complexity - Available computational resources

Accuracy Metrics:

MAE - Mean Absolute Error (average absolute deviation)
RMSE - Root Mean Square Error (penalizes large errors)
MAPE - Mean Absolute Percentage Error (percentage-based)
Theil's U - Compares model to naive forecast

4.5 CB Predictor in Excel

Description: CB Predictor is an Excel add-in that automates time-series forecasting and model selection, making advanced forecasting accessible to business users.

Features:

Automatic data analysis and pattern recognition
Testing multiple forecasting methods automatically
Generating forecast confidence intervals
Creating forecast charts and reports
Integration with Excel for easy data import/export

Workflow: Enter historical data → Select forecasting parameters → CB Predictor tests methods → View results and select best model → Generate forecasts

Unit 3: Questions & Answers

Q1: What is the difference between trend and seasonal patterns?

Answer: Trend is a long-term consistent upward or downward movement (e.g., increasing sales year over year). Seasonality is a repeating pattern at fixed intervals (e.g., higher sales in December each year).

Q2: When would you use exponential smoothing instead of moving averages?

Answer: Exponential smoothing is preferred when recent data points are more important than older ones. It adapts better to recent changes and requires less historical data than moving averages. Use moving averages for more stable, established patterns.

Q3: What does MAPE of 8% indicate about forecast accuracy?

Answer: MAPE of 8% means forecasts deviate from actual values by an average of 8%. Generally, MAPE less than 10% is considered good accuracy, 10-20% is acceptable, and above 20% indicates poor accuracy.

Unit 4: Introduction to Optimization (9 Hours)

Overview: Optimization involves finding the best solution from all feasible alternatives. This unit covers linear programming and multi-objective optimization techniques used in business decision-making.

5.1 Linear and Multi-objective Optimization Models

Description: Linear programming (LP) finds the optimal value of a linear objective function subject to linear constraints. Multi-objective optimization handles problems with multiple, often conflicting goals.

Linear Programming Components:

Decision Variables (Xⱼ) - What we want to determine
Objective Function - Linear function to maximize/minimize
Constraints - Linear inequalities limiting solutions
Non-negativity - Variables must be ≥ 0

Standard LP Form:

Maximize/Minimize: Z = c₁X₁ + c₂X₂ + ... + cₙXₙ
Subject to: a₁₁X₁ + a₁₂X₂ + ... ≤ b₁
Xⱼ ≥ 0 for all j

Multi-objective Optimization:

Maximize profit AND minimize risk
Minimize cost AND maintain quality
Pareto frontier - set of non-dominated solutions

5.2 Modeling Optimization Problems in Excel

Description: Excel provides an accessible platform for building and solving optimization models without programming, making OR accessible to business users.

Excel Setup:

Organize data in tables with clear labels
Define decision variable cells (input values)
Create objective function formula cell
Write constraint formulas in separate cells
Set up Solver parameters referencing these cells

Best Practices:

Use named ranges for clarity
Document assumptions and model logic
Include sensitivity analysis worksheets
Color-code input vs formula cells

Picture Reference: Excel Optimization Model Layout - Input Cells, Constraints, Objective Function Structure

5.3 Building Linear Programming Models

Description: Building LP models requires translating business problems into mathematical form while identifying all relevant constraints.

Model Building Steps:

Problem Definition: What business decision needs optimization?
Variable Definition: What quantities are we deciding?
Objective Clarification: What do we want to maximize/minimize?
Constraint Identification: What limitations exist?
Assumption Statement: What assumptions simplify the problem?

Example: Product Mix Problem

A factory makes chairs (X₁) and tables (X₂):
Maximize: Profit = 50X₁ + 100X₂
Subject to: Wood constraint: 2X₁ + 5X₂ ≤ 1000
Labor constraint: 4X₁ + 6X₂ ≤ 1200
X₁, X₂ ≥ 0

5.4 Solving Linear Programming Models

Description: LP models are solved using algorithmic methods to find the optimal solution at the corner point of the feasible region.

Solution Methods:

Graphical Method: Plotting constraints and finding optimal corner (for 2 variables)
Simplex Method: Iterative algorithm moving toward optimality
Interior Point Methods: Modern algorithms for large-scale problems
Excel Solver: Built-in optimization engine (GRG Nonlinear, Simplex LP)

Solution Types:

Unique optimal solution - Single best answer
Multiple optimal solutions - Multiple solutions with same objective value
Unbounded solution - Objective can be infinitely improved
Infeasible solution - No solution satisfies all constraints

5.5 Interpreting Solver Results and Sensitivity Analysis

Description: After solving, Excel Solver provides detailed reports on the solution and how sensitive it is to parameter changes.

Solver Reports:

Answer Report: Shows optimal values, constraints status (binding vs non-binding)
Sensitivity Report: Shows reduced costs, shadow prices, allowed ranges
Limits Report: Shows lower/upper limits for decision variables

Sensitivity Analysis:

Reduced Cost: How much objective value changes if variable forced to 1 unit
Shadow Price: Value of one additional unit of constrained resource
Allowable Range: Range where objective coefficients/constraints can change without affecting solution

Decision Support: Shadow prices identify profitable bottlenecks to address

5.6 Multi-objective Optimization

Description: Many business problems have multiple competing objectives. Multi-objective optimization finds compromises between them.

Approaches:

Weighted Sum Method: Combine objectives: Z = w₁f₁ + w₂f₂ + ... (weights sum to 1)
Lexicographic Method: Prioritize objectives by importance
Goal Programming: Achieve target levels for multiple objectives
Pareto Frontier: Generate non-dominated solutions for decision-maker choice

Example: Portfolio selection balancing return (maximize) and risk (minimize)

5.7 Premium Solver and Advanced Features

Description: Premium Solver extends Excel's built-in solver with more powerful algorithms and larger problem-solving capabilities.

Features:

Faster solution of large-scale problems
Support for integer and nonlinear programming
Multiple solution methods (LP, SOCP, SDP)
Parallelization for multi-core processors

Unit 4: Questions & Answers

Q1: What are the key differences between linear and nonlinear programming?

Answer: Linear programming has linear objective function and constraints, making it simpler and more scalable. Nonlinear programming involves curved relationships, is more complex to solve, but models realistic business problems better. Linear is preferred when accurate approximation is possible.

Q2: What is the significance of shadow price in sensitivity analysis?

Answer: Shadow price tells the decision-maker the value of obtaining one more unit of a scarce resource. For example, if a constraint has shadow price of 50, acquiring one additional unit of that resource would increase profit by 50 units, guiding resource allocation decisions.

Q3: How would you approach a product mix problem with multiple objectives?

Answer: Use weighted sum method: assign weights based on objective importance (e.g., 70% profit, 30% market share), create combined objective function, solve using Solver, generate Pareto frontier showing profit-market share trade-offs, and present options to management for strategic choice.

Unit 5: Decision and Risk Analysis (6 Hours)

Overview: This unit covers systematic frameworks for making decisions under uncertainty and managing risk effectively in business contexts.

6.1 Applications of Decision and Risk Analysis

Description: Decision and risk analysis applies structured frameworks to complex decisions where outcomes are uncertain. It helps quantify risks and make better decisions under uncertainty.

Business Applications:

Capital Investment Decisions - Evaluating projects with uncertain returns
Product Launch Strategy - Assessing market reception and demand
Insurance and Claims - Pricing policies based on risk assessment
Supply Chain Decisions - Managing supplier risk and disruption
Technology Adoption - Evaluating emerging technologies with uncertain outcomes
Merger and Acquisition - Valuing targets with uncertain synergies

Benefits: Reduced downside risk, improved decision quality, better resource allocation

6.2 Structuring Decision Problems

Description: Proper problem structuring is critical for sound decision analysis. It involves identifying alternatives, outcomes, and stakeholder preferences.

Decision Tree Structure:

Decision Nodes (□): Points where decision-maker chooses actions
Chance Nodes (○): Points where uncertain events occur
Outcomes: Final payoffs or consequences of decision + chance combination

Problem Structuring Steps:

Define decision to be made (Do it? When? How much?)
Identify feasible alternatives (mutually exclusive options)
Specify uncertain events and their probabilities
Quantify outcomes (payoffs, costs, benefits)
Build decision tree or payoff table

Picture Reference: Decision Tree Diagram Example - Investment Decision with Uncertain Market Scenarios

6.3 Understanding Risk in Decision-Making

Description: Risk arises from uncertainty about future events. Understanding and quantifying risk helps decision-makers balance potential gains against potential losses.

Types of Risk:

Market Risk: Uncertainty about customer demand, competition
Operational Risk: Uncertainty about costs, quality, processes
Financial Risk: Uncertainty about interest rates, exchange rates, credit
Strategic Risk: Uncertainty about long-term business direction

Risk Metrics:

Standard Deviation - Measure of outcome variability
Variance - Squared standard deviation
Coefficient of Variation - Risk relative to mean return
Value at Risk (VaR) - Maximum potential loss at confidence level

Risk Tolerance: Different decision-makers and organizations have different risk appetites

6.4 Expected Value Decision-Making

Description: Expected value provides a systematic way to compare alternatives by calculating weighted average of possible outcomes, weighted by probabilities.

Expected Value Formula:

EV = Σ(Probability × Outcome)
EV(Alternative) = P₁×O₁ + P₂×O₂ + ... + Pₙ×Oₙ

Decision Rule: Choose alternative with highest expected value (if maximizing) or lowest (if minimizing costs)

Example: Investment Decision

Project A: 60% chance $100K profit, 40% chance $20K loss
EV(A) = 0.6(100) + 0.4(-20) = 60 - 8 = 52
Project B: 50% chance $80K profit, 50% chance 0
EV(B) = 0.5(80) + 0.5(0) = 40
Decision: Choose Project A (higher EV)

Limitations: EV ignores risk; doesn't account for decision-maker risk aversion

6.5 Optimal Expected Value Decision Strategies

Description: Advanced decision strategies incorporate risk preferences, additional information, and optimal sequencing of decisions.

Decision Strategies:

Expected Utility: Incorporating risk preference through utility functions
Value of Information: How much is additional information worth?
Conditional Strategies: Different actions depending on uncertain event outcomes
Sequential Decisions: Making decisions stage-by-stage as information arrives

Expected Value of Perfect Information (EVPI): Maximum willingness-to-pay for information eliminating all uncertainty

Expected Value of Sample Information (EVSI): Value of imperfect but realistic information (market research, testing)

Video Reference: Decision Analysis with Excel - Building and Solving Decision Trees - https://youtu.be/decision-analysis-excel

Unit 5: Questions & Answers

Q1: What is the difference between decision analysis and risk analysis?

Answer: Decision analysis is the broader framework for choosing among alternatives under uncertainty. Risk analysis is a component focusing on quantifying, measuring, and managing the uncertainty aspect of the decision.

Q2: When would you use expected value of perfect information?

Answer: Use EVPI to determine if it's worthwhile to conduct research, testing, or gather more information before making a decision. If EVPI exceeds the cost of information gathering, pursuing additional information is economically justified.

Q3: How does risk aversion affect decision choices?

Answer: Risk-averse decision-makers prefer certain outcomes over gambles with same expected value. They discount high-risk options and favor safer alternatives even if expected values are lower. Expected utility theory incorporates risk preferences into decision-making.

Unit 6: Monte Carlo Simulation (9 Hours)

Overview: Monte Carlo simulation uses random sampling to model complex systems and understand probability distributions of outcomes under uncertainty.

7.1 Applications of Monte Carlo Simulation

Description: Monte Carlo simulation numerically evaluates complex models by generating random scenarios and analyzing resulting outcomes. Valuable when analytical solutions are difficult or impossible.

Key Applications:

Financial Risk Analysis - Portfolio value at risk, option pricing
Project Management - Project duration and cost uncertainty
Operations - Inventory levels, queue lengths, service times
Manufacturing - Yield analysis, quality control
New Product Development - Market success probability
Environmental - Pollution exposure, climate scenarios

Advantages: Handles complex relationships, incorporates multiple uncertainties, provides probability distributions of outcomes

7.2 Building Monte Carlo Simulation Models

Description: Building simulation models requires specifying input distributions, model logic, and output metrics to analyze.

Model Building Steps:

Input Variables: Identify uncertain parameters (costs, demand, prices)
Probability Distributions: Specify distribution for each input variable
Model Logic: Write formulas calculating outputs from inputs
Simulation Runs: Generate thousands of scenarios randomly
Output Analysis: Analyze distribution of outcomes

Simulation Workflow:

Define uncertain input variables and their distributions
Build spreadsheet model with formulas
Designate output cells to analyze
Run 10,000+ iterations with random inputs
Analyze output distribution (mean, std dev, percentiles)
Perform sensitivity analysis on inputs

Picture Reference: Simulation Model Structure - Input Variables, Process Logic, Output Metrics

7.3 Probability Distributions in Simulation

Description: Choosing appropriate probability distributions for input variables is critical for simulation accuracy. Different distributions model different real-world uncertainty patterns.

Common Distributions:

Normal Distribution: Bell curve, many natural phenomena (heights, errors). Parameters: mean, std dev
Uniform Distribution: Equal probability across range (a to b). When minimum and maximum known equally likely
Triangular Distribution: Three parameters (min, most likely, max). Useful when limited data available
Exponential Distribution: Decay pattern (service times, inter-arrival times). One parameter: mean
Lognormal Distribution: Right-skewed, positive values (project costs, stock prices)
Discrete Distribution: Specific values with probabilities (demand scenarios, outcomes)

Distribution Selection Guidance:

Use historical data analysis to fit distributions
Use expert judgment when data unavailable
Test sensitivity to distribution choice
Prefer distributions with interpretable parameters

7.4 Building Simulation Models with CRYSTAL BALL

Description: CRYSTAL BALL is an Excel add-in providing comprehensive simulation capabilities including distribution specification, Monte Carlo running, and output analysis.

CRYSTAL BALL Workflow:

Step 1: Define uncertain variables (Assumptions) by selecting distribution type
Step 2: Define output cells (Forecasts) to analyze
Step 3: Run simulation (iterates model thousands of times)
Step 4: Analyze forecast distributions, statistics, percentiles
Step 5: Sensitivity analysis showing variable impact on output

Key Features:

Distribution gallery with visualization
Automatic distribution fitting to data
Forecast charts and statistics
Sensitivity tornado charts
Correlation handling between variables
OptQuest optimization integration

Video Reference: CRYSTAL BALL Tutorial - Building and Running Simulations - https://youtu.be/crystal-ball-tutorial

7.5 Sensitivity Analysis in Simulation

Description: Sensitivity analysis identifies which input variables have greatest impact on outcomes, guiding focus on most important uncertainties.

Sensitivity Analysis Types:

Tornado Chart: Horizontal bars showing variable impact on output (largest to smallest)
Scatter Plot: Showing relationship between specific input and output
Correlation Analysis: Correlation coefficient between inputs and outputs
What-If Analysis: Testing specific input value scenarios

Insights from Sensitivity: Focus risk management on high-impact variables; reduce uncertainty in key drivers; allocate resources to most important factors

7.6 Optimization and Simulation - OptQuest

Description: OptQuest combines simulation with optimization to find decision variable values that optimize expected outcomes under uncertainty.

Optimization + Simulation:

Simulation models outcomes given decision variables and uncertainties
Optimization searches for best decision variable values
OptQuest iteratively tests combinations seeking optimal configuration

Applications:

Inventory optimization balancing stockouts and holding costs
Workforce scheduling optimizing service level and labor cost
Portfolio optimization maximizing return given risk constraints

Example: Determine optimal safety stock level by simulating demand uncertainty and optimizing total cost (shortage + holding)

Unit 6: Questions & Answers

Q1: Why is Monte Carlo simulation useful when analytical solutions exist?

Answer: Monte Carlo is useful because it handles complex, realistic models that analytical methods cannot solve. It provides probability distributions of outcomes rather than single point estimates, offers sensitivity analysis, and can model non-linear relationships, dependencies, and complex constraints that are intractable analytically.

Q2: How do you choose a probability distribution for an input variable?

Answer: Consider: (1) Historical data - fit distributions to observed data; (2) Type of variable - positive only (lognormal), bounded range (triangular/uniform); (3) Expert judgment when data unavailable; (4) Physical constraints; (5) Test sensitivity to distribution choice. Triangular is good default when only min/likely/max estimates available.

Q3: What does a tornado chart reveal in sensitivity analysis?

Answer: A tornado chart ranks variables by their impact on output. Longer bars indicate greater influence on results. It shows which variables are most critical to monitor and manage, guiding focus on uncertainty reduction efforts.

Unit 7: Systems Modeling and Simulations (6 Hours)

Overview: This unit covers dynamic system modeling, queueing systems, and inventory models that evolve over time, reflecting real-world business complexities.

8.1 Application of Dynamic System Models

Description: Dynamic systems change over time in response to inputs, feedbacks, and internal state. Dynamic models capture these temporal behaviors essential to operations management.

Dynamic System Applications:

Supply Chain - Bullwhip effect, inventory cascading through network
Production - Building/depleting inventory in response to demand
Customer Service - Queue dynamics, wait times
Healthcare - Patient flow, bed occupancy over time
Resources - Population dynamics, resource depletion

Key Characteristics: State variables that evolve; feedbacks (positive or negative); time delays; accumulation effects

Picture Reference: Supply Chain Bullwhip Effect Diagram - Demand Propagation Through Multiple Stages

8.2 Queueing Systems and Analysis

Description: Queueing systems analyze situations where customers or items arrive, wait in queue, and receive service. Critical for service operations management.

Queueing System Elements:

Arrival Process: How customers arrive (Poisson, deterministic)
Service Process: How long service takes (Exponential, constant)
Number of Servers: Single server, multiple parallel servers
Queue Discipline: FIFO (first-in-first-out), priority, other
System Capacity: Unlimited or limited queue length

Queueing Notation (M/M/1 Example):

First M - Poisson arrivals
Second M - Exponential service time
1 - Single server

Performance Metrics:

Average queue length (Lq)
Average time in system (W)
Server utilization (ρ)
Probability of having k customers (Pn)

Applications: Call centers, hospital emergency rooms, bank tellers, checkout lanes, technical support

Video Reference: Queueing Theory and M/M/1 Systems - https://youtu.be/queueing-theory

8.3 Modeling and Simulating Dynamic Inventory Models

Description: Inventory models determine optimal order quantities and reorder points balancing holding costs, ordering costs, and shortage costs as demand and supply vary over time.

Inventory Decision Variables:

Order Quantity (Q): How much to order when reordering
Reorder Point (R): Inventory level triggering new order
Lead Time: Delay between order and receipt
Safety Stock: Buffer against demand/supply uncertainty

Key Inventory Costs:

Holding Cost: Per-unit cost of carrying inventory (warehousing, insurance)
Ordering Cost: Fixed cost per order (processing, shipping)
Shortage Cost: Cost of stockouts (lost sales, customer goodwill)

Economic Order Quantity (EOQ):

Q* = √(2×D×S/H)
Where D=annual demand, S=ordering cost, H=holding cost

Dynamic Simulation: When demand is uncertain or time-varying, simulation evaluates performance of different order policies over time

Example Simulation Output: Average inventory, stockout frequency, total cost for (Q, R) policy under uncertain demand

Unit 7: Questions & Answers

Q1: What is the bullwhip effect and why does it occur in supply chains?

Answer: The bullwhip effect occurs when small demand fluctuations at the retail level cause progressively larger fluctuations at upstream supply chain levels (wholesalers, manufacturers). Causes include: demand forecast updates, batch ordering, price fluctuations, and information delays. Effects: inefficient inventory, increased costs, poor service.

Q2: In an M/M/1 queue, what does server utilization ρ = 0.8 mean?

Answer: ρ = 0.8 means the server is busy 80% of time. Queue lengths and wait times increase dramatically as ρ approaches 1. At ρ = 0.9, system congestion becomes severe. At ρ ≥ 1, queue grows indefinitely. Queuing management focuses on keeping ρ at reasonable levels (0.7-0.8).

Q3: How would you determine optimal safety stock level?

Answer: Use simulation: model demand uncertainty, simulate inventory dynamics for different safety stock levels, calculate total cost (holding + shortage), find level minimizing total cost. Also consider service level targets - maintain probability of meeting demand without stockout. Simulation allows realistic analysis of lead-time demand variability.

Semester-End Examination Questions

Instructions: Answer any 4 of the following 5 questions. Each question carries equal weightage. Provide detailed explanations with examples where applicable.

Question 1: Comprehensive Problem - Linear Programming and Optimization

A manufacturing company produces two products: P1 and P2. The company has the following constraints:

Raw material availability: 800 kg per week
Labor hours available: 400 hours per week
Machine time available: 200 hours per week
P1 requires: 2 kg material, 1 hour labor, 0.5 hour machine time, and yields profit of Rs. 100
P2 requires: 1 kg material, 2 hours labor, 1 hour machine time, and yields profit of Rs. 120

Required:

a) Formulate a linear programming model to maximize profit
b) Solve the model and identify optimal production quantities
c) Perform sensitivity analysis - what is the shadow price of machine time?
d) If machine time increased to 250 hours, how would optimal solution change?

Question 2: Forecasting and Time Series Analysis

A retail store has collected monthly sales data (in Rs. 1000s) for 24 months. Visual analysis shows both upward trend and seasonal pattern (peaks in months 3, 6, 9, 12). Current methods used are simple moving average and exponential smoothing.

Required:

a) Explain why simple methods may be inadequate and suggest appropriate forecasting methods
b) Discuss decomposition approach - how would you separate trend, seasonal, and random components?
c) Describe how CRYSTAL BALL or CB Predictor could automate this process
d) What accuracy metrics would you use to select the best forecasting method?

Question 3: Decision Analysis Under Uncertainty

A startup company must decide between launching Product A or B. Market research indicates:

Product A: 60% chance of high demand (profit Rs. 5 lakhs), 40% chance of low demand (loss Rs. 1 lakh)
Product B: 50% chance of high demand (profit Rs. 6 lakhs), 50% chance of low demand (loss Rs. 0.5 lakh)
Market research costing Rs. 50,000 can reduce uncertainty, changing probabilities to 80/20 for correct scenario

Required:

a) Calculate expected value for each product and recommend choice
b) Draw decision tree for problem without additional research
c) Calculate expected value of perfect information (EVPI)
d) Is the market research investment justified? Explain your reasoning

Question 4: Monte Carlo Simulation and Risk Analysis

A construction project has three cost components with uncertain values:

Labor cost: Triangular distribution (Min: Rs. 50 lakh, Likely: Rs. 70 lakh, Max: Rs. 95 lakh)
Material cost: Normal distribution (Mean: Rs. 40 lakh, Std Dev: Rs. 8 lakh)
Equipment cost: Uniform distribution (Min: Rs. 30 lakh, Max: Rs. 50 lakh)

Required:

a) Design a Monte Carlo simulation model for total project cost
b) What information would you gather in 10,000 simulation runs?
c) How would sensitivity analysis identify which cost component is most critical to control?
d) If management requires 90% probability of staying within budget, how would you use simulation results to set realistic budget?

Question 5: Integrated Case Study - Systems Modeling and Optimization

A hospital emergency department faces challenges with patient flow. Peak arrival rate is 60 patients/hour with exponential inter-arrival times. Average service time is 15 minutes (exponential), with capacity to add additional doctors (servers).

Required:

a) Analyze current M/M/3 system (3 doctors) - calculate utilization, average wait time, average queue length
b) What system performance problems would occur if demand increases to 70 patients/hour?
c) Using queueing theory and optimization, determine optimal number of doctors to maintain 95% service level
d) Discuss tradeoffs between service level, cost, and patient satisfaction - how would you recommend balancing these objectives?

Semester Examination Questions - Set A

Instructions: Answer any 4 of the following 5 questions. Each question carries equal weightage. Provide detailed explanations with examples where applicable.

Question 1: Data Analysis, Regression, and Forecasting Integration

A e-commerce company has collected data on advertising spend and website sales for 36 months. Monthly advertising spend (X) ranges from Rs. 5 lakhs to Rs. 25 lakhs, and corresponding sales (Y) range from Rs. 20 lakhs to Rs. 120 lakhs. Analysis shows linear trend with some seasonal variation (peaks in months 4, 8, 12).

Required:

a) Perform exploratory data analysis (EDA) - describe what statistical analysis you would conduct and what insights you expect
b) Build a multiple regression model incorporating advertising spend (X₁), trend (X₂), and seasonal component (X₃). Write the regression equation
c) Interpret regression results - what do coefficients represent? How would you assess model quality (R², p-values)?
d) Use the regression model for forecasting - predict sales for next month given advertising budget of Rs. 20 lakhs, and explain confidence interval interpretation
e) Compare regression forecasting with time-series methods (moving average, exponential smoothing) - which approach is preferable and why?

Question 2: Multi-objective Optimization and Decision Making

A manufacturing company must determine product mix between three products (A, B, C) while balancing competing objectives: maximizing profit, maximizing market share, and minimizing environmental impact.

Product A: Profit Rs. 80/unit, Market share points 2/unit, Pollution index 3/unit
Product B: Profit Rs. 120/unit, Market share points 1/unit, Pollution index 1/unit
Product C: Profit Rs. 50/unit, Market share points 4/unit, Pollution index 5/unit
Constraints: Maximum 500 units total capacity, minimum 1000 market share points, maximum pollution index of 2000

Required:

a) Formulate the multi-objective optimization problem with all three objectives and constraints
b) Using weighted sum method, solve for three scenarios: (i) 50% profit, 30% market share, 20% environment; (ii) 33% each; (iii) 20% profit, 20% market share, 60% environment
c) Generate Pareto frontier showing trade-offs between profit and environmental objectives
d) Recommend optimal product mix based on management priorities and explain sensitivity to weight changes

Question 3: Monte Carlo Simulation for Project Management

A software development project has 8 major tasks with uncertain durations. Management requires 95% confidence that project completes within 180 days. Current estimates:

Requirements: Triangular(20, 25, 35) days
Design: Triangular(25, 30, 40) days
Development: Triangular(60, 75, 95) days
Testing: Triangular(30, 40, 55) days
Integration: Triangular(15, 20, 30) days
Documentation: Triangular(10, 15, 25) days
UAT: Triangular(20, 25, 35) days
Deployment: Triangular(5, 7, 10) days

Tasks are sequential (each starts when previous ends).

Required:

a) Design Monte Carlo simulation model for total project duration (explain why simulation needed vs. simple summation)
b) What would 10,000 simulation runs reveal about project completion probability by day 180?
c) Perform sensitivity analysis - which task(s) have greatest impact on total duration? Which should management focus on for risk reduction?
d) If current 180-day deadline appears risky (say only 72% confidence), recommend contingency buffer and explain reasoning

Question 4: Dynamic Inventory Optimization and Supply Chain

A retail chain faces demand uncertainty for a fast-moving consumer good. Current inventory policy uses fixed order quantity Q=500 units and reorder point R=200 units. Annual demand is normally distributed: Mean 5000 units, Std Dev 800 units. Lead time is 2 weeks. Ordering cost is Rs. 250 per order, holding cost is Rs. 20 per unit per year, and stockout cost is estimated at Rs. 150 per unit.

Required:

a) Calculate Economic Order Quantity (EOQ) and compare with current policy Q=500. What is cost difference?
b) Determine optimal reorder point R considering demand during lead time uncertainty (use normal distribution)
c) Design simulation model to evaluate current policy vs. optimal policy - what performance metrics would you track?
d) Discuss bullwhip effect - if retailer increases order sizes during peak season, how would this propagate to supplier inventory needs? What mitigation strategies would you recommend?

Question 5: Strategic Risk Analysis and Expected Value Decision

A pharmaceutical company must decide whether to invest Rs. 10 crores in developing a new drug. Market scenarios:

Blockbuster Success: 10% probability, NPV Rs. 150 crores
Moderate Success: 30% probability, NPV Rs. 40 crores
Marginal Success: 35% probability, NPV Rs. 5 crores
Failure: 25% probability, NPV -Rs. 10 crores (development cost lost)

Company can conduct clinical trials (Rs. 2 crores) to refine probability estimates. With trials, probabilities change to: Blockbuster 15%, Moderate 40%, Marginal 30%, Failure 15%.

Required:

a) Calculate expected NPV for investment without trials. Should company invest?
b) Calculate expected NPV with clinical trials. Is trial investment justified? Show expected value of sample information (EVSI)
c) Draw complete decision tree showing: (i) Option to conduct trials or not; (ii) outcomes with/without trials; (iii) optimal decisions at each node
d) Perform sensitivity analysis - at what probability of blockbuster success does decision change? How does company risk tolerance affect choice?
e) Discuss limitations of expected value criterion - when might company choose lower expected value option and why?

Semester-End Examination Questions - SET B

Instructions: Answer any 4 of the following 5 questions. Each question carries equal weightage. Provide detailed explanations with examples where applicable.

Question 1: Comprehensive Problem - Linear Programming and Optimization

A manufacturing company produces two products: P1 and P2. The company has the following constraints:

Raw material availability: 800 kg per week
Labor hours available: 400 hours per week
Machine time available: 200 hours per week
P1 requires: 2 kg material, 1 hour labor, 0.5 hour machine time, and yields profit of Rs. 100
P2 requires: 1 kg material, 2 hours labor, 1 hour machine time, and yields profit of Rs. 120

Required:

a) Formulate a linear programming model to maximize profit
b) Solve the model and identify optimal production quantities
c) Perform sensitivity analysis - what is the shadow price of machine time?
d) If machine time increased to 250 hours, how would optimal solution change?

Question 2: Forecasting and Time Series Analysis

Required:

a) Explain why simple methods may be inadequate and suggest appropriate forecasting methods
b) Discuss decomposition approach - how would you separate trend, seasonal, and random components?
c) Describe how CRYSTAL BALL or CB Predictor could automate this process
d) What accuracy metrics would you use to select the best forecasting method?

Question 3: Decision Analysis Under Uncertainty

A startup company must decide between launching Product A or B. Market research indicates:

Product A: 60% chance of high demand (profit Rs. 5 lakhs), 40% chance of low demand (loss Rs. 1 lakh)
Product B: 50% chance of high demand (profit Rs. 6 lakhs), 50% chance of low demand (loss Rs. 0.5 lakh)
Market research costing Rs. 50,000 can reduce uncertainty, changing probabilities to 80/20 for correct scenario

Required:

a) Calculate expected value for each product and recommend choice
b) Draw decision tree for problem without additional research
c) Calculate expected value of perfect information (EVPI)
d) Is the market research investment justified? Explain your reasoning

Question 4: Monte Carlo Simulation and Risk Analysis

A construction project has three cost components with uncertain values:

Labor cost: Triangular distribution (Min: Rs. 50 lakh, Likely: Rs. 70 lakh, Max: Rs. 95 lakh)
Material cost: Normal distribution (Mean: Rs. 40 lakh, Std Dev: Rs. 8 lakh)
Equipment cost: Uniform distribution (Min: Rs. 30 lakh, Max: Rs. 50 lakh)

Required:

a) Design a Monte Carlo simulation model for total project cost
b) What information would you gather in 10,000 simulation runs?
c) How would sensitivity analysis identify which cost component is most critical to control?
d) If management requires 90% probability of staying within budget, how would you use simulation results to set realistic budget?

Question 5: Integrated Case Study - Systems Modeling and Optimization

Required:

a) Analyze current M/M/3 system (3 doctors) - calculate utilization, average wait time, average queue length
b) What system performance problems would occur if demand increases to 70 patients/hour?
c) Using queueing theory and optimization, determine optimal number of doctors to maintain 95% service level
d) Discuss tradeoffs between service level, cost, and patient satisfaction - how would you recommend balancing these objectives?

Semester-End Examination Questions - SET C

Instructions: Answer any 4 of the following 5 questions. Each question carries equal weightage. Provide detailed explanations with examples where applicable.

Question 1: Data Analysis, Regression, and Forecasting Integration

Required:

a) Perform exploratory data analysis (EDA) - describe what statistical analysis you would conduct and what insights you expect
b) Build a multiple regression model incorporating advertising spend (X₁), trend (X₂), and seasonal component (X₃). Write the regression equation
c) Interpret regression results - what do coefficients represent? How would you assess model quality (R², p-values)?
d) Use the regression model for forecasting - predict sales for next month given advertising budget of Rs. 20 lakhs, and explain confidence interval interpretation
e) Compare regression forecasting with time-series methods (moving average, exponential smoothing) - which approach is preferable and why?

Question 2: Multi-objective Optimization and Decision Making

Product A: Profit Rs. 80/unit, Market share points 2/unit, Pollution index 3/unit
Product B: Profit Rs. 120/unit, Market share points 1/unit, Pollution index 1/unit
Product C: Profit Rs. 50/unit, Market share points 4/unit, Pollution index 5/unit
Constraints: Maximum 500 units total capacity, minimum 1000 market share points, maximum pollution index of 2000

Required:

a) Formulate the multi-objective optimization problem with all three objectives and constraints
b) Using weighted sum method, solve for three scenarios: (i) 50% profit, 30% market share, 20% environment; (ii) 33% each; (iii) 20% profit, 20% market share, 60% environment
c) Generate Pareto frontier showing trade-offs between profit and environmental objectives
d) Recommend optimal product mix based on management priorities and explain sensitivity to weight changes

Question 3: Monte Carlo Simulation for Project Management

A software development project has 8 major tasks with uncertain durations. Management requires 95% confidence that project completes within 180 days. Current estimates:

Requirements: Triangular(20, 25, 35) days
Design: Triangular(25, 30, 40) days
Development: Triangular(60, 75, 95) days
Testing: Triangular(30, 40, 55) days
Integration: Triangular(15, 20, 30) days
Documentation: Triangular(10, 15, 25) days
UAT: Triangular(20, 25, 35) days
Deployment: Triangular(5, 7, 10) days

Tasks are sequential (each starts when previous ends).

Required:

a) Design Monte Carlo simulation model for total project duration (explain why simulation needed vs. simple summation)
b) What would 10,000 simulation runs reveal about project completion probability by day 180?
c) Perform sensitivity analysis - which task(s) have greatest impact on total duration? Which should management focus on for risk reduction?
d) If current 180-day deadline appears risky (say only 72% confidence), recommend contingency buffer and explain reasoning

Question 4: Dynamic Inventory Optimization and Supply Chain

Required:

a) Calculate Economic Order Quantity (EOQ) and compare with current policy Q=500. What is cost difference?
b) Determine optimal reorder point R considering demand during lead time uncertainty (use normal distribution)
c) Design simulation model to evaluate current policy vs. optimal policy - what performance metrics would you track?
d) Discuss bullwhip effect - if retailer increases order sizes during peak season, how would this propagate to supplier inventory needs? What mitigation strategies would you recommend?

Question 5: Strategic Risk Analysis and Expected Value Decision

A pharmaceutical company must decide whether to invest Rs. 10 crores in developing a new drug. Market scenarios:

Blockbuster Success: 10% probability, NPV Rs. 150 crores
Moderate Success: 30% probability, NPV Rs. 40 crores
Marginal Success: 35% probability, NPV Rs. 5 crores
Failure: 25% probability, NPV -Rs. 10 crores (development cost lost)

Company can conduct clinical trials (Rs. 2 crores) to refine probability estimates. With trials, probabilities change to: Blockbuster 15%, Moderate 40%, Marginal 30%, Failure 15%.

Required:

a) Calculate expected NPV for investment without trials. Should company invest?
b) Calculate expected NPV with clinical trials. Is trial investment justified? Show expected value of sample information (EVSI)
c) Draw complete decision tree showing: (i) Option to conduct trials or not; (ii) outcomes with/without trials; (iii) optimal decisions at each node
d) Perform sensitivity analysis - at what probability of blockbuster success does decision change? How does company risk tolerance affect choice?
e) Discuss limitations of expected value criterion - when might company choose lower expected value option and why?

Semester-End Examination Questions - SET D

Instructions: Answer any 4 of the following 5 questions. Each question carries equal weightage. Provide detailed explanations with examples where applicable.

Question 1: Operations Research Modeling - Manufacturing Optimization

A textile factory produces three fabric types: Cotton (C), Polyester (P), and Blend (B). Each requires different resources and yields different profit:

Cotton: Rs. 150/meter profit, requires 2 kg raw material, 0.5 hours labor, 0.3 hours machine time
Polyester: Rs. 180/meter profit, requires 1.5 kg raw material, 0.4 hours labor, 0.4 hours machine time
Blend: Rs. 160/meter profit, requires 1.8 kg raw material, 0.45 hours labor, 0.35 hours machine time
Available weekly: 1000 kg material, 500 labor hours, 350 machine hours
Minimum production: 200 meters Cotton, 150 meters Polyester, 100 meters Blend

Required:

a) Formulate the linear programming model with objective function and all constraints
b) Solve using Excel Solver and identify optimal production quantities
c) Generate sensitivity report - which resource has highest shadow price? What does this mean for management?
d) If polyester profit increases to Rs. 200/meter, how does optimal solution change? Explain the trade-offs
e) Recommend whether factory should invest in additional machine capacity given current constraints

Question 2: Advanced Forecasting - Seasonal Demand Planning

A beverage company has 48 months of quarterly sales data showing strong seasonality and slight upward trend:

Q1 sales average: Rs. 50 lakhs (lowest demand - winter)
Q2 sales average: Rs. 70 lakhs
Q3 sales average: Rs. 90 lakhs (peak - summer)
Q4 sales average: Rs. 75 lakhs
Average growth rate: 2% annually

Required:

a) Calculate seasonal indices for each quarter and explain their interpretation
b) Deseasonalize the data and identify trend - is linear, exponential, or quadratic trend appropriate?
c) Forecast sales for next 4 quarters (Q1-Q4 of next year) using multiplicative seasonal model
d) Compare accuracy: which method - Holt-Winters, ARIMA, or simple moving average - would be most appropriate? Why?
e) Discuss inventory and production planning implications based on seasonal forecasts

Question 3: Risk-Based Decision Making - Technology Investment

A company must choose between three technology platforms for digital transformation (5-year horizon):

Platform A (Low Risk): Cost Rs. 2 crores, 80% success probability yielding Rs. 12 crores, 20% failure (total loss)
Platform B (Medium Risk): Cost Rs. 1.5 crores, 60% success probability yielding Rs. 15 crores, 40% failure (total loss)
Platform C (High Risk): Cost Rs. 1 crore, 40% success probability yielding Rs. 18 crores, 60% failure (total loss)

Company can hire consultants (Rs. 20 lakhs) to improve assessment accuracy.

Required:

a) Calculate expected NPV for each platform and rank by expected value
b) Calculate standard deviation and coefficient of variation for each platform (risk measure)
c) If company is risk-averse, which platform would you recommend? If risk-neutral?
d) Build decision tree incorporating consultant information - estimate expected value of information
e) Discuss strategic considerations beyond expected value that should influence the decision

Question 4: Simulation-Based Supply Chain Analysis

A 3-tier supply chain has: Manufacturer → Distributor → Retailer. Retail customer demand is normally distributed with Mean 100 units/week, Std Dev 20 units. Each tier places orders based on observed demand from downstream, but with lead time delays.

Current Policy: Each tier orders to maintain 4 weeks of inventory.

Required:

a) Design Monte Carlo simulation model to demonstrate bullwhip effect in 3-tier supply chain (52-week horizon)
b) Calculate coefficient of variation in orders at each level - how much does variability amplify upstream?
c) Perform sensitivity analysis: how does bullwhip worsen if inventory targets increase from 4 to 6 weeks?
d) Test mitigation strategies: (i) Information sharing, (ii) Vendor-managed inventory, (iii) Order batching restrictions. Which is most effective?
e) Recommend supply chain policy changes to reduce bullwhip effect and inventory costs

Question 5: Queueing Systems - Service Design and Optimization

A bank has redesigned its service operations. Customers arrive according to Poisson process at mean rate 30/hour. Service time follows exponential distribution with mean 3 minutes. Currently, bank has 2 tellers (servers). Management wants to maintain average wait time ≤ 5 minutes.

Additional Data: Teller cost Rs. 400/hour, customer waiting time cost estimated at Rs. 100/hour per customer

Required:

a) Calculate system metrics for current M/M/2 configuration: utilization, average queue length, average wait time
b) Does current system meet the ≤ 5 minute wait time target? Show calculations
c) Determine optimal number of tellers that minimizes total cost (teller wages + customer waiting time cost)
d) Simulate queue dynamics with current 2-teller system - what percentage of customers experience wait time > 10 minutes?
e) Discuss service quality vs. cost trade-offs - how would you present options to management?

Course Summary

Key Learning Outcomes:

Upon completion of this course, students will be able to:

Develop mathematical models to represent real-world business problems
Perform data management, analysis, and visualization for business insights
Apply forecasting methods to predict future values with confidence intervals
Formulate and solve linear programming optimization problems
Conduct decision analysis under uncertainty and manage organizational risk
Build and interpret Monte Carlo simulation models
Model and optimize dynamic systems including queueing and inventory
Use analytical software tools (Excel Solver, CRYSTAL BALL) effectively

Integration of Topics:

Throughout this course, the seven units integrate seamlessly: data analysis (Unit 2) feeds forecasting (Unit 3), which informs optimization (Unit 4). Uncertainty quantified in decision analysis (Unit 5) becomes simulation input (Unit 6), supporting complex systems analysis (Unit 7). All units rely on the modeling frameworks introduced in Unit 1.

Practical Application:

These analytical techniques apply across functional areas - operations, finance, marketing, supply chain, and strategic planning. Organizations increasingly demand decision-makers with quantitative skills to optimize operations, reduce costs, manage risk, and gain competitive advantage. This course equips you with foundational knowledge and practical tools to address real business challenges systematically.

Continuing Learning:

This course provides foundation for advanced topics: advanced optimization (integer programming, nonlinear optimization), advanced simulation (system dynamics, discrete event simulation), machine learning, and data science. Continued practice with real datasets and business problems will deepen expertise in applying these analytical methods.

Software Tools Recommended for Further Learning:

Python libraries: NumPy, SciPy, Pandas for data analysis and optimization
R packages: Tidyverse for data management, Forecast for time series
Advanced optimization: GUROBI, CPLEX for large-scale problems
Simulation platforms: AnyLogic, SimPy for complex discrete event systems

Operational Research -Management Information Engineering

Contents

Introduction to Management Information Systems

Unit 1: Introduction to Modeling for Decisions (3 Hours)

Unit 1: Questions & Answers

Unit 2: Data Management and Analysis (6 Hours)

Unit 2: Questions & Answers

Unit 3: Forecasting Methods (6 Hours)

Unit 3: Questions & Answers

Unit 4: Introduction to Optimization (9 Hours)

Unit 4: Questions & Answers

Unit 5: Decision and Risk Analysis (6 Hours)

Unit 5: Questions & Answers

Unit 6: Monte Carlo Simulation (9 Hours)

Unit 6: Questions & Answers

Unit 7: Systems Modeling and Simulations (6 Hours)

Unit 7: Questions & Answers

Semester-End Examination Questions

Semester Examination Questions - Set A

Semester-End Examination Questions - SET B

Semester-End Examination Questions - SET C

Semester-End Examination Questions - SET D

Course Summary

Post a Comment

Contact form