1.

Based on the readings to support the course, as well as those references that you may
investigate, give answers to the following questions:

A. What are the fundamental characteristics of a simulation model?

A simulation model typically has the following fundamental characteristics:

Abstraction: It simplifies real-world systems by representing key elements while ignoring irrelevant
details.

Stochastic vs. Deterministic: Some models incorporate randomness (stochastic), while others
operate with fixed parameters (deterministic).

Discrete vs. Continuous: Models may track changes at specific events (discrete event simulation)
or continuously over time (continuous simulation).

Input and Output Variables: These define system parameters (inputs) and performance measures
(outputs).

Time Handling: The model must define how time progresses (event-driven, time-stepped, or
hybrid).

(Reference: Banks et al., Discrete-Event System Simulation, 2010)

B. What is the process of doing a simulation study?

Problem Definition – Identify objectives and scope.

Conceptual Model Development – Define relationships, entities, and constraints.

Data Collection & Model Formulation – Gather empirical data and formulate the mathematical
model.

Model Implementation – Code the model using a simulation tool.

Verification & Validation – Ensure the model behaves correctly and represents reality.

Experimentation & Analysis – Run simulations and analyze results.

Interpretation & Reporting – Present insights and recommendations.

(Reference: Law & Kelton, Simulation Modeling and Analysis, 2014)

C. What factors should be considered in selecting a simulation tool and why?

Factors include:

Modeling Capability: Ability to handle complexity and scalability.

User Interface & Ease of Use: Visual modeling and scripting options.
Integration & Interoperability: Compatibility with databases, CAD, ERP, etc.

Performance & Scalability: Ability to handle large-scale simulations e iciently.

Cost: Licensing, training, and maintenance costs.

(Reference: Fishman, Monte Carlo: Concepts, Algorithms, and Applications, 1996)

D. What are the future trends in simulation software development within the
manufacturing area?

AI and Machine Learning Integration – Predictive analytics and optimization.

Cloud-Based Simulation – Accessibility and scalability.

Digital Twins – Real-time synchronization with physical systems.

Hybrid Modeling – Combining discrete event, agent-based, and system dynamics models.

VR/AR Integration – Enhancing visualization and training.

(Reference: Negahban & Smith, Simulation for Manufacturing System Design, 2014)

E. In which areas is computer simulation applied and for what purposes?

Manufacturing – Process optimization, supply chain modeling.

Healthcare – Hospital resource planning, epidemic modeling.

Finance – Risk analysis, stock market forecasting.

Defense & Aerospace – Combat simulation, logistics planning.

Urban Planning – Tra ic flow, infrastructure development.

(Reference: Sargent, Verification and Validation of Simulation Models, 2013)

F. What considerations should be taken into account when contracting the services
of simulation consulting? What are the relevant costs?

Consultant Experience: Domain knowledge and technical expertise.

Cost Structure: Development, execution, and maintenance fees.

Data Requirements: Availability and accuracy of input data.

Validation Procedures: Ensuring model reliability.

Deliverables & Support: Reports, software, and training.

(Reference: Robinson, Simulation: The Practice of Model Development and Use, 2014)

11. How many types of simulation are there and which of all those types will be emphasized in
this course?

12. What other methods exist, besides the inverse transformation method, to generate
random observations?

Rejection Sampling (Acceptance-Rejection Method)

Box-Muller Method (for Normal Distribution)

Ziggurat Method (for E icient Normal Sampling)

(Reference: Devroye, Non-Uniform Random Variate Generation, 1986)

13. Why is it said that random numbers generated by digital methods are not really random?
Why are they still used in stochastic simulation despite this?

Deterministic Algorithms: They rely on pseudorandom number generators (PRNGs) like the
Mersenne Twister.

Periodicity: They repeat after a long sequence.

Statistical Properties: Designed to mimic true randomness.

Despite this, they are widely used in stochastic simulations because:

They are reproducible, aiding debugging and verification.

They can be optimized for e iciency.

(Reference: Knuth, The Art of Computer Programming, Vol. 2: Seminumerical Algorithms, 1997
14. In an inventory simulation problem of a probabilistic nature, under what conditions can it
be said that a policy is better than all the others? What conditions must that policy satisfy to
be optimal?

A policy is better if it minimizes long-run expected costs while meeting demand.

Conditions for optimality:

Minimization of Total Cost (Holding + Shortage Costs)

Service Level Constraints Met

Stability in Performance Across Demand Variations

(Reference: Silver et al., Inventory Management and Production Planning, 1998)

15. Why is the computer important in discrete event simulation? What advantage of this
technique is seriously reduced if the computer is not present?

Fast Processing of Events

Complex Model Handling

Large-Scale Simulation Feasibility

Without computers, real-time analysis and optimization become impractical.

(Reference: Law & Kelton, Simulation Modeling and Analysis, 2014)

16. How can discrete simulation facilitate decision making in the management of human
activity systems?

Scenario Testing – Evaluating di erent strategies.

Bottleneck Identification – Improving workflows.

Resource Allocation Optimization – Sta ing, scheduling.

(Reference: Pidd, Computer Simulation in Management Science, 2004)

17. What type of problems can discrete simulation solve in the field of human activity
systems?

Healthcare Sta ing & Scheduling

Public Transportation Planning

Warehouse and Logistics Optimization

(Reference: Banks et al., Discrete-Event System Simulation, 2010)

18. What techniques or models of Operations Research are integrated into Discrete Event
Simulation?

Queueing Theory – For modeling waiting lines.

Markov Chains – For probabilistic decision-making.

Optimization Algorithms – Like Genetic Algorithms or Linear Programming.

(Reference: Winston, Operations Research: Applications and Algorithms, 2003)

19. Why is the mean of means a better estimator of the population mean, compared to the
simple sample mean?

Less Variance: The variance is reduced when averaging across multiple independent sample
means.

Law of Large Numbers: More stable convergence to the population mean.

(Reference: Montgomery, Applied Statistics and Probability for Engineers, 2013)

20. In the inventory simulation example, when does the average expected cost per week
become stationary? What must be observed to identify that this variable has reached the
steady state?
Long-Run Stability: When variance and mean stabilize over successive runs.

Performance Indicators: Plot cost values over time to observe steady-state behavior.

(Reference: Banks et al., Discrete-Event System Simulation, 2010)