Description
Supply chain :
CASE 1
Describe the scenario you propose to analyze. Clarify your creativity in this example.
Develop a model in Arena that reflects your scenario. There are no limits on how you choose to model the system. The model must have at least one of each of the following components (you may have more of each component): o Arrivals
o Decision node or nodes
o Process
o Assign
o A schedule
o Optional1 components are not required, but are worth 0.5 points each A variable
A Record module
Explain metrics used to evaluate the system. These could be the default metrics from Arena, or new metrics. You should discuss at least 4 metrics.
You may (but are NOT required to) develop a creative metric. This can be a combination of current metrics from Arena, or something else you measure in the simulation.
Run the model for the base case, with at least 100 trials. Discuss key results from the Arena analysis. Make sure to highlight the metrics you proposed earlier.
Suggest a recommendation for improving performance of the system. This recommendation must involve a policy class related to a key parameter or input of the model. Explain the rationale for using this policy class. The purpose of this policy class is to identify the relationship between one aspect of the model and various performance metrics you suggested earlier. (As an example, the policy class might be to change the arrival rate. The levels of the policy are the different arrival rates. This is probably a bad example, because arrival rates are not under your control.)
Analyze this policy class using the Arena PAN tool. The policy class must have at least 10 levels for the policy parameter. (This is called a Control in PAN.)
Develop a table to clearly identify the relationship between different levels of the policy class and the metrics proposed earlier.
Present at least 3 trade-off curves between important metrics of the model.
Make a recommendation to management. Provide arguments justifying the recommendation.
CASE 2
Activities A and B start immediately. The time for each activity follows a Normal distribution with mean 50 days, and standard deviation 10 days. o The completion time for activities A and B is correlated, and the correlation parameter is . We will investigate how varying the correlation between these activities affects completion time of the project.
Activity C begins only after both Activities A and B are completed. The time to complete Activity C follows a Uniform distribution between 20 and 40 days. Time for Activity C is not correlated with the other activities.
Formally, we have:
o Time-A ~ N(50 , 102)
o Time-B ~ N(50 , 102)
o Time-C ~ U[20 , 40]
o Corr(Time-A , Time-B) =
Develop the analysis in Crystal Ball (or @Risk or Python)
Evaluate completion time for the project, as a function of different levels of correlation between times of Activities A and B. Remember that correlation can be between -1 and +1. Include a printout of the completion times.
If the objective is shortest completion time, what’s the best form of correlation between activities? Explain this result.
Unformatted Attachment Preview
SCOT 500M – Final Exam
Operations Analytics: Simulation
SCOT 500M, Spring 2024
Individual Final Exam: Due Friday, March 1, 11 p.m.
On Canvas
A few guidelines for the exam:
This assignment is to be done entirely individually.
You may discuss it only with the professor or the TA.
There are 2 different cases included in the exam. 1 requires Arena, and 1 uses Crystal Ball
(or @Risk or Python). Please make sure you submit all 2 cases. Some cases are easier than
others.
You do not need to develop a comprehensive report for each case. It is best to copy the
questions and submit answers clearly for each question. Please make sure to answer each
part clearly.
Include printouts or screen shots of any models you develop and simulation analyses. (I do
not need to see all trials, though)
To facilitate grading, please adhere to the following:
o Please submit each exam separately, on Canvas
o Include your Student ID on each case
Discrete Event Simulation using Arena
Case I: Creative Analysis of a Service System (13 points)
Background: in this part you are expected to show your creativity and understanding of queuing
processes, and use discrete event simulation to develop and analyze a relatively simple system.
This is an open-ended case. You are not being asked to model a specific system. You need to come
up with the example to analyze, and analyze the system you describe using Arena.
You may come up with the idea from any source or example you like, but are not allowed to
discuss this with anyone else.
Part of your grade (2 points) on this assignment is based on your creativity. Creativity is measured
on 3 dimensions:
1. A more creative scenario will earn more credit for creativity. Please note that creativity
does not require more components in the Arena model. You can be creative with few
modules.
2. Creative metrics for evaluating the scenario will also earn credit for creativity
3. Creative solutions for the problem you propose is also valued
[1]
SCOT 500M – Final Exam
Assignment: there are many ways to improve a system that involves queueing systems. Frequently,
these systems have customers waiting for service. We discussed quite a few in class during the
semester. You are tasked with evaluating one (new) system and evaluating performance of the
system under various scenarios.
1. Describe the scenario you propose to analyze. Clarify your creativity in this example.
2. Develop a model in Arena that reflects your scenario. There are no limits on how you choose
to model the system. The model must have at least one of each of the following components
(you may have more of each component):
o Arrivals
o Decision node or nodes
o Process
o Assign
o A schedule
o Optional1 components are not required, but are worth 0.5 points each
A variable
A Record module
3. Explain metrics used to evaluate the system. These could be the default metrics from Arena,
or new metrics. You should discuss at least 4 metrics.
4. You may (but are NOT required to) develop a creative metric. This can be a combination of
current metrics from Arena, or something else you measure in the simulation.
5. Run the model for the base case, with at least 100 trials. Discuss key results from the Arena
analysis. Make sure to highlight the metrics you proposed earlier.
6. Suggest a recommendation for improving performance of the system. This recommendation
must involve a policy class related to a key parameter or input of the model. Explain the
rationale for using this policy class. The purpose of this policy class is to identify the
relationship between one aspect of the model and various performance metrics you suggested
earlier. (As an example, the policy class might be to change the arrival rate. The levels of the
policy are the different arrival rates. This is probably a bad example, because arrival rates are
not under your control.)
7. Analyze this policy class using the Arena PAN tool. The policy class must have at least 10
levels for the policy parameter. (This is called a Control in PAN.)
8. Develop a table to clearly identify the relationship between different levels of the policy class
and the metrics proposed earlier.
9. Present at least 3 trade-off curves between important metrics of the model.
10. Make a recommendation to management. Provide arguments justifying the recommendation.
Your write-up should clearly explain the scenario being analyzed, metrics, recommendation, and
key conclusions. Highlight aspects of the analysis that you think are creative.
1
Each of these optional components is worth 0.5 points. If you do not have one of them, your total grade will be
reduced by 0.5 points, for each one that is missing.
[2]
SCOT 500M – Final Exam
Operations Analytics: Simulation
Monte Carlo Simulation using Crystal Ball / @Risk / Python
Case II: Project Management with Correlated Activities (2 points)
Consider the following rather simple project.
Activities A and B start immediately. The time for each activity follows a Normal
distribution with mean 50 days, and standard deviation 10 days.
o
The completion time for activities A and B is correlated, and the correlation parameter
is . We will investigate how varying the correlation between these activities affects
completion time of the project.
Activity C begins only after both Activities A and B are completed. The time to complete
Activity C follows a Uniform distribution between 20 and 40 days. Time for Activity C is
not correlated with the other activities.
Formally, we have:
o
Time-A ~ N(50 , 102)
o
Time-B ~ N(50 , 102)
o
Time-C ~ U[20 , 40]
o
Corr(Time-A , Time-B) =
Develop the analysis in Crystal Ball (or @Risk or Python)
Evaluate completion time for the project, as a function of different levels of correlation
between times of Activities A and B. Remember that correlation can be between -1 and +1.
Include a printout of the completion times.
If the objective is shortest completion time, what’s the best form of correlation between
activities? Explain this result.
[3]
Purchase answer to see full
attachment
