COMP4038 Reassessment Spring 2023 v1 (28/07/2023)
Management Decision Support for BABIES FORTUNE Charity
BABIES FORTUNE is a new charity to support the Hope House Babies initiative in Nairobi/Kenya. There
are five charity shops of the charity distributed across different locations in the Port of Hamburg.
Although, each store operates in a similar manner, each of them has their own particular features
related to the size, location, etc. SIMOPT is a business intelligence company that provides IT systems
for the charity to assist in the operation of individual stores. The focus of their provision is a system
that allows store managers to estimate the goods demand based on historical records, capture sales
data, and manage workforce, among other functionalities.
SIMOPT wants to improve their business intelligence services by incorporating modern computational
techniques into their systems. They want to explore data analytics, optimisation, simulation, etc. They
have decided to start by developing an optimisation model to schedule employees working in an
individual store based on the estimated demand. Most employees in this optimisation scenario are
part-time, have mixed-skills and require flexible shifts. Then, given a set of shifts that need to be
covered, the problem is to assign employees to each individual shift while satisfying various
constraints. For each shift the length, start/end times, number of required workers and required skills
or roles are given.
You have been assigned as the lead consultant to take this project further by applying a combination
of simulation and optimisation that provides more insights and tools for an effective and efficient
management of the store's workforce. The simulation and optimisation should also provide metrics
and statistics for understanding the operational conditions over time, for example productivity,
utilisation, etc. as well as for exploring variations on the operational conditions. Furthermore, donated
goods delivery management is an add-on that could also be considered in the simulation/optimisation
scenarios (but only attempt this if you after a very high mark).
2
NB: As this is an illustrative case study, please use realistic guesses and information from the internet,
if you require any operational information (e.g. arrival rates; staff roles; opening hours; service times)
or other kind of data that you would normally get from the client.
Split of Marks:
For this coursework, you are asked to focus on the following (related % of 100 marks in orange):
1. Conceptual model (following the instructions given in Lecture 3A): 20%
2. Implemented model (allowing simulation and simulation-optimisation experiments): 40%
3. Simulation-optimisation experiments (including statistical output analysis): 30%
4. Demonstration video: 10%
The marks are given by assessing quality and completeness of the individual tasks (1-4) listed above.
Please note that quality entails that you provide sound explanations of your activities and justifications
for the decisions you made.
The requirements for getting a pass mark are:
• Choice of a meaningful scenario (at least one service centre)
• A simple discrete event or agent-based model that displays some meaningful statistics during
runtime (providing metrics and statistics for understanding the operational conditions of the
system over time)
• A meaningful experiment (exploring variations on the operational conditions)
• Some form of output analysis
If you are aiming for higher marks, consider some (or all) of the following:
• Innovative scenario choice (including implementing multiple service centres)
• Demand variability (e.g. on/off peak arrival rates); some locations are busier than others
• Staff optimisation using a central workforce pool
• Using a hybrid DES/ABM modelling approach (adding state charts to your agents; considering
staff and customer stereotypes, etc.)
• Informative GUI displaying relevant outputs (e.g. satisfaction; utilisation)
• Adding a prototype smart app to your model and explore its impact
• Adding an external optimiser (e.g. HeuristicLab) or writing your own optimisation algorithm
• High calibre optimisation experiment(s), assessing performance and potential improvement
on the operational conditions
• Scholarly written report
• High quality video
Reassessment Coursework Submission Guidelines:
The submission deadline is 23 August 2023, 3pm. Work submitted after the deadline will be subject
to a penalty of 5 marks (the standard 5% absolute) for each late working day out of the total 100
marks. Late submission deadline is 30 August 2023, 3pm. Submissions after this date will only be
accepted through the extenuating circumstances procedure.
You are asked to submit:
• A written report in PDF format (2500 words +/-10%) capturing a description and explanation
of the first three points listed in the beginning of section "Split of Marks"
• Your final AnyLogic PLE simulation-optimisation model (we need to be able to reproduce your
experiment runs with the help of the submitted model using AnyLogic PLE) as a ZIP archive
• A video (4 mins +/- 0.5 mins; not more than 100mb; MP4, MPG, or AVI format), explaining
the model's design and demonstrating its simulation-optimisation capabilities (i.e. showing it
running and showing the output it produces).
3
The reassessment coursework mark accounts for 100% of the 20-credit module mark. I would
recommend to reserve about 40-60 hours for this coursework. For getting some inspiration when
starting your coursework please have a look at related models in the AnyLogic Help menu. If you are
stuck or have any questions, please feel free to ask via private Teams chat.
When you start the implementation, remember to follow the KISS principle - start with something very
simple and then extend it to add functionality and complexity, as demonstrated in the two live coding
sessions.
Students are reminded of the Policy on Academic Misconduct (and in particular plagiarism) and must
ensure that all material from other sources is clearly quoted and acknowledged.
For more information, please see https://www.nottingham.ac.uk/qualitymanual/assessment-awards-
and-deg-classification/pol-academic-misconduct.aspx
Limitations of AnyLogic PLE Software:
AnyLogic PLE has some limitation. Many elements in the Pedestrian Library, Rail Library, and Road
Traffic Library only allow simulation runs of one hour, which is not sufficient for running the required
experiments. Please do not use these libraries. The only exception are space markup elements from
these libraries, which can be used for any model. Other restrictions are that the number of agents you
can create is limited to 50.000, the number of classes you can create is limited to 10, and the number
of decision variables you can use for optimisation is 7. Regarding agent numbers, you learned how to
overcome the restriction by recycling agents. Regarding the optimisation limitation, you could encode
the solution as a bit string where different sections of the string represent separate decision variables.