HeaderSIS.jpg

IS480 Team wiki: 2012T1 Bumblebee Project Overview

From IS480
Revision as of 15:15, 6 October 2012 by Yosin.a.2009 (talk | contribs)
Jump to navigation Jump to search
YosinA.png
Home Project Overview Project Documentation Project Management Learning and Growth
Project Description X Factor Stakeholders Motivation Deliverables and Scope

Project Description

Project Description

Singapore Airport Terminal Services (SATS) is the leading provider of the gateway services and food solutions in the region. It has a staff of 800 to handle more than 35 airlines. Every year, there are two major airline flight schedule change - summer (April) and winter (November). Each change can be a drastic one and has great impact on staff roster. One of SATS’s goals for its staff roster is to meet all its airlines’ requirements. However, there are many uncertainties such as staff members falling sick, staff resigning and flight delay, that cause the roster planned to be ineffective. Thus, Duty Manager in SATS often has to make last minute changes to the rosters and incur costs such as Staff Recall Cost, Over-Time cost and Meal Allowance Compensation (MAC). To help improve the efficiency involving various costs and consequences of the planned staff roster, our project aims to create a Staff Deployment Simulation Software (SDSS) which will first deploy staff based on flight schedule, flight requirements, staff records and staff roster and then simulate the roster plan by taking into account all the various uncertainties and forecast the various costs that the management would have to incur. Understanding the cost and consequences of a given roster, SATS would then be able to make necessary adjustments to avoid high expenses to the company. The diagram below explains the flow of the software.

IPO.png

To predict the uncertainties, we will take in the mean and standard deviation of the listed simulation parameters in diagram above(MC Rate, Flight Delay Rate, Staff Resign Rate, etc). Having those values, our software will then generate a normal distribution of each uncertainty. After such, we will be applying the uncertainties at the planned roster plan such as removing staff from the roster plan if he/she is taking MC. The software will also try to mimic the actions of Duty Manager who does the necessary adjustment to the staff roster to fulfill the airlines requirements i.e. to imitate human decision making process to come out with an optimal staff roster.

At the end of the day, our software will generate a management report which will reveal the cost and consequences of implementation of a given roster. The output given will be Total Staff Working Hours, Flight Demand Coverage, Meal Allowance Compensation (MAC), Over Time (OT) Cost, OT hours, Staff Utilization Rate, Unproductive Hours, Recall Cost and Recall Hours. Understanding the cost and consequences of a given roster, SATS would then be able to make necessary adjustment to avoid high expenses and thus, increase cost savings to the company.

X Factor

X Factor

Efficiency

Our system is able to handle a highly complex, huge domain size, scheduling problem efficiently in comparison to SATS’ current manual scheduling process. It is able to schedule 800 staffs to fulfill approx. 33 airlines in less than 1 hour duration, which is currently manually done in days.


Algorithmic complexity

Optimal scheduling

We provide our client an optimized staff schedule to effectively reduce their overheads from non-optimal scheduling practices. In essence, we developed an optimal scheduling algorithm to maximise SATS’ resources to meet all airline requirements. The optimal scheduling is written based on Greedy Algorithm concept.


Probabilistic Simulation

Our system gives our client a clear perspective of cost loss profile from human resource planning uncertainties. To be specific, we use normal distribution and Kolmogorov-Smirnov test to forecast the uncertainties (flight delay, staff resign rate, etc) which SATS could face. After which we run a simulation to plot our client’s cost profile.


Aviation Knowledge

To build this system, our group has to understand aviation industry practices. Thus, our group did a two month weekly attachment to SATS officers to understand how they schedule staffs to fulfill all airline requirements.


To understand the complexity and the abundance in probabilities of scheduling, try to solve this:
We want to color countries on a map so that no two adjacent areas have the same color. How many colors do you need at the most? Map Coloring Teaser
Courtesy of IS103 Computational Thinking

Stakeholders

Stakeholders

The Bees

BumblebeeTeam4.png

Sponsors

B Sponsor.jpg


Motivation

Motivation

There are 2 main motivations for working on this project:
1. The industry that we are working with - it's aviation industry, in which we have minimum exposure in school! We would love to explore and understand more about what aviation industry is and how IT can help to make a difference in it!
2. Our software can make a significant difference as it will be implemented in SATS' daily operation. =D

Deliverables and Scope

Deliverables and Scope
S/N Component Modules Description Requirements
1 Bootstrap Data Import data: flight schedule, staff records and roster plan from CSV files to the application database. CSV imported has to follow certain format. This function also does intelligent data processing. Support multiple format (txt,csv).

Notify user if there is an error in importing.

2 Manage Airline Requirements Set special requirement such as how many CSA, CSO needed for a combined gate for each airline. How many minutes before scheduled time departure should a counter opens for specific airlines. Able to cater to different combination of flight sizes.
3 Optimised Scheduling Schedule staff roster based on employees’ working hours, system qualifications and other factors. Optimised scheduling is done by selecting cheapest alternatives to fulfill airline requirements. Application development methodology should be agile iterative process coupled with test-driven development. Greedy Algorithm is employed.
4 Simulation Perform simulation for optimised schedule developed in previous module. Simulation parameters such as flight delay generated based on normal distribution would be applied for simulation. Include all generated uncertainties/parameters in the simulation.
5 View result Measure the effectiveness of the roster by simulating the Unproductive Time, Airline Demand Coverage, Over-time Cost, Staff Recall Cost, Meal Allowance Cost, and many others. Simulation results should be credible and realistic.
6 Manage past results Save, delete and view past results. Past results could be exported as PDF report file. GUI should be user-friendly and simple

Please refer to use case description for greater details.