Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Term Project

Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Term project case

The Manila International Airport Authority (MIAA) wants to study Terminal 3 of NAIA with an
eventual eye towards improvement. The first step is to model it as it is during the 8 hours through
the busiest part of a typical weekday. As the analysts hired for the project, we will model the
check-in and security operations only, that is, once passengers get through security they’re on
their way to their gate and out of our model.

Passengers arrive one at a time through the front door from curbside ground transportation with
interarrival times distributed exponentially with mean 0.5 minute (all times are in minutes unless
otherwise noted). Of these passengers, 35% go left to an old-fashioned manual check-in counter,
50% go right to an automated check-in counter, and the remaining 15% don’t need to check in at
all and proceed directly from the front door to security (it takes these latter types of passengers
between 3 and 5 minutes, uniformly distributed, to make the walk from the front door to the
entrance to the security area; the other two passenger types move instantly from their arrival to
the manual or automated check-in counter as the case may be).

There are two agents at the manual check-in station, fed by a single FIFO queue; manual check-
in times follow a triangular distribution between 1 and 5 minutes with a mode of 2 minutes. After
manual check-in, passengers walk to the security area, a stroll that takes them between 2.0 and
5.8 minutes, uniformly distributed. The automated check-in has two stations (a station consists of
a touch-screen kiosk and an employee to take checked bags; view a kiosk-employee pair as a
single unified unit, that is, the kiosk and its employee cannot be separated), fed by a single FIFO
queue, and check-in times are triangularly distributed between 0.5 and 1.5 with a mode of 1. After
automated check-in, passengers walk to the security area, taking between 1 and 3 minutes,
uniformly distributed, to get there (automated check-in passengers are just quicker than manual
check-in passengers at everything).

All passengers eventually get to the security area, where there are six stations fed by a single
FIFO queue; security-check times are triangularly distributed between 1 and 6 with a mode of 2
(this distribution captures all the possibilities there, like x-ray of carry-ons, walking through the
metal detector, bag search, body wanding, shoes off, laptop checking, etc.). Once through the
security check (everybody passes, though it takes some longer than others to do so), passengers
head to their gates and are no longer in our model.

1. Simulate this system for one replication of an 8-hour period and look at the average queue
lengths, average times in queue, resource utilizations, and average total time in system of
passengers (for all passenger types combined).
2. Suppose you could hire one more person, and that person could be assigned either to
manual check-in, automated check-in, or security. Where would this additional person be
best used? As an overall criterion of “goodness” of a system, use the average total time
in system of all types of passengers combined. Make 50 replications of each scenario and
perform statistical analysis to justify your conclusions.
3. The airline noticed that a lot of the people who opt for the manual check-in really don’t
need the extra services there and could have used the automated check-in. Instead of the
original 35% and 50% that go to manual and automated, respectively, suppose that
through more effective “encouragement” from people with red blazers and loud voices, it
could be made 10% and 75% instead. (There probably are 10% of passengers who
genuinely need the extra services of manual check-in.) How would this shift in passenger
attitudes (or maybe bravery) affect average total time in system of passengers? Base your
comparison on 200 replications of the model in each configuration. Once again, perform
statistical analysis to justify your conclusions.

You might also like