Sie sind auf Seite 1von 46

SYSTEM MODELING

PRESENTATION
SLIDES

ALIM
Introducing our team members
Consists of 6 Industrial Engineering female students

Salma Nabila Latifa Ayu Sarah Marsha

Givanny Permata Aisha Adilla Hanny Riana

2
First, we divide this presentation into some parts
Based on its contents: there are 8 parts

Problem Definition Literature Review Novelty

Validation & Verification Data Collection Model Conceptualization

Running the Model Result & Discussion Conclusion

3
01
PROBLEM DEFINITION
We determined the problem at Halim Airport
Defining the symptoms & effects

Explanation Output Outcome

Halim Perdanakusuma Airport has Long queues at the check-in Passengers dissatisfaction
served increasing numbers of counters in the airport
domestic flights

2500000 2200000

The Symptoms 2000000

1500000
2013
1000000 2014

500000 200000
0
Total Passenger

If Halim Perdanakusuma Airport …more dissatisfaction by the …decreasing number of


ignores this problem airport’s quality of service passengers departing from
the airport
The Effects

5
We determined the problem at Halim Airport
Defining the problem statement and the project scope

Problem Statement Project Scope

Who When Sunday, the


• An increasing number of flights and passengers airport’s peak
departing from Halim Perdanakusuma Airport has
day
caused long queues at the check-in counters,
including those of Batik Air, on peak days
(Sundays).
• The average of queueing time on Mondays- What Reducing Why Increasing
Saturdays is 2,5 minutes, whereas on Sundays it number of
is 20,79 minutes. queuing time
flights and
• If this condition keeps on, there can be a passengers
decreasing number of passengers departing from
the airport due to dissatisfaction, causing Halim
Perdanakusuma Airport’s and Batik Air Where Batik Air check-in How
Enterprise’s profits to drop. counter at Halim Discrete-Event
Airport Simulation

6
Then, we decided on the objectives of this project
It is also called the model objectives

• Interview with problem owners Reducing the queueing


• Questionnaire surveys time at check-in counters
by 50%
• Direct observation

7
02
LITERATURE REVIEW
We analyzed journals that used for project references
Explaining the method used on 1st journal

• Linear programming method


• Solving of waiting line models in The Airport Using Queuing Theory Model and Linear
Programming The Practice Case: A.I.M.H.B
Authors: Houda Mehri, Taoufik Djemel, Hichem Kammoun

Advantages Disadvantages

Help managers evaluate the cost and Use quantitative analysis technique
effectiveness of service systems
Describe several common queueing situations & There has to be specific assumption to construct
present mathematical models model
Particularly suitable for queueing theory because Provide the equation needed to compute the
the solution required may be too long even on operating characteristics of queueing lines
the fastest computer
Describe the characteristic of queueing lines Distributions has to be Poisson & exponential

9
We analyzed journals that used for project references
Explaining the method used on 2nd journal

• Operation research method


• Optimizing the Airport Check-In Counter Allocation Problem
Authors: Gerson E. Araujo; Hugo M. Repolho

Advantages Disadvantages

Uses optimization model to determine the There has to be an equation derived from
minimum number of check-in desks to be constraints and level of service
opened, so that operational costs and
quality of service are balanced
Uses simulation to assess if the first step
results meet the overall service level
Uses an optimization model to enforce the
adjacent constraint

10
We analyzed journals that used for project references
Explaining the method used on 3rd journal

• Discrete event analysis method


• Discrete-Event Simulation of Queuing Systems
Authors: Zhang Laifu Joel, Ng Wen Wei Jonathan Louis and Tay Seng Chuan

Advantages Disadvantages
To model the queuing systems and to Uses a single-line multi-server
analyze the side effects when one configuration since it seems faster for
system is changed to the other customers to get to the servers, but
this is not proven by either the model
made by Zhang, Ng, and Tay nor our
model
To avoid costly design errors and to
analyze the behaviors of the existing
systems
To predict the performance of the
existing systems when the input
parameters are changed
11
After that, we choose which method is the best
Which the best method is Discrete-Event Simulation

• Discrete-Event Simulation (DES) is more effective than analytical models


such as Linear Programming (LP) and Operational Research (OR) in
modeling environments. Our research proves it can be applied in many
situations in operations management.
• In real-life situations that are portrayed by analytical queueing theory
models involve a lot of assumptions. Unfortunately, arrivals in real-life
situations may not be correctly approximated by distributions such as
Poisson.

12
03
NOVELTY
After that, we know what is new about our research
It is also called the novelty

Current Research Journal References

• We use DES to examine airport check-in • Mehri, Djemel, Kammoun (2006) used
queues which has advantages: LP
Analytical methods like LP and OR • Gerson and Hugo (2015) used OR
involve a lot of assumptions (constant
service rate, a certain distribution)
Unfortunately, real-life situations may
The not be correctly approximated
Novelties

• There are various variables influencing • Zhang, Ng, and Tay (2000) used DES
queueing time, such as amount of to examine queues at McDonald’s, but
baggage there are no other influential variables

14
04
MODEL
CONCEPTUALIZATION
We had to know the current condition first
Analyzing it through direct observation and put it into ProModel

• In this current condition, there are:


• 9 queues
• 7 conveyors
• 1 entrance
• 1 exit

• 9 Batik Air check-in counters:


• A1 – A3:
no baggage
• A4:
business class
• A5 – A9:
with baggage

16
We simplified the real system for the model
Using flowchart as the representative of real system at Halim check-in counter

Counter Type Opening Hour

A1 Without baggage 02.00 PM – 05.00 PM

A2 Without baggage 02.00 PM – 05.00 PM

A3 Without baggage 09.00 AM – 05.00 PM

A4 Business Class 09.00 AM – 05.00 PM

A5 Multiple baggage 09.00 AM. – 05.00 PM

A6 Multiple baggage 09.00 AM – 05.00 PM

A7 Multiple baggage 12.00 PM – 05.00 PM

A8 Multiple baggage 12.00 PM – 05.00 PM

A9 Multiple baggage 12.00 PM – 05.00 PM

17
In real system, the condition can change any time
Thus, we made assumptions and boundaries for the model

Assumptions Boundaries

The weight of
baggage do not Peak Time
exceed the limit
given by the airline
Service based on
FIFO Availability
No technical check-in space
difficulties e.g: area
server error,
baggage cargo
problem, etc. Single line and Amount of
single server queue check-in counter
Departing line layout
passenger has Amount of
prepared the check-in counter
obligatory document person
before doing check
in

18
05
DATA COLLECTION
We collected data to be programmed into the model
Gathering many types of data through several methodologies

Direct Observation Papers/Journals

Online
Questionnaire Interview SECOND
PRIMARY
ARY
DATA
DATA

Historical Data Books

20
We collected data to be programmed into the model
Gathering many types of data through several methodologies

7. Probability of
passengers going to
1. Layout of check-in each counter
area 4. Arrival rate 8. Percentage of
10. Queuing time
2. Check-in flow 5. Number of available passengers based
11. Service time
3. Customer check-in counter on party size
12. Peak time
satisfaction of 6. Number of baggage 9. Percentage of
queuing time passengers based
on whether they
bring baggage or not

Types of Data

21
We conducted an online survey
352 respondents answered with their opinions of ideal service and queue times

Ideal Check-In Counter Service Time Ideal Check-In Counter Queue Time
6% 2% 4%
12%
5%

1 minute
27% <= 10 minutes
2 minutes
21% 11-15 minutes
3 minutes
28% 16-20 minutes
4 minutes
21-25 minutes
5 minutes
61% 26-30 minutes
> 5 minutes
4%

30%

Conclusion
Most respondents want to be served at a check-in counter for as
long as only 3 minutes and queue at a check-in counter for as long
as 10 minutes.

22
There are many airlines that operating at Halim
We searched the airline with the most influential in check-in service system area

There are 7 airlines that operating …which we minimzed again with


at Halim airport with various Pareto Diagram to find the most
number of flight… influential airline

Airlines
• Batik Air, Citilink, Wings Air, Trinusa, Pareto Diagram of Number of Flights for Each
Susi Air, Airfast, and Pelita Air. Day in Halim Perdanakusuma Airport, Jakarta
100
60
80
With total number of flight per day is 50
40 60
68 flights. 30 40
20
10 20
Accumulation 0 0
Name of Number of Percentage of Batik Air Citilink Wings Air Trinusa Susi Air Airfast Pelita Air
Percentage of
Enterprise Flight per Day Flight per Day
Flight per Day
Batik Air 30 44.11764706 44.11764706
Citilink 29 42.64705882 86.76470588 From the Pareto Diagram, shown that Batik Air
Wings Air 3 4.411764706 91.17647059
Trinusa 2 2.941176471 94.11764706 Enterprise gives 80% of the problem in check
Susi Air 2 2.941176471 97.05882353
Airfast 1 1.470588235 98.52941176
in counter of Halim Perdanakusuma Airport,
Pelita Air 1 1.470588235 100 Jakarta.
TOTAL 68 100

23
We had to find some information for the observation
The data that we need is the peak day and total sample of Halim Airport

We used historical data from Halim


Perdanakusuma information center to After we got the peak day, we calculated the sample for our
know when is the peak day of Halim observation. The calculation we used is Solvin Formula
Airport which showed from the picture below.

From the historical data, we got Sunday as a peak day also


proved by result of advanced judgement

Number of Passengers Flight We used the confidence level = 95%


5000 4170.25
3737 3580 3609.5
3927.75 3776 Based on historical data from Halim Perdanakusuma
4000 3388
Airport, we got the average population number of Batik Air
3000 check in counters from 9 AM to 5 PM = 2900 passengers.
2000
1000 From the formula and numbers above, we got
sample = 352 passengers
0

24
We grouped the observation data into some criteria
Based on the total baggage and total passengers that come

Percentage of Carrying Baggage and Not Percentage of Single Passengers and


Carrying Baggage Group Passengers
Carrying baggage Not carrying baggage Single Passenger Group Passenger

20%

35%

65%

80%

25
Afterwards, we got data that will put into ProModel
They are arrival rate, service time, and probability for passengers

Hours Number of Arrival (%) Number of Baggage Service Time (Seconds)


09.00 am-9.59 am 4.90 0 52.62
10.00 am-10.59 am 5.21 1 87.57
11.00 am-11.59 am 4.97 2 105.92
12.00 pm-12.59 pm 4.79
3 120.71
13.00 pm-13.59 pm 13.48
14.00 pm- 14.59 pm 27.93 >3 282.44
15.00 pm-15.59 pm 18.97

PROBABILITY OF SINGLE PROBABILITY OF GROUP


PASSENGERS FOR EACH COUNTER PASSENGERS FOR EACH COUNTER
A1 Counter A2 Counter A3 Counter A4 Counter A5 Counter A1 Counter A2 Counter A3 Counter A4 Counter A5 Counter
A6 Counter A7 Counter A8 Counter A9 Counter A6 Counter A7 Counter A8 Counter A9 Counter

6% 6%
6% 20% 10%
24%
5%
4% 8%
5% 5%
2% 16%
9%
25%
23% 17% 9%

26
07
RUNNING THE MODEL
06
VALIDATION & VERIFICATION
VALIDATION
The first validity method we used is face validity
Checking validity of model by asking people who know system well and trusted

We asked one of Operation Manager


of Batik Air called Abdul Karim

We asked one of Staff of Service of


Halim Perdanakusuma Airport called
Radynal Manurung

29
The second method is comparing with other model
Comparing output from the simulation with spreadsheet

Average of Counter Queueing Time (Minutes)


A1 A2 A3 A4 A5 A6 A7 A8 A9
Promodel 5.15 10.04 15.47 3.91 25.16 25.32 24.94 36.69 40.08
Spreadsheet 5.23 9.21 15.18 3.34 25.05 25.13 25.12 37.16 40.23

30
We also used extreme condition test for validity
Testing the model using 2 extreme conditions

1st Extreme Condition:


Total Entities 0

31
We also used extreme condition test for validity
Testing the model using 2 extreme conditions

2nd Extreme Condition:


Total Entities 12000

32
We also used extreme condition test for validity
Processes are traced using processing logic to be compared with actual model

33
VERIFICATION
The verification method used is trace & debugger
Enabling us to look deeper what’s happening in the simulation

35
08
RESULT & DISCUSSION
After we ran the model, we checked its statistic
Analyzing the statistic result

The schedule Time for Each Counter


on Actual Condition
• A1, A2: Open at 02.00 pm- 05.00 pm (3 Hours)
• A3, A4, A5, A6: Open at 09.00 pm- 05.00 pm (8 Hours)
• A7, A8, A9: Open at 12.00 pm-17.00 pm (5 Hours)

The percentage of carrying baggage


and not carrying baggage
• 65% for carrying baggage
• 35% for not carrying baggage

The Average of Queueing Time

• The Average of Queueing Time is 20.77 minutes

38
From the result, we made some alternative solution
Here is our 1st alternative solution

• Opening Time :
o 09.00 AM for all counters
• Configuration :
o Single line – single server queue
o Counters A1 to A3 for passengers with no baggage
o Counter A4 for business class passengers
o Counters A5 to A9 for passengers with baggage

• The Advantages
(+) Bigger probability for the passengers to choose a
counter
• The Weaknesses
(-) Accumulated number of passengers on some
counters

39
From the result, we made some alternative solution
Here is our 2nd alternative solution

• Opening Time :
o 09.00 AM for all counters
• Configuration :
o Single line – multi server queue
o Counters A1 to A3 for passengers with no baggage
o Counter A4 for business class passengers
o Counters A5 to A9 for passengers with baggage

• The Advantage
(+) Customer of the counters is evenly distributed
• The Weakness
(-) Customer has a lower probability to choose a
counter

40
From the result, we made some alternative solution
Here is our 3rd alternative solution

• Opening Time :
o 09.00 AM for counters A3, A4, A5, A7, and A9
o 12.00 PM for counter A2
o 01.00 PM for counters A6 and A8
o 02.00 PM for counter A1
• Configuration :
o Single line – single server queue
o Counters A1 to A3 for passengers with no baggage
o Counter A4 for business class passengers
o Counter A5 for passengers with one baggage
o Counters A6 to A7 for passengers with two baggages
o Counters A8 to A9 for passengers with more than two
baggages

• The Advantage
(+) Customer on the counters is evenly distributed
(+) Queing time is directly proportional to the
number of baggage
• The Weakness
(-) Customer has few counter choices
41
Finally, we compare the statistical result of them
Here is our 3rd alternative solution

Alternative 1 Alternative 2 Alternative 3

The Average Time for The Average Time for The Average Time for
Alternative 1 is 26.43 Alternative 2 is 21.50 minutes Alternative 3 is 5 minutes
minutes
42
Then we choose the best solution among them
We choose the 3rd solution and made a discussion about it

Changing the schedule and line configuration will


not contribute much to the solving of the problem,
rather there should be a classification of counters
according to the number of baggage.

This further proves that the decision taken by


McDonald’s in Zhang, Ng, and Tay (2000) to use
a single-line multi-server configuration instead of
a single line single-server one is inappropriate for
that particular situation.

The model objective, which is to reduce the


overall average queueing time by 50%, has been
reached.

43
09
CONCLUSION
In the end, we concluded the overall of this project

We recommend to Halim We also made an efficient schedule We suggest in further


Perdanakusuma Airport is to for the counters to be opened. researches and studies for
open counter A1 to A3 for The chosen optimal schedule is to Discrete-Event Simulation to
passengers with no baggage; open A3, A4, A5, A7, and A9 at 9 be used in problems with a
counter A4 for business class AM; A2 at 12 PM; A6 and A8 at 1 wider scale, such as in an
passengers; counter A5 for PM; and A1 at 2 PM. The new airport larger than Halim
passenger with one baggage; average queue time obtained after Perdanakusuma Airport, to
counters A6 and A7 for implementing the chosen solution further validate the usefulness of
passenger with two baggages; alternative is 5 minutes. It shows Discrete-Event Simulation in
and counters A8 and A9 for that we have reached our goal such situations.
passengers with more two statement to reduce waiting time by
baggages. 50%.

45
THANK YOU
- Aisha Adilla (1406606152)
- Givanny Permata Sari (1406606070)
- Hanny Riana (1406606341)
- Latifa Ayu Lestari (1406606354)
- Salma Nabila Hadi (1406553133)
- Sarah Marsha Davinna (1406553285)

ALIM

Das könnte Ihnen auch gefallen