By :DICKY ARISIKAM, ST., MT.

(Ph.D. Student of Faculty of Civil Engineering and Environment – ITB)

Supervisor :Prof. Ir. Harun Al Rasyid Lubis, M.Sc., Ph.D.

Aine Kusumawati, Ph.D.

Advisor :Ir. Idwan Santoso, M.Sc., Ph.D.

Sony Sulaksono, Ph.D.Kristianto Usman, Ph.D.

RAILWAY RESEARCH

Transportation EngineeringFaculty of Civil Engineering and Environment

TRACK SAFETY ASSESSMENT MODELINGIN INDONESIAN RAILWAY

01Railway Safety in Indonesian Railway

Conceptual Framework

Hypothesis

Objective and Benefit of Research

Scope of Research

Literature Review (State of The Art)

Methodology

02

03

04

05

06

07

TRACK SAFETY ASSESSMENT MODELINGIN INDONESIAN RAILWAY

01. Railway Safety in Indonesian Railway

Railway is everything related to rolling stock, infrastructure, and its supporting facilities for the operation of train in one system (Indonesian Regulation No. 23-2007).

Transportation safety is a condition that occurs from the implementation of good transportation based on the SOP and technical requirements ofrolling stock, infrastructure,and its supporting facilities(Lubis, 2019).

Safety Standard of IndonesianRailway (Regulation of MoTNo. 24-2015) :1. Infrastructure2. Rolling Stock.3. Operational.4. Human Resource. Railway Safety = f (Infrastructure, Operational, Human Resource, Rolling Stock, Environment)

01

02. Conceptual Framework 02

One of the consideration for people in choosing a transportation mode is Safety (Blagojevic, et al,

2020) Research Questions :1. What are the factors of train safety in

Indonesia?2. Which the track segment are riskiest

for train safety in Indonesia?3. How to assess the safety level of track

segment in Indonesia?

Transportation Business =

Safety Business

Train accident can cause in financial lossand damage

the image

Goal : Increase in safety and

zero accident

One of the factors of

train safety is infrastructure

Infrastructure must be maintained

regularly and based on its priority scale

One of the methods to determine the priority

scale is by train accidentprediction

How to predict train accident?

It’s need to develop a model and method for prediction of

train accident in track segment

The development of the model with

Generalized Linear Model (GLM)

Model : The relation between train accident factors (exposures, risks) as independent variables and train

accident prediction in track segment as dependent variable

Safety ranking of track segment with Data

Envelopment Analysis (DEA)

Method : Using train accident factors (exposures, risks) as input data and number of train accident in track segment as output data

Output : Train accident prediction

in track segment

Safety ranking of track segment by

the model

Safety ranking of track segment by

DEA

Compared

Gap

Conceptual Problem

Objective

03. Hypothesis

No Symbol Factors Unit

1 Y Number of Train Accident Accident

2 X1 Track Length (Exposure) Km

3 X2 Train Frequency (Exposure) Train/Day

4 X3 Passing Tonnage (Risk) Ton/Year

5 X4 Number of Level Crossing (Risk) Number

6 X5 Vulnerable Location (Risk) Number

7 Xn ...................... ……….

Number of Train Accident, Y = f (X1, X2, X3, X4, X5, Xn)

To prove the hypothesis is by make a model with GLM that relates to the train accident factors and train accident prediction

General equation : Accident = f (Exposures, Risks)

03

Hypothesis : There is a relationship between train safety factors (exposures, risks) such as track assets, environment, and track maintenance with the number of train accidents in Indonesia

The factors will be validated in data processing and analysis

𝝁𝒊 = 𝒆𝒙𝒑 𝑿𝒊𝜷 = 𝒆𝒙𝒑

𝟏

𝒑

𝒙𝒊𝒋𝜷𝒋

𝝁𝒊 = 𝒆𝒙𝒑 𝑿𝒊𝜷 = 𝒆𝒙𝒑

𝟏

𝒑

𝒙𝒊𝒋𝜷𝒋

𝝁𝒊 = 𝟏 − 𝒒𝒊 𝝀𝒊

𝝀𝒊 = 𝒆𝒙𝒑 𝑿𝒊𝜷 = 𝒆𝒙𝒑

𝟏

𝒑

𝒙𝒊𝒋𝜷𝒋

𝝁𝒊 = 𝟏 − 𝒒𝒊 𝝀𝒊

𝝀𝒊 = 𝒆𝒙𝒑 𝑿𝒊𝜷 = 𝒆𝒙𝒑

𝟏

𝒑

𝒙𝒊𝒋𝜷𝒋

1. Poisson

2. Negative Binomial

3. Zero-Inflated Poisson

4. Zero-Inflated Negative Binomial

Priority 2 Priority 1 Priority 3Priority 4 Priority 5

Objective :

To Get a Model of Track Safety Assessment in Indonesian Railway.

Benefit :

1. To estimate the train accident of track segment with input data based on the model.

2. To analyze the safety ranking of track segment based on estimation of train accident.

3. To determine the priority scale of track maintenance based on analysis of safety ranking.

04. Objective and Benefit of Research 04

Source: Lubis, 2019

04. Objective and Benefit of Research 05

Number of Train Accident, Y = f (X1, X2, X3, X4, X5, Xn)

Website and Application : Safety Rail Map (SRM)

Safety Ranking of Track Segment

04. Objective and Benefit of Research 06

Research :1. Track Safety Assessment Modeling (Arisikam, 2020)2. Operational Safety Assessment Modeling3. Rolling Stock Safety Assessment Modeling4. Level Crossing Safety Assessment Modeling5. Station Safety Assessment Modeling6. …………………………………..7. …………………………………..8. …………………………………..

Railway Safety ManualBook for Indonesian

Railway

05. Scope of Research

Scope :

1. This research will be conducted in Indonesian Railway with the segmentation is between stations.

2. The data needed for this research is train accident factors (exposures, risks) of track segment. The category of accidents are crash between train, overturning, and derailment.

3. The development of model is using Generalized Linier Model (GLM), where the output is train accident prediction in track segment. GLM model :

a. Poisson.b. Negative Binomial.c. Zero-Inflated Poisson.d. Zero-Inflated Negative Binomial.

4. Safety ranking analysis of track segment based on the estimation of train accident from the model.

5. Safety ranking analysis with the Data Envelopment Analysis (DEA) method.

6. Safety ranking comparison from the GLM model and DEA method.

07

06. Literature Review (State of the Art) : Research About Railway Safety Modeling 08

Research About Railway

Safety

1. Blagojevic, et al (2020)Track safety assessment in Bosnia and

Herzegovina Railway using the New Entropy, Fuzzy PIPRECIA and DEA

methods

5. Ioannidou, dkkCalculating the risk level of safety and risk loss (case study at level crossing)

using the RAMS method

6. Zhang, et al (2019)Selection of risk response of safety in

train derailment with MCDC integration (Dematel, ANP, Topsis, Delphi)

2. Wemaker, et al (2017)Determine the relation between railway safety and operational performance on several European Railway Companies

using a linear regression model

3. Fukuda, et al (2002)Study on accident analysis methods for a

railway safety management system

7. Utomo, et al (2019)Derailment analysis of the Bima Train on

the Wonokromo - Waru line with structural calculations

8. Backman (2002)Study on the safety of Swedish railways

on risk and economics with risk management and engineering economics

4. Fukuoka (1999)Uncertainty analysis for the assessment of railway safety using the Probabilistic

Risk Assessment method

How about research of track safety modeling or assessmentin Indonesian railways ???

NO ONE….

12. Miaou, et al (1993)Evaluation of the Linear Regression

model to assess the suitability of the relation between traffic accidents and

road geometry

10. Kusumawati (2008)Study on traffic safety at road

intersections by comparing the suitability of the GLM model

11. Lazuardi, et al (2017)Prediction model of bicycle accidents in

Malang and Blitar with GLM

9. Rakhmat, et al (2012)Development of a prediction model of

traffic accidents on the Purbaleunyi toll road with the Generalized Linear Model

(GLM)

13. Namnuusi, et al (2008)The study of accident prediction models

at highway intersections with GLM

14. Salifu (2003)Prediction model of accidents at urban

intersections without traffic signs in Ghana with GLM

15. Zahiya, et al Motorcycle accident prediction model in

Surabaya City with GLM

16. Bolla, et al (2014)Development of a motorcycle accident

model on the road in Atambua City with the GLM

Research About Railway Research About Roadway

Researcher Description Method Model Input Output

Rakhmat, et al

(2012)

Development of a prediction model of traffic accidents on the Purbaleunyi toll road

Generalize Linear

Model (GLM)

(1) Number of accident, (2) AADT, (3) Bridge, (4)

Ramp, (5) Pavement, (6) Skid resistance, (7)

Unevenness, (8) Slope, (9) Length of median,

(10) Special median, (11) Curve, (12) Number of

curve, (13) Curvature

The accident rate of the

selected segment on the

Purbaleunyi toll road

Kusumawati

(2008)

Comparing the

suitability of GLM types

in the case of traffic

accidents at highway

intersections

Generalize Linear

Model (GLM)

(1) Technical data of road, (2) History data of

accident

Negative Binomial

Regression Model is

better than other models

when compared

statistically

Wemaker, et

al (2017)

Development of risk

models for railway

safety and operational

performance in several

European railway

companies

Linear Regression (1) Train delay, (2) Train frequency, (3)

Operational performance, (4) Train

cancellation, (5) Signal data indicating danger

The efficiency level of the

selected railway

companies

Blagojevic, et

al (2020)

Safety assessment of

tracks on the Railways of

Bosnia and Herzegovina

New Entropy, Fuzzy

PIPRECIA, Data

Envelopment Analysis

(DEA)

(1) Track length, (2) Number of stations, (3)

Number of crossings, (4) Number of rolling

stock, (5) Number of trains, (6) Number of staff

The safety level of the

selected track

Azadeh, et al

(2008)

Optimization and

improvement of the rail

system in Iran Railway

Analytical Hierarchy

Process (AHP), DEA

Track data The efficiency level of the

selected track

Arisikam

(2020)

The development of

model for track safety

in Indonesian Railway

Data Envelopment

Analysis (DEA)

Generalize Linear

Model (GLM)

(1) Track length, (2) Passing tonnage, (3) Track

maintenance cost, (4) Number of level crossing,

(5) Vulnerable location, (6) ………….

The safety level of the

selected track

Didn't develop a model

The rankings are not compared to existing methods

0906. Literature Review (State of the Art) : Researches Related to This Research

07. MethodologyStep 1

Step 2

Step 3

Step 4

Step 5

10Start

Literature Review and Conceptual Problem

Collecting secondary data:

(1) Track asset data, include its defect

(2) Maintenance cost of track, include IMO

Primary data:

(1) Data of train accidents

Collecting primary data:

(1) Data of train accidents

Initial analysis to determine the factors of train accident Factors of train accident

(Independent Variable)

Secondary data:

(1) Track asset data

(2) Maintenance cost of track

Analysis and model development using GLM

Interview

Test the model and determine the best model Model

Train accident estimation of the selected

track segment based on the model that has

been made

Independent

Variable

Train accident rate

(Dependent Variable)

Safety rangking calculation of selected

track segment based on the model that

has been made

Safety ranking calculation of selected

track segment using the DEA method

Safety ranking of

segment (model)Safety ranking of

segment (DEA)

Compare the safety ranking of selected track segments based on analysis of the model

and DEA

Conclusions and Suggestions

Finish

Scope of Research

Thank You