HARYADI, Bambang (2011) PENGARUH VOLUME LALU LINTAS DAN FAKTOR GEOMETRI TERHADAP KECELAKAAN LALU LINTAS DI JALAN TOL ANTAR KOTA INDONESIA. PhD thesis, Program Pascasarjana Undip.
PDF Restricted to Repository staff only 1628Kb |
Abstract
Kecelakaan lalu lintas merupakan dampak samping transportasi jalan raya, tetapi tidak mendapatkan perhatian secara eksplisit dalam proses perencanaan transportasi. Penelitian ini dilakukan dengan maksud untuk mengembangkan model frekuensi kecelakaan yang terjadi pada ruas jalan tol dengan memperhitungkan karakteristik lalu lintas dan geometri. Data penelitian diambil dari empat jalan tol: Jakarta-Cikampek, Jagorawi, Palikanci, dan Padaleunyi. Data volume lalu lintas, karakteristik geometri, dan kecelakaan diperoleh dari kantor pusat dan kantor cabang pengelola jalan tol yang bersangkutan. Data volume lalu lintas meliputi volume harian rata -rata bulanan untuk masing-masing segmen jalan homogen selama periode 2002 hingga tahun 2008. Selain itu dikumpulkan sampel data volume lalu lintas per jam untuk mendapatkan profil lalu lintas per jam harian. Data kecelakaan dikumpulkan untuk lokasi dan periode waktu yang sama. Data kecelakaan mencakup lokasi dan waktu kejadian, tipe kecelakaan, dan kategori tingkat keparahan kecelakaan. Pengembangan model dilakukan dengan menggunakan teknik generalized linear modelling (GLM). Analisis regresi binomial negatif dilakukan untuk memperoleh model- model kecelakaan terpisah menurut jumlah kendaraan yang terlibat (kecelakaan ganda dan kecelakaan tunggal) dan tingkat keparahan kecelakaan (kecelakaan fatal, berat, ringan, dan rusak materi), selain model untuk kecelakaan total. Variabel terikat (respon) dari model adalah frekuensi kecelakaan sedangkan variabel bebas (prediktor) adalah karakteristik geometri ruas jalan dan volume lalu lintas per jam. Frekuensi kecelakaan mempunyai hubungan yang kuat dengan karakteristik geometri ruas jalan dan lalu lintas sebagaimana ditunjukkan dengan nilai-p. Volume lalu lintas per jam, panjang ruas, lebar jalur lalu lintas, dan lebar bahu luar mempunyai koefisien positif dengan frekuensi kecelakaan, artinya semakin besar nilai variabel tersebut semakin besar pula frekuensi kecelakaan. Dilain pihak, lengkung horizontal (dalam rad/km) dan lebar bahu dalam mempunyai koefisien negatif. Tipe pemisah jalur mempunyai pengaruh yang signifikan pada frekuensi kecelakaan di jalan tol. Secara umum dapat dikatakan bahwa dinding beton merupakan tipe pemisah jalur yang paling rawan kecelakaan, disusul oleh tipe median berupa ruang terbuka, dan guardrail merupakan tipe pemisah jalur yang paling aman. Selain itu, periode waktu yang berbeda dalam suatu hari mempunyai pengaruh yang berbeda pada jumlah kejadian kecelakaan. Secara umum dapat dikatakan bahwa periode waktu pukul 0-8 merupakan periode waktu yang paling rawan kecelakaan, sedangkan periode waktu pukul 16-24 relatif paling aman. Hasil penelitian ini mempunyai implikasi yang penting. Karena, dari penelitian ini, hubungan antara berbagai parameter disain jalan bebas hambatan dengan frekuensi terjadinya berbagai kategori kecelakaan dapat diungkapkan baik secara kualitatif maupun kuantitatif. Secara praktis pengetahuan ini memberi peluang untuk mengembangkan metode baru untuk perancangan jalan bebas hambatan yang secara eksplisit memperhitungkan keselamatan. Dimana ukuran keselamatan disini adalah besarnya frekuensi kecelakaan, menurut kategorinya, yang diharapkan terjadi pada suatu ruas jalan per satuan panjang per satuan waktu. Kata kunci: Model prediksi kecelakaan, Keselamatan jalan tol, model binomial negatif. Traffic accident is a by-product of highway transportation that not get explicit consideration during the transportation planning process. The objective of this research is to develop models of the effects of highway geometric and traffic factors on accident frequency that occurred on the Indonesian inter-urban toll roads. Data was obtain from four inter-urban toll roads: Jakarta - Cikampek, Jagorawi, Palikanci, and Padaleunyi toll-roads. Traffic volume, geometric characteristics, and accident data were obtained from the toll-roads operators. Traffic volume data consist of monthly average daily traffic from the year 2002 to 2008 period. In addition, hourly traffic volume survey was conducted to obtain hourly traffic profile. Accident reports was collected for the same time period and location. Geometric characteristics of each toll-road segments were obtained from as built drawings and maintenance records. Models were developed using generalized linear modeling technique. Poisson and Negative binomial regression analysis were conducted to obtain separate accident models base on the number of vehicle involved (single and multi-vehicle accidents) and the level of accident severity (fatal, heavily injured, lightly injured, and property damage only accidents) in addition to total accident. The response variables were accident frequency, and the predictors were some highway geometric characteristics and hourly traffic volume. The results of the study revealed that negative binomial models were better than Poisson models for modeling toll-road traffic accident frequency. Moreover, it was revealed that accident frequency was significantly related to hourly traffic volume and geometric factors, but the effect of each of those factors were different for different accident category. In general, hourly traffic volume, section length, traveled way width, and outside shoulder width had positive effect on accident frequency, meaning that the number of accident was bigger as the values of those factors were bigger. On the other hand, horizontal curve and inside shoulder width had negative effect on accident frequency. The effect of longitudinal grade on accident frequency was inconclusive as the value of this factor on toll-road usually small. Median types had significant effect on inter-urban toll-roads. In general, concrete wall was the worst median type with regard of safety, followed by flush unpaved, and guardrail was the safest type. The study also found that different period of the day had different effect on accident occurrence. In general, accident occurred most during period from 0 to 8 o’clock, and occurred least during period from 16 to 24 o’clock. The results of the study had significant implication, as it revealed the relationship between freeway design parameters and traffic accident frequency qualitatively as well as quantitatively. This new knowledge gives the opportunity to develop new design methods which consider safety explicitly. Safety is measured by the number and severity of accidents, by category, that were expected to occur within the highway section in a unit time. Keywords: accident prediction model, toll road safety, negative binomial model.
Item Type: | Thesis (PhD) |
---|---|
Subjects: | T Technology > TF Railroad engineering and operation |
Divisions: | School of Postgraduate (mixed) > Doctor Program in Civil Engineering |
ID Code: | 40860 |
Deposited By: | INVALID USER |
Deposited On: | 02 Dec 2013 10:04 |
Last Modified: | 02 Dec 2013 10:04 |
Repository Staff Only: item control page