IMPACT OF SEVERAL ROUTE CHOICE MODELS ON THE ACCURACY OF ESTIMATED O-D MATRICES FROM TRAFFIC COUNTS
S U M M A R Y IMPACT OF SEVERAL ROUTE CHOICE MODELS ON THE ACCURACY OF ESTIMATED O-D MATRICES FROM TRAFFIC COUNTS, WIRADAT ANINDITO, 2000, Program Magister Sistem dan Teknik Jalan Raya, Program Pascasarjana, Institut Teknologi Bandung. Travel is part of human life in order to fulfill their basic requirements that are not available at their original place. In any particular study area, this necessity to travel will create problem. The effort to solve the problem is to understand the pattern of people or goods movement that is called the Origin-Destination Matrix. There are two methods to obtain the matrix, conventional and unconventional. The conventional method is the most commonly used. The unconventional method is another method in obtaining origin-destination matrix using traffic count data that need relatively lower cost than the conventional method with undoubted accuracy level. This thesis describes the effect of factors such as delay in intersection, one-way road, turning banned and route choice model in building an origin-destination matrix based on traffic counts. In addition, this thesis also discusses the difference result of estimated 0-D matrices that were built using the Gravity Opportunity Model (GO Model) and Maximum Entropy Matrix Estimation Model (ME2 Model). For the Bandung area case study, the following conclusions were reached: (1) The influence of delay in intersections was more important than route choice model and turning banned. (2) The comparison between estimated O-D matrix were built by equilibrium assignment on network exclude turning banned and include turning banned, for GO Model, value of RMSE=0.196 R2=0.999 and ST=1.18%, and for ME2 Model, value of RMSE=2.799, R2=0.914 and ST=11.43%. (3) The estimated 0-D matrix were produced by ME2 model and Burell assignment on complete had a good fit with the estimated O-D matrix was built by GO Model and equilibrium assignment on complete network than other estimated were built by ME2 Model, value of RMSE=11.085, R2=0.105 and ST=127.57%. (4) The similarity between result of estimated flow and observed traffic counts, for GO Model, the estimated 0-D matrix was produced by equilibrium assignment on complete network have value of Sg=63.21%, for ME2 Model, the estimated 0-D matrix was produced by equilibrium assignment on network without intersection delay and turning banned have value of Sg=9.10%
v
S U M M A R Y IMPACT OF SEVERAL ROUTE CHOICE MODELS ON THE ACCURACY OF ESTIMATED O-D MATRICES FROM TRAFFIC COUNTS, WIRADAT ANINDITO, 2000, Program Magister Sistem dan Teknik Jalan Raya, Program Pascasarjana, Institut Teknologi Bandung. Because one's basic requirements are often not available at the place of origin, travel is a necessary part of human life. In any particular study area, this necessity to travel will create problems. The effort to solve the problem is to understand the pattern of people or goods movement that is called the Origin-Destination Matrix. There are two methods to obtain the matrix, conventional and unconventional. The conventional method is the most commonly used. The unconventional method is another method of obtaining the origin-destination matrix using traffic count data at relatively lower cost than the conventional method but with improved accuracy level. This thesis describes the effect of factors such as delay in intersection, one-way road, turning banned and route choice model in building an origin-destination matrix based on traffic counts. In addition, this thesis also discusses the difference in the estimated 0-D matrices that were built using the Gravity Opportunity Model (GO Model) and Maximum Entropy Matrix Estimation Model (ME2 Model). For the Bandung area case study, the following conclusions were reached: (1) The influence of delay at intersections was more important than route choice model and turning banned. (2) The comparison between the estimated O-D matrix built by equilibrium assignment on a network excluding banned turns and including banned turns, is as follows: for GO Model, value of RMSE=0.196 R2=0.999 and ST=1.18%, and for ME2 Model, value of RMSE=2.799, R2=0. 914 and ST=11.43%. (3) The estimated 0-D matrix produced by ME2 model and Burell assignment on the complete network had a good fit with the estimated 0-D matrix built by GO Model and equilibrium assignment on complete network than other estimated values. For ME2 Model, the values of RMSE=11.085, R2=0.105 and ST=127.57%. (4) The similarity between the results of estimated flow and observed traffic counts, for GO Model, the estimated O-D matrix was produced by equilibrium assignment on the complete network. The parameter value is Sg=63.21%, for ME2 Model, the estimated 0-D matrix was produced by equilibrium assignment on network without intersection delay and banned turn has value of Sg=9.10%
v
R I N G K A S A N PENGARUH MODEL PEMILIHAN RUTE TERHADAP AKURASI MATRIKSASAL-TUJUAN YANG DIPEROLEH DARI INFORMASI DATA ARUS LALU LINTAS, WIRADAT ANINDITO, 2000, Program Magister Sistem dan Teknik Jalan Raya, Program Pascasarjana, Institut Teknologi Bandung. Perjalanan merupakan bagian dari kehidupan manusia dalam usaha memenuhi kebutuhan hidup yang tidak tersedia di tempat mereka berada. Pada tingkat tertentu, kebutuhan perjalanan ini akan menimbulkan masalah. Usaha untuk mengatasi masalah tersebut adalah memahami pola pergerakan orang atau barang yang dikenal dengan nama Matriks-AsalTujuan (MAT). Metoda tidak konvensional merupakan suatu metoda untuk mendapat matrik asal tujuan yaitu dengan menggunakan data arus lalu lintas, dimana membutuhkan biaya relatif lebih murah dibandingkan dengan metoda konvensional dan tingkat keandalannyapun tidak diragukan lagi. Tesis ini membahas seberapa besar pengaruh tundaan di persimpangan, manajemen satu arah, larangan belok dan bermacam model pemilihan rute dalam membentuk MAT berdasarkan informasi arus lalu lintas. Selain itu, tesis ini juga mencoba memahami perbedaan hasil MAT yang diperoleh dari Model Gravity Opportunity (Model GO) dan Model Estimasi Matriks Entropi Maksimum (Model EMEM). Untuk kasus Kota Bandung, beberapa kesimpulannya yaitu: (1) Tundaan di persimpangan lebih berpengaruh dalam mengestimasi MAT dibandingkan model pemilihan rute dan larangan belok. (2) Perbandingan antara hasil estimasi MAT yang dibentuk menggunakan model keseimbangan pada jaringan jalan yang mengabaikan larangan belok dengan yang menyertakan larangan belok tidak terlalu jauh berbeda, baik untuk Model GO (RMSE=0,196, R2=0,999 dan ST=1,18%), maupun Model EMEM,(RMSE=2,799, R2=0,914 dan ST=11,43%). (3) Untuk estimasi MAT yang dibentuk dari Model EMEM, estimasi dengan model Burell pada jaringan jalan yang lengkap mempunyai bentuk yang mirip dengan estimasi matrik yang dibetuk dengan Model GO dengan model keseimbangan pada jaringan jalan lengkap (RMSE=11,085, R2=0,105 dan ST=127,57%). (4) Dengan membandingkan estimasi hasil pembebanan dengan arus dari pengamatan, untuk Model GO, estimasi matrik yang dibentuk dengan model keseimbangan pada jaringan jalan lengkap lebih mirip dengan nilai Sg=63,21%, untuk Model EMEM, estimasi matrik yang dibentuk dengan model keseimbangan pada jaringan jalan yang tidak menyertakan tundaan di persimpangan dan larangan belok lebih mirip dengan nilai Sg=9,10%. vi
