Model Prediksi Cost Overrun Berbasis Genetic Programming dan Monte Carlo Simulation Pada Proyek KPBU Sektor Jalan Tol

Raditya, Akmal (2026) Model Prediksi Cost Overrun Berbasis Genetic Programming dan Monte Carlo Simulation Pada Proyek KPBU Sektor Jalan Tol. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6012231059-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Proyek jalan tol dengan skema Kerjasama Pemerintah dan Badan Usaha (KPBU) di Indonesia sering mengalami kelebihan biaya yang signifikan akibat kompleksitas aspek teknis, keuangan, dan regulasi. Studi ini bertujuan untuk mengembangkan model prediksi kelebihan biaya berbasis Genetic Programming (GP) yang terintegrasi dengan Monte Carlo Simulation (MCS) untuk memberikan prediksi probabilistik yang memperhitungkan ketidakpastian dalam faktor risiko. Data dikumpulkan dari 25 proyek jalan tol KPBU yang melibatkan 54 ahli proyek yang diwakili oleh kontraktor dan Kementrian PU. Lima faktor dominan teridentifikasi berdasarkan penilaian risiko yaitu, masalah pembebasan lahan, kondisi lapanga diluar perkiraan, kesulitan keuangan kontraktor, sering terjadi perubahan desain, dan perubahan peraturan/regulasi. Model GP dikembangkan menggunakan software GeneXproTools dengan input penilaian risiko dan data kelebihan biaya proyek aktual. Model regresi non-linier yang dihasilkan menunjukkan korelasi kuat antara faktor risiko dan kelebihan biaya dengan akurasi tinggi, R² = 0.946 untuk data validasi. Model ini juga menunjukkan kesalahan prediksi yang rendah, dengan RMSE = 0.025 dan MAE = 0.019. Persamaan yang dihasilkan dari GP kemudian digunakan sebagai fungsi deterministik dalam Simulasi Monte Carlo dengan 10.000 iterasi. Hasil simulasi menunjukkan rata-rata kemungkinan kejadian kelebihan biaya dari anggaran awal proyek sebesar 29,38%, dengan kelebihan biaya maksimum tidak akan melebihi 0.52% dengan tingkat kepercayaan 99%. Analisis sensitivitas (tornado chart) mengidentifikasi perubahan peraturan sebagai faktor paling berpengaruh, diikuti oleh perubahan desain, dan kondisi lapangan. Studi ini menyimpulkan bahwa pendekatan hibrida GP–MCS efektif dalam menangkap interaksi non-linier dan ketidakpastian dalam proyek jalan tol KPBU sehingga dapat dijadikan sebagai alat bantu pengambilan keputusan bagi pemangku kepentingan proyek infrastruktur, khususnya dalam penentuan alokasi kontinjensi biaya, perencanaan risiko keuangan, serta evaluasi kelayakan proyek jalan tol berbasis KPBU.
===================================================================================================================================
Toll road projects under the Public-Private Partnership (PPP) scheme in Indonesia often experience significant cost overruns due to the complexity of technical, financial, and regulatory aspects. This study aims to develop a cost overrun prediction model based on Genetic Programming (GP) integrated with Monte Carlo Simulation (MCS) to provide probabilistic predictions that take into account uncertainty in risk factors. Data was collected from 25 PPP toll road projects involving 54 project experts represented by contractors and the Ministry of Public Works. Five dominant factors were identified based on risk assessment, namely land acquisition issues, unexpected field conditions, contractor financial difficulties, frequent design changes, and changes in rules/regulations. The GP model was developed using GeneXproTools 2023.3 with risk assessment inputs and actual project cost overrun data. The resulting non-linear regression model showed a strong correlation between risk factors and cost overruns with high accuracy, R² = 0.946 for validation data. This model also showed low prediction errors, with RMSE = 0.025 and MAE = 0.019. The equation generated from GP was then used as a deterministic function in Monte Carlo Simulation with 10,000 iterations. The simulation results show an average probability of cost overruns from the initial project budget of 29.38%, with maximum cost overruns not exceeding 0.52% with a 99% confidence level. Sensitivity analysis (tornado chart) identifies regulatory changes as the most influential factor, followed by design changes and field conditions. This study concludes that the GP–MCS hybrid approach is effective in capturing non-linear interactions and uncertainties in PPP toll road projects so that it can be used as a decision-making tool for infrastructure project stakeholders, particularly in determining cost contingency allocations, financial risk planning, and evaluating the feasibility of PPP-based toll road projects.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Cost overrun, Genetic Programming, Monte Carlo Simulation, KPBU, Jalan tol, Manajemen Risiko Proyek Cost overrun, Genetic Programming, Monte Carlo Simulation, PPP, Toll Road, Project Risk Management
Subjects:	T Technology > T Technology (General) > T56.8 Project Management T Technology > TA Engineering (General). Civil engineering (General) T Technology > TE Highway engineering. Roads and pavements
Divisions:	Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Civil Engineering > 22101-(S2) Master Thesis
Depositing User:	Akmal Raditya
Date Deposited:	27 Jan 2026 02:43
Last Modified:	27 Jan 2026 02:43
URI:	http://repository.its.ac.id/id/eprint/130564

Actions (login required)

View Item