Integrasi Inversi Probabilistik Dan Pemodelan Rock Physics Secara Statistik Berbasis Gaussian Mixture Model (GMM) Untuk Karakterisasi Reservoir Formasi Plover, Lapangan Poseidon, Northwest Shelf Australia

Prasetyo, Maulana Wahyu (2026) Integrasi Inversi Probabilistik Dan Pemodelan Rock Physics Secara Statistik Berbasis Gaussian Mixture Model (GMM) Untuk Karakterisasi Reservoir Formasi Plover, Lapangan Poseidon, Northwest Shelf Australia. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5017221041-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (17MB) | Request a copy

Abstract

Karakterisasi reservoir gas Formasi Plover di Lapangan Poseidon, Browse Basin, menghadapi ambiguitas litofluida yang signifikan akibat tumpang-tindih respons elastik antara shale, wet sand, dan gas sand pada zona transisi fluvial-deltais bertekanan tinggi di kedalaman lebih dari 4.600 m. Pendekatan inversi simultan deterministik yang mengasumsikan distribusi prior Gaussian unimodal tidak mampu merepresentasikan karakter multimodal litofasies sehingga menghasilkan pemulusan berlebih dan kuantifikasi ketidakpastian yang lemah. Penelitian ini mengimplementasikan kerangka inversi petrofisika probabilistik Bayesian–Gaussian Mixture Model (GMM) Grana dan Della Rossa (2010) secara penuh pada data Sumur Poseidon-1 dan Boreas-1 serta tiga partial-angle stack seismik pada sudut datang 12°, 24°, dan 36°. Alur kerja mencakup pemodelan rock physics berbasis Hertz–Mindlin, batas Hashin–Shtrikman, semen kontak Dvorkin–Nur, dan substitusi fluida Gassmann; pelatihan GMM enam dimensi tiga komponen pada ruang gabungan elastik–petrofisika; pembangunan model frekuensi rendah melalui ordinary kriging horizon-guided dengan pseudo-wells; dan inversi probabilistik linearized AVO dengan 50.000 realisasi Monte Carlo per trace. Hasil substitusi Gassmann memetakan gas sand pada Vp/Vs ≈ 1,51 ± 0,03 dan wet sand pada Vp/Vs ≈ 2,05 ± 0,15, sementara GMM mencapai akurasi klasifikasi 90% dengan macro F1-score 0,89 dan separabilitas LDA 95,6% terhadap label rule-based. Distribusi posterior petrofisika menunjukkan anomali porositas efektif 0,10–0,13, kandungan lempung rendah, dan saturasi air 0,20–0,40 yang terkonsentrasi di interval Plover Reservoir, dengan probabilitas hydrocarbon sand melampaui 70% pada zona pay yang menerus lateral antara kedua sumur. Penelitian ini menegaskan bahwa kerangka Bayesian–GMM secara kuantitatif mereduksi ambiguitas litofluida dan mendelineasi prospek hidrokarbon dengan kuantifikasi ketidakpastian yang konsisten secara stratigrafi.
================================================================================================================================================
Reservoir characterization of the Plover Formation in the Poseidon Field, Browse Basin, is challenged by substantial litho-fluid ambiguity arising from the overlapping elastic responses of shale, wet sand, and gas sand within transitional, high-pressure fluvial-deltaic intervals located below 4,600 m. Deterministic simultaneous inversion approaches, which assume unimodal Gaussian prior distributions, fail to capture the multimodal nature of the underlying lithofacies and consequently yield over-smoothed solutions with limited uncertainty quantification. This study implements the full probabilistic petrophysical inversion framework of Grana and Della Rossa (2010) on log data from the Poseidon-1 and Boreas-1 wells, together with three partial-angle seismic stacks at 12°, 24°, and 36° incidence angles. The workflow integrates rock physics modeling via Hertz–Mindlin contact theory, Hashin–Shtrikman bounds, Dvorkin–Nur contact cement, and Gassmann fluid substitution; six-dimensional three-component GMM training in the joint elastic–petrophysical feature space; low-frequency model construction through horizon-guided ordinary kriging supported by pseudo-wells; and Bayesian linearized AVO inversion coupled with 50.000 Monte Carlo realizations per trace. Gassmann substitution maps gas sand to Vp/Vs ≈ 1.51 ± 0.03 and wet sand to Vp/Vs ≈ 2.05 ± 0.15, while the trained GMM achieves 90% classification accuracy with a macro F1-score of 0.89 and an LDA separability of 95.6% relative to rule-based labels. Posterior petrophysical distributions reveal effective porosity anomalies of 0.10–0.13, low shale volume, and water saturation of 0.20–0.40 concentrated within the Plover Reservoir interval, with hydrocarbon sand probabilities exceeding 70% across a laterally continuous pay zone connecting both wells. These results demonstrate that the Bayesian–GMM framework quantitatively reduces litho-fluid ambiguity and delineates hydrocarbon prospects with stratigraphically consistent uncertainty quantification.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Inversi Probabilistik Bayesian, Gaussian Mixture Model, Rock Physics Statistik, Karakterisasi Reservoir, Bayesian Probabilistic Inversion, Gaussian Mixture Model, Statistical Rock Physics, Reservoir Characterization.
Subjects:	Q Science > QE Geology
Divisions:	Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geophysics Engineering > 33201-(S1) Undergraduate Thesis
Depositing User:	Maulana Wahyu Prasetyo
Date Deposited:	17 Jul 2026 06:40
Last Modified:	17 Jul 2026 06:40
URI:	http://repository.its.ac.id/id/eprint/135323

Actions (login required)

View Item