Date of Publication


Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Chemical Engineering

Subject Categories

Chemical Engineering


Gokongwei College of Engineering


Chemical Engineering

Thesis Advisor

John Frederick D. Tapia
Raymond Girard R. Tan

Defense Panel Chair

Arnel B. Beltran

Defense Panel Member

Gian Paolo O. Bernardo
Kathleen B. Aviso


Carbon capture and storage (CCS) and negative emissions technologies (NETs) such as bioenergy with carbon capture and storage (BECCS) and direct air capture (DAC) systems are necessary for mitigating climate change and attaining net-zero carbon dioxide (CO2) emissions. CCS, BECCS, and DAC rely on geological reservoirs with minimal risk of leakage to permanently store CO2. Machine Learning (ML) algorithms have been implemented to characterize potential storage sites using geological data which can aid decision-makers in implementing large-scale CCS, BECCS, and DAC. In this study, a naive Bayes classifier is developed using published data of 76 natural CO2 reservoirs from geological surveys for identifying secure reservoirs in the form of interpretable rules. In a trial-and-error process, the data from literature is cleaned and discretized using a 1:1 and 2:1 split. The model is then trained and validated until an optimal model with an accuracy of 90-95% is attained. As a result, a set of IF-THEN rules are generated and can be readily interpreted by decision-makers. Despite the independence assumption (which is hardly true in reality), the Naive Bayes classifier is able to combine simplicity with efficiency and produce accurate predictions. In this work, the model using the 2:1 split has an accuracy of 92.00% with a false positive rate of 33.33% and false negative rate of 4.55%. The rules generated in this study suggest that secure CO2 storage depends on reservoir depths between 1000 and 2500 m, high density CO2, thick caprocks, and no presence of fault. At depths less than 1000 m, the buoyancy pressure exerted on the caprock is more likely to exceed the caprock capillary entry pressure and therefore act as a driving force for CO2 migration. Since CO2 density increases with increasing depth, a site with a greater reservoir depth holding higher CO2 densities is likely more secure. Reservoirs with thick caprocks and without faults can handle greater pressure gradients and prevent fluid movement, hence indicating successful CO2 retention. Adopting these criteria will likely increase the security of engineered CO2 sites.

Abstract Format







Carbon sequestration

Upload Full Text


Embargo Period


Available for download on Friday, January 05, 2024