Hydrocarbon production in the reservoir depends on fluid flow through its porous media, such as fractures and their physical parameters, which affect the analysis of the reservoir’s physical properties. The fracture’s physical parameters can be measured conventionally by laboratory analysis or using numerical approaches such as simulations with the Lattice Boltzmann method. However, these methods are time-consuming and resource-intensive; therefore, this research explores the application of machine learning as an alternative method to predict the physical parameters of fractures such as permeability, surface roughness, and mean aperture. Synthetic three-dimensional digital fracture data that resemble real rock fractures were used to train the machine learning models. These included two convolutional neural networks (CNNs) designed and implemented in this research—which are referred to as CNN-1 and CNN-2—as well as three pre-trained models—including DenseNet201, VGG16, and Xception. The models were then evaluated using the ${\mathrm{R}}^2$ and mean absolute percentage error (MAPE). CNN-2 was the best model for accurately predicting the three fracture physical parameters but experienced a drop in performance when tested on real rock fractures.

Oil losses is a problem that often arises in oil and gas industries either in onshore or offshore area. There is a loss discrepancy between total quantities from shippers and measurement in the storage tanks; the total sending volume is lower than the measured volume in the mixing tank in a gathering station; this is known as oil losses. When this occurs, an agreement to determine a fair share of the losses must be made. There are two categories of oil losses, they are individual and group losses. Individual loss occurs when oil from one shipper has not been mixed yet with other oils. This includes emulsion and evaporative losses. Group loss occurs during mixing oils in the same storage tank or pipeline. Furthermore, by knowing the causes of oil losses, a way to minimize oil losses can be determined.

Hydraulic fracturing is one of the stimulation treatment in oil and gas well by creating a fractured through a proppant injection to the formation. A most critical problem in the actual oil and gas industry is that the fracturing engineers could not forecast approximately post-production performance after fracturing the job, which is a severe problem. This problem phenomenon has occurred in some cases and significantly impacts production such as oversizing or lower sizing of pumping rate setting. Integrated analysis for post job hydraulic fracturing production based on the geometry model iteration and Productivity Index (PI) comparison in the conventional oil sand reservoir is simply a method to analyze and forecast approximately incremental production performance. The fractured software generates a fractured geometry model that considers half-length of fractured parameters, width in front of perforation, average width, fractured height, and pressure net. Then we compare the Productivity Index's prediction value through the method of Cinco-Ley, Samaniego and Dominguez. A case study in the well of TM#2 (conventional oil sand reservoir) was conducted as the comprehensive study to provide the data and proceed analysis for production forecast. We found that the geometry model and iteration of PKN 2D method generated a small fractured geometry model compare to fracCADE software. The cooperation between PKN 2D method and Cinco-Ley, Samaniego, and Dominguez concept successfully predict post-production forecast. This concept could be proposed as a quick look measurement for production scenarios to overcome pump sizing.

One of the Enhanced Oil Recovery (EOR) strategies in the petroleum industry is CO2 injection using the huff and puff method. The method is performed on one well that acts as an injection and a production well. The method works by injecting a certain volume of carbon dioxide (CO2) gas into the reservoir and then closing the well for a period of time. This injection cycle can take place over several cycles. Production can be carried out after one or more cycles according to the design. In this study, CO2 injection optimization with the huff and puff method is carried out with reservoir simulation (GEM-CMG) by taking data from one of the oil and gas wells in Indonesia, with carbonate rock characteristics that are water wet. The simulation work steps include inputting data (fluid, rock properties, and production), initialization, history matching, and CO2 injection optimization with the huff and puff method. The optimization scenarios include optimization of injection pressure and number of cycles. The injection pressure scenario uses a range of 500 - 3000 psi, based on the simulation results obtained that the injection pressure of 500 psi produces the highest recovery factor (RF) of 22.2%. Then, the cyclic scenario was carried out at the optimum injection pressure (500 psi) with the number of cycles 2 - 6 cycles. From the simulation results, it is found that the number of cycles for this carbonate reservoir condition does not have a significant effect, as evidenced by the RF values ranging from 22.1 - 22.3%.

Carbon dioxide (CO2) gas injection is one of the carbon capture utilization and storage (CCUS) programs implemented in the Jatibarang oil field, not only for reducing the effects of emissions on the environment but also for increasing oil recovery. CO2 injection scenarios are simulated herein, considering carbonate reservoirs with a high degree of heterogeneity. By modeling the reservoir, the behavior of the reservoir fluid is observed, and the optimal injection scenario that produces the highest oil recovery is determined. CO2 injection involves several key parameters, including the injection pattern and injection pressure, which are varied to improve the performance and increase oil recovery. Injection pressures are assessed in the range of 500–3000 psi, and five patterns are considered: staggered drive, line drive, direct line, four spot, and five spot. From the simulations, a staggered drive pattern with an injection pressure of 2500 psi is optimal, reaching a recovery factor of 27.24%. The changes in the oil viscosity and oil saturation distribution before and after CO2 injection are also obtained.