Iranian Journal of Oil and Gas Science and Technology

Abstract Accurate pore pressure estimation is a critical component of geomechanical modeling, essential for maintaining wellbore stability and optimizing drilling fluid density. This study proposes a Grey Wolf Optimizer-Supported Vector Machine (GWO-SVM) workflow to predict pore pressure in a complex carbonate reservoir in southwestern Iran. While traditional empirical correlations like Eaton and Bowers are widely used, their reliance on continuous calibration can be challenging. To address this, discrete Repeating Formation Test (RFT) pressure points were utilized to calibrate baseline empirical trends, creating continuous reference profiles for the entire depth. The GWO-SVM model was then deployed to automate the replication of these calibrated baselines from standard petrophysical logs (sonic, density, resistivity, porosity, and shale volume). Using a dataset from five wells (four for training, one for blind validation), the GWO optimally tuned the Support Vector Regression (SVR) hyperparameters. The model demonstrated exceptional fidelity in replicating the calibrated baseline on the blind test well, achieving an RMSE=37.96psi and an R^2=0.998. Finally, the 1D well-based predictions were integrated into a 3D geostatistical model using co-kriging to visualize pressure compartmentalization influenced by the local tectonic stress regime. This workflow offers a replicable, high-precision alternative for pre-drill pore pressure modeling in data-limited reservoir settings.

Abstract The primary purpose of this paper is to estimate and evaluate the petrophysical properties of the Ghar formation in the Hendijan and Bahregansar oilfields using a combination of seismic and well-log data. This study follows a step-by-step regression approach. First, sonic, density, and porosity well-log data are collected. Second, seismic attributes, including amplitude, phase, frequency, and acoustic impedance, are extracted from the seismic lines intersecting the wellbore locations. Then, using the MFLN and PNN intelligent systems, a relationship between porosity, shale volume, saturation, and seismic attributes is established. Using this relationship, the physical and petrophysical properties of the reservoir in the Ghar formation are estimated and evaluated. We estimated the reservoir porosity to be between 15% and 20%, higher in the Hendijan oilfield than in the Bahregansar oilfield. The water saturation in the Ghar formation varied from 25% to 30%. On the other hand, the clay content and shale volume of the Ghar formation in the Hendijan field were higher than those of the Bahregansar oil field.

Abstract This study uses the kernel principal component analysis (KPCA) feature extraction method for facies classification to extract new features from the measured well logs. After applying the principal component analysis (PCA) and KPCA feature extraction approaches, the classification was made using three robust classifiers: multilayer perceptron neural network (MLP), support vector machine (SVM), and random forest (RF). Finally, the predicted results for the test data that were not included in the training process were evaluated with the F1 score criterion. The PCA method did not significantly affect the classification performance due to the nonlinear structure of the facies. Our results show that the KPCA improves the performance of facies classification. Compared with the conventional approach based on well-log data, our new approach improves the classification accuracy for each classifier algorithm. In the RF results, the classification accuracy has increased by about 6%, while using the KPCA feature extraction approach raises classification accuracy from 52% to 58% compared to the original well-log data.

Abstract The petrophysical parameters of the Ghar Formation are characterized in this study. A combination of pre-stack seismic data gathers and well-log data is used to estimate water saturation and shale volume in the Ghar reservoir. First, the highest possible correlation between the well logs and the seismic inversion data was established for this purpose. After extracting the optimal wavelet, an accurate relationship between the estimated values and the core data was obtained. Next, using data from another well, the validity of the constructed model was examined. The results showed that the combination of three attributes—instantaneous cosine of phase, , and —can accurately estimate the shale volume of the reservoir, with a correlation coefficient of approximately 70%. Although the two layers in the Ghar section have a shale volume of about 10%, the overall shale volume in the reservoir area is negligible. The logarithm of the ratio of compressional wave velocity to shear wave velocity attribute exhibits the highest correlation, approximately 62%. Finally, validation using previously unutilized well-log data demonstrated an accuracy of about 90% in predicting these properties.

Abstract This paper investigates the role of the adequate thickness of the Asmari reservoir formation zones on oil production in one of the Iranian carbonate oil fields. Adequate thickness is a term that includes the total gross thickness of rocks by lithofacies for a selected wellbore. The lithology of the Asmari formation in the studied area consists of dolomite, sandstone, lime, dolomitic-lime, sandstone-shale, and shale limestone dolomites. Based on the existing well-logs, the average shale volume, the effective arithmetic means of porosity in the gross intervals, and average water saturation or hydrocarbon-bearing increments of the studied field are calculated using well-logs. In wellbore #A, a depth interval of 2214 to 2296 m shows 9.6% average shale volume, 27.2% average water saturation, and 20.9% average porosity. A depth interval of 2213 to 2280 m, in wellbore #B, shows 6% average shale volume, 21.25% average water saturation, and 28.5% average porosity. Based on our petrophysical assessments, we divide the Asmari reservoir in the studied field into eight zones. Zone 1 is made of carbonate (calcareous and dolomitic), and zones 2–5 are mainly sandstone; zones 7 and 8 are calcareous and shale, and zone 6 is a mixture of all the rocks mentioned above. Among these eight zones, there are two primary hydrocarbon productive zones. The numerical calculation of in situ oil volume showed that zone 2 contains 65% of oil volume in this reservoir. With more than 80% sand, this zone has the highest net hydrocarbon column.

Abstract Seismic well tying is a crucial part of the interpretation phase in exploration seismology. Tying wells usually involves forward modeling a synthetic seismogram from sonic and density logs and then matching the obtained synthetic seismogram to the seismic reflection data. A huge amount of time is required to deal with it, yet the outcome signal may not be satisfying and may be suffering a low cross correlation between the seismic signal and the synthetic one; it also requires a high quality synthetic trace. Another problem with the so-called manual tying is that the tying process is not repeatable, indicating that one can rarely obtain the same stretched and squeezed signal if the tying procedure is repeated. In recent years, some researchers have used the dynamic time warping (DTW) method to address well tying problems. They have obtained good results according to the correlation between the seismic signal and the warped synthetic signal. This research demonstrates that the result will be better if filtering is applied before tying, and then the warped signal is smoothed. We also propose a simpler algorithm for extracting a warped signal from the warping curve and the original synthetic trace, which gives rise to better performance for well tying.

Abstract It is difficult to identify the carbonate reservoirs by using conventional seismic reflection data, especially in cases where the reflection coefficient of the gas-bearing zone is close to that of the carbonate background. In such cases, the extended elastic impedance (EEI) as a seismic reconnaissance attribute with the ability to predict fluids and lithology can be used. It allows for a better distinction between seismic anomaly caused by lithology and the one caused by the fluid content. The EEI attribute extends the available reflection angles and applies different weights to the intercept and gradient values so as to extract the petrophysical properties of the rock at a specific incident angle. Using the EEI attribute, we can estimate the elastic parameters such as shear impedance; the ratio of the compressional velocity to shear velocity; Poisson’s ratio; and bulk, Lame, and shear moduli, and petrophysical properties, including porosity, clay content, and water saturation. The known reservoirs in the study area are three oil-bearing formations namely, Surmeh (Arab), Gadvan (Buwaib), and Dariyan (Shuaiba), and three gas-bearing formations, including Kangan, Dalan, and Faraghan. The Dehram group is composed of Kangan (Triassic), Dalan, and Faraghan (Permian) formations. Permian carbonates of Kangan–Dalan and its equivalent Khuff have regionally been developed as a thick carbonate sequence in the southern Persian Gulf region. In this paper, parameters 𝜆𝑝 and 𝜇𝜌 extracted from the EEI method are used to characterize a carbonate reservoir. Our results show that the EEI can highlight the difference between the reservoir and non-reservoir formation to identify the gas-bearing areas.

Abstract Random or incoherent noise is an important type of seismic noise, which can seriously affect the quality of the data. Therefore, decreasing the level of this category of noises is necessary for increasing the signal-to-noise ratio (SNR) of seismic records. Random noises and other events overlap each other in time domain, which makes it difficult to attenuate them from seismic records. In this research, a new technique is produced, by joining FX deconvolution (FXD) and a special kind of median filter in order to suppress random noise from seismic records. The technique is operated in some stages; firstly, FXD is tried to eliminate the Gaussian noise, and the median filter is fixed to diminish the spike-like noise. The synthetic dataset and field data examples (from an oil field in the southwest of Iran) have been employed to demonstrate that random noise reduction can be attained, while the signal content will not be destroyed considerably. The final results indicate the authority of the proposed strategy in suppressing random noises, whereas signal information is almost protected during the filtering.

Abstract Intelligent reservoir characterization using seismic attributes and hydraulic flow units has a vital role in the description of oil and gas traps. The predicted model allows an accurate understanding of the reservoir quality, especially at the un-cored well location. This study was conducted in two major steps. In the first step, the survey compared different intelligent techniques to discover an optimum relationship between well logs and seismic data. For this purpose, three intelligent systems, including probabilistic neural network (PNN),fuzzy logic (FL), and adaptive neuro-fuzzy inference systems (ANFIS)were usedto predict flow zone index (FZI). Well derived FZI logs from three wells were employed to estimate intelligent models in the Arab (Surmeh) reservoir. The validation of the produced models was examined by another well. Optimal seismic attributes for the estimation of FZI include acoustic impedance, integrated absolute amplitude, and average frequency. The results revealed that the ANFIS method performed better than the other systems and showed a remarkable reduction in the measured errors. In the second part of the study, the FZI 3D model was created by using the ANFIS system.The integrated approach introduced in the current survey illustrated that the extracted flow units from intelligent models compromise well with well-logs. Based on the results obtained, the intelligent systems are powerful techniques to predict flow units from seismic data (seismic attributes) for distant well location. Finally, it was shown that ANFIS method was efficient in highlighting high and low-quality flow units in the Arab (Surmeh) reservoir, the Iranian offshore gas field.

Abstract The prediction of lithology is necessary in all areas of petroleum engineering. This means that to design a project in any branch of petroleum engineering, the lithology must be well known. Support vector machines (SVM’s) use an analytical approach to classification based on statistical learning theory, the principles of structural risk minimization, and empirical risk minimization. In this research, SVM classification method is used for lithology prediction from petrophysical well logs based on petrographic studies of core lithology in a heterogeneous carbonate reservoir in southwestern Iran. Data preparation including normalization and attribute selection was performed on the data. Well by well data separation technique was used for data partitioning so that the instances of each well were predicted against training the SVM with the other wells. The effect of different kernel functions on the SVM performance was deliberated. The results showed that the SVM performance in the lithology prediction of wells by applying well by well data partitioning technique is good, and that in two data separation cases, radial basis function (RBF) kernel gives a higher lithology misclassification rate compared with polynomial and normalized polynomial kernels. Moreover, the lithology misclassification rate associated with RBF kernel increases with an increasing training set size.

Abstract Seismic inversion is a method that extracts acoustic impedance data from the seismic traces. Source wavelets are band-limited, and thus seismic traces do not contain low and high frequency information. Therefore, there is a serious problem when the deterministic seismic inversion is applied to real data and the result of deterministic inversion is smooth. Low frequency component is obtained from well log data; however, but when well log and seismic data are used together, it faces a problem which is a function of the support of scale of measurements. Well log data have a high vertical resolution while seismic data represent low details in vertical direction. Geostatistical seismic inversion (GSI) is a method to overcome the aforementioned limitations. GSI uses well log and seismic data together in the geostatistical frameworks. In this study, a new approach of geostatistical inversion based on spectral geostatistical simulation is used. This approach is performed in frequency domain and stochastic framework. Distinct from sequential simulation, spectral simulation method is a direct method, which does not require an acceptance/rejection step. Hence, GSI algorithm based on spectral simulation is fast. This approach is performed in a case study of an Iranian gas field in the Persian Gulf basin. The upper-Dalan and Kangan are two main formations of this field. The results of GSI are compared with deterministic inversion and it is concluded that, as opposed to deterministic inversion, GSI can recover low frequency components.

Abstract Seismic facies analysis (SFA) aims to classify similar seismic traces based on amplitude, phase, frequency, and other seismic attributes. SFA has proven useful in interpreting seismic data, allowing significant information on subsurface geological structures to be extracted. While facies analysis has been widely investigated through unsupervised-classification-based studies, there are few cases associated with supervised classification methods. In this study, we follow supervised classification scheme under classifiers, the support vector classifier (SVC), and multilayer perceptrons (MLP) to provide an opportunity for directly assessing the feasibility of different classifiers. Before choosing classifier, we evaluate extracted seismic attributes using forward feature selection (FFS) and backward feature selection (BFS) methods for logical SFA. The analyses are examined with data from an oil field in Iran, and the results are discussed in detail. The numerical relative errors associated with these two classifiers as a proxy for the robustness of SFA confirm reliable interpretations. The higher performance of SVC comparing to MLP classifier for SFA is proved in two validation steps. The results also demonstrate the power and flexibility of SVC compared with MLP for SFA.

Abstract Reservoir characterization plays an important role in different parts of an industrial project. The results from a reservoir characterization study give insight into rock and fluid properties which can optimize the choice of drilling locations and reduce risk and uncertainty. Delineating hydrocarbon bearing zones within a reservoir is the main objective of any seismic reservoir characterization study. In the current study, using limited well control and seismic data, an attempt was made to predict the productive zones of a reservoir using elastic impedance inversion. Elastic impedance logs at near and far angles of incidence have been crossplotted to find the desired productive parts of the formation. Two partial angle stack seismic data have been inverted using a model-based post-stack seismic inversion. The crossplot of the two inverted volumes is interpreted based on the results from the well location. Finally, the hydrocarbon bearing zones of the reservoir was delineated according to the seismic crossplot analysis.