A pre-exploration well was drilled in the Xihu Sag of East China Sea basin, and commercial oil and gas flow had been achieved. But the oil and gas bearing trap had a big depth with low closure height and small area. The resolution of seismic data acquired by towed streamer is low, so it's difficult to obtain seismic velocity precisely. There were great risk and uncertainty in description of the trap and distribution of gas-bearing sandstone, reservoir prediction of sweet spot, direct hydrocarbon indication, and reserves assessment.
In consideration of the drilling platform on the trap, seismic acquisition technique of walkaway VSP and walk around VSP were introduced, meanwhile some innovative methods in source, receivers and geometry were applied. Twenty three-component hydrophones were composed as signal receivers which had a sample interval of ten meters in the well, two straight shot lines and two loop shot lines were designed around the drilling platform. Besides, volume and depth of air gun array were optimized, and the sailing route of seismic source vessel was planned properly in order to improve the efficiency of collecting work.
The collecting work of walkaway VSP and walk around VSP was accomplished efficiently, and more than seventy kilometers VSP seismic data was achieved. Afterwards, the new data was processed finely in company with zero offset VSP data, so high resolution VSP profiles and accurate seismic velocity were obtained. Reprocess to original seismic data acquired by towed streamer was implemented on the basis of walkaway VSP and walk around VSP data. The quality of normal seismic data was improved through reprocess constrained by walkaway VSP data, and S/N and resolution were much higher than old data. So it would be credible to research the distribution of gas-bearing sandstone and direct hydrocarbon indication using the reprocessed seismic data.
It was the first time to use joint acquisition technique of walkaway VSP and walk around VSP in offshore China which was an important breakthrough. High resolution VSP seismic profiles and precise seismic velocity could be acquired, and the data was important basis for refined evaluation of pre-exploration targets. It's very necessary to popularize and utilize these new techniques further.
Lee, Wei Yi (Centre of Subsurface Seismic Imaging, CSI, Universiti Teknologi PETRONAS) | Hamidi, Rosita (Centre of Subsurface Seismic Imaging, CSI, Universiti Teknologi PETRONAS) | Ghosh, Deva (Centre of Subsurface Seismic Imaging, CSI, Universiti Teknologi PETRONAS) | Musa, Mohd Hafiz (Centre of Subsurface Seismic Imaging, CSI, Universiti Teknologi PETRONAS)
Noise is the unwanted energy in a seismic trace opposed to the signals corresponding to reflected energy from the subsurface features. Since it can overlap with the main signals' energy and conceal the geological information, noise attenuation is one of the most important steps in seismic data processing. The most common method is frequency filtering. However, due to its limitations on separating the noise from signals, this method usually results in hurting the signal. Hence, it is important to develop an alternative method that can attenuate the noise without affecting the signal. Filters based on time-frequency analysis of the data can have a better separation of the noise from signal as they maintain the time localization of events while presenting their frequency content simultaneously. One of the recent approaches to time-frequency analysis of signals is the Empirical Wavelet Transform (EWT) which provides adaptive wavelet filter bank for signal analysis. In this paper, a filter is designed based on EWT for random noise attenuation and is applied on both synthetic and real data.
Bao, Yi (Exploration and Development Research Institute of Daqing Oilfield Company Ltd. CNPC.) | Wang, Cheng (Exploration and Development Research Institute of Daqing Oilfield Company Ltd. CNPC.) | Chen, Shu-min (Exploration and Development Research Institute of Daqing Oilfield Company Ltd. CNPC.) | Wang, Jian-min (Exploration and Development Research Institute of Daqing Oilfield Company Ltd. CNPC.) | Chen, Zhi-de (Exploration and Development Research Institute of Daqing Oilfield Company Ltd. CNPC.) | Pei, Jiang-yun (Exploration and Development Research Institute of Daqing Oilfield Company Ltd. CNPC.) | Wu, Jia-yi (Exploration and Development Research Institute of Daqing Oilfield Company Ltd. CNPC.)
The existing seismic data in the deep layer of Songliao Basin have low vertical resolution, poor imaging accuracy and weak ability to depict anisotropy of seismic data, so the seismic data can not meet the geological requirements of fine target characterization. In order to identify thinner reservoirs and smaller faults in deep and complex structural areas, to complete sequence subdivision on volcanic rocks of Yingcheng formation and the dense sand of Shahezi formation, BWH seismic data acquisition is deployed in Anda sag. Aiming at the characteristics of clear description of BWH wave field but low SNR. A Full-frequency Fidelity and Amplitude preserving processing Technology (Flow) supported by surface consistent time-varying pulse deconvolution and viscoelastic medium prestack time migration and depth migration techniques is formed. Compared with the old data, the target band width of the BWH data has been widened by 15 Hz, the imaging quality of the complex structural area is improved obviously, and the prediction coincidence rate of the thin sand body over 8m is increased by more than 10 percentage points. The thin interbed is developed in continental sedimentary basin, and the horizontal heterogeneity is serious. The BWH acquisition and full frequency amplitude preserving processing technology will bring a new solution for fine target exploration in deep and complex structural area of Continental Sedimentary basin.
The present paper is concerned with an experimental study of the acoustic signature of phase inversion in an oil-water mixture system. The system studied was used to correlate the process of phase inversion with the acoustic field generated during the two fluid mixing. The experimental results revealed that the relation between the acoustic fields produced by a water continuous dispersion and the phase inversion has a clear and different signature from an oil-continuous system using a batch mixing system. This dynamical characteristic of the phase inversion phenomenon could be of use in practical systems to detect phase inversion when it occurs based on the acoustic field measured in the subject process.
I examine the basis of slow convergence of tomographic full waveform inversion (TFWI) and discover that the reason behind it is the unbalanced effects of amplitudes and phase in the design of the regularization term. This imbalance results in a strong reliance of the kinematic updates on the amplitude fitting, slowing down the convergence. To mitigate the problem I propose two modifications to the tomographic inversion. First, by modifying the regularization term to focus more on the phase information, and second, simultaneously updating the source function for modeling. The adjustments reduce the gradient artifacts and allow for explicit control over the amplitudes and phases of the residuals.
Tomographic full waveform inversion (
The modeling operator is able to match the observed data by extending the velocity model with the proper axis, no matter what the accuracy of the initial model is, by using kinematic information from the extended axis with disregard to the occurrence of cycle skipping. The inversion is set up to extract all the essential information from the virtual axes and smoothly fold them back into their original, nonextended form of the model. The kinematic and dynamic information of the data were successfully inverted with exceptional robustness and precision.
Even though cycle-skipping is not an issue with TFWI, this method creates its own challenges, which are; its high computational cost and the big number of iterations that it needs (
Two adjustments to TFWI are proposed to reduce the slow convergence and allow for more control of the ratio between amplitude and phase. These adjustments are consistent in the framework of TFWI and allow for an accurate calculation of the gradient in the data space. The adjustments were tested and resulted in a reduction in the kinematic artifacts in the gradient.
In this paper we focus on electrical-submersible-pump (ESP) failure caused by scale buildup. Weak fluctuations recorded in the motor current signals several weeks before a failure indicate a change in the motor load. Advanced signal analysis of the motor current data reveals the presence of a dynamic characteristic in the ESP signal during rapid scale buildup in the pump stages. On the basis of the raw data from the motor current draw, a dynamic cascade can be identified in the current marked with the superimposition of several characteristic frequencies added over time that develop into a chaotic trend. Our analysis was conducted with different signal-processing tools, such as Fourier transform, wavelet transform, and chaotic attractors, which described the nature of the scale signature in the current logs. This analysis was the first step toward developing a real-time diagnostic tool for predicting ESP failures.
Due to the shift from conventional reservoirs towards unconventional, ultra-low permeability reservoirs in the last decade, Diagnostic Fracture Injection Test (DFIT) has become one of the dominant and economically practical pressure transient tests. It is crucial to analyze and interpret DFIT data correctly to obtain essential fracture design and reservoir parameters. This study presents the application of wavelet analysis to DFIT falloff pressure data to determine fracture closure pressure and time, to ultimately improve the overall efficiency of hydraulic fracturing designs.
In this study, DFIT pressure is treated as a non-stationary signal and analyzed by one of the signal processing techniques which is wavelet transformation. The purpose of signal analysis is to extract relevant information from a signal by transforming it. Firstly, the signal is transformed into wavelet domain by Discrete Wavelet Transformation (DWT) to calculate high-frequency wavelet coefficients (details), then change-point detection technique is applied to distinguish major changes within the coefficients trend to determine fracture closure pressure and time.
DFIT pressure decline data from different wells were analyzed by wavelet transformation. Detail coefficient demonstrates different patterns depending on the formation analyzed and near wellbore activities. This is expected because wavelet analysis is sensitive to any physical changes within the system. From the amplitude changes of the coefficients, wavelet tool demonstrates the fracture closure as a continuing process.
Because wavelet is sensitive to changes in the system, it detects the fracture closure unambiguously by amplitude change, as compared to slope changes in other conventional methodologies. A comparison with some of the most commonly used diagnostic techniques, conventional log-log diagnostic plot, square root time, G-function and its derivative analysis are also provided in this study.
There have been several publications discussing various techniques analyzing DFIT pressure decline in unconventional formations and yet there is relatively high uncertainty in before-closure-analysis. However, this methodology is more sensitive to fundamental changes in the system, so application in detecting closure pressure and time decreases the uncertainty compared to other conventional tangential methodologies.
In this paper, we use a combination of acoustic impedance and production data for history matching the full Norne Field. The purpose of the paper is to illustrate a robust and flexible work flow for assisted history matching of large data sets. We apply an iterative ensemble-based smoother, and the traditional approach for assisted history matching is extended to include updates of additional parameters representing rock clay content, which has a significant effect on seismic data. Further, for seismic data it is a challenge to properly specify the measurement noise, because the noise level and spatial correlation between measurement noise are unknown. For this purpose, we apply a method based on image denoising for estimating the spatially correlated (colored) noise level in the data. For the best possible evaluation of the workflow performance, all data are synthetically generated in this study. We assimilate production data and seismic data sequentially. First, the production data are assimilated using traditional distance-based localization, and the resulting ensemble of reservoir models is then used when assimilating seismic data. This procedure is suitable for real field applications, because production data are usually available before seismic data. If both production data and seismic data are assimilated simultaneously, the high number of seismic data might dominate the overall history-matching performance.
The noise estimation for seismic data involves transforming the observations to a discrete wavelet domain. However, the resulting data do not have a clear spatial position, and the traditional distance-based localization schemes used to avoid spurious correlations and underestimated uncertainty (because of limited ensemble size), are not possible to apply. Instead, we use a localization scheme that is based on correlations between observations and parameters that does not rely on physical position for model variables or data. This method automatically adapts to each observation and iteration.
The results show that we reduce data mismatch for both production and seismic data, and that the use of seismic data reduces estimation errors for porosity, permeability, and net-to-gross ratio (NTG). Such improvements can provide useful information for reservoir management and planning for additional drainage strategies.
Zhu, Liping (China University of Petroleum-Beijing) | Li, Hongqi (China University of Petroleum-Beijing) | Yang, Zhongguo (China University of Petroleum-Beijing) | Li, Chengyang (China University of Petroleum-Beijing) | Ao, Yile (China University of Petroleum-Beijing)
Lithology interpretation is a key component of well-log interpretation, which can be viewed as a supervised classification problem from the perspective of machine learning. Recently, various machine-learning algorithms have been applied for borehole lithology interpretation as an alternative way. Convolution neural network (CNN) is a class of deep, feed-forward artificial neural networks, which has been applied to visual imagery analysis successfully. As one of the most popular and effective deep learning structures, CNN has been widely applied in computer vision problems, image target detection, recommendation systems, and natural language processing. However, the machine-application of CNN on logging-curve-based reservoir evaluation bas not received nearly as much attention as expected. The main reason is that CNN is designed to accept input in the form of multilayer images, meanwhile, it's hard to construct the nonsequential curve values into image-style input. In this article, we propose a wavelet-decomposition-based method to construct multilayer image-style input for each logging point, which makes it possible to convert the problem of logging lithological interpretation into a supervised image-recognition task. The proposed method is applied to the wells of the Daqing Oilfield and achieves excellent application effect.
We present a case study that demonstrates the use of our robust Seismic-Well Tie (SWT) process and seismic attributes to validate the added resolution from Seismic Spectral Blueing (SSB) on the carbonate Mishrif reservoir in the Rumaila oil field. Our SWT process included Vertical Seismic Profile (VSP) corridor stack traces and Reflection Coefficient Modelling (RCM). Seismic attributes generated following the interpretation of the SSB data, revealed geological features that weren’t previously visible on the full- stack seismic. All of these provide validation that the extra wiggle from the SSB is real in this case study.
SSB outputs bandlimited reflectivity traces derived from shaping the amplitude spectra of the input seismic to that of the well log-based reflectivity series. SSB adds seismic bandwidth to the full-stack data that is expressed as an extra trough within the Mishrif reservoir in certain parts of the field. Three-way SWTs, achieved by including a VSP corridor stack trace to a more conventional tie between well log synthetic and seismic trace, is typically seen as a thorough approach. It can help to reinforce confidence in seismic events observed in all three data types and to highlight events or intervals where well logs or seismic may contain significant anomalous data. Three-way SWTs tying full-stack synthetic, full-stack seismic and 8-12-30-45 Hz VSP corridor stack traces, as well as SSB synthetic, SSB and 8-12-50-75 Hz VSP corridor stack traces are of good-quality, with a comparable extra trough also identified on the broader bandwidth VSP corridor stack trace.
Reflection Coefficient Modelling (RCM), a part of the SWT process, is a way of deconstructing a synthetic seismic trace by looking at the intermediate step in wavelet convolution to isolate the contributions of individual Reflectivity Coefficient (RC) contrasts to the resulting seismic event, often referred to as a ‘wiggle’. RCM suggests that the extra trough observed on the SSB data is associated with the development of a rudist-dominated grainstone shoal body. VSP data was used to generate both conventional primary reflectivity response, as well as multiple corridor stacks based on key interbed multiples to understand their generation and kinematics. Different wavefields were generated to allow the discrimination between surface and interbed multiples. This provides support for amplitude fidelity for multiple events and helped identify the adverse effect of multiples on a different reservoir interval trough.
Due to the large well stock, with over 700 wells with porosity logs penetrating the Mishrif reservoir, this case study is peculiar in the sense that the previous Geomodel had no direct seismic attributes used in property distribution. Therefore, seismic attributes generated were compared to the Geomodel properties, such as porosity to see if geological features were identifiable on seismic. A grainstone shoal body on a Geomodel average porosity map, also clearly delineated on the SSB sections and attributes, was only subtlely expressed and not properly identifiable on the full-stack data. One of several sinuous features, interpreted as grainstone-dominated tidal channels, targeted using seismic attributes was recently drilled and encountered good reservoir quality channel facies.
This case study shows how a SWT process (three-way tie, RCM), seismic attributes and results from a recently drilled well provide validation of the authenticity of the added SSB resolution.