|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
ABSTRACT The industry is facing significant challenges due to the recent downturn in oil prices, particularly for the development of tight reservoirs. It is more critical than ever to 1) identify the sweet spots with less uncertainty and 2) optimize the completion-design parameters. The overall objective of this study is to quantify and compare the effects of reservoir quality and completion intensity on well productivity. We developed a supervised fuzzy clustering (SFC) algorithm to rank reservoir quality and completion intensity, and analyze their relative impacts on wells' productivity. We collected reservoir properties and completion-design parameters of 1,784 horizontal oil and gas wells completed in the Western Canadian Sedimentary Basin. Then, we used SFC to classify 1) reservoir quality represented by porosity, hydrocarbon saturation, net pay thickness and initial reservoir pressure; and 2) completion-design intensity represented by proppant concentration, number of stages and injected water volume per stage. Finally, we investigated the relative impacts of reservoir quality and completion intensity on wells' productivity in terms of first year cumulative barrel of oil equivalent (BOE). The results show that in low-quality reservoirs, wells' productivity follows reservoir quality. However, in high-quality reservoirs, the role of completion-design becomes significant, and the productivity can be deterred by inefficient completion design. The results suggest that in low-quality reservoirs, the productivity can be enhanced with less intense completion design, while in high-quality reservoirs, a more intense completion significantly enhances the productivity. Keywords Reservoir quality; completion intensity; supervised fuzzy clustering, approximate reasoning,tight reservoirs development
Abstract A giant gas field that covers 75 km in length and 15 km in width has been producing since 1990 from approximately 1,200 wells which are located in 34 platforms. Deposited within a deltaic environment with enormous multi-layer sand-shale series, the wells undergo commingled production with an average of more than 30 reservoirs per well. With a total of approximately 700 perforation jobs included in more than 4000 well intervention jobs per year, the field is considered as the most complex field in the PSC Block in terms of operations. Prioritization of these perforation jobs are based on the perforation gain of each job. Therefore, properly estimating the perforation gain is crucial in order to efficiently and effectively manage and prioritize well intervention jobs. Hypothetic approaches, for instance productivity index driven from Darcy's equation, may not be straight-forward due to incomplete and imprecise data measurement. Overwhelming operations workload in the field limits the number of data acquisition jobs performed. Consequently, required data to estimate perforation gain such as skin, pressure and drainage radius becomes limited. An alternative approach using artificial intelligence called fuzzy logic was introduced. Being a soft-computing pattern recognition method that allows imprecise input to yield output, fuzzy logic fits well with the nature of high uncertainty in geosciences data. The one-year study is conducted on reservoir basis using well monitoring results to split well level gas rate into individual reservoir gas rate. In order to ensure that proper data are incorporated in the model training, processes of data filtering must be undertaken. Therefore, implementing fuzzy logic to estimate perforation gain includes 3 main steps: (1) Preparing and Filtering Training Data Set; (2) Building the Fuzzy Model; and (3) Performing Blind Test. After series of trial and error process, the model has reached its minimum error without compromising sense of engineering and generality. The fuzzy model results in 960 fuzzy rules and 5 input parameters: netpay, porosity, drawdown, mobility and water risk. Afterwards, the blind test shows that the resulting output from fuzzy logic correlates well with the realized gas rate both on reservoir level and well level, with maximum R-squared value of 0.7. The study is limited within the scope of current best practice for unperforated reservoirs and further study would be required to estimate the perforation gain from unconventional perforation methods and re-perforations. This method of estimating perforation gain using fuzzy logic has been implemented on daily basis with the aim to improve the efficiency and effectiveness of managing and prioritizing well intervention jobs in such a complex environment.
Prediction of the hydrodynamic efficiency of a Wave Energy Converter (WEC) device is crucial to evaluate the design and the concept of the device. Experimental and numerical techniques are the main tools currently available for WEC designers; however, these techniques are still costly and too time expensive to be used for optimisation and commercial purposes. It is, therefore, important to develop an efficient and cost/time-effective technique in order to investigate the hydrodynamic characteristics of WEC devices. In this work, an Adaptive Neuro-Fuzzy Inference System (ANFIS) technique was developed to predict the hydrodynamic efficiency of WEC devices. ANFIS models were designed, trained and tested using published experimental datasets for the hydrodynamic efficiency of fixed Oscillating Water Column (OWC) devices, and different types of membership functions were examined to develop the best accurate model. ANFIS technique was found to provide good estimates in comparison with experimental results and can be used to predict the hydrodynamic efficiency of WEC devices during the early stages of design.
Wave energy is one of the most significant component of the renewable energy sources worldwide (Jin et al., 2010). Wave Energy Converters (WECs) are devices usually categorised by their location from the shoreline; onshore, near shore or offshore. The basic principle of a WEC device is that the incident wave excites the system such that it can make force between an absorber and a reaction point, which acts directly on, or drives a working fluid through a generator or a pump. Accurate prediction of the hydrodynamic efficiency of a WEC device is essential to evaluate its concept and feasibility. Among many types of WECs is the Oscillating Water Column (OWC) device which has been widely under research (Thorimbert et al., 2016, Simonetti et al., 2016, Ning et al., 2016, Luo et al., 2014, López and Iglesias, 2014, Teixeira et al., 2013, Fleming et al., 2013, Ásgeirsson, 2013, Morris-Thomas et al., 2007). However, the OWC technology has not been fully commercialised yet (Ning et al., 2016). The main reason is that the hydrodynamics of the OWC devices has not been fully understood which requires further investigations using theoretical, numerical and experimental approaches (Ning et al., 2016). Various numerical models have been established based on either potential flow or viscous flow models. Using potential flow based techniques (Evans, 1982), the hydrodynamic efficiency of an OWC device is often over-predicted (Ning et al., 2016, Morris-Thomas et al., 2007). Model testing and Computational Fluid Dynamics (CFD) approaches are the main tools currently available for WEC designers. Model testing is arguably the best approach for estimating wave-induced loads and response of fixed and floating objects (Abdussamie et al., 2017c, Banks and Abdussamie, 2017), however, the experimental approach is costly, time-consuming and involves several other constraints such as scaling effects. According to ITTC (ITTC, 2014), large-scale models are required to provide realistic simulations of a power take-off system (Stratigaki et al., 2012) thereby encountering the limitation of existing facilities. Therefore, some distortion in the modelling of WEC systems is unavoidable.
To solve the problem of speed and robustness of small Unmanned Surface Vehicle (USV) heading control in the inland water navigation environment, the paper designed a pod propulsion USV heading control system based bipolar fuzzy controller. The system comprises of heading controller, steering mechanism, electronic compass and rudder angle measuring instrument. Heading controller includes two fuzzy controllers, which can meet the demands of large-angle steering control and small-angle course keeping. The simulation results show that the control system is more to meet the conditions of inland navigation than fuzzy PID autopilot.
Inland USV is mainly used in search and rescue, emergency, cruise, measurement etc. Due to the large density of ship navigation, channel bending, shallow narrow leg more, meteorology, hydrology and other environmental complexity of inland waters, so the requirements of USV heading control flexibility and robustness are higher.
In the USV heading control, there is reported and literatures at home and abroad. The PID controller’s structure is simple, easy to use, and applied on ships heading control at first, but the lack of ability to adapt to vessels working conditions and environment, and its performance is not satisfactory (Cheng and Huang, 1997). Yang and Yu (1999) designed a robust PID control law autopilot, and it has robustness to some extent in ship speed-changing and with disturbance. Based on generalized predictive theory, Peng, Wu and Liu (2014) designed a GPC-PID cascade controller with good control accuracy and robustness. It was noted (Healey and Lienard, 1993; Song, Li and Chen, 2003;) sliding mode controller to achieve the control of the ship's heading or track, but it didn’t solve the high frequency chattering phenomenon. Zhang, Lv and Guo (2007) combined the neural network structure and the closed loop gain shaping algorithm to form a closed loop control system, with strong robustness, and the algorithm was simple and has clear physical meaning. Using the method of adaptive Backstepping, Godhavn, Fossen and Berge (1998) designed a nonlinear controller for keeping ship’s course system with a good control effect. And an integral term was added to improve the course-keeping performance of the control system (Fossen and Storand, 1999). The neural network approach was combined with an adaptive Backstepping method (Xu, Li and Li, 2012).
First break picking is a pattern recognition problem in the area of seismic signal processing, a problem difficult to automate and that requires a lot of human effort to be solved. The objective of this project was to develop an algorithm, based on Fuzzy Inference Systems, using fuzzy rules of the form If - Then to perform automatically first break picking in VSP data. The results indicate that by the implementation of this algorithm the maximum error obtained is of 2ms compared to first breaks calculated manually. However, when the signal to noise ratio (S/N) is low, the algorithm''s accuracy decreases.
Chen, Yie-Ruey (Department of Land Management and Development, Chang Jung Christian University) | Hsieh, Shun-Chieh (Department of Land Management and Development, Chang Jung Christian University) | Liang, Heng-Yu (Department of Land Management and Development, Chang Jung Christian University) | Tsai, Chih-Geng (Department of Land Management and Development, Chang Jung Christian University)
Abstract For oil or gas fields with stratified reservoir layers, detailed productioncontribution for individual layer is always desired.Unfortunately, insome particular cases, production wells are completed following commingledscheme. This is worsened further if only very few production tests arerun for the field.This is the case for the Central Sumatera field withits 95 commingled production wells, among which only a few had undergoneproduction tests and none of them have ever undergone productionlogging.Problems rise when the occassion came in which detailedproduction contribution from individual reservoir layer is required for thefield's reservoir simulation modeling and productionevaluation/prediction. This paper presents an approach to solve the problem.The approach isbasically based on the application of soft computing (Fuzzy Logic) toinvestigate pattern of relationships between production contribution of layersin commingle wells and rock petrophysical data as well as other relevantgeological/engineering data.For the purpose, thirteen wells (key wells)that have production tests are assigned, among which three wells are assignedfor checking the validity of the recognised pattern.Using the validatedmost valid pattern, individual layer's production allocation for other wellsare determined with well-log analysis data as the major input. Result estimates for the candidate wells are better compared to resultsproduced by the conventional method of productivity index (PI)analogy.The resulted variation in water cut and separate oil and watersplit factors appear to be more realistic from any point of view. Introduction In managing a commingle production well, knowledge over productioncontribution of individual sand layer is always desired.The commonpractice performed during drilling and production activities of a productionwell is through the use of well testing/production testing and/or productionlogging. From the test, fluid dynamic data such as total liquid rate, water cut, and gas cut of an individual layer are produced.However, costand time efficiency is always used as the reason for not conducting suchtests. Therefore, even though such tests are always regarded as theprimary source of proof, an alternative means that can be used to provideestimates is always desired. Ideas of establishing a method that can provide illustration over productioncontribution of all layer(s) always exist.Certainly, there are approachesto serve the purpose such as productivity index (PI)/transmissibility analogyand petrophysical approach through fractional flow measurement in corelaboratory. However, those approaches are often considered inadequate foraccommodating various factors that may influence production contribution of aproductive layer. To materialize the requirement stated above, an indirect approach in theform of pattern recognition/modeling was taken.This approach was taken inorder to model relations between various factors in wellbore and productioncontribution of reservoir layers without being trapped by the certaincomplexity that may occur in any mathematical expressions trying to explain therelationships.For the purpose, fuzzy logic (a form of artificialintelligence) has been used.The choice is actually based on its capacityto accommodate both numeric and non-numeric data, since it is considered thatsome non-numeric data such as lithology and pore system also have someinfluence on production contribution.