| Electrical log is the most widely used method of oil saturation logging. It provides the oil saturation and its movability through detecting the inhomogeneity of the formation resistivity and permittivity. The criterion of evaluating the method and the tool performance is the capability of detecting the inhomogeneity and data interpretation. In the actual logging environment, there exist the vertical and radial formation inhomogeneity. The former is formed while the formation aggraded, and the latter is worked while the mud of artesian well invaded. Since logging curve is the tool response along vertical depth, the vertical formation inhomogeneity can be shown directly while the radial formation inhomogeneity can not. Therefore it is very important to improve the ability of investigating the radial formation inhomogeneity and data interpretation. Generally for the single structural logging tool, if its resolution is high, its investigation depth is not enough; if its investigation depth is deep, its resolution is low. So it is very difficult to synchronously detect the vertical and radial inhomogeneity using the single structure logging tools. The array electromagnetic logging tools and its data processing method could provide an efficient way to resolve this issue. There exit two main array electrical log methods, e.g. the array induction log and the array electromagnetic propagation resistivity log (EPRL). The source frequency of EPRL is much higher than that of the induction log. It has one transmitter coil and two receiver coils. Measuring the phase shift and amplitude ratio of the induced voltage in the receiver coils, its resolution is much higher than that of the induction log. It is very suitable for the oil exploitation in our country. The west companies had made their own EPRL array tools. They all use one or two frequencies (0.4-2MHz) and have several transmitter-to-receiver structures (0.23-1.4m). They synchronously measure phase shift and attenuation. Some of them have ten logging curves. VIKIZ is the most popular EPRL array tool in Russia. It uses five different frequencies (0.875-14MHz) and five transmitter-to-receiver structures (1.8-0.45m), and only the phase shift is measured. So it has five phase shift logging curves. They are much difference in designing concepts and realization approaches between the West and Russia array tools and each has its own good qualities and shortages. In all, VIKIZ is better than the West array tools. In present, their method of data processing and interpretation is inefficient. The potential application has not been exerted. Our country has imported VIKIZ and the West array tools, and the CNOOC is developing EPRL array tools now. So the inverse method of EPRL and the optimal combination of the array tools are selected to be study as the main content of this thesis. The oil saturation is judged by the formation resistivity for the EPRL. Sometimes this judgment is invalidated for the change of the complex lithology and formation water salinity. It troubles the interpretation of resistivity logging. For example, the evaluation of low resistivity oil, gas reservoir and freshwater flooded oil reservoir are difficult problems for a long time. The way to solve these problems is to develop the log method that influenced lightly by the lithology and water salinity and can directly shows the oil reservoir. Two matters should be mentioned here. One is that low resistivity annulus is an apparent indication of the movable oil in formation. Another is that the permittivity of the pore formation can show the oil reservoir directly, which is determined by the water ratio of pore liquid and is independent of the water saline and the lithology. The EPRL is very useful in the two aspects. Thus we will pay special attention on how to recognize the low resistivity annulus and acquire the permittivity. The knowledge about the investigating characteristics of tools is an important basis to design tools and to develop the data processing method, especially for the array tools. Tool response function in uniform media is investigated as usual. Also, the tool detective characteristic in inhomogeneous formation is given with the first order relatively partial derivative curves (sensitivity curves) of log response with respect to formation parameters. The sensitivity is the Jacobi matrix element of the inversion target function. Information abundance of the array logging data and the sensitivity of response to formation parameter are obtained through and . Using them the detective characteristic of the VIKIZ, MPR and EWR tool was investigated and analyzed systematically. Some valuable results were gained. It provides the basis of selecting logging data, finding effective inverse method, and improving array tools. C = JJTD =JTJThe EPRL tools have several sub-arrays with different resolution and exploring depth. The influences of layer width, mud invading and formation dielectric permittivity on logging response are severe nonlinear. These factors exist together and have relations each other. The good interpretation results can be given only through doing the inversion using all the parameters. Because of the great number of inversion parameters and the severe non-linear, sometimes the inversion is unconformable. The inversion is instability and has multiple solutions. It strongly depends on the initial values. The damping regularized Gauss-Newton method is used to do the inversion, and the damping factor was modified to improve the inversion quality. We begin it by picking the layer's boundaries from the highest resolution curve using the characteristic recognition method. It then performs vertical one-dimensional inversion to the shallow detection curve and the deep one. The results can be used for interpretation of the thick layers and can provide initial value for two-dimensional inversion. The constructed initial value has fine vertical edges, so it can show the mud invasion correctly. It provides a good basis for two-dimensional inversion. In the first step of two-dimension inversion, parameters with high sensitivity were inversed. The smaller were joined to inverse step by step. By this strategy, the inverse is stable, and the convergent speed and precision are high. VIKIZ has good performance through analyzing its detecting characteristic and its inversion results. It has high resolution. The detection depth of the sub-arrays changes monotonously and distributes reasonably. All these respects are very useful for detecting vertical and radial inhomogeneity. For thick layers, its apparent resistivity interpretation is reasonable. It has good capability to show the low resistivity annulus, thin layer and thin interbeded formation apparently. The quantitative interpretation for the resistivity distribution of thin interbeded formation, high resistivity layers and low resistivity annulus was improved greatly through inversion. The influence of inhomogeneous dielectric permittivity to VIKIZ logging response and resistivity inversion precision was also obtained. It should not be neglected for high resistivity layers and thin formation. Give the inhomogeneous dielectric permittivity an equivalent value, first do two-dimension resistivity inversion, then do resistivity and permittivity inversion together. The high precision results of resistivity and useful dielectric permittivity information were gained subsequently. This is a new result that has not been reported. The conclusion was drawn that it is impossible to get some useful dielectric permittivity information through low frequency logging such as 10 MHz. It is a new idea to dielectric permittivity logging. The disadvantage of VIKIZ is that itsdetection depth is not enough when the formation resistivity is low or the mud invasion is deep. The data process and two-dimensional inversion of the West EPRL tools such as MPR and EWR was studied. There is overlap between the detection depths of the sub-arrays. It is not convenient to use all the data for apparent resistivity interpretation. Five curves are selected out from the eight curves according to their detecting characteristic as the basis for interpretation and inversion. Their detection depth changes monotonously and covers all the detection range. It is very useful to do the apparent resistivity interpretation and improve the inverse efficiency. With the five curves, low resistivity annulus can be detected apparently. The quantitative interpretation for the resistivity distribution of thin interbeded formation, high resistivity layers and low resistivity annulus is improved greatly through inversion. For the low resistivity formation, the data interpretation of MPR and EWR is reasonable. Because of their low source frequency, the influence of dielectric permittivity is little, but the resolution is decreased. So the interpretation for high resistivity formation and thin layers is unreasonable. There is much difference between the designing ideas of Russia VIKIZ and the West EPRL array tools. Through analyzing their detective characteristics and the inverse result of the logging data, the principles of optimal combination of the array tools were presented as follows. 1. The sensitive detection area of the sub-arrays should be divided along the radial. The detection depth of the array tools should cover all the range that needed to be detected. Their distribution along the radial should be reasonable. The Russia VIKIZ tool that designed with the isoparametric condition accords with the above principles. However, the isoparametric condition is not necessary. 2. In the precondition of reasonably distributing the detection depth of the array tools, sub-arrays with higher sensitivity and resolution should be used, especially there should be one shallow detective sub-arrays that has very high resolution. It will be very useful to acquire the invaded zone resistivityies apparently and get the accurate layer boundaries. It is very important for the data inverse. 3. Except in the low resistivity formations ( Ï<1 ?. m), the detective sensitivity and resolution of phase shift are much higher than that of amplitude ratio (more than 100 times). When the detection depth of... |
PDF Full Text Request