| The production profile well logging technology for multiphase flow is a necessary approach to analyses the producing degree of reserves, it can give the foundation of oil field dynamic monitoring and the establishment of all kinds of measures, it will play an important role in the efficient development of oilfields. If there are some slipping velocities, Optimum interpretation technology is the only way which could used in the case. Optimum interpretation technology in multiphase flow logging (OITMFL) can make full use of logging data. Even some of the information is incomplete, or its error is considerable, the optimum interpretation technology also can improve the interpretation accuracy to a certain degree, so it will get a better result.The accuracy of OITMFL is influenced by optimized objective function itself and the performance of optimization algorithm.Objective function for optimization is based on theoretical response equations of logging instruments, the theoretical response equations try to show the relationship between volume flow rate of different phase in the well and response value of logging instruments. Through literature survey, we find that there are an in-depth studies on the response equations of logging instruments for vertical wells and low-inclination wells, there are a steady relationships between phase volume flow rate and response value of logging instruments. But in horizontal sections and nearly horizontal sections, there is a poor relationship between flow rate and response value because of the influence of gravitational differentiation, as a result, the reliability of interpretation results of a production profile is decreased. There are only two kinds of objective function for optimization:weighted square sum form and weighted absolute value sum form. Two kind of function normalize the measured data by using resolution ratio of tool for to weigh the error between measured data and response value equitably. But, if the measured data and response value are different type, resolution ratio of tool can’t normalize the measured data perfectly. Firstly, the theses present an objective function based on relative error method, which is trend to balance the different data by using the reciprocal of measured data to weight the error. There is a weight parameter correspond to reliability level of tools in original objective function, but I can’t find how to set the parameter with theoretical foundation. At the same time, it is usually to get measured data in one trip by using tool string nowadays, so we can believe that tools in one string have the same reliability level. And then, the new construction method don’t considered the reliability parameter, it is helpful to generate the objective function.Because some response equation are complex, absolute value form in objective function lead to poor function properties, and the sum in objective function is depend on the actual logging tool string, we consider deterministic optimization algorithm is not fit OITMFL. The writer introduces to use random global optimization algorithm. After analyses the theory and characteristic of basic particle swarm optimization (PSO) and ant colony optimization (ACO), the writer present a new algorithm, it integrate PSO, ACO and chaos disturbance, named CPA, C indicate chaos disturbance, P indicate PSO, A indicate ACO. CPA attempts to make the best of the advantages of PSO, ACO and chaos disturbance. PSO has a fast convergence at earlier stage, ACO has strong search ability at later stage. Chaos has pseudo-randomness, unpredictability, ergodicity. Chaos disturbance should increase the global search ability of CPA. For objective function, CPA is fit to OITMFL because it only use function value, and it don’t use function characteristic and form. The author select16benchmark functions, analyses the geometric feature when function is2two dimension. And use the16functions to compare the searching ability of three algorithms:PSO (algorithm one), PSO with chaos disturbance (algorithm two), PSO-ACO with chaos disturbance (algorithm three). The results indicate that:if the actual global optimal value is near the searching space boundary enough, the searching ability of algorithm one is the worst, the ones of algorithm two is the best, but algorithm two needs so long computing time, if the actual global optimal value is at some distance from the searching space boundary, the searching ability of algorithm three is the best, and it spend little computing time with high precision solution. The author examines the performance of the new model and new algorithm by programming to the practical data.The influence of gravitational differentiation lead to phase separation between heavy phase and light phase, so the multiphase flow situation in horizontal sections is very different from in vertical well. It is necessary to study flow pattern firstly for studying OITMFL in horizontal sections. The author present the results and conclusions come from the experiment of oil-water two phase flow in horizontal sections and nearly horizontal sections.For flow pattern,(1)The range of well deviation angle is±5°(denote the angle between well-bore and horizontal plane, degree is positive when fluid flow upward, and negative when fluid flow downward), under the experiment condition used, flow pattern is laminar flow regime, the difference are the smooth degree of phase boundary, the fluctuation regularity of phase boundary, the fluctuation range of phase boundary; Along with the increase of the whole quantity of flow, the phase boundary become wave from smooth, and become chaos from wave;(2)When well deviation angle is negative, heavy phase residence is weakened, so the holdup of heavy phase is less than phase cut. When well deviation angle is positive, heavy phase residence is enhanced, so the holdup of heavy phase is larger than phase cut;(3) There is a big influence on the phase boundary from well deviation angle. Especially, the position of phase boundary is very different between0°and±1°(4) When well deviation angle is0degree, for the experimental flow rate, there is a distinguishable relationship between the water depth in the side view (indicate water holdup) and water cut;(5) In the range of±5°, if whole flow rate reach400m3/d, especially reach500m3/d, well deviation angle influent little on the water depth in the side view.For well deviation, water holdup, water cut and whole flow rate,(1) Under the experimental condition, there is a determinate relationship among well deviation, water holdup, water cut and whole flow rate. If well deviation, water cut and whole flow rate are constants, there is almost a one-one corresponding functional relationship. It is possible to get water cut from water holdup;(2) If whole flow rate is more than50m3/d, in the giving well deviation range, we can identify water cut based on water holdup; but if whole flow rate is less than 30m3/d, the reliability level is significantly impacted by well deviation angel. When well deviation angel is0°, even the whole flow rate is10m3/d,it can be distinguished. When well deviation angel is positive,30m3/d is the lower limit flow rate which can be identified; When well deviation angel is negative, less than or equal to30m3/d is hard to be identified;(3)In horizontal well-bore, the most important factor which impact water holdup is well deviation angle. The little the whole flow rate is, The more important the well deviation angle is;(4)If production condition like the experimental condition, it can get water cut from water holdup by using experiment plate.For well deviation, slip velocity, water cut and whole flow rate,(1) If well deviation angle is negative, for a certain water cut, slip velocity increase as the whole flow rate increase, the law show a trend: down-up-down-up, some like a transformative letter "W";(2)If whole flow rate is near or more than500m3/d, the slip velocity is close to0cm/s. So,we can ignore slip phenomenon under a big enough flow rate.(3)Although there is a fuzzy regularity shown, but it is far away from clear. The reasons maybe include:the consistency of the experimental condition, the accuracy of the characteristic data etc. As a fact, the experiment don’t consider viscosity which is depend on temperature, but the experiment spent too much time, it’s started in spring, and finished in autumn. We get characteristic data only use our eyes and side digital image come from digital camera, it will decrease the accuracy of data. It is worth doing some continue study to discuss the relationship among these factors.I researched and found that Capacitance Array Tool (CAT) is more suitable to measure the phase holdup in horizontal section than other logging instruments. Some manufacturers are about to made CAT themselves, they need the interpretation method for the instrument. The author analyses the principle of measurement, and introduce the existing interpretation algorithms. The author presents the structural features of the tool, and some matters need attention in practice. The author presents two new algorithms to calculate phase holdup based on optimal method, and a new imaging algorithm for cross section image, then the author write programme to test the new algorithms under Matlab, then get the phase holdup and the cross section image. |
PDF Full Text Request