Many factors influence the evaluation process of thief zones. The evaluation index contains very complex information. How to quickly obtain effective information is the key to improve the evaluation quality for thief zones. Considering that the correlation and information redundancy among the evaluation indexes will seriously affect the evaluation results for the thief zone, based on the principal component analysis (PCA) method, this paper proposes a multi-layer weighted principal component

more »

... lysis method (MLWPCA). Firstly, factor analysis is performed on the original data to obtain the plurality subsystems of the evaluation index. Then, a principal component is analyzed through the subsystems of the evaluation index PCA to obtain the principal component score. Finally, the subsystem is weighted by the factor score and the comprehensive thief zone score is obtained by combining the subsystem weight and the subsystem score. A case study on the Daqing oilfield shows the effectiveness of the method, verified by tracer tests when applying the MLWPCA method to evaluate the thief zone. The thief zone of the Daqing oilfield is obviously affected by effective thickness, coefficient of permeability variation and interwell connectivity. At present, there are 10 well developed thief zones and eight medium developed thief zones in Daqing oilfield. The accuracy of this method is 94.44%. Compared with PCA, this method has better pertinence in evaluating thief zones, and is more effective in determining the principle influencing factors. Up to now, there are several approaches to identify thief zones. Chetri et al. presented production logs combined with dynamic data providing inferences on water breakthrough trends thus helping to identify thief zones with high permeability [2] . Al-Dhafeeri et al. identified thief zones using core data and production logging tests (PLTs) [3] . They found these zones contributed more than 50% of the total well production with permeability greater than 20 Darcy. Li et al. described a method of identifying thief zones using integration of production logging test (PLT), nuclear magnetic resonance (NMR) and high-resolution image logs [4] . John et al. detected for the first time the location of thief zones using distributed temperature sensing (DTS) technology combined with production logging tests (PLTs) and water flow logs (WFLs) [5] . Chen et al. studied the thief zones using production logging tests (PLTs) and summarized different types of thief zone distribution [6] . These methods are simple and easily identify thief zones, however, the well logging can only analyze the situation in the near well region and the tests will affect the normal operation of the well. Feng et al. applied interference well tests to determine the thief zones and calculated the permeability and thickness of thief zones based on a semilog method [7] . Feng et al. characterized thief zones in mature water flooding reservoirs using pressure transient analysis and established a mathematical model for a well intersected by a high-permeability streak. The solution in Laplace space is derived by Ozkan's source function [8] . It is relatively cheap to identify the thief zone by well testing, but these methods must be based on an ideal model, which is quite different from the actual reservoir situation. Watkins found thief zones through analyzing the time required for tracer-tagged liquids to flow from injection wells to production wells [9]. Ravenne et al. identified the size and distribution of thief zones through building the 3D geocellular model and nested pixel simulations [10]. Shawket et al. investigated the evolution of an injected water front and the effect of different reservoir heterogeneity parameters and gravity on the thief zones [11]. Izgec et al. used modified-Hall analysis (MHA) to discern the characteristics of the thief zone [12]. Ajay et al. took advantage of production logging test (PLT) data, streamline trajectories and tracer data to form an efficient assisted history-matching (AHM) workflow to identify a thief zone [13]. Although these methods are able to quantitatively describe thief zones accurately, they are always time-consuming and expensive. Wang et al. first used the ISODATA clustering analysis method to determine the thief zone [14] . However, this method only describes the situation near the wellbore, and the selection of the evaluation index is too artificial, lacking a corresponding selection method. Ding et al. presented a methodology of determining the thief zone by using automatic history matching and fuzzy mathematics [15] . This method takes into account the uncertainty of the geology, but the parameters are difficult to obtain. In order to improve the recognition accuracy of thief zones and to eliminate the interference of human factors on the recognition process, the principal component analysis (PCA) method is proposed to identify thief zones. However, due to the complicated factors which influence the thief zone, when different properties and levels of the indicators are directly evaluated the distinction of recognition results is low. To solve this problem, in this paper we suggested to layer the evaluation indicators and construct a multi-layer weighted principal component analysis (MLWPCA) method based on the PCA method. Combining the reservoir geological data and production monitoring results, the thief zone of the Daqing oilfield is quantitatively evaluated by the above method. The accuracy of the evaluation results is varified by the tracer method. Analysis Method Principal-Component-Analysis The basic idea of the PCA method is to reduce the dimensions of the original data, converting multiple variables into several independent composite variables (principal components), and selecting the number of principal components according to the principle that the contribution rate of variance is more than 85%. These principal components can reflect most of the information of the original