Computational Geometry Approach On Automatic Identification Of Reservoir Interwell Flow Stratification

The goal of this dissertation is to apply the computer technology into study of the reservoir interwell flow patterns. Currently since the majority of oilfields in China reach water flooding development stage, it becomes a bottleneck problem for urgent resolution on how to accurately and quantitatively identifying the stratified flow patterns between the reservoir injectors and producers. It is valuable to understand the underground fluid distribution and flow characteristics at late stage of oilfield production, analyze the remaining producing potential and its distribution, adjust the position of oil-water wells in production formation, remodel well networks, improve water flooding effectiveness.The dissertation focuses on the study of water flooding oil reservoir and has utilized reservoir static geological data and large amount of dynamic and monitoring data from multiple producing wells. On the basis of computational geometry, subsequently constructing geometric model of flow patterns, developing the interwell visibility graphs, identifying the shortest resistant path, analyzing various sandstone and mudstone areas via Boolean algorithm, a new process flow and approach was developed on automatic identification and quantitative analysis of stratified flow patterns between injectors and producers.First, an algorithm was developed to recognize the stratified flow patterns between the reservoir injectors and producers. The mapping relationship between petroleum domain objects and computational geometric elements was constructed. On the basis of reservoir data combined with geometric primitives from Boolean operation algorithm, point positioning in geometric search, visualization, range search, linear programming in geometry optimization, various techniques in computational geometry and reservoir percolation mechanics were applied to evaluate whether oil-water wells were positioned on the same sandstone, where was the shortest path with the minimum resistance, etc. The judgement criteria of interwell stratified flow patterns were presented, and consequently a computational workflow was developed to recognize the stratified flow regime. After validating the result from this algorithm against 149 tracer test data, the coincidence rate is higher than 90%.Secondly, a quantitative analysis procedure was implemented for oil-water well flow patterns. Based on the interwell stratified flow mechanism and using water-electricity similarity principles, the volume and direction of water injection in each layer can be calculated. As a result, the overall water injection volume can be obtained. The outcome was compared to the actual data from 103 water injection profiles, which shows the coincidence rate within 10% error is greater than 85%. It further testifies the accuracy of this algorithm to identify the stratified flow patterns between injectors and producers.Thirdly, in line with the above algorithm, a method was further established to compute reservoir edge water influx volume from each layer and water flooding reserve level of control. It effectively implemented the quantitative description and assessment of reservoir vertical natural energy supplement as well as water flooding affected area and its regional planar distribution feature. By applying virtual injectors to simulate the reservoir edge water influx process, calculating the stratified flow patterns between virtual injectors and producers, and thus splitting the edge water influx volume from each layer, it can be more scientific and objective to evaluate the development effectiveness of reservoir with natural edge water. In addition, by using Boolean algorithm in computational geometry, the water flooding affected area can be calculated and water flooding reserve level of control can be derived. It is the first real application to use the non-simplified formula of water flooding reserve level of control to compute the above indicators and it turns out the result is more accurate.The software package developed from this methodology has been successfully deployed in over fifty reservoir blocks’development and reformation plans in fourteen domestic and three overseas oilfields. It overcomes some challenges faced by other traditional methods or manual analysis which have narrow range of applications, only analysis of simple mechanism, poor analysis efficiency, and inaccurate result, etc. It has effectively guided the secondary exploration, deep profile control and oil displacement, optimization of regulation and reformation in the mature oilfields. The research findings in this dissertation also support further study of splitting oil and fluid production in the producers as well as optimizing perforation parameters in reservoirs, which have wide and promising application prospects.