Development of Comprehensive and Efficient DTS Interpretation Method for Fracture Diagnosis

Abstract

Listen

Translate article

Distributed temperature sensing (DTS) by fiber optic technology has been commonly used to evaluate the fracturing treatment efficiency. Once DTS data is quantitively interpreted, we can estimate injected fluid volume distribution along each stimulated stage interval, and it allows us to analyze and optimize completion design such as injection schedule, fluid/proppant types, and perforation design. In general, the DTS data can be interpreted by conducting temperature profile matching. After pumping is stopped, the perforation clusters that take more fluid during injection warm up more slowly. Based on this physical feature, the injected fluid volume can be estimated by reproducing the temperature profile during warm-up using a numerical thermal model as a forward model. The temperature matching technique still contains several uncertainties. One of them is the transient temperature behavior as a function of shut-in time. While the DTS data is continuous with time and space, the temperature profile at one time slice is generally interpreted. However, as presented by several authors, the trend of temperature profile often changes with time during warm-up, which could be caused by the fracture closure or small amount of fluid flow induced by adjacent stage treatments. If degree or locations of cooling spots are different with time, this thermal behavior also needs to be incorporated into the mathematical model to correctly evaluate the fracturing stimulation efficiency. In this study, we extend the existing DTS interpretation method to be able to match the temperature profile at multiple time slices. While the degradation of efficiency is expected due to increase of data to be matched and model parameters to be tuned, we apply high-efficiency inversion techniques which have been established in the regular history matching methods to improve the interpretation procedure. Also, this study explains the connection and conflict between fracture stimulation efficiency and productivity. As previously presented, the estimated injected fluid volume distribution may not be perfectly matched with production inflow distribution. If more reliable injected fluid volume distribution can be estimated by considering the temperature behavior at multiple time slices, we can better diagnose the fracturing treatment performance.