Log simulation is critical for understanding and interpreting logging tool responses. It helps the log analyst understand near-wellbore measurements in complex environments, and particularly in anomalous situations. Simulated logs can also be used in "what-if" scenarios or as part of an iterative scheme to invert for the formation's geometry and material properties. We describe our implementation of a forward modeling and inversion environment with a multitier, Web-based architecture. The Web platform offers universal access from the user's desktop through a common Web browser to the simulator engine, running on a high-performance compute server. The application also allows the user to invert for near-wellbore rock properties from a set of logs. In addition to user-accessible Web pages for interactive use of the application, a programmatically accessible log simulation Web Service is created. Regardless of the host platform, this allows any networked application to access the simulator library without the need to replicate code, thus facilitating the development of formation evaluation applications. Because the library is hosted on a high-performance cluster (HPC), the computational engine always runs in the shortest possible time. Introduction The essence of log interpretation is the solution of the inverse problem; that is, the determination of formation parameters from logging data. This task becomes difficult when the logging environment departs from simple well-understood geometries. Many tools are designed to make deep-reading measurements, and therefore are susceptible to formation heterogeneities. The most common environmental effects are formation dip, invasion of borehole fluids, and the influence of neighboring beds. Frequently, a combination of these effects can obscure the true formation properties (e.g., formation resistivity and porosity), making it necessary to construct a complex 3D model of the formation to obtain an accurate solution. The only way to validate these complex models is by tool response simulation or inversion. Some logging tools (such as dual induction and dual laterolog) do not provide sufficient information to perform accurate inversion. Iterative forward modeling of what-if scenarios is routinely used in these cases. Iterative modeling simulates the exact tool response in a series of given formations. However, this process requires a high level of user interaction. The initial formation model must be constructed from logs, cores and geological information, and then updated manually. A solution is reached when a computed log matches the field log. Automated inversion is more efficient because no interaction is required after the simulation is launched. During the past decade, the introduction of array measurements with greater information content has made it possible to routinely use inversion in log interpretation. In fact, some modern tools record so much data that interpreting their response without the use of inversion presents a considerable challenge. Our interactive Web-based interface provides a convenient and universal user access to the forward modeling and inversion software, while the Web-based programming interface to the simulator makes it available as a library to other oil and gas applications. Fundamentals of 3D Modeling and Inversion Electromagnetic and sonic tool modeling in 3D geometries is performed primarily by using finite difference (FD) or finite element (FE) methods. These codes solve partial differential equations in terms of a large number of simultaneous linear equations. The equations are solved by matrix methods to yield simulated tool response at discrete points in space. Here, we use the finite difference method to model resistivity1,2 and sonic3 tool response. The finite difference discretization of a problem leads to a regular grid representation. Although this grid usually does not conform to the formation geometry, it can be made to approximate any geometry by using material averaging techniques. The matrix equations resulting from the finite difference discretization are usually well structured because of the regularity of the grid, and they are always sparse because the derivatives are approximated locally. Thus, the equations can be easily solved by fast, specialized computational methods.
Sonic Scanner logging tool collects a wealth of data about the geological formation. The drawback of this is that we are now discovering new features on the acoustic logs that have never been observed before; only rigorous modeling can help properly interpret the data. Invariably, it is difficult to learn quickly how to run a modeling code, set the parameters properly, and be able to detect possible errors in the input. In addition, complex modeling requires high power computing resources, which are not always easily accessible the user. To address these problems we developed a multi-tier Web-based log modeling environment where the Sonic Scanner simulator is easily accessible from the common Web browser. The user builds the model in an intuitive AJAX-like interface and submits the simulation to a remote High Performance Cluster. The computed waveforms are played back in the browser using Scalable Vector Graphics in a variety of customizable displays. The Web application is easily available to any user with an Internet access. In addition, a programmatically accessible Web service is available to application developers who desire to build their own interpretation applications using the Sonic Scanner simulator engine.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.