Reservoir Characterization. Группа авторов

Чтение книги онлайн.

      Foreword

      What is reservoir characterization? As you will see from this book, this is a very advanced topic so let’s break it down a bit and start form the basics. What is a reservoir? This is ‘a place where something is kept in store’. And what is characterization? That is ‘to describe the character or quality’ all according to the Webster dictionary. So, we are arrived at: ‘describe the character of something that’s kept in store’. It seems relatively benign and easy but ‘the devil is in the details’ is perhaps the best way to get the readers intrigued and immersed in this topic. So, we are left wondering what are these details where the devil resides? And here starts the story…..

      In fact, a better wording would be ‘Subsurface Reservoir Characterization’ or SRC. There have been on the order of thousands of studies in reservoir characterization over the life time of this field. As such, this topic has evolved and matured with many learnings. As illustrated in this book, there are now well established and tested workflows SRC and I‘d like to go over some aspects of these understandings and workflows.

First, it is key to understand that SRC is a continuously changing, multi-discipline and multi-scale topic. For continuously changing a good example would be the recent impact of say machine learning methods. I have learned that if our data quality is good enough and there are physical relationships between reservoir data and properties, machine learning can be an excellent way to quickly uncover relationships in a multi-variable universe. However, once again, even here, the devil is in the details…. Multi-discipline is a word we easily use but have difficulty implementing. In many projects the geologist is tasked with building a static reservoir model and then passing it on to the reservoir engineer to build a dynamic model and history match production. However, it has been challenging to form a loop versus a linear workflow or for the dynamic model to be updated with new static information or cover a range of possible models that fit the data…… As for multi-scale, the discipline involves integration of data from a wide range of data, say, nanometer (electron microscope), to centimeter (cutting and core samples), to decimeter (well log), to meter (seismic) scale. Spatially most of these data are acquired within a small portion of one or several wells and geophysical data gives the capability to extrapolate away from the wells with lower resolution. Due to uncertainties in the data, rapid variations in the subsurface, and sparse sampling multi-scale integration can be a challenging task. There is a good discussion of “SURE Challenge” in the book where the author addresses the above mentioned challenges of integration incolving multitude of data set with different Scale, Unvertainty, Resoultion and Environemnet. It is suggested that different AI and Data Analytics techniques may be best equipped to handle the SURE Challenge.

      The second component can be categorized into input data quality (informally ‘garbage in, garbage out’). Any workflow that is lets say cutting edge cant work without high quality input data. Further, it may cause mis-interpretation that a workflow is ‘not’ a good workflow or appropriate simply because the input data was the culprit. The input data in fact starts from data acquisition, then to data processing and finally to data interpretation and integration. One of the pitfalls along the way is to simply obtain the data as an interpreter and not be aware of lets say the ‘history’. An example would be to apply amplitude based seismic analysis to data that non-amplitude preserving processing was applied to (Automatic Gain Control or AGC would be a simple example). However, the same could be happening with say well-log or production data. The good news is that over time in every SRC related discipline data quality has been improving with not only better tools but also more frequent data acquisition during the life of a reservoir. Further, over time we have learned to build much better processing tools that provide high quality data for the integration component. The net result of this has improved our ability to conduct integrated studies and quantitative products. One example of this near to my heart is joint seismic inversion of PP reflected waves with PS (or converted) reflected waves from a reservoir. We have seen that with improved acquisition and processing, the joint PP/PS inversion can substantially improve pre-stack seismic inversion providing a stable S-impedance as well as a P-impedance that can provide valuable information such as formation properties, porosity, Total Organic Carbon (TOC,) fluid types, and time-lapse reservoir pressure and saturation changes over the life of the reservoir. Such improvements are going on in all the subsurface disciplines thanks to modern acquisition and more diverse data with higher quality.

This book is an excellent resource for beginners in SRC to get an overview of the topic and for expert to study most recent advances in their own and related disciplines. The book covers a wide range of topics from conventional to unconventional reservoirs, from geology to geophysics to petroleum engineering, from laboratory measurements to field applications, from deterministic to statistical methods, from primary depletion to EOR with CO2 injection, from static to dynamic SRC, as well as use of AI for reservoir charcaterization. In the end, SRC requires best practices to be implemented to be value generating. This books certainly provides the necessary best practices.

       Dr. Ali Tura

      Professor of Geophysics

      Co-director of Reservoir Characterization Project (RCP)

      Colorado School of Mine,

      Denver, Colorado