Digital Oil: Machineries of Knowing

INFRASTRUCTURES SERIES

Edited by Geoffrey C. Bowker and Paul N. Edwards

A list of books in the series appears at the back of the book.

DIGITAL OIL

Machineries of Knowing

ERIC MONTEIRO

The MIT Press

Cambridge, Massachusetts

London, England

© 2022 Massachusetts Institute of Technology

This work is subject to a Creative Commons CC-BY-ND-NC license. Subject to such license, all rights are reserved.

The MIT Press would like to thank the anonymous peer reviewers who provided comments on drafts of this book. The generous work of academic experts is essential for establishing the authority and quality of our publications. We acknowledge with gratitude the contributions of these otherwise uncredited readers.

Library of Congress Cataloging-in-Publication Data

Names: Monteiro, Eric, author.

Title: Digital oil : machineries of knowing / Eric Monteiro.

Description: Cambridge, Massachusetts : The MIT Press, [2022] | Series: Infrastructures. | Includes bibliographical references and index.

Identifiers: LCCN 2022003259 (print) | LCCN 2022003260 (ebook) | ISBN 9780262544672 (paperback) | ISBN 9780262372282 (pdf) | ISBN 9780262372299 (epub)

Subjects: LCSH: Petroleum industry and trade—Norway. | Oil fields—Norway—Data processing. | Oil field equipment and supplies industry—Norway—Technological innovations.

Classification: LCC HD9575.N62 M66 2022 (print) | LCC HD9575.N62 (ebook) | DDC 338.2/72809481—dc23/eng/20220124

LC record available at https://lccn.loc.gov/2022003259

LC ebook record available at https://lccn.loc.gov/2022003260

d_r0

Contents

Preface

Acknowledgments

1 INTRODUCTION

I SETTING

2 CONTEXT

3 APPARATUS

II CASES

4 DATA

written with Marius Mikalsen

5 UNCERTAINTY

written with Marius Mikalsen

6 KNOWING

written with Thomas Østerlie

7 POLITICS

written with Elena Parmiggiani

III IMPLICATIONS

8 CONCLUSION

Appendix: A Note on Method

References

Index

List of Figures

Figure 1.1

Work practices in oil: traditional roughneck (top) versus control room based (bottom).

Source: Reproduced by permission from Husmo Foto/Norsk Oljemuseum and Shadé B. Martins/Norsk Oljemuseum, respectively.

Figure 1.2

Subsea installation for oil production.

Source: Reproduced by permission from TechnipFMC and Shell.

Figure 1.3

Compressor component for subsea factory.

Source: Reproduced by permission from TechnipFMC and Shell.

Figure 1.4

Map of the oil licenses on the Norwegian continental shelf.

Source: Reproduced by permission from the Norwegian Petroleum Directorate.

Figure 2.1

A geological trap with the three necessary conditions of a source rock, a migration path, and a seal.

Source: Reproduced by permission from the Norwegian Petroleum Directorate.

Figure 3.1

Hand-drawn sketches illustrate nondigital tools in geology. Picture is taken from a visit to a geological analogy, an accessible location somewhere else in the world similar in important ways to the area you are principally interested in.

Source: Photo by Irina Pene.

Figure 3.2

Seismic horizons from the Sleipner field.

Source: Reproduced by permission from the Volve open data set.

Figure 3.3

Map of seismic surveys on the southern part of the Norwegian continental shelf. The surveys (blue line) are so dense in large areas that they fill the map.

Source: Reproduced by permission from the Norwegian Petroleum Directorate.

Figure 3.4

Permanent reservoir monitoring of the Johan Sverdrup field. A network of seismic sensors cover about 120 square kilometers of the seabed.

Source: Reproduced by permission from Equinor.

Figure 3.5

A core sample.

Source: Reproduced by permission from the Norwegian Petroleum Directorate.

Figure 3.6

A well log.

Source: Reproduced by permission from the Norwegian Petroleum Directorate.

Figure 3.7

Production data sampled from a few days in April 2014 from the Volve field. The parameter registers include downhole pressure, downhole temperature, average choke size, average weight of mud, bore oil volume, and bore gas volume.

Source: Reproduced by permission from the Volve open data set.

Figure 5.1

An overview of the funnel model for the life cycle of prospects together with their underpinning data and associated evaluation routines.

Source: Artwork produced by Marius Mikalsen.

Figure 5.2

Simulation-based geological modeling of an oil reservoir.

Source: Reproduced by permission from the Norne open data set.

Figure 5.3

Linking well log data (detailed, narrow) with seismic data (crude, broad) by visually linking different rock combinations likely to correspond to each other in the two data sets.

Source: Reproduced by permission from the Volve open data set.

Figure 5.4

Input to well tie-in where two wells marked 15/9–19SR and 15/9–11 are calibrated against the seismic.

Source: Reproduced by permission from the Volve open data set.

Figure 7.1

Remaining oil reserves relative to initial (100 percent) resources. The size of the circles indicates the size of the reserves.

Source: Reproduced by permission from the Norwegian Petroleum Directorate (2019).

Figure 7.2

Map of existing seismic surveys completed in the Lofoten-Vesterålen-Senja area. Each straight gray line indicates a seismic survey, and those boxed in purple represent 3D, not just 2D, seismic.

Source: Reproduced by permission from the Norwegian Petroleum Directorate.

Figure 7.3

Outline of the sea observatory, with a camera installed on a crane on the seafloor in Venus to detect marine resources in the proximity of a coral reef (red rectangles, figure on the left). Photographs are taken every thirty minutes, transferred via a fiber-optic cable (center), and visualized via a web portal in real time (right).

Source: Reproduced by permission from the MAREANO/Institute of Marine Research, Norway. Art design by Elena Parmiggiani

Figure 7.4

Outline of the web portal developed as part of the Venus project.

Source: Reproduced by permission from the MAREANO/Institute of Marine Research, Norway. Art design by Elena Parmiggiani.

Figure 7.5

Map of the four different ice edge definitions listed in table 7.1.

Source: Reproduced by permission from Norsk olje og gass (see Johnsen 2020).

List of Table

Table 1.1

A road map for how and where the three aspects of knowing digital oil (objects, modes, and machineries of knowing) are addressed in part II of the book, Cases. Dark gray indicates the main focus, while light gray indicates a supplementary but not main theme.

Table 2.1

Summary of the economic significance of oil and gas in Norway along with the number of people employed in the oil and gas industry in Norway. Source: SSB.no and Norsk Petroleum.

Table 3.1

Summary of key digital tools for different geoscience disciplines.

Table 4.1

Overview of the project data managers’ tools for navigating for information.

Table 5.1

Overview of the three analytic modes of explorationists’ work practices when grappling with geodata.

Table 6.1

An overview of the different periods of digital sand monitoring.

Table 7.1

Four different methods of defining the ice edge, each relying on different data sets, as evident in discussions in 2017. See Rommetveit et al. 2017.

Preface

This book represents a long journey. It mirrors travels that make up much of my academic biography. Ostensibly, with a chair in information systems in a computer science department at an engineering-dominated university, I have developed perspectives in this book that engagements along this journey have shaped in important ways. I dwell on a select few to create a backdrop for what follows.

Trained as an engineer, I was drawn to logic for my graduate studies at the intersection of the humanities and informatics at the University of Oslo. Crucially, however, this interest was tied to logic as a language, not a purely technical discipline. Particularly influential were Husserl’s formulations of constructive mathematics and logic, based on his phenomenological perspectives. At the Norwegian Computing Centre, Oslo, I found myself in the middle of critical, socially informed discourses on the conditions, manifestations, and consequences of Scandinavian-based participatory modes of technology development. Disciplinary boundaries were porous. The field of science and technology studies (STS) had a formative influence on me, first through the Centre for Technology, Innovation and Culture, Oslo, and then at the Centre for Technology and Society, Trondheim. Alongside a theoretical curiosity about STS, I developed a growing empirical interest in large-scale (infrastructure) technology efforts with implicated standardization as, seemingly, this went beyond the existing participatory methods for technology development. Relocating to the Norwegian University of Science and Technology, Trondheim, I became attracted to the perspectives of—and, not least, experiences with—the politics of participatory and interventionist forms of organizational change being pursued at the Department of Industrial Economics and Technology Management. Cultural perspectives on standardization, objectification, and quantification out of the Department of Anthropology were important in broadening my notion of standardization.

My academic coming of age, then, is the result of stitching together a network of colleagues and collaborators from a variety of disciplines and camps. It has been driven by the instinct to challenge my own intellectual comfort zone, wary of growing too comfortable in any one place.

In a final comment on the theme of data science and artificial intelligence (AI) emerging in this book, I must admit that I had no intention whatsoever of revisiting AI; I had my fill a couple of decades ago. But recent demand from external partners from the private and public sectors in research projects—within oil, as I report from here, but also from my research stream in health care—nudged me toward the theme of datafication and data-driven approaches. Having spent much of my professional career explaining why various technology efforts had failed, I was intrigued by how data-driven data science, for particular purposes, apparently works in ways beyond what we have presently accounted for theoretically. In short, my curiosity was stirred by a works in practice, not in theory situation.

Trondheim, Norway, Autumn 2021

Acknowledgments

I have benefited greatly from interactions with a lot of people over the long period I have pondered and explored parts of this book. Providing an exhaustive list is prohibitive.

A number of close colleagues provided invaluable comments based on snippets and arguments, small and large, throughout the process of developing this book. With or without realizing it, their conversations supplied much-needed sparring about early ideas: Petter Almklov, Michael Barrett, Bendik Bygstad, Samer Faraj, Ole Hanseth, Vidar Hepsø, Jannis Kallinikos, Neil Pollock, Knut H. Rolland, Susan Scott, Georg von Krogh, and Robin Williams.

Several scholars contributed to my research during the seminars, visits, and events I attended in locations including Ascona, Barcelona, Cambridge, Copenhagen, Edinburgh, Oslo, London, Seattle, Trondheim, Umeå, Warwick, and Zurich: Christina Alaimo, Panos Constantinides, Ola Henfridsson, Jonny Holmström, James Howison, Steve Jackson, Alexander Kempton, Davide Nicolini, David Ribes, Susan Scott, Geoff Walsham, and Youngjin Yoo. By challenging my arguments, they helped clarify otherwise muddled thinking.

Many colleagues have provided inspiration, energy, and indirect support in ways I struggle to account for adequately: Margunn Aanestad, Jørn Braa, Kristin Braa, Gunnar Ellingsen, Øystein Fossen, Morten Hatling, Roger Klev, Tord Larsen, Morten Levin, Emil Røyrvik, Jens Røyrvik, Sundeep Sahay, Knut H. Sørensen, and Arild Waaler.

I am deeply indebted to Marius Mikalsen, Elena Parmiggiani, and Thomas Østerlie. They are coauthors of the series of published papers that make up the point of departure for the chapters in part II of this book. In reworking and molding these articles to fit within the book’s broader arguments, I have been in dialogue with and responded to feedback from Marius, Elena, and Thomas. Thus, it is reasonable to acknowledge their role in the respective chapters in part II as written with me.

I also want to thank the coauthors of other papers related to the arguments put forward here: Petter Almklov, Vidar Hepsø, Gasparas Jarulaitis, and Knut H. Rolland.

The book contains a number of figures and images, without which much would be lost. I thank these companies, agencies, institutions, colleagues, and open data repositories for granting permission to reproduce: Norsk Oljemuseum (Shadé B. Martins), Equinor, Irina Pene, Norne open data set, Norsk Olje og Gass, Norwegian Petroleum Directorate, Elena Parmiggiani, Shell, TechnipFMC, MAREANO/Institute of Marine Research, and Volve open data set.

I am grateful for help from the MIT Press, not least for the formative comments from reviewers. I also want to thank Geof Bowker and Paul Edwards for support and encouragement throughout the years. Justin Kehoe was a great help in navigating the final hurdles at the publisher.

Finally, I am thankful for the patience, space, and support granted by my family throughout this whole journey.

Part of the research in this book has been supported by Research Council of Norway grants 163365 Aksio, 213115/O70 Digital Oil, and 237898/O30 Centre for Research-Based Innovation (SFI) Sirius.

1 INTRODUCTION

Action and knowing are situated. Coined more than thirty years ago (Suchman 1987), the situated nature of our engagement with digital technologies has shaped many of the socially informed empirical accounts. Early and influentially, Zuboff (1988) studied the digitalization of work in industrial and office settings. She underscored the tactile and embodied competence of predigital daily work routines. For instance, in her study of an industrial pulp mill she emphasized the importance of the operators dipping their fingers into the pulp and tasting (!), smelling, and feeling the temperature and texture of the pulp in order to competently engage in the everyday running of the mill. In a similar vein, practice theory–based accounts underscore how our engagement with digital technologies is "emergent (arising from everyday activities and thus always ‘in the making’), embodied (as evident in such notions as tacit knowing and experimental learning), and embedded (grounded in the situated socio-historic contexts of our lives and work (Orlikowski 2006, 460; emphasis in the original). Hence, knowing is a situated knowing constituted by a person acting in a particular setting and engaging aspects of the self, the body, and the physical and social worlds (Orlikowski 2002, 252). Within the industrial sector empirically examined in this book, offshore oil and gas exploration and production, the typical image of an operator is a roughneck: smeared in oil and grease, hard hat on his (not her"!) head, wrenching loose a 31.6-foot drill pipe.

A broadly practice-oriented perspective—underscoring qualitative, embodied, situated action—has been highly influential and underpins many of the critical, socially informed studies of work and technology. It resonates deeply with my own perspective. However, I do have issues with what such a perspective risks leaving out. In this book I analyze digitalization as ongoing attempts, regularly met with opposition and setbacks, to quantify the qualitative. In other words, I explore whether the above outlined practice-based perspective might have overstated the role and scope of the qualitative in present-day digitally enabled practices of knowing (see figure 1.1).

Figure 1.1

Work practices in oil: traditional roughneck (top) versus control room based (bottom).

Source: Reproduced by permission from Husmo Foto/Norsk Oljemuseum and Shadé B. Martins/Norsk Oljemuseum, respectively.

Empirically, the transformation of work practices has been in full swing for quite some time in the oil and gas industry (Autor 2015; Thune et al. 2018). Roughnecks are increasingly rare. The majority of hydrocarbons produced on the Norwegian continental shelf is by subsea production facilities residing on the bottom of the sea, untouched, as it were, by human hands and remotely operated from onshore control rooms based on real-time sensor streams measuring temperature, pressure, and volume.

The digitally enabled transformations in offshore oil and gas of work practices, roles, and organization that I analyze in this book are not the result of radical, discontinuous change. This book does not promote the imagery of a great digital divide between, on the one hand, early or predigital practices and, on the other hand, more recent forms of digitalization with their emphasis on intelligent systems (artificial intelligence, or AI), blockchain, digital platforms, social media, and the Internet of Things (IoT) as found in more managerial strands of digitalization (cf. Brynjolfsson and McAfee 2014; McAfee et al. 2012; Davenport 2014). On the contrary, this book firmly subscribes to a perspective of technological change emphasizing evolutionary, small-step, socially negotiated change; ongoing experimentation regularly meeting with setbacks; and opposition or outright failure. Still, the cumulative changes over the last couple of decades, the time frame of this book, have significantly changed work routines and roles. The relevance and significance of prominent forms of digitalization during this period, notably that of IoT and data-driven approaches, is not their novelty per se (new digital technologies come and go) but the way they allow attempts (again, subject to negation, conflict, and opposition) at quantifying other kinds of tasks imbued with qualitative and/or tactile qualities. In other words, a fascination with the ongoing attempts at pressing the limits or scope of digitalization animates my analysis. The ambition of this book, then, is to balance a healthy skepticism of proclamations for revolutionary or radical change against an empirical, phenomenon-oriented openness to interestingly different aspects of the new in the old. To that end, a rough historic outline of digitalization within the industrial context under study is helpful.

Digitalization of the Norwegian offshore industry started modestly in the 1980s and early 1990s. The emphasis was on largely stand-alone process control systems. Picking up speed through the 1990s and 2000s, digitalization efforts shifted to enabling data communication, notably between on- and offshore operators, as well as access to sensor data (including downhole). These efforts, known alternatively as eField, intelligent fields, or integrated operations (Rosendahl and Hepsø 2013), challenged existing organizational routines and division of labor. During the last couple of decades, offshore personnel and tasks have been shifted onshore. Videoconferencing, email, and instant messaging are frequently used for communication and collaboration between offshore installations and the mainland. With the availability of real-time sensor data and new engineering applications for visualizing and manipulating this data, onshore engineers can also actively participate in monitoring, diagnosing, and controlling offshore processes (see figure 1.2).

Figure 1.2

Subsea installation for oil production.

Source: Reproduced by permission from TechnipFMC and Shell.

What this implies is that current, everyday offshore oil and gas work is significantly and qualitatively different from the outlined practice-oriented position with its emphasis on the tactile, embodied work characteristic of practices twenty-five years ago. Not in a radical move, but evolutionarily and cumulatively over a couple of decades, the content and context of work practices have profoundly changed. An offshore oil platform off Norway with its subsequently connected subsidiary production platforms as well as unmanned subsea facilities is today a massively instrumented production facility. For instance, the Ekofisk field in the southwestern part of the North Sea has about ten thousand real-time sensor feeds from different depths down in the well and from the topside, valves, processing, and transportation equipment. Sensor-based, IoT-rendered reality is a sine qua non for the work practices constituting everyday offshore operations. Compounded by a sharp drop in oil prices in June 2014, automating and/or shifting more and more functions to digitally enabled, remote operations is a chronic cry.

For instance, an actively pursued vision is that of the unmanned subsea factory. Until recently, oil and gas production on the Norwegian continental shelf has involved processing facilities, pumps, separators, and compressors located on floating platforms to inject water and gas into the geological reservoir. This is needed as, after an initial period of sufficient pressure in the reservoir, the production of hydrocarbons gradually releases and hence decreases the reservoir’s pressure. Left unchecked, a significant portion (more than 70–80 percent) of the oil and gas reserves could not be sucked up to the platform. After the initial period, when the reservoir pressure will push hydrocarbons to the surface only when punctured by a drill string, the pressure must be maintained during production by injecting gas or water into the reservoir. In this manner, oil recovery may be increased to 50 percent of the oil reserve of the reservoir, from a typical global rate of 25–30 percent. The processing facilities, separators, injectors, compressors, and pumps required for this must fight for the precious little space available on a floating platform. They are energy consuming (hence polluting). The vision of the subsea factory, then, is to move away from this situation. The ongoing aim is to fully automate the process, relying on electrical rather than today’s fossil power.¹ In parallel, existing manned platforms are increasingly being operated remotely, as announced by the dominant operator, Statoil (2017; now named Equinor), in 2017 and reports in the news (Johansen and Kristensen 2017; see figure 1.3 depicting a compressor component).

Figure 1.3

Compressor component for subsea factory.

Source: Reproduced by permission from TechnipFMC and Shell.

Accordingly, the issue in offshore oil and gas is not so much one of whether to substitute the manual, tactile, and embodied work practices of roughnecks with IoT-rendered, digitally enabled ones. The issue is one of sequence, scope, and pace. This transformation from manual to automated and/or remotely operated operations, however, is gradual and negotiated at every step along the trajectory. It is subject to contestation from unions and national safety authorities. The ongoing efforts of automation are hardly expressions of technological determinism. For instance, one union leader commenting on the reduction of offshore employees resulting from automation argues that they "request more compelling