Reservoir Engineering in Modern Oilfields:

Vertical, Deviated, Horizontal and Multilateral Well Systems

by

Wilson C. Chin, Ph.D., M.I.T.

Stratamagnetic Software, LLC

Houston, Texas and Beijing, China

Table of Contents

Preface, v

Acknowledgements, viii

 

Preface

Computational simulation has been heralded as the most significant advance in modern engineering and analysis, bringing untold increases in productivity and cost-effectiveness to the design process. However, its capabilities and potential are often misunderstood. For example, in high speed aerodynamics where the environment – namely, simple dry air – is very well characterized, state-of-the-art partial differential equation solvers and grid generation algorithms can predict properties like lift and drag well. Nonetheless, when this author, on earning his Ph.D. from M.I.T. and joining Boeing’s well known C.F.D. group, asked his new Manager, "What types of answers can computers produce?" the response was sarcastic. This member of the National Academy of Engineering and a founding father of the profession, would reply, "Any answer you want." And to be sure, hundreds of three-dimensional simulations would be performed for every set of available wind tunnel data – and only carefully calibrated runs were used to "predict" flow consequences at off-design conditions. Boeing planes fly reliably and efficiently, but modeling provides only a guarded window to engineering design.

Now consider reservoir engineering or flow simulation from huge underground reservoirs. Grid blocks are typically hundreds of feet across in each and every direction. Properties like porosity and anisotropic permeability are inferred from core level data obtained in widely separated delineation wells. Unseen faults, shale streaks, fractures and undulating layers may lurk beneath the surface. Drive mechanisms and reservoir boundaries may not be known accurately. Multiphase effects and coning are possible, which may completely invalidate baseline single phase flow simulations. Well radii cannot be resolved on the scale of large grid block analyses and "productivity indexes" (or "fudge factors," in the colloquial) are typically used. So can any normal person seriously expect usable predictions, let alone numbers that might guide investment decisions that routinely put billions of dollars at risk? Certainly, one cannot abandon simulation and turn back the clock – but its limitations and roles must be prudently understood and a practical philosophy put into place.

In reservoir engineering, as opposed to airframe aerodynamics, one must be careful to understand that small-scale events will likely be predicted inaccurately. Simulation should be used to understand large-scale consequences associated with dominant parameters, e.g., the pressure levels in the well and in the farfield, multilateral well topology, the locations of specific wells. Results should be used qualitatively. It would be unrealistic to assume that production differences, say less then twenty percent, would be even credible. However, if one scenario proved twice as productive as another, well maybe that one is worth a second look. In taking this approach, we are not promoting a doomsday mentality. But the rock permeability predicted, say from formation testing pressure transient log responses, is unlikely to be too correct for simulator input – it is corrupted by small depths of investigation, mud invasion and hosts of other uncontrolled effects. However, the performance of a long horizontal well or an all-encompassing multilateral relative to a traditional vertical well is likely to be correct, even if simpler inputs like viscosity or porosity are not accurate.

This author has found this philosophy useful in other petroleum applications. For instance, in studying cuttings transport in horizontal wells, it is unreasonable to model the fluid mechanics past every conceivable piece of moving rock – chips which may be spinning, tumbling and falling in space. Modeling the flow past a stationary Boeing wing in clean air is difficult enough – so our suggestion derives from experience and not pessimism. However, effective hole cleaning correlates with the mean viscous stress at the low side of the borehole annulus – that high stresses actually "rub away" debris provides the correct physical explanation and guiding design principle. So in cuttings transport, our simulations help control stresses themselves, in order to understand how these are affected by rheology, annular geometry and flow rate.

In reservoir engineering, we define our approach by asking, "How do key parameters like well location, multilateral topology, pressure levels and drive mechanisms affect production? We should and will be concerned with large-scale qualitative consequences as opposed to "detailed results" which cannot be entirely accurate given errors due to uncertain input data. In our work, we thus focused on developing a simulator that offers a reasonable number of layers and a respectable level of grid density – one that runs rapidly and stably all of the time but is not in itself excessive – and a useful product that produces the simple credible suggestions needed for what must ultimately be subjective decisions.

This philosophy has guided the author’s work in numerous petroleum disciplines over the past three decades, for instance, in formation testing, electromagnetic logging, Measurement-While-Drilling design, and drilling and cementing rheology modeling. Ease of software use, low licensing costs and reduced barriers to entry are also paramount objectives. In reservoir simulation, it is not uncommon for oil companies to spend tens of thousands per license, purchase sophisticated hardware and resource-consuming graphics, and provide weeks upon weeks of training. But, as explained, the returns are often limited.

And as of this writing, few user manuals are written with illustrative calibration examples and even fewer will display real well systems with their computed pressure distributions and production rates. The reservoir simulator discussed in this Handbook, the first of several from Wiley-Scrivener, provides capabilities that no other commercial product offers. It is not a "black box" with all results to be taken at face value. Computed results must be prudently judged. But the theory and algorithms are fully explained in several publications – the methods have won numerous awards from leading organizations over the years.

A significant contribution, however, is the user interface developed under multiple operating company funding that allows engineers and novices alike to "sketch any well" and see large-scale flow consequences almost immediately – not crude answers but computed results grounded in rigorous documented and validated theory. And to ensure that the methods are useful immediately to readers, almost two dozen examples are introduced which clearly highlight the capabilities of the tools newly available. It is my hope that MultisimTM will make a difference to small companies as well as large, to students as well as engineers, and to doubters as well as experts. Like other projects that this author has published during the past two years, the work has long been a labor of love and an obsession to do it right. And doing it right and explaining the problem clearly and simply are more important now than ever before.

Wilson C. Chin, Ph.D., M.I.T.

Houston, Texas and Beijing, China

Email: wilsonchin@aol.com

United States cell: (832) 483-6899

Acknowledgements

The author wishes to express his appreciation to his colleagues at Schlumberger, Halliburton, BakerHughes, British Petroleum, GE Oil & Gas and other companies for their insights and suggestions over the years and for shaping his approach to reservoir engineering and simulator development. He is especially grateful to Phil Carmical, Acquisitions Editor and Publisher, for providing him the platform to communicate important ideas in highly technical books and the opportunity to serve as Series Editor for Advances in Petroleum Engineering. Our new Handbook Series, however, serves more practical but related objectives. Content is developed that enables users to solve important and modern problems using the latest technology available and in "plain English" terms. The methods are not "me too" in nature; they are unique, but no attempt is made to be all things to all people. It is often said that "one does not understand, until one can compute it." Few recent oilfield achievements have approached horizontal and multilateral well design in economic and practical significance, the subject of this first Handbook – and with the methodologies presented here, we believe that we have closed an important gap in petroleum engineering by reducing planning and modeling costs by several orders of magnitude. Again, I thank Scrivener Publishing and John Wiley & Sons for this important opportunity to make a real difference.