As an industry, E&P rarely relies on rigorous reservoir modeling in the development phase, and even less during exploration. Reservoir modeling is not just for mature reservoirs anymore. Technology exists today that will bring E&P modeling into the 21st century.
Exploratory wells today are typically drilled with a “best effort” model: maps representing interpretations of seismic data and logs from nearby wells or analogous reservoirs. These maps may appear to be definitive descriptions of reservoir boundaries and rock and fluid properties, but important uncertainties and flaws are always present. Even with the best seismic and petrophysical technology and expertise, other significantly different interpretations are always possible. Explorationists will tell you that they carry multiple different hypothetical models in their heads. Without a rigorous way to manage these mental models, typical industry practice is to test these hypotheseswith the drill bit.
At one time in the medical field, surgery was done with best-effort models: x-rays created maps of the patient’s interior and the surgeon made incisions to reach the problem area. Today, true 3D imaging and 3D computer modeling isalways done for complex procedures, and doctors experiment with the model, not with the patient.
The medical industry has a great advantage over ours, though. Human anatomy is highly predictable, but every O&G reservoir is different. Surprise is the rule, not the exception. We have excellent 3D imaging technology, and even our version of laparoscopic surgery, but our computer modeling technology is decades behind other industries. The O&G industry continues to use an outdated approach (finite difference modeling) that medicine, manufacturing, aerospace, and other industries abandoned long ago for being too slow and imprecise.
3D seismic surveys provide knowledge about prospect location, but fall short in predicting important factors such as recoverable hydrocarbons, faults, fluid contacts, and reservoir drive mechanism. Modeling allows for sound contingency planning for the inevitable surprises. The underlying physics governing fluid flow in porous rock has been studied for centuries. The problem is, for the most part, we haven’t seen real modeling work done before wildcat drilling starts and little until most development drilling is finished. Until recently, the “test” part of modeling has been too cumbersome and imprecise. Studies usually take months, and fine details of a precise model had to be glossed over to get any results.
All this is changing. Finite Element technology available today gives E&P professionals a rapid-cycle modeling tool without compromising essential precision and model fidelity. Modeling with Resolve 3.4 can begin during exploration, even before the exploration well is drilled. A virtual reservoir is described by a set of untested hypotheses and corresponding models. This is where data is incorporated and available knowledge/information goes into an initial, computer-based characterization. It is crucial at this and every future stage that model variations can be created and tested quickly - in minutes and hours, rather than days and months - to support pending time-critical decisions.
As exploration and development proceeds,virtual aspects of models will be updated as each component becomes actual. The first really critical time comes when the exploration well reaches target depth. If no fatal surprises have been encountered to this point, logs, cores, fluid samples, and pressure measurements provide the first direct subsurface measurements that can be used to substantially refine hypothetical alternatives.
Uncertainty limits narrow, and formation and fluid understanding in the pay zone can be better characterized, especially near the discovery well. With refined expectations over the entire reservoir, forecasts and decisions about completions and facilities design improve.
After the well is completed, the first far-field detection of reservoir character becomes possible. Flowing pressure and rate profiles illuminate growing expanses of the reservoir. This is where the hypothetical models begin to diverge seriously in forecasts. Withpredictive fluid flow modeling, implausible models are discarded - those with incorrect drive mechanism or fault transmissibility or frac efficiency. New model variations may be required.
In this early phase, the reservoir learning rate is greatest. The virtual reservoir has become a highly predictive model, with one or two remaining acceptable models and much tighter uncertainty ranges. The model and forecasts can be checked/updated, and the last modeling phase begun. In surveillance, most original interpretations have been discarded, and a model consistent with all of the data has emerged.
The model is now the center of financial decisions. Revenue and cash flow forecasts, drilling and operational needs, etc. can be made quickly, and improved. If reservoir performance begins to diverge from the model’s predictions, this can be detected quickly - often this is the earliest sign of well or reservoir performance degradation. Ideally, model-based surveillance should continue until plug and abandonment.
A number of operating companies are adopting thislife-cycle modeling approach and are carrying out reservoir experimentation without risking expensive mistakes in the midst of current challenges, including smaller, more complex targets and rapidly depleted finds. Decision cycle times today are highly compressed, and the expert resources to make those decisions are becoming scarcer. Life-cycle modeling can be the salvation - reducing F&D costs while easily fitting within those time, and people, constraints.
John Mouton
Chairman of the Board, Object Reservoir
Gene Ennis
President & CEO, Object Reservoir
This page reflects viewpoints on the political, economic, cultural, technological, and environmental issues that shape the future of the petroleum industry. Offshore Magazine invites you to share your thoughts. Email your Beyond the Horizon manuscript to Eldon Ball at[email protected].
