Different approaches to reservoir modeling have divided geoscientists in the past. New technology attempts to cross that divide.

The word "coach" has different meanings on different sides of the Atlantic. In Britain, most people think of a type of bus; in the United States, it commonly means a football trainer. The term "reservoir modeling" too has different meanings. But unlike the English language, which is slowly separating along the Mid-Atlantic Ridge, advances in technology mean the different perceptions of reservoir modeling are coming closer together.
Modeling
To some, reservoir modeling means modeling reservoir architecture and fluid flow through the reservoir. For most people, the term reservoir simulation describes the fluid flow, so we'll concentrate on using reservoir modeling to describe the static properties of the reservoir, its architecture.
Structural models
Even within the subset of the static model, reservoir modeling can mean different things to different people. To a structural geologist, or any geoscientist or engineer who works with significant structural complexity, it frequently means the ability to design and interact with a 3-D model that deals with the interplay of the main structural elements in a reservoir: surfaces and faults.
Models of this type have proved invaluable in the interpretation of structurally complex fields. These models allow a variety of users the ability to work with the same model at the same time. Thus, because of good visual tools, a drilling engineer can more easily understand why a well trajectory needs to bend around a fault to reach its target. These models also allow for a variety of surfaces to be visualized at the same time. This helps with quality control. For example, crossing surfaces can be easily seen and corrective action quickly taken.
Heterogeneous models
To another group of people - geologists, geostatisticians and engineers who work with heterogeneous reservoirs - the term reservoir modeling has meant a series of tools to help understand heterogeneous reservoirs with significant internal complexities. These typically are more mature reservoirs with significant production history. It is not that mature fields are more geologically complex than recently discovered fields; it's just that more data is available, so these fields are much more complex than initially believed.
Until fairly recently, serious software tools that were good at helping users understand internal reservoir heterogeneity were difficult to use. So specialists in oil company research groups and central technical services departments did much of this type of work. Not only were these tools not very user-friendly, but frequently they produced their results in a "black box" - a regularly spaced grid that either does not incorporate faults or does not rigorously deal with stratigraphic continuity and volumetrics across faults.
The best of both worlds
Technology developments incorporated in software tools such as Roxar's IRAP RMS suite allow users to deal with heterogeneous models in the context of their structural environment. Thus the two elements of static reservoir modeling are coming together. In addition, they are coming together in a user-friendly way such that the entire asset team can readily work with the reservoir model. Geoscientists can model internal reservoir heterogeneity in a structurally complex environment. This provides a deeper understanding of how fluids can move through a reservoir. For example, knowledge of how sands are juxtaposed on opposite sides of a fault allows users to put better transmissibility multipliers across faults in dynamic simulation models.
A reservoir model will never represent reality, merely an understanding of a reservoir given the available information at the time. Decisions about how best to manage the reservoir are based on the output of dynamic flow simulators. It is therefore in the operator's best interests to put the "best" static reservoir model into the flow simulator. This model may be a range of models, as in an uncertainty study. These may incorporate a range of scenarios or a stochastically generated series of realizations of the same range of input data, or both.
Workflow management tools in reservoir modeling packages facilitate these types of study.
The future
In the broader sense of static and dynamic reservoir modeling, streamline technology allows geoscientists to easily and rapidly model fluid flow in models containing many millions of cells. Users are able to easily investigate fluid movement at the scale with which reservoir heterogeneity is modeled. This eliminates the need for upscaling, with its accompanying data loss. Because of some assumptions inherent in the streamline method, the method is unlikely to be a full replacement for finite difference simulators; nevertheless, streamline techniques are a quick and useful method for obtaining valuable information on dynamic reservoir performance.
What does the future hold? The future is likely to contain better links for getting data in and out of models, more comprehensive toolboxes for structural and stratigraphic complexity, the ability to deal with bigger models faster, and ultimately a better understanding of the term reservoir modeling.