Reservoir Engineers are taking good advantage of the power of Digital, High Speed, Large Memory computers to process Mathematical Reservoir Models. These are used to explain the behaviour of, and describe existing reservoirs, through use of field data. They also predict production changes which would result from the use of different production schemes.
An allied discipline within the Energy Engineering field that of Nuclear Reactor Power Plant design makes use of similar simulation models for prediction of system responses. While significant differences between the time spans studied by these two engineering sections exist, similar mathematical equations, numerical approximations, error minimization methods and presentations of results prevail.
The paper briefly discusses Mathematical Reservoir simulation techniques and their shortcomings. It describes certain specific Nuclear Engineering Modelling techniques in greater depth. Comparisons between the techniques of the two disciplines are drawn. 1 t is suggested that confidence level estimation approaches used in Nuclear Engineering could be applied to Reservoir modelling. It is further suggested that simulation procedures are available from other disciplines, such as Nuclear Engineering, which can be employed by the Petroleum engineer as the scope of systems covered by simulation expands.
It is often difficult to test, directly, hypotheses which seek to explain or predict complex processes. The systems to be studied may be inherently dangerous, sensitive, or have unmanageable physical dimensions, time and cost requirements. One useful method of exploring relationships and assumptions and analyzing and explaining in such cases, is to simulate the process with a model. In simulation modelling descriptions or analogies are used to develop logical representations of the object of study. These approximate its behaviour or characteristics. They ease visualization and can be more conveniently manipulated than the actual process or object.
Such models may be theorized concepts, or physical emulations, either scaled or analogous, or they may be symbolic mathematical descriptions. Physical models are descriptive and the least abstract. They resemble the real prototype. Analog models are more abstract and use a set of properties which are different from but correlative with those of the process in question. Models which characterize a system through the use of mathematical relationships are the most abstract or symbolic or dissociated in form from the object of inquiry. They are, nevertheless, general, precise and can be manipulated exactly by utilizing the laws of mathematics. Their functions are designed empirically or conceptually to apply appropriate transformations to input data. The results are steered to match the process or object under study or to reconstitute options under varying conditions. Further distinctions can be made among models by degrees of linearity, stability, constraint and transience, etc. In the case of mathematical models, if the processes described are simple and precise they are Deterministic. If they are complex and predictable only within degrees of probability they are Stochastic or Hybrid. They may also be static or dynamic and within these categories steady and non-steady cases will be considered.