Incomplete or ambiguous knowledge about communication is an important area of uncertainty in reservoir simulation. We document several cases in which a new quantitative and deterministic method, strontium fingerprinting, can be used to give exact predictions of reservoir communication and flow units. In all cases, the isotopic composition of strontium (87Sr/86Sr) dissolved in the formation water was used to monitor compositional variability of formation water. This natural isotopic tracer is particularly useful as it can be measured simply from conventional core samples by extracting residual salts which have precipitated in the pore spaces as a result of formation water evaporation during storage. We have termed this technique "Residual Salt Analysis" (RSA). It enables detailed sampling in any part of a well for which there is core available.
So far, RSA data have tended to yield rather ambiguous predictions. A quantitative method for treating the data has been lacking, as well as an empirical foundation of interpretations. However, we have tested the following hypotheses with empirical data: Intrafield variability of formation water chemistry may be closely correlated with reservoir compartmentalization. If two reservoir units are in good flow communication, their formation waters are likely to have been homogenised by flow and diffusion. On the other hand, a lack of flow communication will inhibit water mixing and thus preserve variations in water compositions. Thus, whatever the cause of intra-field variations in water composition, two reservoir units that have a similar water chemistry are more likely to be in good flow communication than two units with different water compositions.
The Sr-ratio values obtained from five wells (n = 117) show a natural tendency to clustering. This contrasts with, for instance, a statistical normal distribution of the data. Moreover, data from distinct reservoir units in different wells are aggregated as distinct clusters in nearly all cases. These observations prove that the hypotheses are generally correct. The second step is to develop a non-parametric method for quantifying and comparing clusters. With such a method, it is possible to make precise predictions of inter-well flow communication. In the present case, the 3-D flow architecture of the Smorbukk South Field (Haltenbanken, Norwegian continental shelf) could be predicted based on core samples from three wells, and the communication between various zones in two wells was predicted in the Smorbukk North Field.
In the Smorbukk Fields the main problems are lateral compartmentalization by small faults combined with extensive quartz cementation and horizontal compartmentalization by areally extensive shales. We predicted the position and lateral extent of the barrier between the reservoir zones Tilje 1 and 2 and the between-well communication of these zones in Smorbukk North, and the same also for Ile and partially Garn formation in Smorbukk South. Using RSA data from the three wells studied in this field, it was possible to predict that the Garn Fm. has good internal vertical communication, but is laterally divided into two or more isolated areas, probably the result of sealing faults. The Ile Fm. forms three laterally isolated compartments over the central part of the field, and Tilje 1 and 2 consist of two lateral compartments each.