**Author details**

146 New Technologies in the Oil and Gas Industry

Criterion of convergence

correlations are needed. Size of the columns

less layer could be retained.

or minimum mismatch as the objective function.

**6. Conclusions** 

Number of facies simulations (N) per iterations

This parameter can have a major influence on the optimization: with numerous simulations the process has a wide variety of possibilities to choose from, and can build a more precise

best-facies cube, but if the simulations are very alike, the choice of N will not matter.

**Figure 24.** Influence of different number of iterations (1 to 6) and simulations (20, 30 and 40).

The criterion of differences is more precise than the correlation, since the correlation gives us a value of the similarity between two sets of data, while the difference is more susceptible to different patterns or small variation in the patterns. But to calculate the probabilities the

In the previous test the cubes were divided in 4 layers with 20 values each column. Since the correlation has some sensitivity to the number of data used, the results also change. For a largest column the convergence will tend to be slower, since it is more difficult to match an entire column than piece-wise matching. In general we recommend that the number of layers chosen depends on the vertical resolution of the seismic data, i.e. for lower resolution

The algorithm proposed uses two different approaches to create a stochastic model of a reservoir property that are conditioned to both well and seismic data, since this two data are totally different, this process was to be an iterative one, that used the maximum correlation Hugo Caetano *Partex Oil and Gas, Portugal* 
