![hampson russell normalize logs hampson russell normalize logs](https://images.slideplayer.com/32/9915075/slides/slide_16.jpg)
Hulten (1984) provided a comprehensive geologic description for the Waseca formation in and around the Pikes Peak field.
HAMPSON RUSSELL NORMALIZE LOGS SOFTWARE
The chosen algorithms were obtained from Hampson-Russell Software (Sparse Spike Inversion and Model Based Constrained Inversion) and the NN method was developed in Mathworks MATLABĀ®.The Pikes Peak oil field is located 40 km east of Lloydminster, Saskatchewan (Figure 1) and produces heavy oil (12*API) from the Waseca sands of the Lower Cretaceous Mannville Group. The second technique, used herein for comparison purposes, is based upon seismic trace inversion. The first technique is based upon the use of genetic algorithm, back propagation neural network scheme, to estimate sonic logs from seismic attributes used as input data to a specially trained neural network. In the pages to follow, two techniques are explored in order to obtain a rational prediction of sonic logs along planned well location using seismic data. The motivation of this paper is to explore the use of seismic data for characterizing the overburden. Even though seismic interval velocities have been, more recently, used to predict pore pressure distribution, sonic logs are the source of data preferred by prediction algorithms. Extensive use of offset wells and a best guess-based sonic log construction are the tools more often used for providing data for well planning. The estimation of sonic log data along a given well trajectory, for drilling planning purposes, is not a straightforward operation and several simplifying assumptions are made. Traditionally, sonic log data is used to estimate rock properties and pore pressure along a proposed well trajectory. Hidden information in the seismic section should be searched for in order to improve the quality of the drilling operation and, consequently, for thereduction of risks. However, the characterization of the overburden can improve very much from an extensive analysis of the seismic data, mainly when used in conjunction with other data such as logs, drilling reports and geological maps. The information conveyed by seismic data is normally used for exploration and production purposes. Seismic data is not used intensively during well design. The result achieved in this research suggests that the application of the proposed methodology when P-wave log information is needed between wells far away from each other leads to acceptable predictions. Recursive inversion of seismic trace was utilized for validation and qualification of uncertainties. This methodology applied to similar data from Gulf of Mexicoproved to be quite satisfactory. The predicted results are compared with the actual sonic log data acquired at the well. Subsequently, the sonic log at a second well was predicted using attributes from seismic traces at the well location as input data to the trained NN. At the first well, the generated sonic log and the basic attributes of the nearest seismic trace were employed for training the NN. The trained system was tested in two well locations. At a certain well location, P-wave logs and basic attributes of a seismic trace were used as the NN training dataset. Well logs and fundamental attributes of seismic traces along a 2-D, migrated seismic line from Namorado Reservoir (Campos Basin, Brazil), were used to create and test the method. A methodology that utilizes neural network (NN) architecture was developed for predicting P-wave sonic logs.