Techsia offers integrated studies for better reservoir characterization, involving a thorough management of uncertainties.

1. Data Loading

Data are loaded from individual files or pre-existing third party software projects. Data can include wireline or LWD/MWD log readings, array-log readings, core plug quantitative data and alphanumeric core descriptive data. Data may come from one or more wells; the Techlog project database is a fully multi-well data store.

For reliable and efficient data loading, Techsia uses the Techlog^©, platform.

2. Core data storage, processing, integration and interpretation

To perform specific processing tasks within one user environment, e.g. the calculation of capillary functions, resistivity index, formation factor, pore-size distributions, relative permeabilities etc., and the indexation of these parameters directly against log data using a self-organized map.

To combine the maximum amount of data available to generate “petrophysical logs”, which are fully representative of the variation in core data upscaled by rigorous methods in order to be compatible with the scale of the wireline or MWD log data. The upscaling is performed by reference to either continuous core description data (possibly created by picking on core images) or to the most finely resolved quantitative measurements on the core (e.g. mini-K or core-gamma) by 1D Kriging interpolation (several external drifts are possible).

To partition the core data into “petrophysical groups” that reflect zones in which storage and transmissivity of fluids are likely to be similar. Techsia’s extensive experience in techniques of classification, correspondence analysis and exploratory statistical tools such as principal components analysis is applied for this task.

To rapidly deliver clear, integrated and comprehensive results , Techsia uses the Techcore module from Techlog^©,.

3. Data QC and data preparation

This is a key point in the overall workflow and its efficient completion can materially influence the success of prediction.

Detailed review of the quality of the log and core data.

Comparative examination of data in numerous multi-variate displays and contexts.

Data editing, repair, reconstruction and depth shifting to correct or remove abnormal data.

Choose a training set of data selected to take account of all the variability present in the core and log data over the interval of interest.

For best results, Techsia uses the Techlog^© platform and the QC Field module.

4. Facies delineation from log data using the Ipsom technology

To investigate data structure inherent in the log data set using “unsupervised” ordering and data partitioning methods.

To perform an initial indexation of the data partitions using the raw core descriptions. Quantify the goodness of fit between raw core descriptions and the resulting predicted facies curve using the Ancor tool for correspondence analysis (observed vs. predicted data analysis mode).

To refine the correspondence of the given geological partitions and observed log data partitions by reference to data viewers and log displays; by this means the user keeps control of the interpretation and need not accept just the statistically based model. Derive an optimal representation of the core facies data within the log data set by iterative adaptation of the core descriptions.

To propagate the final indexed log predictions based on the refined model into test wells.

To achieve the most robust prediction models, Techsia uses the Ipsom module and the Ancor tool.

5. Parameter: prediction (K, phi etc.) and data repair using K.mod

To develop one or more neural network models as appropriate for predicting (inter alia) the formation permeability and porosity distribution. Classic techniques such as Linear Discriminant Analysis can provide guidance on the best log data to use to predict any particular parameter of choice from a network model. The quantitative assessment of the quality of the network prediction and the management of the associated uncertainties is available by use of a combination of technologies (the indexed and probabilised self organising map and the multi-layer perceptron).

To model log responses in intervals affected by eg. poor borehold quality, back propagation NN are used to utilize all available data. Automatic patching of good and bad data provides an efficient and easy-to-use solution.

For high-end results, Techsia uses the K.mod module.

6. "Conventional" log interpretation with uncertainty analysis

To evaluate log data for (inter alia) lithology, porosity and fluid saturation using conventional log interpretation algorithms as appropriate. Interpretation may be by sequential analysis or simultaneous probabilistic analysis of various available log data.

To investigate model uncertainties through Monte Carlo analysis and inversion.

To achieve clear results and appreciate the uncertainties, Techsia uses the Quanti module.

7. Hydraulic units elicitation

To delineate “hydraulic units” through evaluation of (inter alia) permeability, porosity and lithology characteristics (e.g. volume of shale in clastic formations). To integrate fully to log and core domains by involving available capillary pressure (Pc) data to the maximum extent.
The partitioning of the formation according to petrophysical characteristics is established by applying the indexed and probabilised self organizing map to these derived logs. Saturation may be a useful arbiter curve in this process.

To refine the observed petrophysical class partitions. Multivariate data displays and full interactivity between processes provide clear insight to the user.

For the most consistent interpretation, Techsia uses the Ipsom module.

8. Analysis of correspondences

To quantify the goodness of fit between the adapted core descriptions and the final facies prediction using the “observed vs. predicted” data analysis mode.

To quantify the relationships between the log-based geological facies prediction and the petrophysical classes of formation characteristics using the “Lithofacies vs Petrofacies” data analysis mode. This will provide associative probability distributions that link log-derived facies to the occurrence of hydraulic units.

To provide unique insights into the results, Techsia uses the Ancor tool.

9. Reporting

To get explanatory compilation of methods and results and to export digital quantitative results.