Quantitative parameter modelling K.mod is designed to extract essential information from log data in order to: predict non-recorded parameters (PHI, K, Sw) , reconstruct missing or poor quality measurements and therefore compensate for bad hole conditions, environmental effects, acquisition problems, etc. bring solutions for scale shift management from core to reservoir scale, potentially reduce the need for coring and plug analysis for the subsequent appraisal wells by comparing well log and core data.

	Parameters can be reconstructed or modelled directly from log data, via an interactive learning process.
	Multi-Layer Perceptron: a powerful non-linear modelling tool that retains all the original variability in the data.

Fully quantified uncertainties

K.mod is not a "black box" tool: the users keep full control of the input parameters and receive clear feedback on the log quality and model quality, at all times.
Uncertainties can be managed:

	on inputs: back propagation method to check the contribution of each input,
	on output: self-organized map is categorizing data samples in the training and validation data for their effectiveness in modeling the target data,
	possibility to weight inputs to force the model to reach extreme values.

Interactive, easy-to-use and very fast

	K.mod is based on a complex technology but remains easy-to-use.
	It is a straightforward but efficient tool that offers a simple and intuitive interface for an enhanced interpretation and a more accurate reservoir characterization.

| Top | Back to Techlog |