INVERSE MODELLING OF HYDRAULIC TESTS WITHIN FRACTURED CRYSTALLINE ROCK BASED ON A MARKOV-CHAIN TRANSITION-PROBABILITY GEOSTATISTICAL APPROACH

The work presented here investigates a series of interference hydraulic tests conducted at the Olkiluoto Island site (Finland), where detailed hydrogeological characterization is underway to establish the suitability of the site for long-term disposal of the Finnish used nuclear fuel. This work is conducted within Task 7 of the Äspö Task Force on modelling groundwater flow and transport of solutes, whose objectives are to quantify the reduction of uncertainty in the properties of the fracture network at Olkiluoto Island and to assess the Posiva Flow Log (PFL) measuring probe, which measures flow rates within boreholes.

The hydrogeological model considered here covers an area of about 400 m x 400 m at the center of Olkiluoto Island, which is composed of fractured crystalline bedrock. In that area, a series of interference hydraulic tests have been conducted in boreholes that were either packed-off or left open. The fractured bedrock is represented as a 3D stochastic equivalent porous medium using a geostatistical approach characterized by a transitional-probability model based on Markov chains. The bedrock is represented as a series of facies defined from fracture density, orientation, and transmissivity obtained from PFL measurements.

Inverse modeling of the interference hydraulic tests is conducted using measured hydraulic heads and PFL flow rates within boreholes as targets for calibration of the hydraulic properties of the bedrock facies. Two main options are considered for inverse modeling: calibration based on hydraulic head alone or on both head and flow measurements. The latter option provides narrower confidence intervals for the calibrated hydraulic properties, therefore reducing the uncertainty associated with these properties. The simulations further demonstrate that the proposed conceptual model based on the definition of fractured rock facies from PFL flow measurements correctly reproduces groundwater flow at the site, providing an alternative to discrete fracture network models.