**8. Discussion and inference**

The standard long term instantaneous shut-in pressure revealed reasonable pact with the magnitude of the vertical principal stress component obtained from overcoring at the site. In all the sites tested, this was the only instance in which a viscous fracturing fluid had to be employed specifically to enable a crack to be initiated.

There was no indication of crack spin on the impression packer images.

The results indicate the effect of test fluid viscosity on the ability to reliably estimate the magnitude of minor stress component in the horizontal plane from the long term instantaneous shut-in pressure when crack initiation under a seal is suspected. While the agreement was acceptable, for practical purposes, for the tests conducted with water (especially considering the relatively severe influence of experimental errors at the absolute stress levels involved) the discrepancy in the case of the tests conducted with oil was disproportionate. It was also noticed that the relative differences between the tangent intersection and tangent divergence estimates for instantaneous shut-in pressure decreased as the viscosity of the test fluid intensified.

Re-pressurization of the test zones originally tested with oil or hydraulic oil produced instantaneous shut-in pressures and crack reopening pressures consistent with the results obtained using oil as the only test medium. Testing using a combination of fluids such as this may represent a practical means. These results have important implications in the field wherever the hydraulic fracturing stress measurements are required in fractured and porous rock mass. It is suggested to have a re-look at the long-standing view that the hydraulic fracturing method is not suitable for fractured and porous rock mass. But this study has disproved this assumption. In situ stress may vary from point to point, and method to method in a rock mass, and may have different values when measured over different volumes. Such variations are intrinsic and should not always be seen as anomalies or errors in the measurement themselves and cannot be concluded that no comparison or correlation can be drawn from different methods [18].
