Evaluation des contraintes en place à partir d'essais de leak-off. Analyse bibliographique et application à des cas concrets
Evaluation of in-Situ Stresses Based on Leak-Off Tests Data. Bibliographic Analysis and Application to Field Examples
Institut Français du Pétrole
L'essai dit de leak-offconsiste à augmenter la pression dans un puits, au droit d'une formation, jusqu'à la limite de rupture de la roche. Cette opération sert principalement à évaluer la pression de boue maximale admissible. Depuis une décennie on essaie d'utiliser la pression en leak-off dans les puits verticaux pour évaluer les deux contraintes horizontales en place. Une variante de l'opération précédente consiste à aller jusqu'à rompre la roche et propager la rupture : c'est la micro fracturation . L'interprétation des pressions obtenues reste basée principalement sur la mécanique linéaire de la rupture : paramètres élastiques et ténacité de la roche sont constants. On montre comment quatre facteurs - rapport des contraintes recherchées, existence ou non de fissure(s) préexistante(s) en paroi, pénétration du fluide dans ces fissures, filtration du fluide - peuvent conduire à des niveaux et des évolutions de pression très variés. Ils sont cependant insuffisants pour expliquer quelques observations courantes : la partie non-linéaire de la courbe pression-temps, la valeur du maximum de pression, sa variation avec le débit injecté en fracture. On montre comment la prise en compte d'effets poroélastiques en bout de fracture et d'un endommagement mécanique en paroi de puits pourrait permettre de faire mieux. Deux grandeurs apparaissent comme essentielles à l'interprétation : la limite supérieure de linéarité en pression, la pression de fermeture de la fracture. L'application à trois cas concrets montre que, dans les couvertures, il faudrait probablement évaluer la pénétration du fluide dans la fracture à l'initiation de celle-ci et modéliser l'initiation de fracture en tenant compte d'un endommagement mécanique de la paroi rocheuse. La deuxième partie de ce programme est en cours.
Abstract
This summary contains formulas (***) which can not be displayed on this screen. The leak-off test technique is used in wells during drilling and/or completion operations. It consists in increasing the fluid pressure in the well in front of a rock formation up to a limit corresponding to the initiation of a rupture in this rock. The main purpose is to estimate the maximum admissible mud pressure during drilling. However through theoretical analyses laboratory experiments and field work, ways have beeen proposed to extract from the leak-off tests performed in vertical wells the values of the horizontal principal stresses. If fluid injection is carried on up to propagation of a fracture in the rock, the operation is then called micro fracturing . Usually a leak-off test or a micro fracturing operation are performed under a low and constantfluid injection rate. The corresponding pressure increase is first linear, then non-linear above a pressure p index (nl). It reaches a maximum p index (b) (break-down pressure) and then a lower and quasi constant level p index (e) (extension pressure, i. e. pressure needed to extend the fracture). If fluid injection is stopped and if the fluid mobility is low enough, the pressure lowers to p index f, the closure pressure. The problem is to derive relations between these characteristic pressures and in-situ stresses. For fractures of limited extent this problem can be limited to two-dimensional analyses. Most analyses are based on linear fracture mechanics : the elastic parameters and the critical stress intensity factors of the rock are considered as constants. The minimum principal stress is taken horizontal and consequently the mode I fracture is vertical. A basic parameter is the ratio of the fracture radius to the well radius (alpha / a). Explicit solutions are available for very short fractures (alpha/a = 1 + epsilon) and for largefractures (alpha / a > 5). Recent work suggests that the pressure corresponding to a very short fracture, or initiation pressure, coincide with the pressure p index (nl). Given the maximum horizontal stress sigma index h and the minimum horizontal stress sigma index (h), this pressure still depends on four parameters : tensile strength of the rock T, penetration of the fluid into the fracture s, mobility of the fluid towards the fracture faces, pore pressure p index (p). The tensile strength T is essentially unknown. A method in use to bypass this difficulty is to carry out several fluid injection/bleed-off phases : then T can be (as least theoretically) set to zero. When the fluid filtration time is short compared to the fracture propagation time, the initiation pressure can be written (Haimson and Fairhurst):(***)ß is a poroelastic constant which depends on the Biot coefficient and the Poisson's ratio of the drained rock. When the fluid filtration time is large compared to the fracture propagation time ( impermeable rock ) the fracture initiation pressure is supposed to be (adapted by Detournay) :(***)Interpretation of fracture initiation pressure in overburden formations can be based on this last formula. Depending on s, which characterizes the access of the fluid to the fracture, this pressure is bounded by a maximum (no fluid penetration) and a minimum (complete fluid penetration) value:(***)For large fractures the best practical data available from micro fracturing operations is the closure pressure p index (f). Indeed under most actual conditions, p index (f) can be considered as equal to sigma index (h). Although a rather large number of parameters is involved in the interpretation of the leak-off pressure the interpretation of the nonlinear part of the pressure-time curve (from p index (nl) to p index (b)) is still matter of controversy. Indeed under the assumptions of linear fracture mechanics there should not be a non linear part in the pressure curve and the breakdown pressure p index (b) should not depend on the injection rate. However, in addition to non-linearity, laboratory experiments and field tests show that p index (b) increases with the injection flow rate up to values which are far over the theoretical prediction. Various ways have been explored to explain this behavior. The first one consists in substituting a non linear fracture mechanics to the linear one : the tensile strength of the rock in the plane of the fracture is not set to zero at the rock rupture but still exists after creation of the fracture. Then it decreases progressively when the fracture thickness increases and, for example, reaches the zero value for a ten microns fracture thickness. Another interpretation of the p index (b) value keeps the linear fracture mechanics assumptions and adds to the problem poroelastic considerations. According to this theory the pore pressure field around the fracture is inhomogeneous and in particular there is a depression zone ahead of the fracture tip. The progression of the fracture is controlled by the progression rate of this depression zone, which itself depends mainly on matrix properties (mobility of the fluid, elastic properties and porcelastic coupling). When this fracture progression rate no more keeps up with the fluid injection rate the pressure exceeds the values predicted by a more classical (non poroelastic) theory. A third way of interpretation has been modeled at Institut Français du Pétrole (IFP). In this model a mechanical damage phase comes before the creation of a single fracture plane. The non linear pressure growth takes place during the damaging phase. The results of three micro fracturing field tests (two tests in overburden layers, the third one in a reservoir) are used to extract from the p index (nl) and p index (f) pressure values the maximum horizontal stress sigma index (h), the interpretation being based on linear fracture mechanics. The exercise is convincing when applied to the reservoir example. It is not satisfactory when applied to the other examples since the fluid penetration parameter needs to be tailored to each case. The measurement of this parameter is recommended in parallel with the development of the above mentioned damage/localization modeling approach.
© IFP, 1995