Traitement des diagraphies acoustiques. Troisième partie : caractérisation d'un réservoir par diagraphies acoustiques obtenues avec un outil dipolaire

Full-Waveform Acoustic Data Processing. Part Three: Reservoir Characterisation Using Acoustic Logs Obtained with the Dipole Sonic Imaging Tool

¹ Gaz de France
² Institut Français du Pétrole

Résumé

En formation lente, l'utilisation d'un outil acoustique équipé d'émetteurs et de récepteurs dipolaires permet d'accéder à la lenteur S de la formation. Nous montrons comment un tel outil peut être utilisé pour caractériser un réservoir en formation lente. Le pointé des premières arrivées des différentes ondes composant un enregistrement acoustique en champ total permet d'accéder aux logs acoustiques classiques (lenteur, fréquence, atténuation) ainsi qu'aux logs de dispersion de mesure qui leur sont associés. Ces logs conduisent à des logs de paramètres mécaniques notamment au coefficient de Poisson. Dans le cas du puits étudié traversant une formation argilo-gréseuse, le log du coefficient de Poisson et le log de fréquence des ondes de Stoneley sont utilisés comme indicateur lithologique d'argilosité et de présence d'hydrocarbures (gaz). Après correction d'argilosité, les logs de lenteur en onde compression et de cisaillement sont utilisés en combinaison avec un log densité de type Gardner pour estimer les paramètres pétrophysiques de la formation (indices de porosité et de saturation en gaz). Les résultats obtenus sont comparés à ceux fournis par l'analyse de logs classiques (neutron, densité).

Abstract

In slow formations, in which the velocity of the compression wave in mud is greater than the velocity of the shear wave in the formation, the slowness S of the formation can be determined only very indirectly (using Stoneley waves) with a conventional monopole tool. For such formations, the use of a dipole tool makes it easier to gain access to this measurement and to obtain more complete information about the formation. This article shows how such a tool can be used for characterizing a slow-formation reservoir. The tool has a monopole transmitter capable of emitting low frequencies so as to favor interface modes and, at high frequencies, head waves (P refraction in the formation). The tool has two dipole transmitters for generating flexural modes having propagation characteristics in the formation enabling the slowness S of the formation to be measured. The picking of the first arrivals of the different waves making up a total acoustic field recording leads to conventional acoustic logs (slowness, frequency, attenuation) as well as to dispersion measurement logs associated with them. The logs are shown in Figs. 5 to 7 for the compression waves, in Figs. 9 and 10 for the shear wave, and in Figs. 13 and 14 for the Stoneley wave. The slowness log of Stoneley waves was used to compute the slowness wave S, using the equation proposed by White (1983) linking the slowness of the S wave (DtS) to the slowness of Stoneley waves (DtSt), to the formation density pb, and to the slowness (Dtf) and density (pf) of the fluid (Fig. 15). Measurement of the slowness of compression and shear waves gives access to the Poisson coefficient (Fig. 16), which is a good lithologic indicator (Fig. 17). It can be used to differentiate compact formations from unconsolidated formations and is a direct indicator of the presence of hydrocarbons in gas sands. For wells crossing through a shaly-sandy formation, the log of the Poisson coefficient and the frequency log of Stoneley waves are used as lithologic indicators of the shaliness (Fig. 18) and of the presence of hydrocarbons (gas). After correction for shaliness, the slowness logs of the compression and shear waves are used in combination with a density log of the Gartner type for estimating the petrophysical parameters of the formation (porosity and gas-saturation indices). Porosity was evaluated by using the slowness log of compression waves with correction for shaliness (Fig. 18) and the Wyllie relationship for formations having a Poisson coefficient higher than 0. 25 by assuming that such formations are water saturated (Dtf = 190 µs/ft). The cutoff made by using the log of the Poisson coefficient is introduced so as to eliminated partially gas-saturated zones in the reservoir. In these partially gas-saturated zone in the reservoir, the porosity was interpolated. The porosity log thus computed is shown in Figure 20. The reservoir zones have a porosity slightly higher than 25%. We tried to obtain an indication of the gas saturation of reservoirs by using the interpretation model proposed by Krief et al. (1990) , based on an equation linking the squared velocities of P waves and S waves and the porosity of the formation, an equation based on theoretical and empirical research. With this method, the gas saturation is estimated to be about 15% (solid-line curve called DtP-DtS in the middle part of Fig. 20). The estimated saturation value is a pessimistic value, which is considerably lower than the real gas-saturation value of reservoirs. Gas saturation Sg was also evaluated by using the density value RhoG computed by a Gardner-type formula in the gas zone. RhoG is deduced from DtP after correction for shaliness. The gas-saturation values estimated by means of a Gartner-type density are higher than those found by using the model proposed by Krief. They reach 45% for the lower reservoir, in the upper reservoir they vary greatly with a mean value of about 30%. The values provided by estimators of shaliness (Vsh), porosity and gas saturation (Sg) were used to compute a synthetic density log that is shown in Fig. 20. Figure 21 shows a comparison of these results and the ones obtained by conventional logs, in particular using the Schlumberger ELAN lithologic processing, with Rhob coming from the curves for density, shaliness, Rhog and gas saturation. Generally speaking, the average trends of the measured parameters (porosity, shaliness, gas saturation, density) as functions of depth are quite effectively appraised. The variation of the major lithologic units can be found on most of the curves, although with less detail inside interbedded layers. In some cases, the approach to the absolute value of the parameters remains fairly approximate, being either over-or underestimated. For example, the main gas levels are clearly detected if saturation is not taken into consideration.