Petroleum engineers use the notion of residual gas saturation (Sgr) to estimate microscopic trapped gas content, corresponding to the end-point of imbibition curves. It has been established [3] that simple experimental conditions may be representative of gas trapping in reservoirs.

Many studies have tried to correlate trapped gas saturation to reservoir characteristics, showing notably [1,4] a rough negatively correlated scattered relationship between Sgr and porosity. Initial gas saturation (Sgi) affects trapped gas saturation, leading to various empirical Sgr–Sgi relationships [1–4,6]. Except for [1,3,4], all the studies are based on a few samples and consider only a few parameters. In these conditions, no conclusion is obvious.

The aim of this work is to establish an experimental database including a large number of water wetted sandstone samples and parameters such as porosity, permeability, porosimetric and NMR curves. This study focuses on the influence of rock characteristics and initial gas saturation on Sgr.

2.1 Core samples

A total of 400 samples were selected from four different sandstone gas reservoirs (M1, M3, I5 and I6), and from Fontainebleau sandstone outcrops (FTB). Cylindrical samples of different lengths, 23 or 40 mm in diameter, were cut from whole core samples. The wide range of porosity and permeability (Fig. 1) shows the variety of intergranular porous media.

2.2 Measurements

The results concern three lines of research: the relation between clay content and microporosity, the influence of microporosity on Sgr values and the shape of Sgr–Sgi relationships.

3.1 Microporosity and clay contents

As bulk and solid clay density cannot be estimated precisely, only an equivalent parameter, S′_arg, based on the three first terms (T_arg, ρ_s, and φ) is used. It is a function only of macroscopic rock characteristics, and proportional to S_arg:

$$\mathrm{S}{\text{'}}_{\arg }=\frac{M_{\mathrm{clay}}}{V_{\mathrm{p}}}={\mathrm{T}}_{\arg }\times {\rho }_{\mathrm{s}}\left(\frac{1}{\varphi }-1\right)\left({\mathrm{m}}^3{\mathrm{kg}}^{-1}\right)$$

The microporosity amount is estimated by CBW, which is a NMR parameter. The part of pore volume quantified by CBW is likely to be equivalent to S_arg. This assumption is validated by the observed linear trend between the equivalent parameter, S′_arg and CBW (Fig. 2).

3.2 Influence of rocks characteristics

3.2.1 Porosity influence.

Fig. 3 shows the relationship between porosity and Sgrm. It illustrates the fact that the variations of Sgrm versus porosity present three major trends:

• – in the low and medium porosity range, as porosity decreases, Sgrm either increases (Fontainebleau Sandstone), or decreases (other sandstones);
• – for the highest porosities, the two above trends merge around an average Sgrm value of 30%.

3.2.2 Influence of microporosity.

The influence of microporosity on gas trapping is illustrated by the relationship between Sgrm and S′_arg and between Sgrm and CBW (Fig. 4). This confirms that the clay influence is due to the induced microporosity.

3.3 Influence of initial gas saturation

3.3.1 Shape of Sgr–Sgi curves.

Examples of experimental relationships are presented in Fig. 5. Sgr–Sgi relationships can be decomposed into two distinct parts: a linear Sgr–Sgi region and a constant Sgr region. In this plateau, Sgr is equal to Sgrm. The breaking point is named Sgo.

The curve shape is important because it controls the analytical expression of the Sgr–Sgi law. The usual Land's law has a hyperbolic shape that induces errors in the Sgr estimates [8]. A better approximation of Sgr with Sgi and Sgrm confirms that a linear-constant form is preferable:

$$\mathrm{if}\mathrm{Sgi}<\mathrm{Sgo},\mathrm{Sgr}=\frac{\mathrm{Sgrm}}{\mathrm{Sgo}}·\mathrm{Sgi},\mathrm{else}\mathrm{Sgr}=\mathrm{Sgrm}$$

But this result raises the problem of Sgo estimation.

The relationship between Sgo and irreducible water saturation (Swi) is clear. But it also shows that Sgo is not the complement of Swi. Another relationship between Sgo and CBW (Fig. 6) is indicative of the microporosity amount. In conclusion, the Sgo parameter is a function of the porous media properties, and more precisely of the amount of microporosity.

We have demonstrated a linear-constant form of Sgr–Sgi experimental relationships. Sgr values are a function of initial gas saturation (Sgi), Sgrm (Sgr obtained on a dried plug) and Sgo (breaking point).

The porous media controls the Sgo parameter, which is not the complement of Swi. It implies that Swi is not a key parameter in imbibition mechanisms.

The amount of microporosity is a rock characteristic that controls Sgrm and Sgo. The clay influence on Sgr is due to microporosity indicated by the S′_arg parameter.

These observations raise the question of possible gas-trapping mechanisms. Usually, authors consider that microporosity, thus clays, is not able to trap gas because of the low pore-size-to-pore-throat ratio [5,7] or gas-diffusion mechanisms [1,2,4,9].

The constant Sgr region of Sgr–Sgi curves confirms that the microporosity does not trap gas. The linear Sgr–Sgi region corresponds to constant gas trapping, in the macroporosity, independently of pores size influence.

It should be noted that parameters controlling gas trapping can be different from parameters that control the content of trapped gas. For example, the pore-size-to-pore-throat ratio probably does not influence Sgr, but is likely to represent only one condition of gas trapping.

Les échantillons sont des grès sélectionnés parmi quatre réservoirs à gaz différents (M1, M3, I6 et I5) et le grès d'affleurement de Fontainebleau (FTB). Les grès étudiés sont variés : quartzeux, argileux, bioclastiques, cimentés… Cette grande variété de faciès induit une très grande diversité dans les milieux poreux intergranulaires étudiés et dans leurs caractéristiques pétrophysiques (Fig. 1).

Les différentes mesures réalisées sont les suivantes :

• – mesures pétrophysiques conventionnelles : porosité, perméabilité, densité du squelette solide, facteur de formation ;
• – approche quantitative du milieu poreux : porosimétrie au mercure et mesures RMN ;
• – composition minéralogique par diffraction X.

La mesure de la saturation résiduelle en gaz (Sgr) a été réalisée par imbibition spontanée co-courant. Le couple de fluides est l'air et une huile de synthèse, afin de limiter les perturbations de la texture des argiles présentes dans les pores. La mesure de Sgr a été effectuée, soit sur un échantillon initialement sec (Sgr est alors nommée Sgrm, saturation résiduelle en gaz maximale), soit sur un échantillon avec une saturation initiale en gaz (Sgi) donnée.

La fraction de micropores liés à la présence d'argiles a été évaluée grâce au paramètre S_arg. Ce paramètre est fonction, d'une part, des caractéristiques de la roche (φ, ρ_s, T_arg), d'autre part, de caractéristiques intrinsèques des argiles (ρ^arg_app, ρ^arg_s) :

Comme les caractéristiques intrinsèques des argiles ne peuvent pas être estimées précisément, on a défini le paramètre S′_arg, proportionnel à S_arg, qui est uniquement fonction des caractéristiques macroscopiques de l'échantillon :

L'utilisation des paramètres S′_arg ou S_arg, permet d'étendre l'étude aux échantillons caractérisés par une analyse par diffraction X. Comme les mesures RMN sont moins communes, ceci augmente considérablement le nombre d'échantillons étudiés.

Sgrm est une valeur spécifique de Sgr obtenue sur un échantillon initialement sec. Ce paramètre est utilisé afin de déterminer les paramètres qui contrôlent la valeur de Sgr, et ceci indépendamment de l'influence de Sgi.

Les mesures réalisées ont été présentées dans le §2.2. Seuls deux des paramètres étudiés ont une influence majeure sur les valeurs de Sgrm : la porosité et la quantité de microporosité.

3.2.1 Relation Sgrm–porosité

3.2.2 Influence de la microporosité

L'influence de la saturation initiale en gaz (Sgi) a été déterminée en établissant 60 relations expérimentales entre Sgr et Sgi.

3.3.1 Forme des courbes