致密多孔介质中气体视渗透率的有效联系外文文献翻译_第1页
致密多孔介质中气体视渗透率的有效联系外文文献翻译_第2页
致密多孔介质中气体视渗透率的有效联系外文文献翻译_第3页
致密多孔介质中气体视渗透率的有效联系外文文献翻译_第4页
致密多孔介质中气体视渗透率的有效联系外文文献翻译_第5页
已阅读5页,还剩17页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

Effective Correlation of Apparent Gas Permeability in Tight Porous MediaFaruk CivanAbstract:Gaseous flow regimes through tight porous media are described by rigorous application of a unified HagenPoiseuille-type equation. Proper implementation is accom-plished based on the realization of the preferential flow paths in porous media as a bundle of tortuous capillary tubes. Improved formulations and methodology presented here are shown to provide accurate and meaningful correlations of data considering the effect of the charac-teristic parameters of porous media including intrinsic permeability, porosity, and tortuosity on the apparent gas permeability, rarefaction coefficient, and Klinkenberg gas slippage factor.Keywords:Tight porous media Apparent gas permeability Rarefaction coefficient Klinkenberg gas slippage factor Tortuosity1 IntroductionDescription of various gaseous flow regimes through tight porous media has drawn con-siderable attention because the convetional Darcys law cannot realistically describe the variety of the relevant flow regimes other than the viscous flow regime. For example, Javadpour et al. ( 2007) have determined that gas flow in shales deviates from behavior described by the conventional Ficks and Darcys laws. Therefore, many attempts have been made in describing the transfer of gas through tight porous media under various regimes. Such efforts are of utmost practical importance when dealing with extraction of hydro-carbon gases from unconventional gas reservoirs, such as shale-gas and coal-bed methane reservoirs. Skjetne and Gudmundsson (1995), and Skjetne and Auriault (1999) theoretically investigated the wall-slip gas flow phenomenon in porous media based on the Navier-Stokes equation, but did not offer any correlation for the Klinkenberg effect. Wu et al. (1998)developed analytical procedures for determination of the Klinkenberg coefficient from laboratory and well tests, but did not provide any correlation. Having reviewed the vari-ous correlations available, Sampath and Keighin (1982) proposed an improved correlation for the Klinkenberg coefficient of the N2 gas in the presence of water in porous media, expressed here in the consistent SI units aswhere bk is in Pa, K is in m2, and is in fraction. The significance of this correlation is that its exponent is very close to the 0.50 exponent value obtained by theoretical analysis in this article.Beskok and Karniadakis (1999) developed a unified HagenPoiseuille-type equation covering the fundamental flow regimes in tight porous media, including continuum fluid flow, slip flow, transition flow, and free molecular flow conditions. Ability to describe all four flow regimes in one equation alone is an outstanding accomplishment. However, the empirical correlation of the available data of the dimensionless rarefaction coefficient is a mathe-matically complicated trigonometric function. As demonstrated in this article, much accu-rate correlation of the same data can be accomplished using a simple inverse-power-law function.Florence et al. ( 2007) made an attempt at utilizing the HagenPoiseuille-type equation of Beskok and Karniadakis ( 1999) to derive a general expression for the apparent gas perme-ability of tight porous media and correlated some essential parameters by means of exper-imental data, including the Klinkenberg gas slippage factor and the Knudsen number by ignoring the effect of tortuosity, although it is an important factor especially in tight porous media. Therefore, although their overall methodology is reasonable, their formulation and data analysis procedure require some critical improvements as pointed out in this article when attempting to apply the HagenPoiseuille-type equation, originally derived for pipe flow to tight-porous media flow. Their treatment neglects a number of important issues. The HagenPoiseuille-type equation of Beskok and Karniadakis ( 1999) has been derived for a single-pipe flow. When the bundle of tortuous tubes realization of the preferential flow paths in tight-porous media ( Carman 1956) is considered, the number and tortuosity of the pref-erential flow paths formed in porous media should be taken into account as the important parameters.Further, the approach taken by Florence et al. ( 2007) for correlation of the Klinkenberg gas slippage factor is not correct and consequently their correlation cannot represent the data over the full range of the gas molecular mass (commonly called weight). These errors are corrected in this article by a rigorous approach which leads to a very accurate correlation of their data with a coefficient of regression almost equal to 1.0. In addition, the present analysis lends itself to a practical method by which tortuosity of tight porous media can be determined using the flow data obtained by conventional gas flow tests. To the authors knowledge, such a method does not presently exist in the literature.Hence, the primary objectives of this article are threefold: (1)Correlation of the rarefaction coefficient (2)Derivation of the apparent gas permeability equation (3)Correlation of the Klinkenberg gas slippage factor These issues are resolved and verified in the following sections by theoretical means and rigorously analyzing experimental data.2 Correlation of the Rarefaction CoefficientBeskok and Karniadakis (1999) derived a unified HagenPoiseuille-type equation for volu-metric gas flow qh through a single pipe, given below:where the flow condition function f (K n) is given bywhere Kn is the Knudsen number given by where Rh and Lh denote the hydraulic radius and length of flow tube, and denotes the mean-free-path of molecules given by ( Loeb 1934)where p is the absolute gas pressure in Pa, T is the absolute temperature in K, M is the molecular mass in kg/kmol, Rg = 8314 J/kmol/K is the universal gas constant, and is the viscosity of gas in Pa.s.Equation 2 describes the fundamental flow regimes, namely the conditions of continuum fluid flow (Kn 0.001), slip flow (0.001 Kn 0.1), transition flow (0.1 Kn 10), and free molecular flow (Kn 10), according to the classification of flow regimes by Schaaf and Chambre ( 1961). However, Beskok and Karniadakis (1999) emphasize that the Knudsen number limits given in this classification are based on pipe flow experiments and may vary by the geometry of other cases.The parameter appearing in Eq. 3 is a dimensionless rarefaction coefficient which varies in the range of 0 o over 0 Kn . Beskok and Karniadakis (1999) provide an empirical correlation as:where 1 = 4.0, 2 = 0.4, and o is an asymptotic upper limit value of as Kn (representing free molecular flow condition), calculated by: Here, b denotes a slip coefficient. They indicate that = 0 and b = 1 in the slip flow condition, and therefore Eq. 7 becomes: The expression of Eq. 6 is mathematically complicated. In the following exercise, it is demonstrated that a simple inverse power-law expression as given below provides a much more accurate and practical alternative to Eq. 6 for the range of data analyzed by Beskok and Karniadakis (1999): where A and B are empirical fitting constants. Note that Eq. (9) honors the limiting conditions of 0 o over 0 Kn . In fact, it can be shown thatFig. 1 Present approach using Eq. 9 accurately correlates the data of both Loyalka and Hamoodi (1990)using the theoretically predicted upper limit value of o = 1.358 and Tison and Tilford (1993) using anadjusted upper limit value of o = 1.205. The present approach yields accurate fit of data with coefficients of regressions very close to 1.0As illustrated in Fig. 1, the present approach using Eq. 9 accurately correlates the data of both Loyalka and Hamoodi (1990) using the theoretically predicted upper limit value of o = 1.358 and Tison and Tilford (1993) using an adjusted upper limit value of o = 1.205. Consequently, the data of Loyalka and Hamoodi (1990) is correlated as Thus, A = 0.1780 and B = 0.4348. On the other hand, the data of Tison and Tilford(1993) is correlated as Figure 2 shows that the data of Loyalka and Hamoodi (1990) can be correlated accurately by both the present correlation approach using Eq. 9 with a coefficient of regression of R2 = 0.9871 and the empirical equation given by Beskok and Karniadakis (1999)with a coefficient of regression of R2 = 0.9697 using the theoretically predicted value of o = 1.358. However, as indicated by the comparison of the coefficients of regressions, the present approach yields a more accurate correlation than that of Beskok and Karniadakis(1999).Fig. 2 Data of Loyalka and Hamoodi (1990) can be correlated accurately by both the present correlation approach using Eq. 9 with a coefficient of regression of R2 = 0.9871 and the empirical equation given by Beskok and Karniadakis (1999) with a coefficient of regression of R2 = 0.9697 using the theoretically predicted value of o = 1.358. However, as indicated by the comparison of the coefficients of regressions, thepresent approach yields a more accurate correlation than that of Beskok and Karniadakis (1999) Figure 3 shows that the present correlation with Eq. 9 using the adjusted value of o =1.205 represents the data of Tison and Tilford (1993) accurately with a coefficient of regression of R2 = 0.9486, close to 1.0. In contrast, the empirical equation given by Beskok and Karniadakis (1999) using the adjusted value of o = 1.19 leads to a lower quality correlation with a coefficient of regression of R2 = 0.7925, less than 1.0. As indicated by the comparison of the coefficients of regressions, the present approach yields a much more accurate correlation than that of Beskok andKarniadakis(1999)Fig. 3 Present correlation with Eq. 9 using the adjusted value of o = 1.205 represents the data of Tison and Tilford (1993) accurately with a coefficient of regression of R2 = 0.9486, very close to 1.0. However, the empirical equation given by Beskok and Karniadakis (1999) using the adjusted value of o = 1.19 leads to a lower quality correlation with a coefficient of regression of R2 = 0.7925, less than 1.0. As indicated by the comparison of the coefficients of regressions, the present approach yields a much more accurate correlation than that of Beskok and Karniadakis (1999) It is concluded that the present simple inverse power-law expression yields more accurate correlation of the dimensionless rarefaction coefficient with coefficients of regressions very close to 1.0 in both cases and therefore more suitable than the empirical equation given by Beskok and Karniadakis(1999).3 Derivation of the Apparent Gas Permeability EquationThe Beskok and Karniadakis (1999) unified HagenPoiseuille-type equation (Eq. 2) derived for flow qh through a single pipe can now be applied for the volumetric gas flow through a bundle of tortuous flow paths as:where Lh denotes the length of the tortuous flow paths and n denotes the number of preferential hydraulic flow paths formed in porous media. The latter can be approximated by rounding the value calculated by the following equation to the nearest integer (Civan 2007)where is porosity and Ab is the bulk surface area of porous media normal to flow direction. The symbol q denotes the total volumetric flow through porous media. It can be described macroscopically by a Darcy-type gradient-law of flow, where the flow is assumed proportional to the pressure gradient given by where K denotes the apparent gas permeability of tight porous media and L is the length of bulk porous media. Note that Eq. 16 is used frequently, although it is not rigorously correct. The corrections required on Eq.16, such as the effect of the threshold pressure gradient (Prada and Civan1999), are provided elsewhere by Civan (2008), but are ignored here to avoid unnecessary complications for purposes of the present discussion and derivation. Nevertheless, Civan(2008) argued that such corrections are usually negligible for gaseous flow, although the validity of such claim for tight porous media needs detailed investigation. The tortuosity factor hof hydraulic preferential flow paths in porous media is defined by Hence, the following expression can be derived for the apparent gas permeability by combining Eqs. 1417:where K denotes the liquid permeability of porous media given by Equation 19 can be rearranged to express the hydraulic tube diameter as Alternatively, it can be shown for a pack of porous media grains that (Civan 2007)Where g denotes the specific grain surface in porous media. Hence, equating Eqs. 20 and 21 yields the well-known Kozeny-Carman equation of permeability as (Carman 1956; Civan2007) The function f (Kn) does not appear in Eq. 22 because the intrinsic permeability K of porous media is only a property of porous media and does not depend on the fluid type and flow condition.The formulation presented by Florence et al. (2007) for the apparent gas permeability of tight porous media considered flow through a single straight pipe and therefore needs corrections, according to the procedure described above.4 Correlation of the Klinkenberg Gas Slippage FactorUnder slip flow conditions, = 0 and b = 1, and therefore Eq. 18 combined with Eq. 3 can be written as: Florence et al. (2007) approximate this equation for Kn 1, upon substitution of the Kn number expression Eq. 4 as: whereas the equation of Klinkenberg (1941) is given by:where bk is the slippage factor. Comparing Eqs. 24 and 25 yields an expression asSubstituting Eqs. 5 and 20 into Eq. 26 results in:where the coefficient is defined by Figure 4 indicates that the present Eq. 28 yields an accurate correlation for the data of Florence et al. (2007) involving the flow of various gases (hydrogen, helium, nitrogen, air, and carbon dioxide) in a given porous medium (sandstone) under isothermal conditions (assumed as 298K) as the following:where is in Pa.m, in Pa.s, and M is in kg/kmol. The results reported in Table 1 indicate that the values calculated by Eq. 29 obtained by correlating the coefficient divided by the viscosity versus the square-root of the molecular mass, according to Eq. 28 accurately match the data of Florence et al. (2007). In contrast, the values calculated using the correlation developed by Florence et al. (2007), also given in Table 1, differ significantly from the same data because they correlated the coefficient directly (i.e. without dividing by the gas viscosity) by an inverse-power-law function of the molecular mass, expressed here in the consistent SI units as:where is in Pa.m and M is in kg/kmol. In view of Eq. 28, their approach is not rigorously correct because the effect of viscosity was ignored in their correlation. Note that Eqs. 2628 can be used to derive an expression for the Knudsen number asFor example, applying the correlation given by Eq. 29 to Eq. 31 yields:ThereforeFig. 4 Present approach yields an accurate correlation of the (/), m/s parameter versus the molecular mass (M, kg/kmol) for the data of Florence et al. (2007) for flow of various gases (hydrogen, helium, nitrogen, air, and carbon dioxide) in sandstone under 298Kisothermal conditions. The coefficient of regression is almost equal to 1.0 indicating that the present correlation approach is rigorousTable 1 Comparison of the values indicate that the present correlation approach reproduces the data with high accuracyIn contrast, the values calculated by the correlation developed by Florence et al. (2007) differ from the data significantlywhere bk is in Pa, in Pa.s, M is in kg/kmol, p is in Pa, K is in m2, and is fraction. When applied for the N2 gas, Eq. 33 yields The coefficient 0.0094 of this equation is in the same order ofmagnitude as the coefficient 0.0414of Eq. 1 of Sampath and Keighin (1982) but differs by a factor of 4.4. This may be explained due to the 0.53 value of the exponent of Eq. 1 being different than the exponent value of 0.50 of Eq. 34 and the N2 gas flow tests conducted in the presence of water in porous media instead of the dry porous media considered by Florence et al. (2007). As a bonus, Eq. 29 reveals that the tortuosity h of the preferential hydraulic flow paths in the porous medium is given by, expressed in the consisted SI unitswhere Rg is 8314 J/kmol-K and T is in K. Equation 35 may be used in determining the tortuosity of porous media. For example, the tortuosity is estimated to be h = 1.0 for the sandstone sample used in the tests assuming a temperature of 298 K, according to Florence et al. (2007). However, Florence et al. (2007)mention that the actual temperaturewas unknown for the data involved in their studies and they simply assumed the value of 298K for their calculations.5 ConclusionsThe present approach rigorously accounts for the effect of the characteristic parameters of porous media including intrinsic permeability, porosity, and tortuosity on the apparent gas permeability, rarefaction coefficient, and Klinkenberg gas slippage factor. Improved formulations presented in this study have been proven to be instrumental in accurate correlation of experimental data for effective description of gas flow in tight porous media. This has been demonstrated by correlating the available data more accurately than the previous attempts.中文译文致密多孔介质中气体视渗透率的有效联系Faruk Civan摘要:通过致密多孔介质的气体流动状态是被标准的哈根泊肃叶典型方程的严格应用所描述。合适的实现是在基于多孔介质中最好的流动路线实现的条件下作为一束

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论