aUniversity of Oklahoma, Suez University, USA
bUniversity of Oklahoma, USA
The development of shale reservoirs has brought a paradigm shift in the worldwide energy equation. This entails developing robust techniques to properly evaluate and unlock the potential of those reservoirs. The application of Nuclear Magnetic Resonance techniques in ﬂuid typing and properties estimation is well-developed in conventional reservoirs. However, Shale reservoirs characteristics like pore size, organic matter, clay content, wettability, adsorption, and mineralogy would limit the applicability of the used interpretation methods and correlation. Some of these limitations include the inapplicability of the controlling equations that were derived assuming fast relaxation regime, the overlap of different ﬂuids peaks and the lack of robust correlation to estimate ﬂuid properties in shale. This study presents a state-of-the-art review of the main contributions presented on ﬂuid typing methods and correlations in both experimental and theoretical side. The study involves Dual Tw, Dual Te, and doping agent's application, T1-T2, D-T2 and T2sec vs. T1/T2 methods. In addition, ﬂuid properties estimation such as density, viscosity and the gas-oil ratio is discussed. This study investigates the applicability of these methods along with a study of the current ﬂuid properties correlations and their limitations. Moreover, it recommends the appropriate method and correlation which are capable of tackling shale heterogeneity.
Since the introduction of Nuclear Magnetic resonance technology in petroleum industry, its applications answered a lot of questions in reservoir engineering and provided unambiguous techniques to evaluate rock, rock-ﬂuid and ﬂuid properties. NMR applications started with a tool to calculate the total porosity independently from rock matrix effects by calculating the intensity of hydrogen protons in formation. Then, the differentiation between the bound and free fraction of ﬂuid in porous medium became obtainable. Further improvements were performed to extend its applicability to determine capillary pressure , wettability , and relative permeability  from NMR measurements. Consequently, NMR became a reliable instrument to diagnose ﬂuid types, properties, and rock properties as well for both conventional and unconventional reservoirs [4-8].
Counting on NMR methods used in conventional reservoir to determine rock and ﬂuid properties as a guide to analyze shale reservoirs response is misleading and will yield unreliable data . In terms of composition, shale is a heterogeneous rock with various clay contents . The clay distribution affects both NMR response and interpretation . Apart from clay content, the presence of the organic matter will add more complexity to the system. Considering the pore size, shale has a wide spectrum of pore sizes ranging from nanometer pores, conventional pores to natural fractures along with the dependence of the organic pores on the maturity of the rock. Moreover, shale has different wettabilities and most of them are mixed . Due to the complexity of the shale system, an extensive study of rock and ﬂuid characteristics impact is required to successfully interpret the NMR response.
NMR measurements procedure starts with exciting the sample by a magnetic ﬁeld, which will polarize the hydrogen protons in one direction. Then, measuring the longitudinal relaxation time (T1) or use Carr Purcell Meiboom Gill (CPMG) sequence and measure the transverse relaxation time (T2) or use pulsed-ﬁeld gradient sequence and measure the diffusion coefﬁcient (D) . After that, an inversion method will be adopted to obtain the decay exponents distribution . This distribution will be the base of all succeeding interpretation methods.
The relaxation of ﬂuids in a bulk state is bulk relaxation and tends to be longer as the relaxation will be due the interactions among ﬂuid protons only. On the other hand, ﬂuids relaxation in porous media will be promoted by the interactions among the ﬂuids and the conﬁning surface protons. Besides, the transverse relaxation time will be more affected by the molecules diffusion. The diffusion relaxation will not affect T1 measurement, but it inﬂuences T2 where there are spin dephasing and refocusing . The relaxation governing equations in porous media is developed in fast diffusion regime as a weighted average between bulk and surface relaxation rate. Therefore, they will be
The validity of fast diffusion regime assumption in shale will determine the applicability of those equations. In addition, the impact of the diffusion coupling, homonuclear dipole coupling, heteronuclear dipole coupling, residual dipole coupling and magnetization transfer as relaxation mechanisms should be considered .
Shale is a clay-rich rock which contains variable content of clay minerals and organic matter. This rock contains different pore sizes which have fractional wettability and host different kinds of ﬂuids. In addition, the Nano-scale pores and adsorption will add more ambiguity to the nature of these rocks. This section will discuss the main characteristics of Shale and their impact on NMR response.
The behavior of ﬂuids in conﬁned porous media diverges from their behavior in bulk state . These deviances will be more signiﬁcant in Nano-scale pores (Fig. 1). Therefore, this entails a study of these deviances effects on relaxation mechanisms.
Fig. 1. Sizes of molecules and pore throats in siliciclastic rocks on a logarithmic scale. Measurement techniques are shown at the top of the graph .
In most conventional pores studies, it is assumed that the main relaxation mechanism is the surface relaxation. Therefore, the position of the ﬂuid peak will be directly proportional to the size of the pore-containing that ﬂuid. With this in mind, many studies managed to estimate pore size distribution from NMR. Moving into shale domain, the main relaxation mechanism will be also the surface relaxation. But knowing that the relaxation interactions scale is comparable to the scale of pores investigated , will render any study of pore size based on NMR relaxation time useless. Since, it is evident that Equation (2) is not valid to calculate the relaxation time of water in Smectite interlayers . Therefore, conventional surface relaxivity concept is not applicable in nano-scale pores anymore.
Furthermore, the heterogeneous distribution of pore size observed in shale would puzzle the interpretation scheme. Shale contains micro-pores in the organic matter and nano-pores in the clay minerals and natural fractures . The presence of these different pore size scales in one rock complicates the interpretation process. As it will be more challenging to decide if the peaks in relaxation time distributions are resulting from different pore sizes or from different ﬂuids in these pores.
The organic matter and clay content are considered as matrix constituents and hydrogen protons populations that affect NMR signal . Therefore, the NMR response is not independent of the rock matrix in this case . The organic matter affects NMR signals in two ways directly as it is one of the proton population investigated and indirectly by controlling the relaxation time of the ﬂuid contained in the organic pores where it acts as the relaxation surface. In both cases, the effect of organic matter depends on its maturity. Considering the direct effect, the degree of maturation is directly proportional to the mobility of protons . Therefore, the organic matter signature of samples from gas, oil or immature windows will not be the same. Unfortunately, the organic matter response may be masked if high hydrogen index ﬂuid is present. On the other hand, the organic matter as a matrix contains organic porosity up to 20.2% and average pore size approximately 100nm . Therefore, the surface relaxivity of organic matter and the interactions between the ﬂuid and the surface need further investigation. As, Washburn  suggested that the surface relaxation in organic pores depends on the homonuclear dipole coupling among hydrogen protons in the ﬂuid and the surface. In contrast to the surface relaxation in conventional pores where it is dominated by interactions among ﬂuid protons and paramagnetic impurities in the surface. Consequently, the sample organic content and maturation should be investigated to properly evaluate their effect. Fig. 2 shows the morphology of conventional and unconventional pores through SEM images along with a schematic representation of shale structure.
Fig. 2. (a) represents a comparison between the conventional sandstone and the unconventional organic pores . (b) Represents schematic illustration of shale morphology .
Clay content and distribution add another degree of complexity for ﬂuid identiﬁcation in these heterogeneous reservoirs. Similarly, clay minerals are one of the proton population, which have NMR response and inﬂuence the responses of the attached ﬂuid in many different ways. Firstly, the NMR response of clay will depend on their distribution and compaction in shale . If the clay particles present in structural or laminar distribution, Their NMR response will be below 0.1 ms and will require high resolution equipment to detect their signal . However, if the clay is dispersed, it will not produce a separate signal, but will affect other ﬂuids responses by providing extra surface area for ﬂuids to relax on. Secondly, the type of clay minerals present will affect the wettability of the pores. Saada et al.  explained that Illite is water-wet while Kaolinite is hydrocarbon-wet. In addition, clay inter-layers ﬂuids will also exhibit restricted relaxation as it will be bonded within the internal structure of the molecule. Interestingly, Zhang et al. 's experimental results suggest that Smectite may absorb hydrocarbons in certain conditions. So, according to clay type, clay will absorb and adsorb different ﬂuids.
Wettability determines which ﬂuid will adhere to the surface and subsequently will be affected more by the surface-ﬂuid interactions. In contrast to conventional reservoirs where hydrocarbons accumulations migrate from the source rock to the reservoirs, Shale is considered self-sourced reservoir. Besides, the heterogeneity reported in shale mineralogy. Fig. 3 highlights the main constituents of shale matrix and pores. Therefore, Shale pores have different wettabilities and most of them are mixed. Subsequently, the determination of ﬂuid location within the pore is indeﬁnite and will depend on the pore type under investigation .
Fig. 3. Schematic representations of different shale system constitutions .
The electrical charge imbalance on the surfaces of both organic matter and clay particles leads to ﬂuid adsorption on these surfaces. There are two kinds of adsorption experienced in Shale: gas adsorption in organic pores and water adsorption on clay particles. Gas adsorption will be more tangible in high pressures as indicated from Langmuir isotherm. The adsorbed ﬂuid signal will be mainly controlled by the interactions with surface molecules and will exhibit relatively faster relaxation compared to the remaining ﬂuid in the pore. The exchange between molecules between the adsorbed phase and the free phase and its impact on the ﬂuid response needs more investigation.
NMR response is assumed to be independent from the rock matrix, however rock mineralogy will control the surface chemistry of the grains and, in turn, the surface relaxation mechanism. Shale is a ﬁne-grained clastic sedimentary rock composed of different portions of clay minerals, quartz, calcite and organic matter. Interestingly, the resin present in the organic pores also affects the surface chemistry of these pores as it coats the grains. Therefore, the heterogeneity in the composition reﬂects on the surface chemistry of the pores which will exhibit different relaxitivity characteristics. Therefore, assigning a single value for surface relaxivity for the whole rock would be misleading.
Paramagnetic impurities are another aspect that should be considered during shale mineralogy discussions. The most common paramagnetic ions present in shale are iron and manganese. The high paramagnetic content may be misinterpreted to high surface relaxivity. But, in reality, most of the iron content in shale is related to the pyrite and small amount is dispersed as an impurity in pore surfaces and very little in organic matter. Therefore, the paramagnetic ions will not be the reason for faster relaxation . Fig. 4 displays the mineralogy of the main shale gas plays in the United States.
Fig. 4. Mineralogy of the gas shale plays in the United States .
The relaxation time of the ﬂuids in bulk volume will depend mainly on the interactions between the protons within the ﬂuid. But, the relaxation of the ﬂuid in shale pores will be affected by several competing factors as indicated in the previous section. The resulting signal will signify massive information about the ﬂuid and the pore size. This section discusses the NMR response of the ﬂuids in shale. Table 1 shows the qualitative response of different ﬂuid in porous media. Table 2 display the ﬂuids signatures reported in literature.
Table 1 Qualitative T1, T2, and D values for different ﬂuids in porous media [modiﬁed after 12].
Table 2 Comparison between the response of the brine, oil and gas in sandstone and shale [6,14,26-29].
The relaxation of the bulk liquids is controlled by dipole-dipole interactions. Consequently, the response is directly proportional to liquid viscosity and inversely to absolute temperature. In addition, the response of brine depends on the type of salts dissolved. As, the presence of small concentrations paramagnetic ions (mn+2) or ferromagnetic ions reduces water response substantially. The combination of the long bulk relaxation time and the tendency of water to ﬁll the small pores are the main reasons for the domination of the surface relaxation mechanism on the brine relaxation in porous media.
Oil is a mixture of different liquid hydrocarbons components and series. Therefore, oil is expected to have board distribution without unique peak characterizing the ﬂuid type as in pure ﬂuids. The more heterogeneous oil composition is the more broad the oil response would be. The fractions of the light and heavy component in oil determine whether the oil will relax fast due to the heavy components or relax slowly due to light components. Moving to the nanoscale pores, the wettability of the surface determines which ﬂuid will be affected by surface interactions. However, Minh et al.  suggested that the relaxation of heavy oil will be dominated by bulk relaxation mechanism regardless of the wettability of the pore due to the short bulk relaxation of heavy oil. Based on their simulation results, they set 100cps as the dividing value. Higher than this value, the wettability effect is considered negligible. Based on experimental study results, Tinni et al.  agreed that the heavy oil response will not be affected by the surface relaxation. On the other hand, the bulk decay exponent is comparable to the surface decay exponent in light oil. So wettability will determine the essence of the relaxation mechanisms involved.
Dunn et al.  showed that the relaxation behavior of gas will not follow the same behavior of liquids. As the bulk relaxation of gas will be dominated by spin-rotation interactions. Therefore, the response is directly proportional to absolute temperature and inversely to ﬂuid viscosity. Moreover, the T1 relaxation dependence on temperature and pressure is opposite to the situation in liquids. T1 is directly proportional to the pressure and inversely to the temperature. Moving to relaxation in shale, the situation is more complicated with the introduction of adsorbed gas to the system. As, gas diffusion coefﬁcient will depend on the diffusion of adsorbed phase, free phase and the exchange between them. Ref.  studied the gas dynamics in Shale. Their study proposes a new restricted diffusion model. However the experimental results do not present the diffusion in conﬁned nanopores due to the limited resolution of the used machines. They claims that T1 and T2 are controlled by dipole interactions between gas protons and paramagnetic impurities in the surface. However, Washburn  suggested that the relaxation might be due to homonuclear coupling between gas protons and protons on the surface.
The complex structure and heterogeneous composition of shale require the manipulation of more than one dimension (1-D) distribution to identify the different ﬂuids present in the pores. In addition, the overlap between the ﬂuid relaxation peaks would confound ﬂuid typing methods, for example, the overlap between heavy oil peak and water peak considering 1-D interpretation. Also, Kausik et al.  reported an overlap between oil signal (6e20 ms) and methane signal (10e20 ms). However, the implementation of two-dimension (2-D) methods provide better differentiation between the ﬂuid peaks and in some cases may capture clay particles and organic matter response. Unfortunately, there are two expected weaknesses in these methods: I) the wide range of oil viscosities leads to indeﬁnite signature. ii) All these methods do not distinguish between free and adsorbed gas fractions. In this section, different interpretation methods will be discussed.
There are two methods counting on only 1-D measurement. The ﬁrst method is simply based on the T1 contrast among the ﬂuids (dual TW method). This method involves polarizing the sample with short and long wait times. Water protons are polarized in the two times, but the hydrocarbons will be polarized only during the long time. So the differential response will deﬁne the hydrocarbon response (light oil or gas). The second method depends on the contrast among the diffusion coefﬁcients of ﬂuids (dual TE method). Based on the Equation (2) if the T2 measurement was performed with different TE the responses will be different. The differential response can be used to identify the ﬂuids. This method identiﬁes the gas response and measures its volume accurately. Xie and Xiao  proved from numerical simulation that water response in large pores will not be identiﬁed by TW method and there will be an overlap between the responses of gas in small pores and irreducible water using TE method. These results suggest the use of 2-D methods for better resolution.
2-D methods will overcome most drawbacks encountered in 1-D methods. These methods include plotting diffusion distributions versus T1 or T2, T1 versus T2 and T2SEC versus T1/T2. These methods yield better ﬂuid typing by providing more contrasts between the ﬂuids peaks reveal the overlaps between the ﬂuids signatures.
The T1-T2 method provides better differentiation between the different ﬂuids responses. A number of experimental studies have been conducted in this area, including the work of Fleury  with low ﬁeld NMR to quantify ﬂuids through T1-T2 maps. The experiments were performed on both the organic matter (immature, oil window and gas window) and shale samples. According to the experimental results, T1-T2 method successfully capture water and methane signature (T1/T2 ~ 2 and 10 respectively) along with a diagnostic map (T1 vs. T2) showing the expected region of the response from the organic matter and hydroxyls groups in the clays as indicated in Fig. 5. Washburn  has studied the T1-T2 response of four shale samples, three from Piceance basin (outcrop, oil shale and well cuttings) and outcrop sample from unita basin. This study conﬁrms the ability of the T1-T2 maps to identify different organic maturities present in shale, mainly solid and liquid-like phases. Xie and Xiao  studied the T1-T2 maps with numerical simulation. Their results are consistent with the previous experimental work. Therefore T1-T2 method is the applicable tool for ﬂuid typing in most situations except differentiating between heavy oil and bound water response.
Fig. 5. T1-T2 diagnostic map [Modiﬁed after 6].
The overlap between the gas and irreducible water signals resulting from the use of Dual TE method can be resolved by the D- T2 method. Experimental investigation conducted by Zielinski et al.  on carbonates introduced the concept of restricted diffusion of ﬂuids in porous media, followed by the application of these restricted diffusion maps to interpret the dynamics of restricted gas in shale , completed by a revised model based on the restricted diffusion to interpret the response from gas and oil shale as shown in Fig. 6. However, it worth noting that all the diffusion-based methods lack real measurements of restricted diffusion coefﬁcients in shale due to the limited resolution of the machines.
Fig. 6. D-T2 method in conventional and unconventional reservoirs .
Daigle et al.  introduced a method based on the secular relaxation (T2sec) to distinguish between the ﬂuids through plotting (T2sec vs. T1/T2). Secular relaxation rate is the difference between transverse and longitudinal relaxation rates. Their experiments were performed in ambient conditions with low ﬁeld NMR on samples from the Bakken, Woodford and shallow Marine mudstone from offshore Japan. This method succeeded to separate signals of the ﬂuids based on viscosity and pore size. They provide a diagnostic map to interpret the NMR responses to seven different scenarios as shown in Fig. 7. Followed by the work of Gips et al.  to evaluate the hydrocarbon characteristics based on the differential response. This differential response is the absolute difference between the NMR response at a higher temperature and at ambient conditions. This differential response may carry valuable information about the ﬂuid molecular size and viscosity if it is correlated to the correlation time. But, it is worth mentioning that the experimental procedures were performed on a grinded sample. Moreover, Washburn  claims that this differential response will be mainly due to the dependence of the surface relaxivity on temperature.
Fig. 7. T2sec-T1/T2 method [modiﬁed after 7].
Another approach based on introducing an artiﬁcial contrast between hydrocarbons and water to separate their signals in the T1-T2 maps which will allow better identiﬁcation of the ﬂuids. Mitchell et al.  have analyzed the effect of the paramagnetic doping agent, especially chelated manganese-EDTA and unchelated manganese on four limestone plugs. This study recommends the use of doping agents for better clear separation among oil and water signals and the mitigation of unchelated manganese with EDTA whenever clay is present. On the other hand, this study opposes the use of any paramagnetic doping agent in any system with PH<9, As, it will result in precipitation of insoluble products which will affect the ﬂow capacity. Unfortunately it does not include experimental work in shale. Using the same concept, Gannaway  succeeds in providing robust method to separate water and oil response. This separation enables to quantify different types of porosities encountered in shale as presented in Fig. 8. However, this study did not consider the formation damage resulting from the interactions between mn+2 and clays and PH effect on doping effect in shale.
Fig. 8. Gannaway results from using doping agent to quantify the different porosities [modiﬁed after 33].
The estimation of the ﬂuid properties from NMR measurements was reported in literature in the experimental work performed by Freedman et al.  in bulk ﬂuid samples, Ref.  in Berea rocks and  in oil sands. Ref.  proposed a mapping function to estimate ﬂuid properties from NMR measurements based on the database that they have established. Ref.  proposed a set of correlation to estimate the ﬂuid properties from NMR signal. On the other hand, in one of the earliest studies to calculate the in-situ viscosity of oil sands based on NMR measurement, Ref.  assumed that the heavy oil relaxation will depend only on internal interactions so they derived a set of correlation to estimate in-situ viscosity of oil sands based on T2 and relative hydrogen index.
The successful estimation of ﬂuid properties is tied to the accuracy of ﬂuid identiﬁcation by previous interpretation methods. The ﬂuid properties that could be estimated from NMR include density, Gas-oil ratio (GOR) and viscosity [38-40].
The ﬂuid density can be estimated after ﬂuid identiﬁcation using hydrogen index equation . This equation has two variables hydrogen index and total porosity. So if the total porosity is known, the ﬂuid density can be estimated.
Dead oil viscosity of pure alkanes exhibits a linear relationship with both T1, 2 and D . Hirasaki and Zhang  proposed a correction for live oil viscosity as a function of Gas Oil Ratio (GOR). However, Chen and Chen  recommended a mixing rule to estimate the viscosity of hydrocarbon mixtures. But, this is not the case in the shale reservoirs where the main relaxation mechanism is surface relaxation. Subsequently, the decay exponent is mainly a function of surface-ﬂuid interactions and does not represent the internal reactions within ﬂuid protons. However, Minh et al.  suggested based on simulation study that the heavy oil (viscosity more than100 cps) relaxation will be controlled by the internal reactions. Therefore, the heavy oil viscosity could be correlated with their relaxation time following the procedure presented in Bryan et al.  to estimate the viscosity of oil sands.
Tinni et al.  used the T1/T2 ratio to differentiate between moveable and non-moveable ﬂuids in both conventional and unconventional reservoirs. The experimental results of this study show that the non-moveable hydrocarbons will not be affected by surface relaxation which was conﬁrmed by the simulation results of . This study proves that NMR can estimate the ﬂuid viscosity at least qualitatively.
Winkler et al.  proposed a model to estimate GOR as the ratio between the numbers of the gas protons to the number of oil protons assuming that oil and gas molecules maintain their identities microscopically and, in turn, oil and gas molecules will maintain their self-diffusivity. So, after the differentiation between gas and oil molecules using their proposed mixing rule, GOR and viscosity could be determined. Chen and Chen  extend this model to mitigate the invasion of oil base mud and provide a method to calibrate NMR log to estimate GOR and viscosity of live oil signatures. But, this approach requires precise differentiation between gas and oil. There is another approach based on the observed deviation from the temperature-viscosity/ relaxation time correlation in the case of live oil. Consequently, this deviation can be correlated empirically to GOR . But, Winkler et al.  disputed the latter approach. As it overestimates GOR and they attributed this to the complexity of the competing factors; spin rotation and intramolecular attraction.
The extension of the capability of NMR in shale to include ﬂuid properties identiﬁcation will make it a stand-alone formation evaluation method. However, the accuracy of the properties derived will be doubtful. As the correlations used were derived based on pure alkanes assuming single decay constant. Also, the used correlations will have limited applicability to the formations and the ﬂuids used to derive them.
The fast relaxation of protons in shale limits the resolution of low-frequency ﬁeld measurements. Therefor, the ﬂuid properties will be better estimated using high-frequency NMR. However, these high ﬁeld measurement are double edged weapons. While it enables capturing the bulk relaxation response, it magniﬁes the heterogeneity in the system. So capturing this response would be more challenging and inaccurate. Moreover, the small size of the probe of the common NMR high-frequency machines will not allow sample measurements. Instead, the measurements will be performed on grinded samples.
There is another approach based on measuring the response from the sample at high-temperatures, then subtract the high temperature response from the ambient temperature response. So the differential response could be correlated to the change of ﬂuid viscosity with temperature assuming that surface relaxation component will remain constant. But this approach is disputed by Washburn . As they proved the dependence of the surface relaxivity of the organic pores on the temperature.
The abundance of different typing methods should boost the identiﬁcation process, however, unfortunately, shale heterogeneity limit the applicability of most of them. The T1-T2 method seems to be the most appropriate method for ﬂuid typing in shale. As, the high internal gradients observed in shale would limit the applicability of all diffusion based methods. Moreover, the accuracy of the ﬂuid properties estimated will depend on the efﬁciency of extracting the bulk relaxation response. Although most current correlations were derived for pure alkane components, the mixing rules proposed in the literature could extend their applicability to the multicomponent systems. The ﬂuid properties estimation from NMR measurements in shale is challenging and needs more experimental work to develop reliable correlations and properly calibrate the devices. The ﬂuid typing in shale is a complicated process, but it would be better performed by well-understanding of the shale system on one side and developing robust identiﬁcation methods on the other side.
 F.J. Slijkerman, J.P. Hofman, W.J. Looyestijn, et al., A practical approach to obtain primary drainage capillary pressure curves from NMR core and log data, Petrophysics 42 (04) (2001).
 E. Odusina, C.H. Sondergeld, C.S. Rai, An NMR study on shale wettability, in: SPE 147371, Presented at the Canadian Unconventional Resources Conference held in Calgery, Alberta, Canada, 15-17, November, 2011.
 T. AlGhamdi, C. Arns, Correlations between NMMR-relaxation response and relative permeability from tomographic reservoir-rock images, SPE Reserv. Eval. Eng. 16 (04) (2013).
 R. Akkurt, H.N. Bachman, C. Cao Minh, Nuclear Magnetic Resonance comes out of shell, Oil Field Rev. 20 (4) (2008/2009 Winter) 20.
 C.C. Minh, S. Crary, L. Zielinski, et al., 2D-NMR applications in unconventional reservoirs, in: SPE 161578, Canadian Unconventional Resources Conference held in Calgary, Alberta, Canada, 30 Octobere1 November, 2012.
 M. Fleury, Characterization of shales with low ﬁeld NMR, in: The International Symposium of Core Analysts, Avignon, France, 8-11 September, SCA2014-014, 2014.
 H. Daigle, A. Johnson, J.P. Gips, Porosity evaluation of shales using NMR secular relaxation, in: Unconventional Resources Technology Conference (URTEC), Denver, Colorado, 25-27 August, SPE1905272-MS, 2014.
 J.P. Gips, H. Dailage, S. Mukul, Characterization of free and bound ﬂuids in hydrocarbon bearing shales using NMR and Py-GC-MS, in: SPE-1917689- MS, Unconventional Resources Technology Conference (URTEC), Denver, Colorado, 25-27 August, 2014.
 R.F. Sigal, F. Odusina, Laboratory NMR measurements on methane satu- rated Barnett Shale samples, Petrophysics 52 (1) (2011) 32-49.
 Carl H. Sondergeld, Unconventional shale analysis, in: Lecture Notes on Unconventional Reservoirs, The University of Oklahoma, Norman, United States of America, 2014.
 V. Anand, G.J. Hirasaki, M. Fleury, NMR diffusional coupling: effects of temperature and clay distribution, Petrophysics 49 (4) (2008) 362.
 G.R. Coates, L. Xiao, M.G. Prammer, NMR Logging: Principles and Applications, Gulf Professional Publishing, Houston, 1999 (Reprint).
 Boqin Sun, Dunn Keh-Jim, NMR inversion methods for ﬂuid typing, in: SPWLA 44th Annual Logging Symposium, Galveston, Texas, 22e25 June, SPWLA-2003-GGG, 2003.
 K.-J. Dunn, D.J. Bergman, G. Torraca, Nuclear Magnetic Resonance: Petrophysical and Logging Applications, Elsevier, 2002 (Reprint).
 H. Nelson, Pore-throat sizes in sandstones, tight sandstones, and shales, AAPG Bull. 93 (3) (2008) 329-340.
 D. Devegowda, Phase Behavior in Shale. Lecture Notes on Unconventional Reservoirs, The University of Oklahoma, Norman, United States of America, 2014.
 M. Fleury, E. Kohler, F. Norrant, et al., Characterization and quantiﬁcation of water in smectites with low-ﬁeld NMR, J. Phys. Chem. C2013 117 (2013)4551-4560.
 E.k Washburn, Relaxation mechanisms and shales, Concept. Magn. Reson. Part A 43 (3) (2014) 57-89. http://dx.doi.org/.
 R. Loucks, R. Reed, S. Ruppel, D. Jarvie, Morphology, genesis, and distribution of nanometer scale pores in siliceous mudstones of the Mississippian Barnett Shale, J. Sediment. Res. 79 (2009) 848-861.
 Q.R. Passey, K.M. Bohacs, W.L. Esch, R. Klimentidis, S. Sinha, My source rock is now my reservoir e geologic and petrophysical characterization of shale-gas reservoirs, Search Discov. (2012). Article #80231.
 F. Civan, D. Devegowda, Rigorous modelling for data analysis towards accurate determination of shale gas-permeability by multiple-repeated pressure-pluse transmission tests on crushed samples, in: SPE 170659-PT Presented at the SPE Annual Technical Conference and Exhibition,
Amsterdam, the Netherlands, October, 2014, pp. 27-29.
 A. Saada, B. Siffert, E. Papirer, Comparison of the hydrophilicity/hydrophobicity of illite and kaolinites, J. Colloid Interface Sci. 174 (1995) 185-190.
 T. Zhang, G. Ellis, S. Ruppel, K. Miliken, R. Yang, Effect of Organic-matter type and thermal maturity on methane adsorption in shale-gas systems, J. Org. Geochem. 47 (2012) 120-131.
 Dyrka, I. Petrophysical Properties of Shale Rocks. http://infolupki.pgi.gov. pl/en/gas/petrophysical-properties-shale-rocks [accessed 26.12.15].
 M. Schreiber, J. Chermak, Mineralogy and trace element geochemistry of gas shales in the United States: environmental implications, Inter. J. Coal Geol. 126 (2014) (2013) 32-44.
 A. Tinni, C. Sundergeld, C. Rai, NMR T1-T2 response of moveable and nonmoveable ﬂuids in conventional and unconventional rocks, in: Presented at the International Symposium of Core Analysts, Avignon, France,8-11 September, SCA2014-014, 2014.
 R. Kausik, K. Fellah, E. Rylander, P.M. Singer, R.E. Lewis, S.M. Sinclair, NMR Petrophysics for tight oil shale enabled by core resaturation, SCA 2014-073 (2014).
 A. Tinni, E. Odusina, I. Sulucarian, NMR response of brine, oil and methane in organic shales, in: Presented at SPE Unconventional Resources Conference, 1e3 April, Woodlands, Texas, USA, SPE-168971-ms, 2014.
 R. Kausik, C. Cao Minh, L. Zielinski, et al., Characterization of Gas dynamics in kerogen nanopores by NMR, in: SPE 147198, Presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, 30 October, 2 November, 2011.
 R. Xie, L. Xiao, Advanced ﬂuid-typing methods for NMR logging, Petroleum Sci. 8 (2) (2011) 163e169. http://dx.doi.org/.
 E.k. Washburn, J.E. Birdwell, A new laboratory approach to shale analysis using NMR relaxometry, in: Unconventional Resources Technology Con- ference (URTEC), Denver, Colorado, 12-14 August, SPE-168798-MS, 2013.
 L. Zielinski, R. Ramamoorthy, C. Minh, et al., Restricted diffusion effects in saturation estimates from 2D diffusion-relaxation NMR maps, in: SPE-134841-MS, Presented at SPE Annual Technical Conference and Exhibition, Florence, Italy, 19-22 September, 2010.
 G. Gannaway, NMR investigation of pore structure in gas shales, in: SPE 173474 Presented at the SPE Annual Technical Conference and Exhibition, Amsterdam, The Netherlands, October, 2014, pp. 27-29.
 J. Mitchell, J. Staniland, E.J. Fordham, Paramagnetic doping agents in magnetic resonance studies of oil recovery, Petrophysics 54 (4) (2013) 349e367.  R. Freedman, et al., Major advancement in reservoir ﬂuid analysis achieved using a new high-performance nuclear magnetic resonance laboratory system, Petrophysics 54 (5) (2013) 439-456 (GE).
 R. Patricia, R. Pedro, Estimation of ﬂuid properties using NMR correlation in Berea rocks, in: SPE 69608, Presented at SPE Latin American and Caribbean Petroleum Engineering Conference, 25e28 March, Buenos Aires, Argentina, 2001.
 J. Bryan, D. Moon, A. Kantzas, In situ viscosity of oil sands using low-ﬁeld NMR, J. Can. Petrol. Technol. 44 (9) (2003).
 M. Winkler, J.J. Freeman, M. Appel, The limits of ﬂuid property correlations used in NMR well logging: an experimental study of reservoir ﬂuids at reservoir conditions, Petrophysics 46 (2) (2004) 104-112.
 J. Chen, S. Chen, A mixing rule for of elf diffusivities in methane hydrocarbon mixtures and determination of GOR and oil viscosities from NMR log data, Res. Eval. Eng. 13 (2) (2008) 275-282. SPE-115510-PA.
 J. Hirasakia, S. Lo, Y. Zhang, NMR properties of petroleum reservoir ﬂuids, Magn. Reson. Imagin. 21 (2003) 269-277.
 S.-W. Lo, G.J. Hirasaki, W.V. House, R. Kobayashi, Mixing rules and correlations of NMR relaxation time with viscosity, diffusivity, and gas/oil ratio of methane/hydrocarbon mixtures, in: SPE-77264-PA, Presented at SPE Annual Technical Conference Proceedings, 1-4 October, Dallas, Texas, 2002.
Received 26 October 2015
Received in revised form
25 January 2016
Accepted 3 February 2016
Corresponding author. (Mohamed Mehana) Tel.: +1 4058873536.
E-mail address: M.Mehana@ou.edu (M. Mehana).
Peer review under responsibility of Southwest Petroleum University.
Copyright © 2016, Southwest Petroleum University. Production and hosting by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).