Saidian A Comparison Of Measurement Techniques For Porosity Compressed

111The fourth sample set consisted of 22 samples taken from the Silurian play in Eastern Europe112(Kuila, 2013). The samples contained up to 52 wt% quartz and up to 57 wt% clay which is113mostly illite (Figure 1b). The kerogen was thermally mature in the gas window and TOC ranged114between 1-6 wt %. No pyrogram peaks were observed in the Rock Eval results up to 550 C115which confirmed the absence of any pyrolyzable kerogen or bitumen in these samples. The HI116varied from 0 to 9 with an average of 3. Figure 3a and b show SEM images of two Silurian117samples. Silt-size particles such as quartz, pyrite, and dolomite are surrounded by a matrix of118illite and chlorite and organic matter is dispersed between clay particles.1193.3 Niobrara Formation12022 marl and chalk samples came from a well in the Berthoud Field, Larimer County, CO, USA121(Kuila, 2013), specifically from the Fort Hays limestone and the overlying Smoky Hill members122of the Niobrara formation. They were calcite-rich rocks with moderate amounts of clay (up to 35123wt% with an average of 16.5 wt%), quartz (average of 11 wt%) and pyrite (Figure 1c). Type II124kerogen was thermally mature and was in the oil window based on average Tmax of 436 C and125HI which varied from 119 to 386 with an average of 306. TOC varied from 0.1 to 5.3 wt% and126RockEval programs between 450-500 C showed an abundance of bitumen in these samples.127SEM images for two samples are shown in Figure 4a and 4b. Figure 4a shows relatively more128homogeneous intercrystalline pore distribution which is filled with organic matter. Figure 4b129shows larger intercrystalline pores filled with organic matter. Note that the larger intercrystalline130pores are within the peloid structures.1313.4 Monterey Formation6

132A combination of 12 sidewall and conventional core samples were taken from an oil producing133well drilled at the western flank of the southern San Joaquin Basin in California. The samples134were predominantly quartz phase porcellanites containing moderate amounts of clay (up to 24135wt% with an average of 8 wt%) and pyrite (up to 7 wt% with an average of 3 wt%). A smaller136sample set (3 samples) were calcite-rich (Figure 1d). TOC varied from 0.85-4.95 wt% and137RockEval S2 data showed an abundance of bitumen (Rivera, 2014). Kerogen was Type II with138thermal maturity in the oil window based on average Tmax of 438 C and the hydrogen index (HI)139varying from 184 to 473 with an average of 374. SEM images of Monterey samples (Figure 5)140show moderate carbonate and dominant quartz grains with the presence of organic matter filling141up the intergranular and intercrystalline pores.1424. Methods143Several methods have been used to measure porosity and pore or throat size distribution. In this144section, we describe the pretreatments, displacement fluids, and methods as well as the145associated challenges for each measurement. In the results and discussion part we will146investigate how these challenges and limitations affect the porosity measurement for different147sample sets.1484.1 Helium Expansion using the Gas Research Institute (GRI) Method and Helium Injection149Under Confining Stress150The porosity was measured by a commercial laboratory using the Gas Research Institute (GRI)151helium porosimetry technique. Due to sample limitations only Silurian and the Haynesville152samples were used for these measurements. The porosity was also measured using helium153injection under confining stress for the Monterey samples. In the GRI technique, bulk rock154volume is measured by mercury immersion using Archimedes’ principle and crushed rock grain7

155volume is measured by Boyle’s Law (Luffel and Guidry, 1992; Luffel et al., 1992; GRI-15695/0496). We measured the porosity of cleaned and dried cylindrical Monterey samples with a157CMS300TM helium injection porosimeter.158Helium expansion, using either crushed or intact samples, measures pores that are connected and159accessible by helium gas. In nanodarcy permeability rocks, temperature fluctuations can160compromise pressure equilibration and decrease the measurement accuracy. On the other hand,161using crushed samples allows the gas to access isolated pores that might be inaccessible162otherwise. Sondergeld et al. (2010b) and Passey et al. (2010) reported very high disparities in the163porosity and permeability values measured by different laboratories using same samples. Kuila et164al. (2014a) provided a thorough discussion of the GRI technique and investigated the effect of165pretreatment and laboratory conditions on the final grain density and porosity results. For166example, the Dean-Stark extraction pretreatment with a hot solvent can create porosity in167thermally mature samples, such as the Niobrara and Monterey, by dissolving bitumen and lead to168porosity overestimation. Luffel and Guidry (1992) suggested 30 minutes as the equilibration time169for shales as opposed to 1 minute for conventional rocks.1704.2 Water Immersion (WI)171The original protocol for WI porosity measurements was recommended by the American172Petroleum Institute (API RP40). In this study, we used the adapted protocol for mudrocks173developed by Kuila et al. (2014a). In this adapted protocol, approximately 5 grams of rock chips174are first dried and weighed in air. They are then saturated and weighed again in water. This175protocol of using intact samples and measuring grain density at low humidity increases the176repeatability of the experiment (Kuila et al., 2013).8

177Immersion porosity measurements assess the pores available to the saturating fluid (water or oil).178Thus, the fluid type, the pore surface wettability, and the saturation method affect the porosity179values. Also presence of expandable clays such as smectite would affect the accuracy of this180technique.1814.3 Mercury Intrusion (MI)182In MI porosimetry, small intact rock chips are heated up to 200 C for 12 hours and degassed for18330 minutes at 50 µmHg to remove water and volatile hydrocarbons. Mercury is then injected in184the sample at uniform pressure steps from 0.14 to 420 MPa. Pressure is considered equilibrated185when the injection rate falls below 0.001 µl/g/s. The Washburn model (Washburn, 1921) is used186to convert the pressure data to pore throat size distribution. A conformance correction, for187example, the Bailey method from Comisky et al. (2011) must be applied to high pressure188measurements to account for mineral compressibility. The porosity is calculated by measuring189the bulk volume of the sample submerged in the mercury and the pore volume measured by the190volume of intruded mercury.191We used MI to measure porosity and pore throat size distributions for all sample sets. The MI192technique measures pore and bulk volume as well as pore-throat size distribution. Since MI can193only measure the volume of pores with throat-diameter larger than 3.6 nm (at 420 MPa), a194significant portion of the pore space in fine-grained rocks is neglected.1954.4 Nitrogen Adsorption (N2)196This technique has traditionally been used to measure total specific surface areas (TSSA) in197conventional reservoir rocks. Due to practical limitations N2 technique can only assess the198volume of pores with a diameter of less than 200 nm (Gregg and Sing, 1983). 1-3 grams of9

199samples are crushed, sieved through a 40 mesh (420 µm) sieve, and then degassed under vacuum200at 200 C until the outgassing rate is less than 0.005 Torr/min over a 15 minute interval. Nitrogen201is injected into the analysis chamber and adsorption of nitrogen to the degassed sample starts.202The adsorption takes place under constant temperature of liquid nitrogen. Pressure and adsorbed203quantity of nitrogen are recorded as isotherms and used to calculate pore size distribution. We204used the Barrett, Joyner and Halenda (BJH) inversion (Barrett et al., 1951) as recommended by205Kuila (2013) for mudrocks. This inversion method calculates the PSD assuming non-connecting206cylindrical pores. The total pore volume is calculated by measuring the amount of nitrogen207adsorbed in the sample. The Harkins and Jura (1944) thickness curve is utilized for both BJH208inversion and micropore (pores smaller than 1.7 nm) volume calculation with t-plot analysis. The209total pore volume is calculated by combining the micropore volume and the total volume210measured from 1.7 nm to 200 nm. We used this technique to measure pore size distribution and211pore volume in all samples.2125. Results213We analyzed samples with representation from major lithology types (Figure 1): predominantly214carbonate (Niobrara); predominantly quartz (Monterey); predominantly clay (Haynesville); and a215mineral mixture (Silurian). Each sample set has varying amounts of clay and TOC contents. This216richness of data allowed us to analyze and explain porosity mismatch between the methods for217mineralogy, pore size distribution, and measurement condition effects.2185.1 Haynesville Formation219Porosity values are measured by N2, WI, GRI and MI techniques (Figure 6). The following220observations can be made from this figure:10

221 MI underestimates the porosity when compared with N2 and WI (Figure 6a and 6b).222 WI and N2 porosities are comparable within 2 p.u. difference (Figure 6c).223 GRI and N2 are comparable within 2 p.u. difference (Figure 6d).224 Except some low content samples that show higher GRI porosity, the WI and GRI225porosity show similar porosities within 2 p.u. difference (Figure 6e). Note that in Figure 6226N2 and WI and GRI experiments are not performed for all samples.227 The presence of clays does not affect the porosity measurements and porosity valueincreases with clay content.228229The PSD was measured using the N2 technique (Figure 7). As shown in this figure the PSD230spectra varies significantly for different samples. Figure 7a and 7b show the PSD colorcoded by231clay and TOC content, respectively. Samples with high clay (low TOC) show large amplitudes at232the small mesopore range ( 10nm). Samples with low clay (high TOC) show large amplitudes at233the big mesopores (10-50 nm) and macropore range ( 50 nm). MI throat size distribution was234also measured for a subset of Haynesville samples (Figure 8). The MI spectra suggests that there235are pores with throats smaller then 3.6 nm which are not assessed by this technique.2365.2 Silurian Formation237We measured the porosity of the Silurian samples using WI, MI and N2 techniques (Figure 9).238The following observations can be made:239 9a to 9c).240241242MI, N2 and WI measurements show an increase in porosity with increase in TOC (Figure MI highly underestimates the porosity compared to WI and N2 techniques (Figure 9a and9b).11

243 WI and N2 porosities are comparable withing 2 p.u. difference (Figure 9c).244 GRI generally underestimates the porosity compared to WI and N2 in these samples(Figure 9d and 9e).245246We measured PSD using N2 (Figure 10) and MI (Figure 11) techniques for all Silurian samples.247The N2 PSD spectra are colorcoded by both clay (Figure 10a) and TOC (Figure 10b) contents.248We make the following observatins:249 Significant variation in amplitude in the small mesopore range ( 10 nm) is observed.250 The big mesopore (10-50 nm) and macropore range ( 50 nm) show similar spectra withsmall variations in amplitude.251252 TOC and clay content do not show a clear effect on the PSD spectra.253 The MI PSD results (Figure 11) suggest that there are pores that are not accessible usingthe MI technique.254255256 Clay content does not show a effect on the MI spectra amplitude (Figure 11a) whereasTOC and spectra amplitude show a direct correlation (Figure 11b).2575.3 Niobrara Formation258Figure 12 shows the comparison between porosity values measured using the N2, MI and WI259techniques. Two distinct groups of data are observed in Figure 12a. Group 1 (data circled by a260blue dashed line in Figure 12a) shows higher WI porosity compared to N2 porosity, whereas261Group 2 (data circled by a red solid line in Figure 12a) shows more comparable WI and N2262porosities (within 2 p.u. difference). We make the following observations for samples with a263wide distribution of pore sizes and high clay content such as the Niobrara samples:12

264 samples and WI overestimates the porosity for some samples (Figure 12a).265266 WI and MI porosities are comparable within 2 p.u. for low clay content ( 10 wt%)samples. WI overestimates the porosity for high clay samples ( 10 wt%) (Figure 12b).267268Comparing N2 and WI samples show that N2 underestimates the porosity for some N2 shows higher porosity for high clay content samples and MI shows higher porosity forlow clay samples (Figure 12c).269270We measured PSD for Niobrara samples using the N2 (Figure 13) and MI (Figure 14)271techniques:272 The shapes of the PSD spectra (Figure 13) for low clay content samples show increasing273PSD amplitude which suggests the presence of pores larger than 200 nm. The high clay274content samples show a dominant pore size of 80-100 nm.275 Similar to N2-PSD data, the MI PSD data (Figure 14) show two different sets of spectra.276One set shows a PSD spectrum with a dominant pore size 80-100 nm (low clay content)277and the other set shows an abundance of small pores smaller than 20 nm (high clay278content) (Figure 14).279 The shape of MI spectra for high clay samples indicates the presence of pores that are notaccessible by mercury.2802815.4 Monterey Formation282Porosity values were measured for Monterey samples using WI, MI, HE and N2 techniques283(Figure 15). We make the following observations:284285 WI, MI and HE porosities show comparable values within 2 p.u. difference (Figure 15ato 15c).13

286 HE and WI show the best correlation (Figure 15b).287 N2 significantly underestimates the porosity for high porosity ( 5 p.u.) samples.288 N2 and HE show comparable values (within 2 p.u.) for low porosity samples ( 5 p.u.).289 Clay content does not affect the porosity measurements.290Pore size distributions were measured for Monetery samples using N2 (Figure 16) and MI291(Figure 17) techniques, colorcoded by clay content:292 and shows no correlation with the clay content.293294 299The MI throat size distributions (Figure 17) also show a significant variation in bothamplitude and throat size.297298The amplitide for small mesopores ( 10 nm) is small except for one high clay contentsample (Figure 16).295296The N2 pore size amplitude (Figure 16) for pores larger than 10 nm varies significantly For these spectra clay content does not show any correlation with the shape andamplitude either.3006. Discussion301The main driver for this comparative study was to analyze the differences in each method and to302exploit these differences to learn more about the samples honoring the mineralogical and303geochemical properties of each sample set. Samples were chosen from different formations.304Haynesville and Silurian formations are highly mature (gas window) with a very low hydrogen305index (average of 36 and 3, respectively). The Niobrara and Monterey samples are less mature306(oil window) with a very high hydrogen index (average of 306 and 374). We discussed the307results considering the effect of thermal maturity, TOC and clay content on the porosity, pore308size distribution and pore types. We worked with the strengths and drawbacks of each method as14

309listed in Table 2 for porosity and pore size distribution measurement. Here, we discuss our310results and evaluate the benefits and applicability of each technique for organic-rich mudrocks.311As mentioned in Table 2 MI technique measures the porosity for pores with a throat diameter312smaller than 3.6 nm. Underestimation of MI porosity values for Haynesville (Figure 6a and 6b)313and Silurian (Figure 9a and b) is because of the high thermal maturity (low HI) and consequently314the abundance of organic matter (OM)-hosted pores in these samples. OM-hosted pores are in315the mesopore (2-50 nm) and small macropore size (50-200 nm) range and not accessible by MI316technique. MI porosity values for low clay content ( 10 wt%) Niobrara (Figure 12b) and317Monterey samples (Figure 15a and c) are comparable with other techniques. The porosity in318these samples are mainly intercrystalline and accessible by mercury.319WI porosity measures the pores that are filled with distilled water. The presence of expandable320clays in the samples causes overestimation of porosity. Mineralogy of Niobrara samples showed321presence of expandable clays (smectite) in these samples. The clay content in other sample sets322mainly consisted of Kaolinite and illite which are less expandable. The effect of expandble clays323in Niobrara high clay ( 10 wt%) samples can be seen when WI is compared against N2 (Figure32412a) and MI (Figure 12b). In more thermally mature samples such as Haynesville (Figure 6c)325and Silurian (Figure 9c) and low clay content samples such as Monterey (Figure 15a and 15b)326the clay swelling is not significant and WI porosity is comparable with other techniques within 2327p.u. difference. As mentioned in Table 2 water might not be able to fill the hydro-phobic organic328hosted pores. We can not directly observe this phenomena in the porosity data presented in this329study but there are some indications of the effect of hydrophobic pores on WI porosity330measurement. There is higher scatter in the WI-N2 porosity comparison for Haynesville (Figure3316c) and Silurian (Figure 9c) samples compared to the HE-WI porosity comparison for Monterey15

332samples (Figure 15b). Monterey samples are in the oil window and intercrystalline pores are the333dominant pore types, but Haynesville and Silurian are in the gas window and OM-hosted pores334form the dominant pore type which is hydrophobic.335Since the N2 technique can not assess pores with a diameter larger than 200 nm, the comparable336N2 porosity for Haynesville (Figure 6c and 6d) and Silurian (Figure 9c) shows the abundance of337pores within the mentioned range in these samples. High Clay ( 10 wt%) Niobrara samples also338show the presence of OM-hosted pores. The N2 technique is able to measure the porosity in339High Clay ( 10 wt%) Niobrara samples while the MI technique underestimates and WI340overestimates the porosity. OM-hosted pores are not abundant in the low porosity ( 5 p.u.)341Monterey samples. The similarity of N2 porosity with other techniques (Figure 15a to 15c), for342these samples is due to the presence of intercrystalline pores with a diameter smaller than 200343nm.344GRI porosity was measured for Haynesville samples and a limited number of Silurian samples.345The higher scatter in low clay ( 35 wt%) Haynesville samples (Figure 6e) is due to hot solvent346extraction prior to GRI measurements. Removal of soluable organic matter increases the porosity347measured by the GRI technique. This effect is ruled out for Silurian samples since no soluble348bitumen was observed in these samples. The underestimation of GRI for Silurian samples might349be due to inacscessibility of gas to the micropores. The HE technique successfully measures350porosity for the Monterey samples since no OM-hosted pores are present in these samples.351Comparison of porosity values measured by different techniques show that when the N2 and MI352techniques measure comparable porosity with other techniques, the pore size distribution353assessed by these techniques can be used to study the pore structure. For example the N2 pore16

354size distribution can be used to study the pore structure

85 values measured various techniques with the pore size distributionwith s measured with three 86 different techniques. Finally, we provide recommendations for a new approach for pore size 87 distribution comparison in mudrocks . 88 In this study we use the pore size classification suggested by et al. (1994). In this Rouquerol