L ow fi eld nuclear magnetic resonance (NMR) has opened a wide spectrum of opportunities for reservoir characterization. NMR logging tools and laboratory systems can explore the physical interactions of proton bearing fl uids when they are exposed to magnetic fi elds at frequencies of 1 to 2 MHz. Several service companies currently market NMR-based logs focusing primarily on determining properties of the reservoir such as rock porosity and permeability, as well as the distribution of mobile and immobile fl uids. Parallel to the development of logging tools, laboratory based bench-top instruments have also been developed for research and calibration purposes.Low fi eld NMR technology was initially developed for formations in Texas and the North Sea. The primary targets for the development of this technique were sandstones in the Gulf of Mexico and Texas Chalks, followed by the large sandstone reservoirs in the North Atlantic. NMR has been used in the reservoir characterization of Canadian formations since the A number of techniques have previously been developed that use low fi eld nuclear magnetic resonance (NMR) relaxometry for conventional and heavy oil reservoir characterization. In the current work, the adaptation of these algorithms for use in the oil sands industry is presented. NMR based methods have been developed for identifi cation of water and bitumen content in ore and froth samples. Consistent algorithms have been used to analyze over 500 ore samples and 50 froth samples from the Athabasca oil sands in northern Alberta. Preliminary analyses are shown, with applications for in-situ fl uid determination using NMR logging tools and improved process control in oil sands processing plants.Plusieurs techniques reposant sur la relaxométrie à résonance magnétique nucléaire (RMN) pour la caractérisation des réservoirs conventionnels et d'huiles lourdes ont été mises au point antérieurement. Dans le présent travail, on présente l'adaptation de ces algorithmes à des fi ns d'utilisation dans l'industrie des sables bitumineux. Des méthodes reposant sur la RMN ont été mises au point pour la détermination de la teneur en eau et en bitume dans des échantillons de minerai et d'écume. On a utilisé des algorithmes consistants pour analyser plus de 500 échantil-lons de minerai et 50 échantillons d'écume venant des sables bitumineux d'Athabasca dans le nord de l'Alberta. Des analyses préliminaires sont présentées dans le cadre de la détermination des fl uides in situ au moyen de sondes de RMN et d'une régulation de procédé amélioré dans les usines de traitement des sables bitumineux.

Alberta contains significant deposits of oil and gas in carbonate formations. Carbonates tend to have fairly tight matrix structures, resulting in low primary porosity and permeability. Laboratory characterization of carbonate properties is a slow and tedious process, however, core data is often collected in order to augment and tune logging tool predictions. In this application, having a good understanding of carbonate pore systems at the core analysis level is key to proper reservoir characterization. Low-field NMR is an emerging technology that shows great promise for rock characterization measurements. In this paper, low-field NMR technology is investigated for determining primary and secondary porosity through the interpretation of NMR spectra. This data was also used to establish the bound and mobile fluid distributions existing in the porous medium. The data set for this experimental work consists of a large collection of core samples from various fields in Alberta and Saskatchewan. CT data were analyzed to obtain the primary and secondary porosity fractions, which were used to find corresponding NMR cutoff values that separate the NMR spectra into primary and secondary porosity. A distinct relationship was observed between the primary porosity fraction and the irreducible water saturation, Swi. The fraction of NMR amplitude in the last peak of the NMR spectra can also be correlated to CT secondary porosity. Another important relationship observed is that the geometric mean relaxation time of the last NMR peak correlates well with the cutoff between primary and secondary porosity. The bound and mobile fluid distributions are generally distinguished through the identification of T2cutoff values. A correlation was found to predict T2cutoff for this wide range of samples. This study shows that information from the fully saturated NMR spectrum can be used to estimate primary and secondary porosity fractions in carbonates, as well as bound and mobile fluid fractions. Introduction Porosity of carbonates is a complex problem that has had only limited attention in the literature(1). In general, carbonate porosity is divided into primary and secondary porosity. These different types of porosity are not easily distinguishable unless the primary pores and the diagenesis processes that occurred are studied(1). Despite these difficulties, it is very important to recognize and attempt to quantify the different porosity types and mobile/immobile fluid fractions in carbonates in order to help in developing carbonate reservoirs and to estimate the pore connections and recovery efficiency in these reservoirs. As various researchers have found, Nuclear Magnetic Resonance (NMR) can capture pore size information of the porous media(2–4). Thus, in theory, it describes both the primary and secondary porosity. However, separating the signal into different porosity components remains a daunting task. Part of this difficulty arises from the fact that there is no clear distinction between primary and secondary pore size distributions, as they overlap with each other. Chang et al.(3) have previously tried to separate the signal of vugs in NMR response. In carbonates, however, even the definition of vugs can be quite different. Chang et al.(3) used the term vugs to describe cavities that are formed in the matrix by diagenesis, with sizes ranging from about 100 μm to cavern size.

Conventional reservoir analysis has always been an extensive process. In order to properly characterize a reservoir, cores and/or logs have to be obtained. Both core and log analysis is expensive and time consuming. NMR is an attractive alternative to these tools due to the fact that in theory, only one measurement is required. However, the conventional methods of interpreting NMR data only seem to work for simple sandstones. A new method of interpreting NMR data is required for complex porous structure such as carbonates. It was found that NMR can predict porosity that is similar to the values obtained by gas expansion. By using the NMR data at fully saturated and irreducible water saturation (Swi), a T2cutoff value was obtained for each sample that separates the bound and movable fluid signals. It was found that T2cutoff for carbonates is not 100 ms as is widely believed by many people who have analyze NMR in carbonates. A correlation for T2cutoff was found as a function of the size of the last peak and its geometric mean. A correlation was also found for Swi, which was a function of the size of the first and last peak. The Free Fluid and the mean T2 permeability models were evaluated. It was seen that the predictions from these models were not adequate. Another permeability model was developed, which is expressed in terms of the size of the first and last peak of the NMR spectrum obtained from the fully saturated sample. It was found that the correlation did a better job of predicting the permeability values. The new model has its own limitations, a method is currently being investigated to overcome these limitations. Despite these limitations, however, the new NMR permeability model provides better estimates of carbonate permeability than any other established methods. Introduction Conventional methods of analyzing the characteristics of carbonate reservoirs usually involve physically analyzing the core samples and/or analyzing the various logs collected from the wells. The important reservoir parameters that are usually investigated are porosity, permeability, and irreducible water saturation. These parameters will give an indication of the amount of hydrocarbons existing in the reservoir and how easy it is to recover them. In order to find these parameters using conventional core analysis, the cores samples taken from the wells first have to be cut and cleaned. The samples are then measured for porosity using one of many options available, and permeability is measured at the dry state. To find Swi, the core samples have to be saturated with brine and spun to the irreducible water condition. Carbonates generally have very tight pore structures, so the process of finding these parameters through core analysis is expensive and time consuming. To determine these reservoir parameters through log analysis, various logs have to be run. Due to limited vertical resolution, the presence of vuggy porosity might not be detected at all1,2. Also, to estimate porosity from logs, lithology components are required3. This causes difficulties in analyzing carbonate reservoirs in which the lithology is quite complex. Nuclear Magnetic Resonance (NMR) is a fairly recent application in reservoir study and it has garnered major successes in characterizing sandstone reservoirs4. From a single NMR spectrum at the fully saturated conditions, porosity, irreducible water saturation and permeability of these reservoirs could be estimated. However, NMR application in carbonates has not been very successful.This is due to the fact that most earlier works assumed simple lithology and attempted to use the same models as for the sandstone reservoir. Thus the traditional method of interpreting NMR data can often lead to erroneous estimations in complicated porous media such as carbonates4. This paper details an attempt to investigate porosity, permeability and irreducible water saturation by using NMR and Computed Tomography (CT) method to provide details on the pore structure of the carbonate samples.

ACKNOWLEDGEMENTS iv LIST OF FIGURES ix LIST OF TABLES xi 24 4.0 NMR APPLICATIONS 4.1 Pore size distribution 4.2 Porosity 27 4.3 Bound and free fluid fractions 30 4.4 Permeability 4.4.1 Free Fluid model 34 4.4.2 Mean T 2 (SDR) model 5.0 PROCEDURE 38 vi 5.1 Sample preparation 38 5.2 Air permeability 39 5.3 Gas expansion 41 5.4 Brine saturation 42 5.5 Brine permeability 44 5.6 NMR measurements 45 5.7 Computerized Tomography (CT) scanning 47 5.8 Centrifuging 48 6.0 POROSITY DETERMINATION 49 6.1 Porosity measurements 49, 6.1.1 Gas expansion 49 6.1.2 Archimedes principle 49 6.1 .3 NMR 51 6.1.4 CT (wet cores) 53 6.1.5 CT (dry cores) 56 6.2 Porosity comparisons 6.2.1 NMR porosity at d i fferent echo spacings 6.2.2 Brine saturation and gas expansion 6.2.3 NMR and brine saturation 6.2.4 NMR and gas expansion 6.2.5 CT (wet) and gas expansion 62 6.2.6 CT (thy) and gas expansion 63 6.2.7 CT (wet) and CT (thy) 63 7.0 EVALUATION OF IRREDUCIBLE WATER SATURATION (55 7.1 Investigation of end effects 65 7.2 S,j determination from mass balance 68 7.3 S determination from NMR 71 7.4 S,j comparison between mass balance and NMR 72 8.0 DETERMINING CORRELATIONS TO ESTIMATE IRREDUCIBLE WATER SATURATION 75 8.1 T2toff determination 75 8.2 Relationships between T2utoff and characteristics of NMR spectra 79 8.3 Relationships between S wi and characteristics of NMR spectra 87 9.0 DETERMINATION OF VUGGY AND NON-VUGGY POROSITY 91 9.1 Vuggy and non-vuggy porosity determination 9.2 Observed relationships between various parameters vii 92 96 10.0 PERMEABILITY 10.1 Air and brine permeability 10.2 Permeability models 10.2.1 Evaluation of existing permeability models using NMR data 10.2.2 Development of a new permeability model 12.0 CONCLUSIONS AND RECOMMENDATIONS 12.1 Conclusions 114 11.2 Recommendations 116

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractAlberta contains significant deposits of oil and gas in carbonate formations. Carbonates tend to have fairly tight matrix structures, resulting in low primary porosity and permeability. As a result, laboratory characterization of carbonate properties is a slow and tedious process. Low field NMR is an emerging technology shows great promise in rock characterization. In a single NMR experiment, rock properties like porosity, permeability and S wi can in theory be measured.Experiments were performed on approximately 80 core plugs from six carbonate formations.Porosity measurements were performed through gas expansion and brine saturation (Archimedes' principle). Air permeabilities were also measured. Cores were also saturated with brine and spun to irreducible water saturation. NMR measurements were taken at both saturation stages (S w = 1.0 and S w = S wi ). NMR data were interpreted using the conventional core analysis results as guides. Observations were made regarding NMR trends and corresponding rock properties.Preliminary analysis of the data shows that NMR can successfully predict the content and distribution of the fluids in the porous media. Also, the NMR spectra of carbonate samples seem to suggest that NMR can be used as a tool for classifying cores into different pore systems.Using T 2cutoff values as a tool, the cores can be divided into groups. Group 1 has T 2cutoff < 80 ms, while group 2 has T 2cutoff in the range of 80 to 200 ms and group 3 has T 2cutoff values > 200 ms. Group 1 cores generally have higher porosity values than groups 2 and 3. Group 1 also has low values of T 2gm , compared to the other two groups. Air permeability was compared to the geometric means and T 2gm_cutoff values of the three groups and some general trends were observed, but more analysis is required before these trends can be quantified.