GB2434875A

GB2434875A - T1 based logging

Info

Publication number: GB2434875A
Application number: GB0706262A
Authority: GB
Inventors: Ridvan Akkurt
Original assignee: Halliburton Energy Services Inc
Current assignee: Halliburton Energy Services Inc
Priority date: 2003-10-03
Filing date: 2004-10-01
Publication date: 2007-08-08
Anticipated expiration: 2024-10-01
Also published as: GB2434875B; GB0706262D0

Abstract

NMR Well logging method providing a distribution of t1 relaxation times obtained with a maximum wait time less than 5 times the maximum T1 component of the T1 spectrum, and computing a correction for insufficient polarization which is based solely on the provided distribution of T1 relaxation times and acquisition parameters of the log. Increased logging speed is possible by this method.

Description

SYSTEM AND METHODS FOR Ti-BASED LOGGING

FIELD OF THE INVENTION

[0001] The present invention relates generally to nuclear magnetic resonance (NMR) well logging and in particular to T relaxation measurements for wireline, logging-while-drilling (LWD) and other applications.

BACKGROUND OF THE INVENTION

[0002] In oil and gas exploration it is desirable to understand the structure and properties of the geological formation surrounding a borehole, in order to determine if the formation contains hydrocarbon resources (oil and/or gas), to estimate the amount and producibility of hydrocarbon contained in the formation, and to evaluate the best options for completing the well in production. A significant aid in this evaluation is the use of wireline logging and/or logging-while-drilling (LWD) or measurement-while-drilling (MWD) measurements of the formation surrounding the borehole (referred to collectively as "logs" or "log measurements"). Typically, one or more logging tools are lowered into the borehole and the tool readings or measurement logs are recorded as the tools traverse the borehole. These measurement logs are used to infer the desired formation properties.

[00031 NMR logging has become very important for purposes of formation evaluation and is one of the preferred methods for determining formation parameters because of its non-destructive character. Improvements in the NMR logging tools, as well as advances in data analysis and interpretation allow log analysts to generate detailed reservoir description reports, including clay-bound and capillary-bound related porosity, estimates of the amounts of bound and free fluids, fluid types (i.e., oil, gas and water), permeability and other properties of interest. In general, NMR logging devices may be separate from the drilling apparatus (in what is known as wireline logging), or they may be lowered into the borehole along with the drilling apparatus, enabling NMR measurement while drilling is taking place.

The latter types of tools are known in the art as logging-while-drilling (LWD) or measurement-whi Ic-drilling (MWD) logging tools [0004] NMR tools used in practical applications include, for example, the centralized MRIL tool made by NUMAR Corporation, a [-falliburton company, and the sidewall CMR tool made by Schlumberger. The MRIL tool is described, for example, in U.S. Pat. 4,710,713 toTaicher et al. and in various other publications including: "Spin Echo Magnetic Resonance Logging: Porosity and Free Fluid Index Determination," by Miller, Paltiel, Gillen, Granot and Bouton, SPE 20561, 65th Annual Technical Conference of the SPE, New Orleans, LA, Sept. 23-26, 1990; "Improved Log Quality With a Dual-Frequency Pulsed NMR Tool," by Chandler, Drack, Miller and Prammer, SPE 28365, 69th Annual Technical Conference of the SPE, New Orleans, LA, Sept. 25-28, 1994. Certain details of the structure and the use of the MR IL tool, as well as the interpretation of various measurement parameters are also discussed in U.S. patents 4,717,876; 4,717,877; 4,717,878; 5,212,447; 5,280,243; 5,309,098; 5,412,320; 5,517,115, 5,557,200; 5,696,448; 5,936,405; 6,005,389; 6,023,164; 6,051,973; 6,107,796; 6,111,408; 6,242,913; 6,255,819; 6,268,726; 6,362,619; 6,512,371; 6,525,534; 6,531,868; 6,541,969; 6,577,125 and 6,583,621. The structure and operation of the Schiumberger CMR tool is described, for example, in U.S. Pat. Nos. 4,939,648; 5,055, 787 and 5,055,788 and further in "Novel NMR Apparatus for Investigating an External Sample," by Kleinberg, Sezginer and Griffin, J. Magn. Reson. 97, 466-485, 1992; and "An Improved NMR Too! Design for Faster Logging," D. McKeon et a!., SPWLA 40th Annual Logging Symposium, May-June 1999. The content of the above patents is hereby expressly incorporated by reference for all purposes, and all non-patent references are

incorporated by reference for background.

[00051 NMR logging is based on the observation that when an assembly of magnetic moments, such as those of hydrogen nuclei, are exposed to a static magnetic field, they tend to align along the direction of the magnetic field, resulting in bulk magnetization. The rate at which equilibrium is established in such bulk magnetization is characterized by the parameter T1, known as the spin-lattice relaxation time. The T1 parameter characterizes the coupling of nuclear spins to energy-absorbing molecular motions like rotation, vibration and translation.

[0006J Another related and frequently used NMR logging parameter is the spin-spin relaxation time 12 (also known as transverse relaxation time), which is an expression of the relaxation due to non-homogeneities in the local magnetic field over the sensing volume of the logging tool. In general, the mechanisms for spin-spin relaxation time T2 include, in addition to those contributing to T, the exchange of energy between spins. Both the T, and the T2 relaxation times provide information about formation porosity, composition and quantity of formation fluid, and other parameters important in oil exploration.

[0007] The pioneers in NMR measurement technologies envisioned the relaxation time T as the primary measurement result because it carries only information about the liquid-solid surface relaxation and bulk-fluid relaxation. In particular, unlike the transverse relaxation time 12, the spin lattice relaxation parameter T is not affected by rock-internal magnetic field gradients or by differences in fluid diffusivity. Moreover, instrument artifacts influence 1, measurements to a much lesser degree than T2 measurements.

[0008] Despite this understanding, modern pulsed NMR logging in the early 1990s was based primarily on 12 measurements, largely because of hardware limitations. Specifically, the construction of the T recovery curve requires data collected with multiple wait times that range from a few milliseconds to several seconds. Acquiring T1 data using tools that operated in single-frequency mode without effective pre-polarization was too time-consuming and not feasible. 12 measurements, on the other hand, were faster and contained information similar to T1 at low resonance frequencies. As a result, 12 CPMG measurements were chosen as the main mode of tool operation.

[0009] One characteristic of NMR logging is that unlike many other logging methods the measurements are not instantaneous. Each measurement cycle, including the wait time needed for polarization, can take several seconds. Frequently, several cycles have to be stacked to achieve adequate signal-to-noise ratio (SNR).

[0010] Thus, if a cycle takes T seconds to complete, and N cycles must be stacked, the vertical resolution of a measurement is inversely proportional to vNT, where i' is the logging speed. Clearly, the longer the cycle times and the higher the logging speeds, the worse the vertical resolution. Therefore, an ever-present challenge in NMR logging is to design tools that can log faster, while retaining acceptable vertical resolution. For practical reasons overcoming this challenge is a important task. Several innovations towards faster logging have been put into practice over the past several years.

[0011] One such innovation was the introduction of multi-frequency logging in the early 1990s. The benefit of multi- frequency logging is that the tools acquire data simultaneously over several frequencies, and the additional SNR available can be used to speed tip logging as well as to obtain higher-quality results. The state-of-the-art in multi-frequency logging is the MR IL Prime tool by Nuniar, a Halliburton Corporation, which currently can operate on 9 frequencies.

[0012] Another innovation was the introduction of simultaneous acquisition of partially and fully polarized echo trains with different SNR. Proper total porosity measurements require: (1) a short interecho time Te to sample fast decays, (2) high SNR to reduce the uncertainty in the estimation of fast decays, (3) long sampling time (Ne T where Ne is the number of echoes) for adequate sampling of longer decays. It is practically impossible to achieve all these objectives with a unique wait time T Te and Ne combination; while maintaining acceptable logging speeds and vertical resolution. Therefore, one solution is to optimize the acquisition by mixing partially and fully recovered data with different measurement parameters T Te, Ne and desired SNR. Another closely related innovation was the concept of simultaneous-inversion, where data acquired with different measurement parameters is inverted simultaneously using forward models that properly account for the differences in fluid NMR properties, acquisition parameters and noise levels.

[0013] Yet another innovation was the use of pre-polarization. In this approach the cycle time for each measurement is shortened, by placing static magnets above the antenna section to realize additional polarization during tool motion. Current generation NMR tools generally contain pre-polarization sections, allowing overall faster logging. Various other approaches have been attempted in the art, including the patents listed above.

[0014] The focus of this application is on novel systems and methods for T NMR logging alone or in combination other logging techniques. As discussed in application serial number 60/474,747, filed on May 03, 2003, to the same assignee as the present application, T logging adds a different dimension to interpretation, sometimes by complementing 12 logs, sometimes uniquely by itself. The 747 application is incorporated herein for all purposes. The novel technical approaches in accordance with the present invention directed to overcoming problems associated with the prior art are discussed below.

SUMMARY OF THE INVENTION

[0015] In accordance with the present invention systems and methods for using nuclear magnetic resonance (NMR) T measurements for wireline, LWD and MWD applications and down-hole NMR fluid analyzers are provided. The Ti measurements are characterized by insensitivity to motion, as the detrimental effects arising from tool motion or fluid flow are effectively reduced or eliminated. T measurements alone or in combination with other standard oil field measurements are shown to provide efficient data acquisition resulting in compact and robust data sets, the potential for substantially increased logging speeds, and simple methods for fluid typing, including direct and robust identification of gas.

[00161 In one aspect, the present application concerns a direct and more robust method for identification of gas. The method is particularly applicable in tight reservoirs, where the longest T for the water phase is known a priori. A simple correction for gas is made possible in such applications, without having to resort to sophisticated processing techniques.

[0017] In another aspect, this application concerns ability to use NMR logs with short wait times, on the order of 2 to 3 seconds, capable of determine total porosity estimates using a self-sufficient correction algorithm that provides compensation for insufficient polarization.

A particularly important result of this approach is the ability to provide overall faster logging, due to shorter cycle times, or higher vertical resolution, dependent on the practical application. Thus, in a specific embodiment using pre-polarization of the NMR logging tool, good vertical resolution can be provided at logging speeds about or higher than 15 feet per minute.

[0018] In particular, in one aspect the invention is directed to a method for identifying gas in a geologic formation, comprising the steps of: providing a distribution of nuclear magnetic resonance (NMR) T, relaxation times corresponding to a NMR log of the geologic formation; selecting a threshold value in the provided T distribution; and integrating the T, distribution above the selected threshold value to obtain an estimate of gas-filled porosity in the geologic formation.

[0019] In another aspect, this application is directed to novel combinations of T and 12 logs, for example, to estimate the diffusivity of low viscosity light hydrocarbon (non wetting phase), and combine the above diffusivity estimates to calculate viscosity, and combine the T1 and D) values to estimate gas to oil ratios (GOR).

[0020] In yet another aspect, the invention is directed to providing a correction for insufficient polarization of NMR T1 and T2 logs based solely on a distribution of NMR T relaxation times and acquisition parameters of the NMR log. In another aspect, the invention is directed to increased-speed Ii logging, based on corrections for pre-polarization magnetization.

BRIEF DESCRWFION OF THE DRAWINGS

[0021] Other objects and many attendant features of this invention will be appreciated and better understood with reference to the following detailed description when considered in connection with the accompanying drawings, wherein: [0022] Fig. 1 is a block diagram of a NMR logging system, which can be programmed for use in accordance with a specific embodiment of the present invention; [0023J Fig. 2 is a partially schematic, partially block diagram of a NMR logging tool and attached electronics used in one embodiment directed to wireline logging; [0024] Figs. 3 and 4 illustrate pulse sequences that can be used in specific embodiments of the present invention; [00251 Figs. 5 and 6 show T1 and T2 logs in a tight gas reservoir to illustrate advantages of T, logging in accordance with this invention; [0026] Figs. 7 is a T model illustrating insufficient polarization and the post-inversion correction method in accordance with one embodiment of the present invention; Fig. 8 illustrates the first echo amplitude from the longest wait time as a function of depth; Fig. 9 shows porosity phenomena due to insufficient polarization; [0027] Fig. 10 is shows apparent porosity, where the polarization factors have been corrected for the longest wait times; the results of post-inversion correction in accordance with this invention, applied to the logs shown in Fig. 10 are shown inFig. 11; [0028J Fig. 12 illustrates polarization profiles used for faster T, logging in accordance with one aspect of the present invention; [0029] Fig. 13 shows porosity logs verifying the quality of petrophysical determinations from T1 measurements by comparison to corresponding T2 logs in simulated-drilling mode;

] Fig. [4 illustrates a series of wireline Ii logs used to assess the operational and petrophysical feasibility of the novel approaches in accordance with the invention; [0031] Fig. 15 shows NMR spectra from two core plugs [0032] Fig. 16 shows the pressure data in a well, confirming the NMR-based interpretation in accordance with the present invention; [0033] Fig. 17 shows estimation of gas to oil ratio (GOR) in accordance with the present invention.

DETAILED DESCRIPTION

Theoretical Background

[00341 The first step in any NMR measurement is to align the magnetic nuclei with a magnetic field. This alignment process, or polarization, is not instantaneous and takes some time, which is associated with the T1 parameter. In reservoir rocks, the value of T depends on the characteristics of the fluids and the confining pore space. Two distinct relaxation mechanisms acting in parallel determine the longitudinal relaxation time: 1 1 1 (1) i 73 where the subscripts B and S correspond to bulk and surface relaxation, respectively.

[0035] Direct measurement of the longitudinal magnetization is not feasible and therefore requires tipping of the spins onto the transverse plane. Relaxation in the transverse plane is generally a more complex process compared to T, relaxation. In general, the transverse relaxation time T2 is equal to or shorter than its longitudinal counterpart T1. This is mostly due to molecular diffusion characterized by T20, resulting in the following equation for the measured T2 response: 1 1 1 1 -=-± -(2) T TB T2 T20 where the extra term containing the subscript D corresponds to diffusion. Given the typical magnetic field gradients of modern logging tools, diffusion can dominate transverse relaxation in the case of highly diffusive fluids, leading to large T11F2 ratios.

[0036] Tipping of the spins onto the transverse plane is accomplished through the application of an RF-pulse at a specific resonance frequency (Larmor frequency). Once in the transverse plane, the tipped protons induce a rapidly decaying signal in the receiver coils of the NMR device. The time constant associated with this so-called Free Induction Decay (FID) is on the order of a few tens of microseconds only, making it hard to measure directly.

Laboratory measurements that aim at quantifying T1 through Saturation Recovery (SR) or Inversion Recovery (IR), often measure (the amplitude ot) this FID. FID measurements have yet to be demonstrated in oil-field logging tools; measurement of T, or T2 in the field is usually achieved through a series of spin-echo sequences. The best known spin- echo sequence is by Carr, Purcell, Meiboom and Gill (CPMG) and has been developed to facilitate easy measurement of the transverse relaxation time T2.

[0037] Although Inversion Recovery is the most popular T1 measurement method in laboratory applications, Saturation Recovery technique is the practical choice in well logging due to its efficiency and shorter cycle times. The SR sequence used for wireline T logging can be considered as a series of CPMG echo trainlets, comprised of a few echoes with distinctly different wait times. Within this analogy, the signal from the jth echo of the ith wait time, given relaxation times Tik and T2k with associated amplitudes ak, is given by y1 =ak(1-e (3) where Ic is the inter-echo time. Note that the response is dependent on both T1 and T2, and the imprint of 12 is eliminated when j=0. In practice, the decoupling of the T2 response is accomplished by using very short Te, and only a few echoes. Proper sampling in T1 acquisition requires at least a couple of recovery measurements per decade. As a rule of thumb, components faster than the shortest recovery time, or components slower than the longest recovery time can not be resolved. In the example logs shown below, the wait times vary from 10 ms to 6.3 s. The shortest inter-echo time is either 0.51 ms, or 0.60 ms. A maximum of 10 echoes are acquired within each trainlet.

[0038] T1 inversion used here is cast as a classic linear least squares problem, where the minimum of an objective function is sought, as described below. By defining the residual or the misfit between the data and the fit by e,, =d -y, (4A) where y, .-xL(1_ewe (4B) and d1 is the amplitude of the jth echo in the ith CPMG trainlet, the objective function to be minimized to determine that unknown vector x is given by (5) 1=1 J=I l=I [0039] Note that the elements of the unknown vector x Contain the amplitudes of the T distribution. While the first term in the objective function represents the sum of the squares of the misfit, the second term represents the penalty, or the regularization term that is applied to prevent spiky or oscillatory distributions. Zeroth order regularization has been used in the above equation only for sake of simplicity, since many alternatives exist for regularization. It should be noted here that as part of solving the linear problem, the relaxation times Tik and T2k are determined apriori. However, the longitudinal and transverse relaxation times can not be treated as independent properties since in general (6) where the explicit dependence may vary with k. The relationship is governed by the relaxivity of the rocks in the case of the wetting phase, and the bulk and diffusivity properties of the fluids in the case of the non-wetting phase.

[0040] Other factors such as internal gradients may also come into play in certain cases.

In the simplest approach, one can define a linear relationship in the following form T,k=TJk/2, (7) where A. is a constant, referred to as the T11f2 ratio, with a typical value of 1.5 to 3.0.

Actually, the first step of the inversion process used here involves the determination of the function that relates 12 to T1, using a model where the relationship is allowed to be non-linear. The information obtained from this initial process is then used in the inversion to set 12k for a given Tik.

[0041] Various techniques for T1 and T2 measurements including pulse sequences have been developed in the art, including, without limitation, the disclosures in U.S. Pat. Nos. 5,309,098; 5,517,115, 5,696,448; 5,936,405; 6,005,389; 6,023,164; 6,049,205; 6,051,973; 6,107,796; 6,111,408; 6,242,913; 6,255,819; 6,268,726; 6,362,619; 6,512,371; 6,525,534; 6,531,868; 6,541,969; 6,577,125; 6,600,316, which are incorporated herein for all purposes.

The System And Underlying Measurements [0042] Fig. 1 is a block diagram of a system, which can be programmed for use in accordance with a specific embodiment of the present invention, which shows individual block components for controlling data collection, processing of the collected data and displaying the measurement results. In Fig. 1 a logging tool 106 comprises an NMR probe controller 30 and pulse echo detection electronics and is lowered in a borehole drilled in the formation 109. The output signal from the tool detection electronics is processed by data processor 12 to record NMR pulse echo data from the tool and analyze the relaxation characteristics of the materials surrounding the borehole. The output of the data processor 12 is fed to parameter estimator 40. Measurement cycle controller 22 provides an appropriate control signals to the probe. The processed data from the log measurements is stored in data storage 60. Data processor 12 is connected to display 50, which is capable of providing a graphical display of one or more measurement parameters, preferably superimposed on display data from data storage 60. The components of the system of the present invention shown in Fig. 1 can be implemented in hardware or software, or any combination thereof suitable for practical purposes.

[0043] Reference is now made to Fig. 2, which illustrates in a semi-block diagram form an NMR logging apparatus, such as the MRIL Prime tool of Numar Corporation (a Halliburton Company), which can be used for NMR measurements in accordance with a preferred embodiment of the present invention. In standard operation, first portion 6 of the tool is arranged to be lowered into a borehole 7 having a borehole longitudinal axis 8 in order to examine properties of the geologic formation in the vicinity of borehole 7.

[0044] The first portion comprises a generally cylindrical permanent magnet 10, preferably having a longitudinal axis 11, which is preferably coaxial with the longitudinal axis 8 of the borehole. Alternatively, a plurality of permanent magnets 10 may be employed.

Permanent magnet 10 preferably has uniform magnetization substantially perpendicular to the longitudinal axis of the logging tool, which is parallel to the longitudinal axis 8 of the borehole 7.

[0045] The first portion 6 also comprises one or more coil windings 16, which preferably are arranged on top of the permanent magnet and form the tool antenna. The magnetization direction 13 created by the antenna is substantially perpendicular to the longitudinal axis 11 of the bore hole. The coil windings 16, together with a transmitter/receiver (T/R) matching circuit 20 define a transmitter/receiver (T/R) circuit. hR matching circuit 20 typically includes a resonance capacitor, a T/R switch and both to-transmitter and to-receiver matching circuitry and is coupled to a first RF power amplifier 24 and to a receiver pre-amplifier 26.

[0046] The permanent magnet 10 and coil windings 16 are preferably housed in a non-conductive, non-ferromagnetic protective housing 18. The housing and its contents will hereinafter be referred to as the probe 19. In operation, the probe along with RF amplifier 24, preamplifier 26 and T/R matching circuit 20, designated collectively as housing 28 are passed through the borehole. Alternatively, some of the above elements may be located above ground in housing 30.

[0047] Disposed in a housing indicated in Fig. 2 by block 30, is a control circuitry, including a computer 32, which provides a control output to a pulse programmer 34. Pulse programmer 34 controls the operation of phase shifter 44, as well as an RF driver 38, which drives RF power amplifier 24. Pulse programmer 34 controls the operation of a variable frequency RF source 36, the output of which is passed through phase shifter 44 to the RF driver 38. The signal from RF driver 38 is amplified in RF power amplifier 24 and passed through hR matching circuit 20 to the antenna 16.

[0048] NMR signals from excited nuclei in the formation surrounding the borehole are picked up by the receiving antenna 16 and passed through TIR matching circuit 20 to RF receiver pre-amplifier 26, the output of which is supplied to an RF receiver 40 which also receives an input from phase shifter 44. Receiver 40 outputs via an AID converter with a buffer 46 to the computer 32 for providing desired well logging output data for further use and analysis.

[0049] While the above description was provided with reference to wireline logging, it will be apparent that the T1 measurement principles of the present invention in a preferred embodiment can be applied to logging-while-drilling (LWD) or measurement-while-drilling (MWD). Further details on the structure of NMR wireline and LWD tools and fluid analyzers that can be used in accordance with preferred embodiments of the present invention can be found, for example, in U.S. patents 4,710,713; 4,717,876; 4,717,877; 4,717,878; 5,280,243; 5,712,566; 6,023,164; 6,107,796; 6,111,408; 6,268,726; 6,362,619; 6,512,371; 6,525,534; 6,531,868; 6,541,969; 6,577,125 and 6,583,621. Additional designs, as shown in U.S. Pat. Nos. 4,939,648; 5,055,787 and 5,055,788 can also be used. As noted earlier, the above patents have been incorporated herein by reference for all purposes.

[0050] In different embodiments, the T1 and 12 measurements in accordance with this invention can be obtained either simultaneously or separately, using the same or different NMR logging tools. In one embodiment, the T1 and 12 measurements are obtained using the LWD-MWD tool disclosed in U.S. Patent Number 6,531,868, which has been incorporated by reference. Such a tool contains two distinct operating modes, one designed for while-drilling operations and the other for wiping trips. In one embodiment, the motion-tolerant T1 measurements are obtained when drilling motion is detected and the tool switches over to T2 once drilling ceases. In another embodiment, T1 and T2 measurements are acquired simultaneously over the same depth interval during a wiping trip.

[0051] In one embodiment, the process for obtaining T1 measurements during a drilling operation is described as follows. At the start of a measurement, one or more radio frequency pulses covering a relatively wide range of frequencies, or using one or more pulses which are frequency swept, are transmitted to saturate the nuclear magnetization in a cylindrical volume around the drilling tool. The range of frequencies can be, for example, 50-IOOkI-lz and is covered in a specific embodiment using a rapid succession of short radio frequency pulses similar to the first pulse in a standard CPMG pulse sequence, or using a single long pulse in a frequency sweep. Changing the range of frequencies used in this step varies the position and the width of the sensitive region in the formation. In a specific embodiment using the tool, a frequency range between 50 and 100kHz saturates the nuclear magnetization in a cylindrical volume around the tool, where the cylinder has a typical diameter of 14", a height of 24", and thickness of between about 1/2" to 1". [0052] Following the step of saturation, which typically takes about 1

ms, in accordance with the present invention a readout pulse is transmitted at a frequency near the center of the range of covered frequencies. In alternative embodiments one or more subsequent readout pulses can also be used. In accordance with the present invention, a readout pulse sequence is comprised of a 90 pulse followed by data acquisition, or of a 90 pulse followed by a 1800 pulse, followed by data acquisition, where the steps of applying a 180 pulse and data acquisition can be repeated. The readout pulse sequence generally follows a predetermined wait time, as explained in more detail below. In a specific embodiment the readout pulse sequence is transmitted at a center frequency of about 500kHz, and is followed by one or more refocusing pulses. An illustration of a pulse sequence used in a specific embodiment of the present invention is shown in Fig. 3. Yet another type of pulse sequence that can be used in accordance with this invention is illustrated in Fig. 4, as discussed in more detail in the 6,531,868 patent.

[00531 Following the readout pulse(s), corresponding NMR echo signals are received, amplified and stored for further processing. In accordance with a preferred embodiment, the amplitude of the retained echo signal is interpreted as the level of nuclear magnetization present after the particular wait time. In the particular example considered above, the center frequency of the NMR echo signals corresponds to about 14" diameter of investigation.

[0054] The measurement process described above is repeated for a series of increasing wait times the values of which can, for example, be equally distributed on a logarithmic scale.

In a specific embodiment, wait times are stepped through the values Ims, 3ms, lOms, 3Oms, lOOms, 300ms, l000ms and 3000ms, and the measurement results are stacked to produce several data points on a multi-component 1, relaxation curve. In one embodiment, only a few echoes are collected for each wait time to compute the T1 relaxation curve. Preferably, about two to five echoes per wait time are retained.

[00551 T2 measurements can be obtained either separately or simultaneously with T1 measurements. In one embodiment, T2 measurements are obtained simultaneously with T1 measurements during one of the long wait times. This is done by acquiring a large number of echoes, preferably 500, during the long wait time and then using the large number of echoes to compute the 12 relaxation curve.

[0056) The T and T2 measurements obtained during the drilling operation are processed to derive petrophysical properties of local geological formations. As known in the art, these measurements can be used to compute distributions ofT1 and T2 relaxation times. The resultant distributions of T1 and T2 relaxation times comprise data points of T1 and T2 relaxation curves. These relaxation curves are further processed to extract the dominant T1 and 12 relaxation modes, from which amounts of bound water, free water and hydrocarbons are estimated. The characteristic l or T2 times of the surface-wetting phase can also be used to estimate formation pore size distributions and formation permeability, as known in the art.

[0057] In a particular application, the T1 and 12 relaxation curves can be used to determine different pore systems residing in carbonate formations and detect the existence of diffusive coupling among different pore systems. In one embodiment, a T1 relaxation curve is obtained and porosity analysis is performed by observing the T1 relaxation curve. The I relaxation curve may contain one or more peaks or modes. As known in the art, each peak or mode is associated with a pore system in the formation being analyzed. The size of each pore system can also be estimated based on the relaxation time associated with each peak. For carbonate formations, the T1 relaxation curve is bi-modal, indicating the existence of both a micro and macro pore systems. Additional details of this embodiment are disclosed in application Ser. No. 60/474,747, filed on May 03, 2003, which is incorporated herein by reference.

Apnlications Direct Identification of Gas [0058] It is known in the art that methane and more generally gas in the formation has long T1 and, given the gradient of the NMR logging tools, a short 12 value, because of its large diffusivity. This observation can be exploited in the detection of gas in the form of a large T1/ T2 ratio. Methods, such as the differential spectrum method or time domain analysis (TDA), have been designed and used to detect and correct gas by making use of the Ii and T2 contrasts. These differential methods have a drawback, however, because they work on the difference of two signals, which difference can be small particularly in those cases where the hydrogen index (HI) is low, the porosity is low, or invasion is deep. In other words, there are limitations to these approaches, which are important in the proper estimation of total porosity using NMR.

[0059] In accordance with the present invention, given knowledge of the rock properties, one can bypass more sophisticated gas identification techniques in certain cases, in particular in applications involving tight gas reservoirs. In a tight gas reservoir, the longest T1 of 100% water saturated rock is no more than a few hundred milliseconds. The T1 of gas, on the other hand, is typically on the order of several seconds. In other words, in a tight gas reservoir, any signal longer than about one second is in all likelihood only due to gas. In accordance with this invention, this observation leads to a novel and simple approach for (i) detecting gas and (ii) correcting the apparent porosity value for gas effects to calculate total porosity. Figs. 5 and 6 show T and T2 logs in a tight gas reservoir, which illustrate this approach. There are seven tracks in each figure, where Track 1 shows Gamma Ray, SP, caliper, PE; Track 2 shows Resistivities; Track 3 shows a VDL of T2 distributions; Track 4 shows Porosities from the T2 Log; Track 5 shows Time Domain Analysis (TDA), generated from T2 logs; Track 6 shows Porosities from T1 log; and Track 7 shows the VDL of the T1 distributions. In each figure the zone of interest is marked by two boxes.

[0060] In particular, the zone of interest in Fig. 5 contains sand with residual-gas.

Residual condition implies that most of the gas has been produced and the reservoir will produce mostly water and some gas, at least initially. The resistivity logs indicate that water saturations are high, confirming the above conclusion. Nevertheless, there is still some gas in the reservoir and this has been confirmed by gas in the mud logs. Notice that time domain analysis (IDA) in Track 5 shows some gas (dark-shaded area). It is important to notice the corresponding signal in the T1 distribution in Track 7 (even if not very strong because of residual conditions in the case). Since the rocks illustrated in the figure are tight (i.e., have low porosity, low permeability, small pore sizes), such long signal in the T1 distribution is a strong indication of gas. At the same time, one can see that the 12 distribution shown in Track 3 can hardly be interpreted to have gas without the help of TDA.

[00611 In Fig. 6, the zone of Lnterest contains no gas, although there are indications in the IDA analysis (Track 5). The gas shown from TDA is not real -it is probably an artifact of dealing with very low porosities in the echo differences. The T1 distribution does not show any gas, especially when compared to Fig. 5. Unlike the section shown in Fig. 5, there were no gas shows in the zone in Fig. 6, confirming the conclusions derived from the T1 log alone.

[0062] Based on the above observations, in accordance with the present invention peaks at shorter T times are interpreted as being due to water, peaks at longer times (around three to four seconds) are attributed to gas. Thus, in a preferred embodiment, the uncorrected gas filled porosity can simply be obtained directly using the T distribution. In particular, selecting a cutoff parameter, which in a preferred embodiment is one second, one can integrate the signal under the T1 distribution to obtain the uncorrected gas filled porosity.

Furthermore, knowing the HI parameter of gas (already established a priori, given temperature and pressure), one can make the HI correction in the form at a simple scalar multiplication to obtain a value for total porosity. HI is defined as the number of hydrogen atoms per unit volume relative to that of freshwater at standard temperature and pressure (75.degree. F., 15 psi). Thus the HI value for freshwater is taken as 1.0 and values of I-il for other fluids are referenced to this value. HI estimates for various fluids, dependent on temperature and pressure conditions, and how such corrections are applied to provide more accurate porosity estimates are known in the art and need not be considered in detail.

Importantly, the above operations do not involve taking the difference of signals, and will thus result in more robust detection since the threshold for detection is much smaller. Also, HI correction for gas in accordance with this invention is very straightforward and can be

applied in the field.

IS

Compensating For Insufficient Polarization [0063] Another application of T1 logging in accordance with the present invention is compensation for insufficient polarization. Assuming a single exponential behavior (which can be generalized for the multi-exponential case), the apparent (or measured) porosity is related to the true porosity by = ,Hl(l_e_TlTI) (8) where HI is the hydrogen index; Tw is the wait time; T is a longitudinal relaxation time of the fluid; and the subscripts a and t refer to apparent and true, respectively. Given HI = I, measuring the true porosity requires 1 -e_TITI = 1 (9) which holds if Tw 5Timax. Satisfying this requirement for all applications requires explicit knowledge of umax, which is not always available a priori. The current practice is to use a very long wait time, sufficient for all cases, typically in the order of 10 to 12 seconds. Using such a longe wait time in well logging applications forces the use of slow logging speeds, so that reasonable vertical resolution can be maintained.

[0064] Although the problem is highlighted for T logging here, the same is true for T2 logging. It is known in the art to use different approaches to compensate for insufficient polarization. One such approach is to use a short wait time, no more than about 3 seconds, and apply a polarization correction assuming a known value for the T1/ T2 ratio. The problem is that there are no criteria for determining this value -in order to make the correction properly, one has to know the T1 spectrum. So, in the prior art case, it translates into a fudge factor often determined by observing other porosities, such as crossplot, unnecessarily complicating the analysis and creating the potential for erroneous readings.

[00651 In a preferred embodiment of this invention, if the longest wait time in a T1 log is not sufficient for full polarization and there is enough resolution in the T1 axis, one can make a polarization correction given the apparent T1 spectrum and knowledge of acquisition parameters.

[0066] In particular, the signal y, from the jth echo of the ith wait time, given relaxation times Tik and T2k with associated amplitudes ak is given by: = a (i -e" /T1)eJ/T2h (10) [0067] where Te is the interecho time. Assuming a single echo acquisition from this point on (for purposes of simplification in the notation, without loss of generality), Eq. (10) can be further simplified to drop any dependence on echo number, resulting in y, = (i /Tfr)e_T2k. (11) [0068] The term in parenthesis is called the pokrization factor. The sum of the amplitudes ak yields the true porosity: (12) Considering I wait times and K T1 components; given the data vector d of dimensions (I by I), and the unknown vector x (I by 1) that contains the amplitudes of the T1 spectrum, the linear system Ax = d is solved using linear least squares techniques to obtain x. The elements of the A matrix (1 by K) is given by A(i,k) = (1-e"'T1k)e_T21. (13) [0069] Note that the relaxation times are fixed a-priori, usually chosen to be equally spaced on a logarithmic grid. Defining Timax as the longest T1 component in the actual T1 spectrum of the fluid, Twmax as the longest wait (recovery) time in the activation (pulse sequence), and the apparent porosity by Oa XA, one can see that I'(I <01, 1 max < a T1, (14) [00701 where a is equal to 3 for practical purposes, and is assumed equal to 5 in theory.

The condition in Eq. (14) defines the phenomenon known as insufficient polarization, which basically means that the apparent porosity will be less than the true porosity, if the longest wait time is not at least 3 times the longest T present in the sample. Thus, the only way to prevent insufficient polarization is to keep Twmax very long, in the order of 12 to 14 seconds.

Using such a long wait time is generally impractical, since the long cycle time will lead to very slow logging speeds. In accordance with a preferred embodiment of this invention, one approach to overcoming the insufficient polarization at reasonable logging speeds is to keep Twmax close to Tinax (i.e., Twmax limax), and then compensate for the insufficient polarization via signal processing. In a preferred embodiment, compensation is done in a separate processing step appi led post-inversion.

[0071] Fig. 7 shows a T, model illustrating insufficient polarization and the post-inversion correction method in accordance with one aspect of this invention. The porosity log and the T, spectrum are shown in the left and right tracks, respectively. The straight line in the porosity track shows the input porosity which, as described below, is 20 Pu (porosity units). The other two illustrated curves correspond to irreducible and micro porosity. These curves are output automatically by the software, but are not of interest in this modeling study.

There are 101 hundred depth points in the model, where the depths range from 4900 ft to 5000 ft. At each point, the input spectrum has two spikes, where each component has an amplitude of 10 pu. While the two components have constant amplitudes, their T1s vary with depth. One component starts at 10 ms at the top of the log, and increases with depth to 5000 ms. The other component starts at 5000 ms at the top of the log, and decreases to 10 ms at the bottom, resulting in a crossing pattern, where a different T, range is simulated by the model at each depth. Based on these values, Tlmax = 5000 ms.

[0072] Based on the model shown in Fig. 7, synthetic logs were generated using 6 wait times ranging from 1 ms to 6300 ms (1=6), and 10 echoes per wait time (J=10) with a Te of 0.5 ms. The acquisition parameters used in the simulations are identical to those used in the real logs presented in the rest of the disclosure. Since each component has a porosity of 10 pu, the apparent porosity from the T1 log, after inversion, should be 20 pu at each depth.

However, given the longest wait time of 6300 ms (Twmax = 6300 ms), the apparent porosity will be less than 20 pu due to insufficient polarization at the top and bottom of the log, since T1 approaches 5000 ms in these end points (Timax = 5000 ms). This can be best seen from the time domain data, as shown in Fig. 8. In Fig. 8, the first echo amplitude from the longest wait time is shown as a function of depth. Also shown as a straight line is the input total porosity of 20 pu. The departure from the 20 pu line, at the top and bottom of the log, is due to insufficient polarization. The curved line indicates the amplitude of the first echo from the longest wait time. The difference between the two lines is due to insufficient polarization, approaching almost 4 pu at the top and bottom of the log. It will be appreciated that prevention of insufficient polarization via acquisition would require at least a Twmax value of 15000 ms, which is impractical.

[0073] The synthetic logs were generated using a conventional linear least squares algorithm, in particular the MATLAB mathematical software program, which is well known in the art. There are 41 bins in the inversion algorithm (K=41), ranging from 1 ms to [0000 ms. Second-order regularization was applied to prevent oscillations in the distributions. As known, regularization may blur the T, distributions because while the position of the peaks and the total area under the curve are generally maintained, the regularization process tends to spread the energy into adjacent bins. This may lead to excessive porosity values when the set of time constants Tik used in the inversion contains elements that are longer than Timax.

This observation is illustrated in Fig. 9. The basis set contains TIk up to 10000 ms, and closer inspection of Fig. 9 shows that there are non-zero amplitudes in bins where Tik > 5000 ms, even though Timax = 5000. Inversion automatically compensates for the low polarization factors by boosting the amplitudes associated with these bins.

[0074] In accordance with a preferred embodiment, one approach to resolving the issues illustrated in Fig. 9 is to constrain the polarization factors in the A matrix and apply a post-inversion correction. In this approach, the polarization factors in matrix A, only for the longest wait time, are set to 1, as shown below: (l_e_TThl)=l, fori=I. (15) [0075] This results in reduced porosities, as shown in Fig. 10, because the amplitudes are not overcompensated. One can see that the apparent porosity is less near the top and the bottom of the log in Fig. 10, similar to the trend observed in Fig. 8. The straight line in the porosity track shows the input porosity (20 pu). The other line is the apparent porosity where the polarization factors are set to 1 for the longest wait time. This results in lower porosity where the T1s are long: at the top and bottom of the log.

[0076] The post-inversion correction factors Ck, given the amplitudes Xk from inversion, are defined in accordance with one embodiment of this invention by: CL -7/T where i=I. (16) The corrected porosity in a preferred embodiment is then given by: q Xk. (17) [00771 The result of the post-inversion correction applied to the logs shown in Fig. 10 are shown in Fig. 11. As before, the straight line in the porosity track shows the input porosity, which is 20 pu. The superimposed line is the post-inversion corrected apparent porosity, obtained in accordance with one aspect of the present invention. Clearly, except in a few points, the corrected apparent porosity agrees well with the input porosity. The deviations from the trend are not systematic, and are probably due to local variations in the random noise.

[0078] A direct benefit of this polarization compensation method applied in accordance with a preferred embodiment of this invention is the potential for increased logging speeds, or increased vertical resolution, since the cycle times are shorter and the data collected are closer spatially. Faster logging applications are considered below.

Faster Logging [0079] NMR logging is relatively slow compared to other open-hole logs. Thus, while some other logging tools are run at speeds exceeding 2500 ft/hr, NMR logs are rarely run faster than 1000 ftlhr, which speeds have been realized only recently with the advent of multi-frequency tools such as the MRIL -Prime, discussed above. One limiting factor on the logging speed is the time it takes for nuclei to polarize. In particular, before an NMR measurement is taken, sufficient time has to be allowed for polarization, because insufficient polarization leads to under-estimation of formation porosity. Accordingly, pre-polarization has become another requirement for fast logging. Modern tools such as the MRIL Prime have sufficient pre-polarization that can be utilized in T2 logging. Coupled with a long wait time in the order of 10 to 14 seconds, one can get full polarization in most formations, and this is what makes logging speeds of about 1000 ft/hr feasible. However, there are limitations on how much faster logging can be done without further modifications.

[0080] To illustrate the problem, consider the spins subjected to the 90 -pulse at the beginning of the measurement that are left out of the sensitive volume as the tool moves during the course of a typical CPMG pulse sequence. The loss of signal from these spins creates an artifact in the form of additional decay in the NMR signal. The number of the spins that are left out is proportional to the product of v*t, where v is the logging speed and t is measurement time. Obviously at faster logging speeds, the affect is more pronounced.

Another problem associated with faster T2 logging is the vertical resolution of the log, which decreases as the logging speed increases. This is especially true for the case of the typical long wait times used in current practice.

[0081] Neither of these problems affects T1 logging significantly, as discussed in this application. Spins being flushed out of the sensitive region is not a problem since only a few echoes per wait time are acquired in T1 logging, and the effect due to the loss of spins left out of the sensitive volume during this short time period is insignificant. The resolution is also not a major problem because T1 measurement cycles are generally shorter than their T2 counterparts.

[00821 An issue that has to be addressed in fast T1 logging applications is the mixed polarization profile where magnetization originating from pre-polarization (B0 >0, B 1=0, referred to as prepol below) is mixed with magnetization originating from the standard magnets where the antenna is located (B0> 0, > 0, referred to as standard-pol, or just standard). In order to understand this phenomenon, consider the two polarization profiles shown in Fig. 12. In this figure, the magnetization distribution in the sensitive volume at the beginning of the CPMG measurement, for a 24-inch antenna, is shown for two different wait times. The formation modeled has a T1 of 4000 ms, the logging speed is 900 ft/br, and the wait times are 2 and 12 seconds, from left to right, respectively. The snapshot of magnetization along the length of the antenna is taken immediately after the 90 pulse, just before the first echo of the CPMG.

[0083] The dark color in Fig. 12 corresponds to the pre-pol component of magnetization, while the gray-color is associated with standard-pol. Note that in the case of the long wait time (Figure 1(b)), there is no gray-shaded region, all of the magnetization is due to pre- pol and polarization is almost 100%. In the case of the shorter wait time, the magnetization originates from both pre-pol and standard. As can be seen from the difference of the magnetization levels, while magnetization due to pre-pol is almost 100%, the standard-pol portion is about 40%. Note that if the tool logged very slowly, the level of magnetization would be about 40%, there is excess magnetization (or polarization) due to tool motion. This poses a problem, because existing models used in inversion in general do not account for pre-polarization contributions.

[0084] To illustrate the problem, consider the following hypothetical case where the porosity of the formation is 100 Pu, T is 4000 ms, and the wait time is 2000 ms. Based on the profiles shown in Fig. 12, the apparent porosity (assuming that the hydrogen index is 1) in this case would be 100k(1_e 5) =39.35.

[0085] The apparent porosity would be similar if the tool were moving very slowly.

However, in the case of fast logging, due to pre-pol, there is more magnetization, and the apparent porosity is almost 55 pu. This is much more than what it is in the stationary case and must be corrected. While the above example illustrates the problem, it also suggests a solution that can be applied in accordance with another aspect of the present invention. In particular, with the addition of a speed related term, the relation between apparent and true porosities can be rewritten as -,iS* * vu -)f Pa Pi.' where po and p,. correspond to polarization factors for the stationary(v=O), and non-zero logging speed cases, respectively. Note that by definition, po=1_e_T. (18) Hence, with the addition of ps,, one can still obtain the true porosity.

[0086] Unlike p0. which only depends on the wait time and the T1, the definition of p. is more complex since it depends not only on the wait time and the Ii; but also on the specifics of the pre-polarization and standard magnetic field distributions, and the logging speed. The values for p,, can be computed in a preferred embodiment given knowledge of tool design parameters. Thus, in the hypothetical example used above, v=l.3924, for v=900 ft/hr and T=2OOO ms, T1=4000 ms.

[00871 If the polarization due to pre-pol is explicitly accounted for in the inversion, the complications due to tool motion vanish. Going back to the linear system defined for the T1 problem (See, Eq. 13): * Ax=d, A(i,k) = (1_e_IT1k)e_lT2k one can redefine the A matrix for the speed effects as below A(i,k,v)= p0(i,k) p1(i,k,v) eT'T1l, (19) where P0 IS defined as in Eq. (18), and the factors dependent on Pv can be determined for a particular tool design a-priori using modeling, or through measurements for a particular tool design and acquisition parameters. Note that the p, terms change depending on the logging speed, which changes can be accounted for using the illustration in Fig. 12. Solution of the linear system with the modified A matrix in Eq. 19 yields the correct porosity.

[0088] The advantage of explicitly accounting for pre-polarization magnetization is that one does not have to seek 100% polarization, as shown in Fig. 12(b). Since the sources of polarization are accounted for explicitly, one can resort to shorter wait times (for example a maximum of 3 seconds instead of 12), and thus increase logging speed significantly. Use of such wait times shorten the cycle time for the T1 measurement, which helps retain good vertical resolution at high logging speeds.

Estimation of Diffusivity and Gas-to-Oil Ratios [0089] Given two 12 logs, acquired with different Te's to (inter-echo times), one can estimate of diffusivity of light hydrocarbons using the relationship.

1 1 + D011.(G.y.TE)2 T,,,. 1, 12 as explained in the examples below. In particular, a T1 log can be considered as a T2 log acquired with an infinitely short Te (no diffusion effects). Hence, the combination of a T1 log and a T2 log can be used in accordance with the present invention to estimate directly the diffusivity of the hydrocarbons phase. Furthermore, in another aspect of the invention once the diffusivity is estimated, one can estimate the gas to oil ratio (GOR), combining the T1 value with the D0 value. An illustration is provided in the following section of the disclosure on Practical Examples.

Practical Examples

T1 versus T2 Logging [00901 The T, inversion used in commercial applications, wireline or LWD, uses generally all the data available, which is typically 10 echoes per CPMG trainlet. Single echo inversion is either limited to qualitative processes, such as the Reconnaissance Mode (described in paper SPE 62981 presented at the 2000 SPE conference), or to those very high SNR applications, such as the MRILabTM tool where SNR is typically in the order of 200 or better. All the results presented herein, as well as those presented in previous publications (paper DDD, presented at the 43rd Annual Logging Symposium, Osio, Japan; paper SPE 77395 presented at the 2002 SPE Annual Technical Conference and Exhibition) use 10 echoes per CPMG. The inter-echo time in all these cases is either 0.5 1ms or 0.6 ms. All papers identified in this paragraph are incorporated herein by reference for background.

[0091] The porosity logs displayed in Fig. 13 verify the quality of petrophysical answers from T1 measurements and demonstrate their robustness and repeatability. Data from eight different runs, in the same well, are presented in this figure including: a wireline MRIL Prime T2 log; an MRILWDTM T1 log acquired while drilling the well; and six MRILWDTM T logs acquired in simulated-drilling mode, where the tool was rotated while going up or down in the well.

[0092] Track 1 in Fig. 13 shows GR, SP and Caliper logs. Track 2 shows the total porosity from all runs (xxPTOTxx), whereas Tracks 3 and 4 show the irreducible and micro porosity curves, respectively (xxT2PIRMXxx, xxPMlCxx). The prefix T1 or T2 is used to distinguish Ti-based curves from T2-based curves. The main reservoir is between 550 to 610 feet, and is logged in each pass. Some of the passes do not cover the shallower portions of the well, and some logs terminate above the washout at 620 feet.

[0093] The standard deviation of the total porosity from the T1 logs is 0.94 pu, and the standard deviation of the irreducible fluid porosity from the same is 1.06 pu. The excellent repeatability among the T1 logs, as well as the close agreement with the benchmark MRIL Prime T2 log should leave no doubt as to the validity and quality of the T1 measurement results. It should be noted also that the comparisons are made using the more challenging LWD logs, not wireline logs.

Case Study [0094] As part of a pilot study, a series of wireline T1 logs were run to assess the operational and petrophysical feasibility of this novel application. This study was assisted by access to a major core study that includes NMR T1 data.

[00951 The logs presented here in Fig. 14 are from the very first well in the program.

This well was drilled with an 8"" bit using salt-saturated mud, which resulted in a fairly smooth well bore with some washed-out zones in the non reservoir carbonate section, and only marginal enlargement and mud-cake buildup in the deeper sandstone reservoirs.

[0096] A full suite of openhole logs was acquired in tool-push mode including caliper, GR, SP, resistivity, density, neutron, acoustic, formation tester (pressures only), and borehole images. A standard MR1L PrimeTM tool was run to acquire a dual wait time T2 log, followed by a T1 log over the same interval. Bore hole enlargement is indicated with dark shading, mud-cake build-up by the gray shading. The second track shows six array-resistivity curves at different investigation depths. Breakdown of the MRIL T, porosity in micro (dark gray), irreducible (dark, striped gray) and moveable (light gray) porosity is presented in the third track. Density porosity is shown as reference. The gray-shaded T1 spectra are on a log scale from 0.5 ms to 5 s. Track 5 presents the breakdown of MR IL T2 porosity, using the same shadings as before with the T2 spectra (gray shading) in the adjacent track.

[0097] NMR core data from an offset well were available from an earlier, large-scale laboratory NMR core study. The main objective of this core study was to improve NMR log interpretation by establishing field specific parameters for e.g. spectral-and cut-off BVI, permeability, etc. T1 measurements were already included in the program, in anticipation of future T1 based applications. The benefits of this core study were immediately realized during the interpretation of the well, particularly in the determination of hydrocarbon types.

[0098] The basic log panel in Fig 14 shows the openhole logs, where the reservoir sands start at a depth of about xx 100 ft. The main zone of interest starts with 90 feet of clean and homogeneous sand, intersected 65 ft from the top by a -most likely sealing-tight streak.

There are several poorer quality wet sands below the main pay zone, followed by another small pay zone located between xx335 to xx350 ft. The pay zones can easily be identified from the difference (under call) between the apparent density and NMR (either T1 or T2) porosity logs. Note that the same logs agree very well in the wet zones. Also, note the resistivity logs show no signs of significant invasion.

[0099] The T1 log exhibits bi-modal distributions in the pay zones, whereas the T2 logs are generally uni-modal. Also, the comparison of uncorrected T1 and T2 porosities in the pay zone reveals that the porosity from the T2 log is higher than T1 porosity. This can be explained by the fact that the T2 log has a longer wait time. The longest recovery time in the T1 log is 6.3 seconds, while the wait time in the T2 log is 12.0 seconds. Obviously, more polarization occurs in the case of the longer wait time when the reservoir fluid has a long T,.

Using the rule-of thumb of three times the T1 for full polarization, one can estimate that the T1 of the hydrocarbon is at least 2.1 seconds.

[00100] While aT1 value longer than 2.1 seconds would be consistent with the expected hydrocarbon type of dry gas, apparent T1 porosity (uncorrected for hydrogen index or incomplete polarization), and the T1 distributions indicate a different fluid, as discussed next.

[00101] The T1 distributions show a very clear bi-modai distribution in the pay zones, where the long T1 peak is centered about 3 seconds. Since such a peak is not observed in the wet zones, this peak can be easily attributed to the presence of gas. However, using a T1 cutoff of 1 second, and applying gas corrections to the log (i.e., accounting for partial polarization and factoring in the Hydrogen Index of 0.48 for natural gas under the pressure and temperature conditions encountered in this well), results in a porosity that is significantly higher than what has been observed field wise. Simply put, the signal associated with the long T1s in the pay zone can not be attributed to dry gas only. There are two possibilities: (1) some of the signal is due to invasion water, (2) the hydrocarbon is not just methane, but contains heavier components (which effectively increases the HE, reduces the HI-correction and hence the apparent porosity).

[00102] NMR spectra from two core plugs that were included in the before mentioned core study, are shown in Fig. 15. These plugs were taken from the core from an offset well. They were first cleaned, then saturated with water (Sw = 100%) and subjected to a series of conventional petrophysical measurements, as well as NMR T1 and 12. There is clearly no support for long T1s in the fully saturated core data: the slowest 1, components relax at a rate much faster than 1 second. This rules out possibility #1 to reduce the apparent porosity, only leaving the possibility that we're not dealing with just dry gas, but also with heavier components.

[001031 Ruling out the presence of dry gas helps explain the trends observed in the 12 log.

Analysis of dual wait time (T2) data shows differential amplitudes in the pay sands centered on T2 values of approximately 200- 300 ms, instead of the 35 ms that would be expected in the case of gas. This confirms the interpretation that the reservoir fluid is not gas, but a very light hydrocarbon instead. The T1 and T2 of the hydrocarbon phase are 3.5 and 0.3 s, respectively. The large T1/T2 ratio indicates a large D0 value, (see the discussion on estimation of D0 and GOR). Due to the large diffusivity of the hydrocarbon phase, the water and hydrocarbon signals overlap in the 12 domain, but are well separated in T1 domain. Lack of diffusion effects in the T log actually result in easier identification of the pay zones for this reservoir.

[00104] Fig. 16 displays the pressure data in the well, confimiing the NMR-based interpretation: the pressure gradient in the pay sand (Group 1; dots in the top left corner) is established at 0.23 psi / ft, which corresponds to a density of 0.53 g / cm3, indicative of a light hydrocarbon indeed. When entering the wet zones, there is a distinct change in the gradient.

[00105] The interval xx 190-xx270 ft is of poor to non reservoir quality with some inter-bedded cleaner layers. The sand body starting at xx270 ft is again of reservoir quality; the NMR logs, however, match the density-derived porosity values much closer, indicating a different (mixture of) reservoir fluid. The much weaker differential signal appears again at T2 200 ms, suggesting that the reservoir is only partially hydrocarbon filled at these depths.

Based on the similarities in T, and T2 characteristics, the hydrocarbon is thought to be the same tight oil as encountered higher up in the well. Log analysis, using conventional data in combination with NMR porosity, confirms this interpretation and shows an abundance of free water with some pockets of gas.

[00106] The MRIL T1 and T2 data corroborate this interpretation. Sands of reservoir quality exhibit T2 peaks well in excess of 50 ms; most center on 200 ms. Wherever these 12 peaks occur in combination with the distinct T peaks of some 2 s, we are dealing with light (high GOR) oil, whereas the other zones are wet.

Estimation of GOR [00107] Under the assumptions that the reservoir is water-wet and that for the hydrocarbon, iI.bulk T2.bulk, equations 1 and 2 can be combined and re-written as: I = 1 + D011.(G.y.TE)2 (20) T21 1,, 12 [001081 Stacking the echo-trains (trainlets) over the entire pay-zone, followed by inversion, yields TJh 3.5 s and T2,hc 300 ms. These values, combined with the tool parameters in eq. (8) indicate that the viscosity of the light oil in this reservoir is on the order of 13 x 10-5 cm2 / s. Using the correlations by Lo et al. (paper SPE 63217 presented at the 2000 SPE Annual Technical Conference and Exhibition, Dallas, Texas) of T1 relaxation times with diffusivity D0 and Gas-to-Oil Ratios (GOR), estimates the GOR in this reservoir at 2500. This solution is indicated by the solid dot in Fig 17. Analyzing the sensitivity of this result with respect to the parameters derived from the logs (mainly TI,h and T2,h) shows that GOR fits in the range 1000 -4000, indicated by the dark-shaded area in Fig 17. For details on computing the GOR estimates the interested reader is directed to Lo et al., paper SPE 63217 presented at the 2000 SPE Annual Technical Conference and Exhibition, Dallas, Texas, 1-4 October, 2000.

[00109] Based on the above, it is apparent that T1 information alone already adds significant value petrophysically and helps delineating the reservoir fluids and establishing fluid contacts. It was further demonstrated that the measurements have excellent robustness and repeatability, similar to wireline 12 logs run under comparable conditions. Given field knowledge, Ii togging can be used very easily to recognize hydrocarbon bearing zones, and simple cutoff techniques can be used to correct for hydrocarbon effects, since the hydrocarbon phase is easily and directly identified. It was demonstrated that when combined with T2 log(s), T1 logs can be utilized to determine diffusivity, GOR and viscosity at reservoir conditions of (light) hydrocarbons. Importantly, T1 saturation recovery logs used in accordance with the present invention are more compact than CPMG 12 logs and can run faster, since their total measurement time is generally shorter compared to T2 logging. The combinations discussed above are believed to be a significant contribution to the art of NMR logging with wide ranging applications involving virtually all NMR tools, and a broad range of practical applications, including both wireline and LWD/MWD.

[00110] While the invention has been described with reference to the preferred embodiments, it will be appreciated by those of ordinary skill in the art that modifications can be made to the structure and form of the invention without departing from its spirit and scope which is defined in the following claims.

Nomenclature D0 =coefficient for molecular self diffusion, 10-5 cm2/ s

G = magnetic field gradient, gauss / cm

GOR = Gas Oil Ratio, v / v OR = Gamma Ray (log) HI = Hydrogen Index NE = number of echoes in CPMG sequence PE = Photo Electric (log) S = surface area of pore space, cm2 SP = Spontaneous Potential (log) Te = echo spacing, ms T = longitudinal relaxation time (distribution), ms 12 = transverse relaxation time (distribution), ms Tw = recovery time for magnetization to build-up V = volume of pore space, cm3 = porosity, p.u.

<p≥ gyro magnetic ratio, 27r*4258 Hz / gauss for protons p = surface.relaxivity, cm / ms References of potential interest 1. Brown, R. J. S., and Gamson, B. W., 1959: "Nuclear Magnetism Logging," Society of Petroleum Engineers, presented at the 34th Annual Fall Meeting, Dallas, Tx, Oct. 4-7.

2. Timur, A., 1968: "Effective Porosity and Permeability of Sandstones Investigated Through Nuclear Magnetic Resonance Principles," Paper K, SPWLA, presented at the 9th Annual Logging Symposium, New Orleans, LA.

3. Kenyon W. E., Howard, J. J., Sezginer, A., Straley, C., and Matteson, A., l989,:"Pore-size Distribution and NMR in Microporous Cherty Sandstones," Paper LL, SPWLA, presented at the 30th Annual Logging Symposium, Denver, CO.

4. Prammer, M. 0., Akkurt, R., Cherry, R., and Menger, S., 2002: "A New Direction in Wjreljne and LWD NMR," paper DDD, presented at the 43rd Annual Logging Symposium, Osio, Japan.

5. Prammer, M.G., Drack, E., Goodman, 0., Masak, P., Menger, S., Morys, M., Zannoni, S., Suddarth, B., and Dudley, J., 2000: "The Magnetic Resonance While Drilling Tool: Theory and Operation" paper SPE 62981 presented at the 2000 SPE Annual Technical Conference and Exhibition, Dallas, Texas, 1-4 October.

6. Prammer, M. 0., Bouton, J., and Masak, P., 2001: "The Downhole Fluid Analyzer," paper N, SPWLA, presented at the 42nd Annual Logging Symposium, Houston, TX.

7. Morley, J., Heidler, R., Horkowitz, J., Luong, B., Woodbum, C., Poitzsch, M., Borbas, T., and Wendt, B., 2002: "Field Testing of a New Nuclear Magnetic Resonance Logging While Drilling Tool", paper SPE 77477 presented at the 2002 SPE Annual Technical Conference and Exhibition, San Antonio, Texas, 29 September -2 October.

8. Appel, M., Radcliffe, N. J., Aadireddy, P., Bonnie, R. J. M., and Akkurt, R, 2002: "Nuclear Magnetic Resonance While Drilling in the U.K. Southern North Sea", paper SPE 77395 presented at the 2002 SPE Annual Technical Conference and Exhibition, San Antonio, Texas, 29 September -2 October, 2002.

9. Lo, S.-W., Hirasaki, G.J., House, W.V. and Kobayashi, R., 2000: "Correlations of NMR Relaxation Times with Viscosity, Diffusivity and Gas / Oil Ratio of Methane / Hydrocarbon Mixtures" paper SPE 63217 presented at the 2000 SPE Annual Technical Conference and Exhibition, Dallas, Texas, 1-4 October.

Claims

CLAIMS

I. An NMR method for analyzing geologic formations, comprising the steps of: providing a distribution of NMR T1 relaxation times corresponding to an NMR log of the geologic formation obtained with a maximum wait time such that T,<aTim, where Timax is the longest component in the Ti spectrum of the NMR signal and a is a constant less than 5; and computing a correction for insufficient polarization of the NMR log based solely on the provided distribution of NMR T1 relaxation times and acquisition parameters of the NMR log.

2. The method of claim I further comprising the step of obtaining an NMR log of the geologic formation.

3. The method of claim I wherein the constant a is between about 1 and 3.

4. The method of claim I wherein for I wait times in the NMR log measurements, and K T1 components, the uncorrected T1 spectrum is computed using a solution to the linear system Ax=d, in which d(/,) is the data vector; A(jK) is a constrained matrix the elements of which are defined as A(i,k) = (1 _e_ThI)e_T. I T and (1 -e_T = 1, for i=I and the vector of uncorrected Ti spectrum components is X(IXI).

5. The method of claim 4 wherein the post-inversion correction factors Ck, given the amplitudes xk from inversion, are defined by: c = , where 1=1.

I -e'' 6. The method of claim I further comprising the step of computing a correction for insufficient polarization of a T2 NMR log based solely on the provided distribution of NMR I relaxation times and acquisition parameters of the NMR log.

7. An NMR system for analyzing geologic formations, comprising: means for providing a distribution of NMR T1 relaxation times corresponding to an NMR log of the geologic formation obtained with a maximum wait time Twmax such that Twmax<alimax, where Tlmax is the longest component in the T spectrum of the NMR signal and a is a constant less than 5; and means for computing a correction for insufficient polarization of the NMR log based solely on the provided distribution of NMR T1 relaxation times and acquisition parameters of the NMR log.