RELIABILITY INDICATION OF QUANTITATIVE CEMENT EVALUATION WITH LWD SONIC

Size: px

Start display at page:

Download "RELIABILITY INDICATION OF QUANTITATIVE CEMENT EVALUATION WITH LWD SONIC"

Helen Kennedy
5 years ago
Views:

1 ELIABILITY INDICATION OF QUANTITATIVE CEMENT EVALUATION WITH LWD SONIC Shin ichi Watanabe 1, Wataru Izuhara 1, Vivian Pistre 2, and Hiroaki Yamamoto 1 1. Schlumberger K.K. 2. Schlumberger This paper was selected for presentation by a JFES program committee following review of an abstract submitted by the author(s). ABSTACT There has been a growing interest in measuring the cement bond quality quantitatively with LWD sonic tools. ecent studies have shown that quantitative cement evaluation with an LWD tool is feasible up to a certain bond index depending on conditions. In this paper we present several methods developed to identify the reliability of bond index computation with the current LWD sonic tools. The methods utilize information on the amplitude, transit time, and attenuation of the sonic waveforms acquired along a receiver array, as well as extensive modeling using the finite-difference method. The theoretical studies have been validated with real data recorded during some field test experiments that are also presented. The answer product based on these methods provides bond index values within the measurement range in conjunction with various quality control plots to provide confidence to the user on the validity of the answer. INTODUCTION Over the past 4 years, there has been an increased focus on well integrity, from operators as well as regulators. One key element is the cement-to-casing bonding condition, as the cement supports the casing and provides the hydraulic seal. Information on the integrity of the cement job is currently provided by wireline services such as ultrasonic azimuthal imaging or a sonic cement bond log (CBL). There is, however, an incentive to provide cement integrity information with logging-while-drilling (LWD) tools which can help save rig time (Longo et al., 2012); the earliest possible identification of poor cement bonding is important to be able to consider potential remedial work and to reduce any risk before drilling the next section. They also allow monitoring of the possible changes of the casing-to-cement bonding with time as, while running in and out of hole, LWD tools can survey the same cased hole section on multiple passes (Blyth et al., 2013). In recent years, the service offered by the LWD acoustic tool has been the top of cement (TOC) indication, which only provides a qualitative analysis of the cement presence. There is a technical challenge for quantitative cement evaluation with LWD acoustic tools due to the rigid steel collar of LWD tools. The desired casing signals mix with unwanted signals propagating through the collar at a similar speed. The evaluation of bond index is possible when the casing signal is large enough in comparison to the collar signal (Kinoshita et al., 2013). Since the casing signal is a function of various factors, such as the acoustic impedance of the cement and the casing size, then the bond index computed with an LWD tool and its associated limitations also depend on these factors. Thus, it is important to know the conditions that can provide a reliable bond index with an LWD and the limitations of this approach. To incorporate this cement evaluation algorithm into an answer product, various quality control (QC) plots are essential to support the validity of the computed bond index values. The purpose of this paper is to present several methods developed to identify the reliability of bond index computation with the current LWD sonic tools. MODELING AND LIMITATIONS Earlier studies have suggested that, aside from some extreme cases such as very thin casing less than 0.3 in, quantitative cement evaluation with an LWD tool is feasible up to a bond index (BI) of near 0.6 with conventional class G cement and it is validated by the case studies using finite-difference modeling (FDM) and real data (Kinoshita et al., 2013). Figure 1 shows an example of the validation and comparison of the modeling work with real data from a test well. In Figure 1(a), correlations between the wireline BI and the normalized casing amplitude in db from a 4.75-in. LWD tool over a certain depth interval are plotted. Cement type used is conventional class G (6.8 Mayl) and casing inner and outer diameters (ID and OD) are 6.09 in. and 7.0 in., respectively. Assuming that the same LWD tool is placed in the same well conditions in terms of acoustic impedance of cement, casing size and thickness, and fluid parameters, FDM -1-

2 analysis results are plotted for seven bonding conditions from 0 (free pipe) to 1 (full bonding) in Figure 1(b). The BI trend from the real LWD data agrees with the modeling results. It seems reasonable to conclude that quantitative evaluation of cement conditions is possible up to a BI of 0.5 or 0.6 in this environment. The measureable limit of BI can vary depending on conditions, such as cement type and casing thickness (Kinoshita et al., 2013). The later part of this paper explains how to identify the limit from QC plots based on acquired data without referring to the modeling results. resultant log and associated QC plots (steps 6 and 7). In step 3, amplitude normalization, one should choose method that either calibrates a BI at a certain depth point (e.g., a free-pipe point with BI=0) or let the program perform a computation based on estimated values from the casing size and fluid properties. In the former case, one should set a BI value and a depth (step 4.1). In the latter case, the free-pipe amplitude is estimated based on a chart embedded in the software (step 4.2). Then, one should set a time window at a specific depth in step 5 as shown in Figure 3. The time window is tuned at each depth based on zero-crossing point of the waveforms and peak-to-peak amplitude is measured for an entire depth range of processing. At the end, the software produces a composite log showing various QC plots in addition to a main BI log in step 7. Figure 1: Validation of the FDM model and comparison with real data. (a) Correlations between the wireline BI and the normalized casing amplitude in db from a 4.75-in. LWD tool. (b) FDM modeling results (cross marks) at different bonding conditions with a tool model of the same LWD tool. A dashed line indicates the trend from modeling conditions without the LWD tool body in the well. COMPUTATIONAL WOKFLOW We have implemented an answer product for quantitative cement evaluation on the Techlog * wellbore software platform. Figure 2 depicts a processing flowchart for the software, which basically comprises three processing tasks: data loading (step 1), parameter setting (steps 2, 3, 4, and 5), and computation followed by generation of a * Mark of Schlumberger Figure 2: Processing flowchart for the quantitative cement evaluation answer product. -2-

impedance are 29 ppf, 6.184 in., and 6.8 Mayl, respectively. The borehole fluid is oil-based mud.

The five tracks, in addition to the depth track, in Figure 4 are explained below.

3 impedance are 29 ppf, in., and 6.8 Mayl, respectively. The borehole fluid is oil-based mud. The modeling study showed that, in this environment, cement evaluation is considered to be feasible up to a BI of near 0.6 (Kinoshita et al., 2013). The five tracks, in addition to the depth track, in Figure 4 are explained below. Figure 3: Time window setting for amplitude detection and amplitude measurements in the time window for a receiver array. QC PLOTS WITH EAL LWD DATA Figure 4 shows an example BI log and associated QC plots from a well with a 7-in. casing in the North Sea. A 4.75-in. LWD tool, which has 12 receiver stations, was run to acquire monopole acoustic data. The weight of casing, the nominal casing ID, and the cement acoustic Track 1 shows a BI log as the main output. The BI is calculated based on peak-to-peak amplitude obtained in a waveform acquired at the nearest receiver from the transmitter by default. In this case, the waveform was acquired at the nearest receiver, whose transmitter-to-receiver (T) spacing was 7 ft. As the BI computation uses modified algorithms from those used in the traditional 3-ft CBL, amplitude normalization is essentially required. In this example, we regarded the bonding condition at XX160 ft as free-pipe (i.e., BI = 0). The circle in track 1 indicates BI = 0 at that depth. Track 2 presents three kinds of logs. The first one is an amplitude-based BI from each of a given receiver set Figure 4: A composite log from the quantitative cement evaluation answer product. A circle in track 1 indicates that the amplitude normalization depth was set to XX160 ft with BI = 0 in the software. -3-

$(BI_QC_x where x = {1, 4, 7}). A BI log is reliable when all amplitude-based bond indices are matched.$ The second one is an attenuation-based BI (BI_ATT). Over the receiver array, an amplitude attenuation trend is used for the BI computation.

The second one is an attenuation-based BI (BI_ATT). Over the receiver array, an amplitude attenuation trend is used for the BI computation.

4 (BI_QC_x where x = {1, 4, 7}). A BI log is reliable when all amplitude-based bond indices are matched. When they are not, the BI is unreliable due to low waveform quality or to LWD measurement limit with the collar arrival as seen in XX095-XX110 ft, XX175-XX190 ft, and XX220-XX250 ft. The second one is an attenuation-based BI (BI_ATT). Over the receiver array, an amplitude attenuation trend is used for the BI computation. The attenuation-based BI is also referred to in order to identify BI limitation. The log presents remarkably large discrepancy from amplitude-based BI when the BI is over the limit as seen in XX095-XX110 ft, XX175-XX190 ft, and XX220-XX250 ft. Another role of the attenuation-based BI is to confirm whether the aforementioned normalization works correctly or not. Figure 5 shows two processing results with amplitude normalization at different depths. Figure 5(a) illustrate an example output due to improper normalization parameters. As indicated in the figure, BI_QC_x is biased from BI_ATT in lower bonding zones. When the normalization parameters are properly set, BI_ATT fits in well with BI_QC_x in the same zones, as seen in Figure 5(b). The last one is a BI noise level (BI_NOISE). To make BI_NOISE, the background-noise amplitude is fed into the BI computation. Any BI values exceeding BI_NOISE would be affected by noise and be invalid. Track 3 in Figure 4 indicates minimum and maximum positions of a time window (TW_MIN and TW_MAX) in which peak detection is made. In addition, transit times for both minimum and maximum peaks (TT_MIN and TT_MAX) in the time window are plotted. These logs are useful to understand the measurement limit and effect of background-noise. When casing amplitude is large enough, time window and transit time curves are stable and vary slowly along the depth direction, whereas when casing amplitude is comparable or smaller than the collar amplitude, detection tends to pick collar propagations that arrive earlier than casing signals and the transit time is shifted. Fluctuations of transit time indicate unreliable measurements of casing amplitude smaller than noise amplitude. Figure 6 shows examples for shift and fluctuation of transit time. Figure 6: (a) Shifts in time due to collar arrival. (b) Fluctuations in time due to background-noise. Figure 5: (a) esults with incorrect amplitude normalization. BI_QC_x is biased from BI_ATT at the low bonding zone. Amplitude normalization is made at XX140 ft with free-pipe (BI=0). (b) esults with correct amplitude normalization. BI_QC_x is matched with BI_ATT at the low bonding zone. Amplitude normalization is made at XX160 ft with free-pipe (BI=0). Track 4 in Figure 4 presents two QC logs. One is an amplitude level at zero transmitter-receiver (T-) spacing (SA0). The other is a threshold for the measurement limitation (THES_FCF). The former acts as a synthetic casing amplitude at zero T- spacing by taking into account both an observed attenuation rate over the receiver array and the T- span. This means that not only amplitude-based but also attenuation-based approaches are jointly utilized. The purpose of SA0 log is to identify collar arrival limit by comparing with the threshold. SA0-4-

5 is relatively high when the casing-to-cement bonding is within the LWD measurement limit, whereas when it exceeds the LWD measurement limit, the SA0 decreases. In track 5 in Figure 4, the input monopole waveforms acquired at the nearest receiver from the transmitter are mapped to logarithmic gray-scale color and are presented as a variable density log (VDL) image. The VDL display is useful to assess the quality of casing-to-cement bonding and cement-to-formation bonding by visually identifying casing arrivals as well as formation arrivals. On top of the VDL, both TT_MIN and TT_MAX channels (as presented in track 3) are overlaid as an aid for QC of the casing signal detection. Comprehensive analyses of these QC plots provide a solid reliability indicator for the main BI log. CONCLUSION We have developed an answer product to evaluate the cement bond quality quantitatively up to a certain threshold utilizing the current generation of LWD sonic tools. To clearly indicate the measurement limitations and the data quality, the software developed produces a composite log that comprises a set of QC plots together with a main bond index log. By analyzing these QC plots in an integrated manner, we can identify the reliability of the computed bond index log and the quality of the bond index computation. EFEENCES Blyth, M., Hupp, D., Whyte, I., and Kinoshita, T., 2013, LWD Sonic Cement Logging: Benefits, Applicability, and Novel Uses for Assessing Well Integrity: SPE MS, presented at the SPE/IADC Drilling Conference, 5-7 March, Amsterdam, The Netherlands. Kinoshita, T., Izuhara, W., Valero, H.P., and Blyth, M., 2013, Feasibility and Challenge of Quantitative Cement Evaluation With LWD Sonic: SPE MS, presented at the SPE Annual Technical Conference and Exhibition, 30 September-2 October, New Orleans, Louisiana, USA. Longo, J., Hupp, D., Blyth, M., and Alford, J., 2012, Logging-While-Drilling Cement Evaluation: A Case Study from the North Slope, Alaska: SPE MS, presented at the SPE Annual Technical Conference and Exhibition, 8-10 October, San Antonio, Texas, USA. ABOUT THE AUTHOS Shin ichi Watanabe is a software engineer working for the sonic program at Schlumberger. Since he joined in Schlumberger in 1984, he has been involved in a broader field of product developments for the oil & gas industry and semiconductor test business in the United States and Japan. He received BSc and MEng degrees, respectively, from Kobe University and Osaka University. He received a Ph.D. (2005) in statistical science from the Graduate University for Advanced Studies (the Institute of Statistical Mathematics, Japan). He is a member of SEG. Wataru Izuhara is a physicist in the sonic department at Schlumberger K. K., Japan. He has 7 years of experience with Schlumberger; he joined Schlumberger in 2007 and worked on development of the latest LWD sonic tools and the early stage of a wireline slim sonic tool until March He is currently working on a sonic well integrity project. He obtained Bachelor and Master degrees from the Department of Civil and Earth esources Engineering, Kyoto University, Japan. Vivian Pistre is a Principal Scientist having worked for Schlumberger since 1981 in a variety of positions for field operations, data interpretation, and engineering. He is currently the Domain Head for Geophysics, Acoustics & Geomechanics for the Drilling & Measurements segment. He holds an Engineer degree and a Master s degree in Computer Science from ENSEEIHT in Toulouse, France, followed by a specialization in artificial intelligence. He is a member of SPWLA, SPE, SEG, and EAGE. Hiroaki Yamamoto is a Senior Engineer at Schlumberger K.K. Since he joined Schlumberger in 1985, he has participated in an acoustic tool development project and conducted acoustic data analysis in Japan, the United States, and France. He received BS and MS degrees in engineering geophysics from Kyoto University, Japan. He is a member of SPWLA and SEG. -5-

Isolation Scanner. Advanced evaluation of wellbore integrity

Isolation Scanner. Advanced evaluation of wellbore integrity Isolation Scanner Advanced evaluation of wellbore integrity Isolation Scanner* cement evaluation service integrates the conventional pulse-echo technique with flexural wave propagation to fully characterize