METHOD FOR IDENTIFYING GEOLOGICAL AND DRILLING PATTERNS IN A UNIDIMENSIONAL SPACE VIA PERIODIC ORTHOGONAL FUNCTIONS APPLIED TO DRILLING PARAMETER DATA

Abstract

The present invention relates to a technique for identifying geological and drilling patterns by analyzing data from drilling parameters, using periodic orthogonal functions applied to such data. The use of averages to convert time data to depth data acts as a low-pass filter, attenuating the textural information from the torque data. Therefore, with the use of simple statistical filters combined with the use of multidimensional data visualization methods, it is possible to discretize drilling and geological patterns in the depth data acquired, aiding in the characterization of the top of the pre-salt carbonate reservoir, minimizing the geological and engineering risk in these operations. It is noted that in some situations of well kill, it is common not to have perceptible contrasts when analyzing the drilling parameters in the transition between the basal anhydrite and the carbonates, with the use of techniques such as PCA (Principal Components Analysis) being necessary in order to increase the method's power of discrimination. The technique disclosed minimizes the exploratory risks in pre-salt kill situations, as it allows precise characterization of the top of the carbonate reservoir.

Claims

1-5. (canceled)

6. A method for identifying geological and drilling patterns in a unidimensional space via periodic orthogonal functions applied to drilling parameter data, the method comprising: defining, as entry variables, MD, MDBIT, TORQUE, RPM, WOB, FLWIN, SPP, and ROP; filtering time records and only keeping events with the status of drilling in a database; eliminating outliers of the time data by using specific algorithms; choosing a function for calculating fx(t) from among Andrews (1972), Khattree & Naik (2002), Embrecht & Hezberg (1991), Wegman & Shen (1993); standardizing the time acquired drilling parameter data and used to calculate fx(t); defining the window for using the filters to capture textural information from the torque data; defining the order and the variables that will be used for entry in the model for calculating fx(t); defining the step for establishing time fx(t) to depth of 1 m or 0.5 m; defining the method for establishing fx(t) between average or median; and using aesthetic filters and defining numerical parameters to show the time and depth ranges of fx(t).

7. The method of claim 6, comprising calculating the PCAs.

8. The method of claim 6, comprising removing outliers from the depth data;

9. The method of claim 6, wherein the matrix used in calculating fx(t) is defined by X=[ROP, WOB, MSE, TORQUE_IQR, TORQUE, TORQUE_STD, TORQU_VAR].

10. The method of claim 6, wherein at least one of MSE, TGAS and MDLAG is used as an entry variable.

11. The method of claim 10, wherein the variable MDLAG is used to synchronize the TGAS data with the drilling parameters.

12. The method of claim 6, wherein the textural information from the torque data comprises interquartile distance, standard deviation, and variance.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The present invention will be described in greater detail below, referencing the attached figures, which clearly and without limitation on the inventive scope, present the results obtained. The drawings show:

[0022] FIG. 1 shows application of the Andrews technique (1972) to the data set Iris Flower Data Set where all possible projections of fx(t), with t varying between −π<t<π, are shown simultaneously;

[0023] FIG. 2 shows a flow chart with the routine for processing drilling parameter data implemented for generation of time and depth ranges, which will be described in detail below;

[0024] FIG. 3 shows the result of using the method in Well 1, shown as an interpolated image (track 1), geological tops/lithological descriptions (track 2) and interpreted lithology taken from the Exata database system (Petrobras) (track 3). The technique allowed the four lithological groupings (halite, anhydrite, intertrap carbonate, and pre-salt carbonate) found in Well 1 to be individualized;

[0025] FIG. 4 shows the result of using the method in Well 2, shown as an interpolated image (track 1), geological tops/lithological descriptions (track 2) and interpreted lithology taken from the Exata database system (Petrobras) (track 3). The technique allowed the three lithological groupings (halite, anhydrite, and pre-salt carbonate) found in Well 2 to be individualized;

[0026] FIG. 5 shows the result of using the method in Well 3, shown as an interpolated image (track 1), geological tops/lithological descriptions (track 2) and interpreted lithology taken from the Exata System database (Petrobras) (track 3). In the case of Well 3, use of the PCA technique was necessary, which increased the discriminating power of the method, allowing the three [sic] lithological groups (halite, anhydrite, intertrap carbonate and pre-salt carbonate) found in Well 3 to be individualized.

DETAILED DESCRIPTION OF THE INVENTION

[0027] First, it is noted that the following description will begin with preferred embodiments of the invention. As will be apparent to anyone skilled in the art, however, the invention is not limited to those particular embodiments.

[0028] The processing routine implemented and applied to time acquired drilling parameter data, is briefly described below: 1) Define the entry variables: MD (Well Depth), MDBIT (Drill Bit Depth), TORQUE, RPM (rotation of drilling string), WOB (Weight on Bit), FLWIN (Inflow), SPP (Injection Pressure), ROP (Rate of Penetration), TGAS (Total Gas, optional) and MDLAG (Gas Return Depth, optional). The MDLAG variable is necessary to synchronize the TGAS data with the time-drilling parameter data; 2) Filter the time records in order to only keep events with drilling status in the database; 3) Eliminate the time data outliers by using specific algorithms, minimizing the influence of these atypical values on the following calculations; 4) Choice of function for calculating fx(t) (options: Andrews (1972), Khattree & Naik (2002), Embrecht & Hezberg (1991), Wegman & Shen (1993)); 5) Standardize (for average zero and standard deviation of one) the time acquired drilling parameter data and used in the calculations of fx(t); 6) Define the window for applying the filters to capture the textural information from the torque data (interquartile distance, standard deviation, and variance); 7) Define the order and the variables that will be used for entry in the model to calculate fx(t); 8) Calculate the PCA (optional); 9) Define the step for establishing fx(t) from time to depth (options: step from 1 or 0.5 m); 10) Define the method for establishing fx(t) (options: average or median); 11) Remove outliers from the depth data (optional); 12) Apply aesthetic filters and define the numerical parameters for exhibiting the time and depth ranges of fx(t). An important point in relation to the processing routine is the fact that the TGAS variable was not used as an entry in the model, although it is available for such use. Use of TGAS was not chosen because in well kill operations an independent valuation of the trinomial parameters of drilling, gases, and trough samples is fundamental. An adequate assessment of this trinomial is essential for building the deductive model that will lead a geologist to indicate whether or not to enter the top of the carbonate reservoir. Therefore, the matrix used in calculating fx(t) in the examples described below was defined as X=[ROP, WOB, MSE (Mechanical Specific Energy), TORQUE_iqr, TORQUE, TORQUE_std, TORQUE_var] where the variables TORQUE_iqr, TORQUE_std, TORQUE_var correspond to the interquartile distance, standard deviation, and variance of the torque, respectively. The variables MSE, TORQUE_iqr, TORQUE_std and TORQUE_var are calculated from the entry parameters. In the study in question, the decision was made to follow the recommendations of Embrechts & Herzberg (Embrechts, P.; Herzberg, A. M. Variations of Andrews' Plots. International Statistical Review, 59(2):175-194, 1991) in which these authors propose grouping the highly correlated variables and placing those considered to have the higher discriminating potential in the extreme frequencies of sines and cosines. FIG. 2 shows the flow chart used as the routine for processing drilling parameter data for generating time and depth ranges. In the present invention, we will only deal with depth ranges.

[0029] The method described above was applied to three wells drilled in the pre-salt in Santos Basin, called Well 1, Well 2, and Well 3. In all three wells, the technique allowed precise characterization of the top of the reservoir, as shown in FIGS. 3, 4 and 5 attached, proving the robustness of the method in well kill operations (track 1). In the case of Well 3, use of the PCA technique was necessary, in order to increase the discriminating power of the method. This allowed individualizing the different lithologies passed through in the well. The tops identified in the three wells from the proposed method are indicated in track 2 of FIGS. 3, 4 and 5, by the abbreviations T. AND (DPspec) (basal anhydrite top), Intertrap (DPspec) (intertrap carbonate), T. BVE (DPsec) (top of the pre-salt carbonate in the Barra Velha Formation). In track 2, these same tops are also indicated, interpreted by the geologist in the field/probe, and that they are available in the (Petrobras) Exata System database (T. AND (Exata), Intertrap (Exata) (intertrap carbonate), T. BVE (Exata)). In track 3, the lithologies passed through (halite, anhydrite and carbonate) in the three wells are indicated.