Kelly-to-Riser Position Determining System With Adjustment for Uncompensated Heave

This invention relates to a system embodied in a data processor for solving depth tracking problems on floating drilling platforms. More particularly, this invention relates to a system for performing a linear regression on a history of accumulated kelly and heave measurements whereby accurate predictions of bit depths are achieved.

Accurate depth tracking (e.g. bit depth) on floating drilling rigs has long been a difficult task. An ocean surface's dynamic qualities (e.g., waves and tidal action) introduce many variables to any depth tracking system that incorporates heave motion. Compensation systems which are attached to a fixed point (the drilling riser) help reduce the degree to which wave motion is transmitted to the drilling mechanism (the kelly). In doing so, they provide a measurable motion (heave). If all motions were ideal, tracking would be simple. Unfortunately, inefficiencies in the compensators, drillstring stretch, and non-periodic heave motions all contribute to producing a high degree of variability in directly measured values. These problems cause a great deal of difficulty when attempting to correctly calculate the total drilling depth (bit depth) at any given time.

Several solutions to this problem have been considered. Most prior art methods introduce inaccuracies due to either invalid assumptions or by not including critical motions. For example, long-term depth changes due to tide action are not taken into account. One prior art system simply assumed that adding heave to kelly readings would cancel out all heave effects. However, this system experienced a repeating oscillation between on-bottom and off-of-bottom drilling states. This resulted in widely varying Rate of Penetration (ROP) and Weight on Bit (WOB) values, and also a loss of plotted MWD (measurement-while-drilling) data due to depth ordering errors. Still another prior art system comprising a time average filter. The input to the filter is the sum of the heave and kelly sensor readings. The output of the filter includes an "on-bottom clamp" that reduces off-bottom time. The time average filter takes a simple average of the last x seconds data and produces an update for each input. However, this simple average can be quite inaccurate.

The above-discussed and other problems and deficiencies of the prior art are overcome or alleviated by the system for solving depth tracking problems on floating drilling platforms of the present invention. The present invention comprises a system embodied in a data processor which removes uncompensated heave from sensor measurements which are used to calculate bit depth on floating drilling platforms. This system is termed a Linear Extrapolative Filter (LEF).

In accordance with the present invention, surface sensors measure heave compensator action and kelly movements. Each heave sensors reading is added to a kelly height sensor reading in order to obtain a value termed kelly-to-riser. Each kelly-to-riser value is fed into the LEF and a filtered reading is returned for use in calculating bit depth and rate of penetration (ROP). The LEF generates filtered readings by performing a linear regression on measurements taken over its pre-specified filter interval. It then predicts the current kelly-to-riser position from the resultant trend line.

Effective use of the LEF depends on proper data coverage of the heave wave forms. Data is acquired from the sensors preferably on one second intervals, which spreads the ten second average (nominal) heave wave over ten readings. The LEF's efficiency and accuracy are maximized if some whole number of exact heave periods is used as its filter interval in order to compensate for small non-periodic variations. For this reason, in a preferred embodiment, three average (ten second) heave periods are included, comprising a 30 second filter interval.

Non-uniform movements of the floating rig can often generate erroneous upward trends in the LEF's output. Accordingly, in a preferred embodiment, a software clamp is used to differentiate between slow movements that are witheld from output (e.g., caused by wave action or rig pitch and roll) and fast movements that are allowed to pass (e.g., the driller pulling the drillstring off of bottom).

Another preferred implementation feature is used to control the LEF's output. When data points exhibit a trend of downward velocities that are slowing (a negative acceleration), the filter would wrongly extrapolate to a point beyond the lowest actual value. Whenever this occurs, a braking function cancels the "momentum" that the unwanted trend creates by purging the LEF's buffer with constant values.

The novel depth tracking system of the present invention also has tide tracking capabilities. In order to correct all depths to mean sea level, which is the standard for most drilling operations, current tide and rig attitude data can be entered during calibration.

Significantly, the LEF is flexible enough to operate with or without a heave sensor input, by virtue of the fact that its multi-period design cancels out most periodical movements.

The above-discussed and other features and advantages of the present invention will be appreciated and understood by those skilled in the art from the following detailed description and drawings.

Referring now to the drawings, wherein the like elements are numbered alike in the several FIGS.:

FIG. 1 is a diagrammatic view of a floating drilling rig;

FIG. 2A-C are graphical plots of Kelly height, Heave and Kelly-to-Riser measurements, respectively, used as data inputs in accordance with the system of the present invention;

FIG. 3 is a flow chart diagram of a general system which utilizes the system of the present invention as a component thereof;

FIG. 4 is a flow chart diagram of the real time inputs acquired from sensors on the floating drilling rig of FIG. 1 which are used with the system of the present invention;

FIG. 5 is a flow chart diagram of the real time data processing system which includes the system of the present invention;

FIG. 6 is a flow chart diagram of the sub-routine RIG CONVERT which shows the system of the present invention;

FIG. 7 is a flow chart diagram of the sub-routine COMP HEAVE which forms a portion of the system of the present invention; and

FIG. 8 is a flow chart diagram of the sub-routine BRAKE HEAVE which forms a portion of the system of the present invention.

Referring first to FIG. 1, a well known floating drilling rig is shown generally at 10. Rig 10 comprises a floating drilling platform 12 having a derrick 14 supported thereon. Derrick 10 supports a drillstring or drill stem 16 which terminates in a drill bit 18. As is well known in the art, the entire drillstring may rotate, or the drillstring may be maintained stationary and only the drill bit rotated, either of which may be the environment of the present invention. The drillstring 16 is made up of a series of interconnected segments, with new segments being added as the depth of the well increases. The drillstring is suspended from a movable or traveling block 20 of a winch system 22, and the entire drillstring may be driven in rotation by a square kelly 24 which slidably passes through but is rotatably driven by the rotary table 26 at the foot of the derrick. A motor assembly (not shown) is connected to operate both winch 22 and rotatably drive rotary table 26. A drilling riser 28 comprises a conduit which is annularly arranged about drill string 16. Riser 28 is fixed in place by direct attachment to a well head 29 which is secured to sea floor 32.

The lower part of drillstring 16 may contain one or more segments of larger diameter and thicker walls (e.g. bottom hole assembly 30) than other segments of the drillstring (known as drill collars). As is well known in the art, bottom hole assembly 30 may contain sensors and electronic circuitry for sensors, and power sources, such as mud driven turbines which drive drill bits and/or generators and, to supply the electrical energy for the sensing elements. Drill cuttings produced by the operation of drill bit 18 are carried away by a mud stream rising up through the free annular space 31 between the drillstring and the wall of the well. The mud column in drillstring 16 may serve as the transmission medium for carrying signals of downhole parameters to the surface. This signal transmission is accomplished by the well known technique of mud pulse generation whereby pressure pulses are generated in the mud column in drillstring 16 representative of sensed parameters down the well. The drilling parameters are sensed in a sensor unit in bottom hole assembly 30 near or adjacent to drill bit 18. Pressure pulses are established in the mud stream within drillstring 16, and these pressure pulses are received by a pressure transducer and then transmitted to a signal receiving unit which may record, display and/or perform computations on the signals to provide information on various conditions down the well.

As mentioned, accurate depth tracking (bit depth) on floating rigs is difficult to measure in real time. An ocean surface's dynamic qualities (e.g., waves, rig attitude, and tides) introduce many variables to any depth tracking system. In order to factor in the dynamic environment of the floating drilling rig relative to the drill bit depth, known compensation systems are used. These compensation systems are attached to a fixed point (the drilling riser 28) to help reduce the degree to which wave motion is transmitted to the drilling mechanism (the kelly 24). In so doing, they provide a measurable motion (heave). In FIG. 1, the compensation system is comprised of the rucker lines 33, the riser tensioners 38 and the compensated traveling block 20. The rucker lines transmit motions of rig 10 relative to the riser 28 to the riser tensioners 38 where they are converted by hydraulic pressure. This hydraulic pressure is transferred to the compensated traveling block 20 which acts on the kelly 24 to move it up or down in relation to vertical displacement caused by the rolling and pitching of rig 10.

Sensors are provided on platform 12 to measure the heave compensation system. These sensors include a heave sensor 34 which is placed on rucker line 33 for measuring the vertical displacement caused by waves; a kelly sensor (not shown) for measuring kelly height 24 and a hookload sensor (not shown) which is mounted on the drill line 22 for determining hookload (tension) on the drill string 16.

Unfortunately, inefficiencies in the compensators, drill string stretch and non-periodic heave motions all contribute to producing a high degree of variability in directly measured values. As a result, the compensation system will not completely compensate heave leading to a certain degree of "uncompensated heave". These problems cause a great deal of difficulty when attempting to correctly calculate the total drilling depth (bit depth) at any given time.

The present invention utilizes a data processing system which removes (e.g. filters) uncompensated heave from sensor measurements so that bit depth can be accurately calculated. Turning to FIGS. 2A-C, surface sensors 34 and 36 measure heave compensator action and kelly movements (FIGS. 2A and 2B). Each heave sensor reading is then added to a kelly sensor reading in order to obtain a value termed kelly-to-riser (FIG. 2C). Each kelly-to-riser value is fed (inputted) into the system of the present invention and a filtered reading is returned for use in calculating bit depth and rate of penetration (ROP). The filtering system is termed as linear extrapolative filter (LEF). The LEF generates filtered readings by performing a linear regression on measurements taken over its pre-specified filter interval. It then predicts the current kelly-to-riser position from the resultant trend line as shown in FIG. 2C. The LEF's efficiency and accuracy are maximized if some whole number of exact heave periods is used as its filter interval. Accordingly, in a preferred embodiment, three average (ten second) heave periods are included, comprising a 30 second filter interval (see FIG. 2C).

Referring now to FIGS. 3-5, the linear extrapolative filter of the present invention is contained in a package of software called M-series (which is part of a measurement-while-drilling (MWD) system employed by Teleco Oilfield Services, Inc., assignee of the present invention). This package consists of background and foreground (real time) systems. The LEF is a part of the foreground system. The general structure of the foreground system which provides the environment for the LEF is shown in FIGS. 3-5.

As can be seen in FIGS. 3 and 4, heave data and other measured inputs 40 (kelly height, heave, speed of rotating drill string (Rpm) and hook load) are received from the external sensors (e.g. kelly height sensor 36, heave sensor 34, hook load sensor 38), collected and routed through a subroutine 42 called DACQMAN (Data Acquisition Manager) to be analyzed for processing in a computer 44. The object code (including explanations thereof) for DACQMAN is attached hereto as Table 1. A sub-routine labeled FOREMAN (Data Control 46) controls all interrupt-initiated activity. As is clear from a review of the object code for FOREMAN set forth in Table 2, this system receives and processes data from DAQ (Data Acquisition Unit) 48 and the MRU (MWD Receiving Unit 50) by calling DACQMAN 42, and then outputs it by calling DARTMAN (Data retrieval manager) 52.

Referring to FIG. 5, a sub-routine called ACQUIRE 54 is used to control the gathering of DAQ sensor readings and to refresh the DAQ instruction sequence. The object code (including explanation thereof) for ACQUIRE is set forth in Table 3. The ANALYZE program 56 then, in turn, calls RIG CONVERT 58 (FIG. 6) which executes the majority of the surface data processing. ANALYZE 56 manages the filtering and interpretation of all acquired data. The object code (and explanations thereof) for ANALYZE is set forth in Table 4.

Referring now to FIG. 6, a flow chart diagram is shown for the RIG CONVERT System. RIG CONVERT contains the actual calls to the LEF and its supporting routines. The objective of RIG CONVERT is to analyze surface sensor readings and produce calculate parameters. The object code (and explanations thereof) for RIG CONVERT is set forth in Table 5. It will be appreciated that those numbers that follow which appear in parentheses (#) reference portions of the object codes and flow charts being discussed.

This procedure (RIG CONVERT) is written in PASCAL, and will process the data passed to it from the BASIC routine named ANALYZE (Table 5). As part of the calling sequence, kelly height and heave data enter as the arrays named heave and khght (1). These arrays and others of the same type that are passed along with them contain five readings each, taken on one second intervals.

After local declarations and initializations, some maintainence is performed on rate-of-penetration control flags, and then the main processing loop is begun (2). This loop will execute the number of times required to process all data passed into the routine.

The first operations involve calculations of each individual measurement time (3) and then conversion of the hookload value into an easily usable form (4).

The first major decision is reached next, and asks whether or not the kelly sensor is dead (not present) (5). If it is not present, then depth calculations are performed and the loop is completed. If it is present, then individual values are pulled from the data arrays for further analysis (6).

The next major decision is reached at (7), where the presence of a heave sensor is checked. If a heave sensor is indeed present, then the LEF is put into use.

Firstly, the heave reading is corrected to mean sea level through CORRECT-HEAVE (8) (see object code and explanations thereof in Table 6.). CORRECT-HEAVE allows the depth tracking routines to follow tidal changes and correct all calculated depths to mean sea level. This corrected value is then forwarded along with the current kelly reading to the LEF, which is located in COMP-HEAVE (9) (see object code and explanation thereof in Table 7 and flow diagram of FIG. 7). The LEF will output a global variable, named Last-position, and will also return the current drilling velocity.

The LEF's output must be controlled under certain circumstances. In order to check for these conditions, several variables are investigated. Firstly, it is important to determine whether or not this set of readings signifies the start of actual forward drilling (10). Next, this Just-hit-bottom condition is checked in conjunction with both the drilling velocity, as calculated by the LEF, and flags indicating recalibration status and tripping (non-drilling) status (11).

If these checks result positively, then the braking function will be used. First, the brake setting to be used is chosen depending upon the recalibration flag (12). Next, the braking function is called via BRAKE-HEAVE (13) (see object code and explanation thereof in Table 8 and the flow diagram of FIG. 8). This function will purge the LEF's data histories with constant values, thereby cancelling any momentum that may have been accumulated to this point. As is clear from a review of FIG. 8 and Table 8, BRAKE HEAVE is a routine that uses a constant and repeatedly calls COMP-HEAVE until the LEF's buffer has been completely flushed.

If the heave sensor was not present at step (7), then two necessary variables are intialized before calling the depth calculation routines.

The depth calculations which will utilize the LEF's filtered output are accessed through CALC-DEPTH (14) (see object code and explanation thereof in Table 9). CALC-DEPTH provides the interface between RIG-CONVERT and DEPTH-STATUS (see object code and explanation thereof in Table 10) for depth tracking and calculations. The primary output returned from CALC-DEPTH will be the value of bitdepth, along with several status flags. These outputs are then transferred into secondary arrays for late usage (15).

The remainder of the code in this (RIG CONVERT) routine consists of calculating several parameters from the data sets, and then updating the appropriate process control areas before completing the loop (16).

As mentioned, the LEF is located in the COMP-HEAVE sub routine (refer to Table 7 and FIG. 7). The objective of COMP-HEAVE is to update kelly and heave data histories with the latest data points and to perform a least squares analysis on these histories which will produce a current filtered projection of kelly-to-riser (FIG. 2C). This resulting value will be used in subsequent depth calculations.

The COMP-HEAVE routine maintains histories of thirty heave and kelly readings each. Every time the routine is called, a pointer is updated in order to allow proper data control within the histories so that the newest data will replace the oldest data (2). Actual replacement will take place later, but first the old values must be retained for removal from the least squares sums.

The current measured kelly-to-riser value (FIG. 2C) is obtained by adding the kelly and heave readings that have been passed in from the calling routine (1). This term is created to simplify usages in the least squares analysis that is forthcoming.

The heart of the Linear Extrapolative Filter lies in the least squares analysis found in the next four stages. Four sums are required to complete the analysis, Sumx, Sumxx, Sumxy, and Sumy. For this application, x-axis will represent time, and y-axis will represent linear displacement. Since the data points are collected regularly on one-second intervals, the x sums can be easily and properly approximated with constants (3). Sumxy and Sumy involve usage of the kelly-to-riser data, and in calculating them a short cut is utilized which is also allowed by virtue of x-axis (time) assumptions (4). The four equations used in performing the least squares analysis preferably comprise the following:

Where:

After all of the sums have been updated, the slope and intercept of the best fit line can be derived (5). These two parameters are subsequently used to obtain a projected kelly-to-riser value (6) (FIG. 2C). This value is projected to the current time, so that averaging lag is reduced to a minimum.

Where:

Slope = slope of trend line

Intercept = intercept of trend line

Before this final value is used, some checks are run to determine whether or not this particular output will be passed along for depth calculations (7). Of primary concern is determining whether or not an apparent upward movement is due to actual drilling activity (such as pulling the drillstring off-of-bottom) or rather is due to uncompensated heave movements. If the checks are passed (motions are true drilling), then the projected value is assigned to Last-position (8). This global variable will carry the filtered value to the depth calculation routines (Tables 9 and 10). Lastly, the current kelly and heave data points are used to update their respective history arrays (9).

As discussed below, the LEF of the present invention has been successfully tested in both the lab and field.

Three potential scenarios were created in lab testing of the routines. Firstly, synchronous heave and kelly motions (180 degrees out of phase) were input for use by the tracking system, representing the ideal situation. Secondly, asynchronous heave and kelly inputs were generated, producing phase differences ranging between 0 and 180 degrees. This tested both the most realistic, with phase differences near 180, and the most strenuous (but rare) cases when the motions were exactly in phase. Next, only heave-influenced kelly motions were input, simulating complete lack of a heave sensor. This last case was the most streneous that was expected in actual use of the tracking system. A multiple of rig activity simulations were conducted under each of these scenarios. Lastly, recordings of actual rig data were run through the system.

Following successful completion of lab testing, the routines were field tested in the Norwegian sector of the North Sea. Testing took place on four floating drilling rigs.

The most strenuous expected scenario devised in the lab was realized immediately. The first two tests began without heave sensors, as a surface hole was being drilled with sea water and no drilling risers were present to which the compensators could have been attached. The LEF worked so well that operators were able to accurately track depth even without installing the heave sensors when the risers were being used for later hole sections. The only disadvantage of operation in this manner was the loss of tide-tracking capabilities.

The third and fourth test were conducted with full compensators and tide tracking, as were the latter part of the first and second tests. The LEF proved itself during these tests as a highly useful method of heave compensated depth tracking. Operators reported only one miscalculated depth value from all four of these tests combined.

In summary, the Linear Extrapolative Filter (LEF) of the present invention is a data processing procedure that allows accurate depth tracking from floating rigs. The LEF has at least the following features:

1. ROP Measurement

A primary goal of the LEF is to properly measure ROP in real time while removing uncompensated heave from the depth measurements. To achieve this goal, the LEF establishes a trend line from measurements taken over the last n (e.g. 30) seconds. The filter then predicts the bit's current position from the trend line. The LEF is designed with no built-in delays, so that every depth reading immediately results in a filtered reading.

2. On-Bottom Clamp

The system incorporates a software on-bottom "clamp" that keeps the depth system on bottom if it sees small, slow movements (e.g., caused by wave action or rig pitch and roll). However, it immediately comes off bottom when it detects quick movements (e.g., the driller pulling off bottom).

3. HEAVE-BRAKE

The LEF system responds to kelly acceleration as well as velocity. The LEF sees any sudden velocity changes (like tagging bottom after a connection) and automatically hits a heave brake to prevent the depth system from "drilling ahead" of the driller (by producing incorrectly projected values).

4. Tidal Compensation/System Calibration

The LEF system is calibrated by inputting tidal data and rig ballast information. Thereafter, the system tracks tidal and ballast changes as drilling progresses and adjusts all depth readings to mean sea level (MSL). The system is periodically recalibrated by updating the tidal and ballast data.

5. Multi-Period Sampling Interval

The LEF samples heave and kelly sensors roughly every second which spreads the nominal heave period over 8 to 12 samplings. Several periods of the heave wave are included in the sample interval in order to cover most small variations from the true average and to increase the accuracy of the filter. ##SPC1##

While preferred embodiments have been shown and described, various modifications and substitutions may be made thereto without departing from the spirit and scope of the invention has been described by way of illustrations and not limitation.