Pipeline upheaval, loss of cover solution for HT/HP fields

As the number of high temperature/high pressure developments involving small diameter subsea flowlines increases, so too does the potential for line upheaval. This is leading to a general tightening of design specifications for out-of-straightness during pipelay operations and also for depth of cover measurement.

One project where these factors are of particular importance to pipeline stability is the ongoing Troll West Field development in the Norwegian North Sea. Accurate depth of cover and out-of-straightness measurements are critical for the trenched lines, necessitating use of an ROV-based mechanical stabbing system. The stabbing interval is 5-10 meters and if cover height and out-of-straightness prove to be outside specified requirements, remedial work by sand/rock dumping is required. The trench is too deep to obtain acceptable results with existing pipe-tracking units.

Although this system has worked well, the method is time consuming and relatively costly. As a result, during installation of the K1 and K2 flowlines in January 1999, operator Norsk Hydro embarked on a unique project to determine whether an inertial geometry inspection tool could - in addition to providing improved pipeline curvature or out-of-straightness data - establish depth of cover by overlaying the measured pipeline shape and position (in three dimensions) on highly accurate seabed topographical profiles. This might allow frequency of the ROV stabs to be reduced significantly, or even eliminated, with consequent benefits to the project cost and schedule.

Installation procedures

The Troll West area has been developed using flexible flowlines hooked up to the floating platforms Troll B and Troll C. All flowlines run as pairs, with one test line and one production line alongside a third integrated service umbilical. The lines are laid for several km on the soft clay seabed out to a No. 1 template structure. Round-trip pigging is then possible. The lines are trenched and partly rock-dumped.

To stabilize and insulate each flowline, a minimum backfill of 0.50 meters has been specified. This is relatively low compared with most North Sea pipelines. Out-of-straightness of less than 30 cm over 30 meters is the established trenching target for preventing upheaval-buckling problems. Finally, the total trench depth requirement is 1.2 meters (from seabed to top of pipe). Actual depth obtained for most of the lines was in the range 1.5-1.8 meters - beyond the range of traditional pipe-tracking devices.

In response, an ROV-based hydraulic mechanical measurement device, constructed by Stolt Comex Seaway, was used to ensure that measurement of backfill, out-of-straightness, and total burial depth was guaranteed to an accuracy of greater than 0.10 meters. This system consists of a hydraulic ram, roughly two meters long, mounted vertically in the front center of the ROV. A 40-cm diameter steel cross at the end of the ram facilitates location of the line in the trench. A wire is attached to the cross and is pulled out as the ram is extended (the measuring wire is divided into 5 cm sections with colored tape).

When the ram is fully retracted, the wire at the measuring point reads zero. Using a camera, ROV personnel look directly at this point to compare the observed color coding with a prepared reference sheet displaying color coding and distances. These readings, along with simultaneous high accuracy profiles of the trench, give cover depth and total depth at each measuring location.

However, with a significant proportion of the planned flowlines on Troll West still to be installed, Norsk Hydro wanted an alternative method that could provide depth information quicker and at lower cost. Following discussions with BJ Pipeline Inspection Services in Aberdeen, it settled on the initial geometry surveying technique. This is a well-established inspection method in the North Sea for determining very accurate pipeline out-of-straightness (although not employed previously for pipeline depth of burial). Fixing the pipeline position relative to the seabed topographical profile presented challenges, as did confirmation of the system's accuracy. The patented BJ Geopig Inertial Geometry Tool has been developed over a 12-year period. It is equipped with the following sensor systems:

The inertial measurement unit comprises angle rate gyros and linear accelerometers. The system measures the precise path the pig has taken during its traversal of the pipeline. It is also used to generate a detailed map of the line, measure curvature, and to identify any significant out-of-straightness features.
Odometers measure the distance moved by the pig along the line plus its instantaneous speed within the line.
Pressure and temperature sensors measure these properties in the line during the pig run.
Mechanical calipers measure pipeline internal diameter, ovality, dent size, shape, and also undertake weld detection.

Pipeline position and bending strain are computed from the odometer and inertial measurements. The odometers measure the distance traveled by the Geopig, while the strap-down inertial system provides acceleration and rotation of the Geopig about three orthogonal axes.

Processed inertial data is rotated into the selected tie-in points on the pipeline with known Universal Transverse Mercator (UTM) coordinates. These tie-points are used to adjust the inertial system biases in such a way that the pipe centerline shape fits the tie-in points in an optimum manner over long distances. When the adjustment is completed, the pipe centerline coordinates are tied to those points, preventing accumulation of the absolute position error and transforming the coordinates into the required UTM mapping projection. For offshore pipelines, the UTM of the tie-in points is normally obtained from the ROV as-built survey.

Minimum spacing between the tie-in points depends on the required absolute accuracy. Specified accuracy is 1:2,000 of the distance to the nearest tie-in point. Azimuth and pitch of the pipe centerline in the local level frame are also computed. The pitch, P(s), describes the pipeline tilt with respect to the horizontal plane at chainage, while the azimuth, A(s), specifies the angle between the pipe direction and north. Changes of pitch and azimuth over a distance along the pipe centerline allow for computation of the pipeline total curvature and its vertical and horizontal components.

Collecting accurate pipeline shape (curvature) with the Geopig is routine. The real challenge with this project was to locate the absolute position of the pipe accurately relative to the 3D seabed topographical profile so that depth of burial could be determined. Simply using the known positions of the riser base connection PRBN (production riser base north) close to Troll B and the K1 (5.9 km from PRBN) and K2 (2.5 km from K1) templates would be insufficient. Cumulative random drift error in the inertial measurements would be significantly greater than the desired burial depth accuracy, and therefore the data in terms of pipeline position would be meaningless.

In view of the need for additional accurate tie-in points, numerous stabs would be required using the current ROV technique to tie the relative Geopig data accurately to the absolute seabed data. A trade-off had to be made between the number and cost of additional ROV stabs required and the use of the ROV for the entire survey.

By virtue of the accelerometer data collected, the Geopig can, under good operating conditions, generate significantly higher accuracy data in the vertical plane by analysis of the additional pitch data resulting from measurement of the gravity vector.

Should an accuracy of greater than 1:2,000 be achievable in the vertical plane, a direct cost and schedule saving could be realized as the number of stabs decreases. Norsk Hydro decided therefore that as part of the K1/K2 flowline survey, accuracy of the data in the vertical plane would be assessed to determine tie-point spacing requirements.

Field pigging operations

The Geopig survey of the 10-in. test and production flowlines was performed successfully using seawater as the propelling medium. The survey was executed in a single run with the Geopig launched from the test line temporary trap on Troll B, then returned through the loop at the K2 template, and finally recovered at the production line temporary receiver on Troll B. The entire inspection operation, including a pre-survey gauging run through to receipt of the Geopig, was completed within 15 hours. On receipt of the Geopig, the data was downloaded. Initial processing demonstrated good data within 24 hours.

For the K1 and K2 flowlines, a stabbing program was completed at 5-10 meter intervals in order to verify accuracy of the inertial data, and also to independently assess the capability of the Geopig for depth of burial measurement. To verify the Geopig's performance, a staged approach to data processing was adopted. Stage 1 inertial data was processed using only the PRBN, K1 and K2 template locations as tie-in points for absolute position, with reference position coordinates being obtained from the as-built ROV survey. Stage 2 inertial data was re-processed with additional stab data at varying intervals.

Following delivery of the results to Norsk Hydro, three sets of stab data were returned to BJ in order to complete Stage 2 processing. One contained tie points spaced every 100 meters, the second at 250-meter intervals, and the third at 500-meter spacing, being a subset of the other two sets. The data for each tie point consisted of three stab points at 10-meter intervals in order to minimize the effect of outliers in the stab data. The initial survey elevation was adjusted at the tie point so that the average height at these three locations was the same as the average height of the corresponding three stab points.

ROV stab data for every 5-10 meters were acquired along the lines and processed onboard the Geofjord survey vessel, then wired ashore via the Internet for further processing and comparison with the inertial pig results. Measurements in the trench gave burial depth, backfill depth, trench width, and kilometer position (Kp) for each stab. Suspicious stab points were removed or replaced by re-stabs down to every 5-meter position. This occurred typically in cases where the stabbing tool had completely missed the pipe (for instance, when the pipe had moved to one side of the trench, such as in bends).

The main problem with the Stage 1 processing is the absolute position drift, or determination of the vertical level of the line. However, local relative measurements of the pipe behavior seem to be reliable. Additional stab points should therefore aid in locking the inertial ROV data to a more correct level.

When all three sets of data from the Stage 2 measurements were re-processed and made available as complete lines, Norsk Hydro constructed maps allowing a direct comparison to be made with the "manually" obtained stab data for every 10 meters. A general conclusion had to be drawn upon studies of the three sets of data, as this would be the only time where manually obtained ROV data would be available for estimating signal drift along a complete line. A subset of the 10-meter interval also had to be chosen in order to reliably tie in future inertial data.

Selection of a suitable density for future stab points is a matter of cost and accuracy. The evaluation process considered the inertial data tied to three 10-meter points every 100, 250 or 500 meters. The 100-meter estimate was seen to be accurate, but also prone to errors in the selected stab data. Such errors were indeed observed, and one, two or even all three data points had to be omitted at some intervals in order to make the line fit with measurements at successive intervals. Using stabs every 100 meters, the stab intensity will still be fairly high and costly.

The 250-meter estimate was deemed sufficiently accurate, with a deviation of max +/- 10 cm from other measurements. Stabs here could also be prone to errors in the selected stab data. An increase of the 10-meter stab points from three stabs to five stabs at every 250-meter interval was recommended in order to minimize the effect of stabbing errors. The 500-meter estimate, meanwhile, still showed signal drift, causing deviations from the other signals of more than +/- 20 cm.

Conclusion

The inertial pig, combined with a number of tie-in points, has given promising results. However, known data points only at the riser base and at the templates several km away do not tie down the inertial pig measurements adequately to determine cover height to the specified accuracy level. A sufficiently accurate vertical position of +/- 0.10 meters is obtained using additional ROV stab points every 250 meters along the pipeline. To minimize the effect of possible errors in the ROV stabs and measurements, a total of five stabs, 10 meters apart, are recommended at every 250-meter stab interval.

The procedure will reduce numbers of mechanically obtained ROV stabs from 100 to 20 for every km of pipeline. This five-fold decrease should in turn lead to a significant reduction in processing turn-around time, and also in time spent offshore.

First results from the mechanical stabbing (250-meter stab interval) last September indicate a survey/stab performance of approximately 8 km/day, compared to the previous 1.5 km/day (5-10 meter stab interval). To illustrate the potential gain in cost, approximately 40 km of flowlines were installed in the Troll West area in 1998, an additional 40 km in 1999, and 37 km more will be installed this year.