Determination of Peaking of Pipes

Abstract

The invention relates to a method for determining peaking of a pipeline, wherein measurement values of the internal geometry of the pipe are recorded adjacent to a searched point P of the peak, wherein the measurement of meaningful measurement values is made more difficult or impossible due to the weld seam that is present there, and wherein a trend is determined based on the measurement values and a value of the peaking is determined based on the trend.

Claims

1. A method for determining the seam peaking of a pipeline, wherein measured values of the internal pipe geometry are recorded adjacent to a sought point of the seam peaking, at which the measurement of meaningful measured values is difficult or impossible due to a weld seam present there, and a trend is determined from the measured values and a value of the seam peaking is determined on the basis of the trend.

2. The method according to claim 1, wherein measured values which are measured directly at the weld seam are not used by the method for determining the seam peaking (h).

3. The method according to claim 1, wherein neighbouring measured values are compared with one another and a search is made for outliers which serve as an indication of the presence of the weld seam at this location and/or the presence of the weld seam at a specific location is inferred from measured data from one or more previous measurements at other axial locations of the pipeline, and the outliers and/or measured values determined in this way at locations of the expected weld seam are not used for the seam peaking determination.

4. The method according to claim 1, wherein a sensor carrier comprises a plurality of sensors and is arranged in the pipe in such a way that at least two measured values of the distance from a sensor to the inner wall of the pipe are recorded on either side of the sought point-in the circumferential direction of the pipe, so that an angular position of the inner wall of the pipe, in particular relative to the sensor carrier, is determined in each case and the seam peaking of the sought point is determined via an intersection of the angular positions.

5. The method according to claim 1 the preceding claims, wherein in each case at least two points of the inner surface of the pipe are determined in the circumferential direction of the pipe on either side of the sought point-of the seam peaking, which are each approximated, in particular interpolated, to a straight line whose point of intersection corresponds to the seam peaking.

6. The method according to claim 4, wherein to form the trend optionally: a) a plurality of measured values is used via a linear regression in order to obtain the angular position or the straight line, b) in each case neighbouring sensors (25) of a sensor carrier (20) each determine an inclination of the inner wall of the pipe and the median value of this inclination is used to determine the angular position or to determine the straight line, and c) spatially neighbouring sensors (25) of a sensor carrier (20) each determine an inclination of the inner wall of the pipe, and it is checked whether these inclinations contain an outlier and only inclinations are used to form a mean value, in particular a median value, which, seen from the outlier, lie adjacent to the sought point.

7. The method according to claim 1, wherein, in the event that the seam peaking is provided with a negative sign and thus corresponds to a retraction, a mirroring of the measured values over the ideal circular arc geometry is carried out, the calculations for determining the seam peaking are subsequently carried out, a mirroring over the ideal circular arc geometry is carried out for the calculated value of the seam peaking and the extent of the retraction, is determined therefrom, whereby the mirrorings in particular are carried out using a Kelvin transformation.

8. The method according to claim 1 with a sensor carrier with sensors, which has a width in the circumferential direction of the pipe, which corresponds to at least 70% of an expected seam peaking width, and/or wherein the width of the sensor carrier corresponds to a maximum of 130% of the expected seam peaking width and in particular to a maximum of 100% of the expected seam peaking width.

9. A pig for carrying out a wall thickness measurement and/or a geometry measurement and/or a crack measurement in a pipe, having a sensor carrier which has sensors, and a control system and/or software which is configured to carry out the method according to claim 1.

10. A computer program product loadable into a program memory and comprising program instructions to carry out all the steps of a method according to claim 1 when the program is executed.

Description

[0019] Preferred embodiments are described below by way of example with reference to the figures. The figures show:

[0020] FIG. 1 a sensor area of a part of a pig,

[0021] FIG. 2 a two-part diagram with model data,

[0022] FIG. 3 a graphic illustration to explain the calculation of the seam peaking height h,

[0023] FIG. 4 a split diagram showing the seam peaking calculation for a negative seam peaking h,

[0024] FIG. 5 an illustration explaining the processing of measurement results from sensors located on different sensor carriers,

[0025] FIG. 6 a detail of FIG. 5,

[0026] FIG. 7 a diagram of the measured values according to FIG. 5 on the left-hand side of the 0 mm position, i.e. the weld seam,

[0027] FIG. 8 a diagram of the measured values to the right of the 0 mm position, i.e. the weld seam of FIG. 5,

[0028] FIGS. 9 and 10 each show gradients of neighbouring measuring points of FIG. 7 and FIG. 8.

[0029] FIG. 1 shows a pig 10 for use inside pipes with a disc-like base body 12 at the bottom, which is smaller by a certain amount than the associated pipe and which serves as a support for sensor carriers 20. For this purpose, arms 22 are used, one end of which is pivotably attached to the base body 12 and the other end of which is connected to a sensor carrier 20. The swivelling is such that the sensor carriers 20 can be moved radially outwards. Springs (not shown) are used for this purpose, which generate an outwardly directed force. The sensor carriers 20 each comprise a centre area in which an array of ultrasonic sensors 25 is arranged. The sensor carriers 20 comprise skids 26, between which the array of ultrasonic sensors 25 is arranged. The ultrasonic sensors are displaced back in relation to the skids, i.e. radially inwards in relation to the pig base body 10, so that when the pig 10 is in the pipe and the sensor carriers are pressed outwards, only the skids 26, but not the ultrasonic sensors 25, are in contact with the pipe. In addition to this swivelling capability, the sensor carriers 20 can be swivelled around the longitudinal axis of the pig so that, if possible, both skids 26 can lie against the pipe even if the inner contour of the pipe is uneven, which is also supported by the aforementioned springs. In addition, in the third of the three Cartesian rotational degrees of freedom, the sensor carriers 20 can optionally be swivelled in such a way that the skids can rest against the inner wall of the pipe along their length. The sensor carriers 20 are arranged evenly in two axially offset rows, with the rows having an angular offset in the pig circumferential direction. This ensures that the inside of the pipe is evenly covered with sensor fields. If the sensors are spaced at a distance of 4 mm in the circumferential direction, for example, the entire inner circumference of the pipe can be analysed at a given axial position at each measuring point. By moving the pig in the pipe's axial direction, it is possible to examine the entire inner surface of the pipe without interruption, in particular with constant measuring point distances all round. The inspection can carry out a pipe thickness measurement in which the measuring direction of the sensors is directed radially outwards and/or a crack inspection can be carried out, and in this case the sensors are preferably orientated at an angle to the inner pipe surface. Instead of the configuration described, any sensor carrier can be used.

[0030] FIG. 2 shows in the upper part the idealised distances of the individual sensors 25 from the inside of the pipe as the so-called SO data (SO=stand off) with a seam peaking of height h. This is centred in the diagram and can result in particular from a longitudinal weld seam of the pipe. The distances from the weld line are plotted in millimetres on the horizontal axis. However, it is practically always the case that the measurement method shows inaccurate and even incorrect values in the direct vicinity of the seam peaking h. One of the reasons for this is that excess weld metal protrudes radially inwards and forms a so-called weld bead. It may therefore be necessary to analyse and select the measured data in order to determine the position of the weld seam. This is done, for example, by comparing neighbouring measured values with a search for outliers that are interpreted as protruding weld metal (or weld bead). And outliers determined in this way are not used to determine the seam peaking.

[0031] This material is used to measure reduced distances from the measured pipe geometry to the ideal circular geometry. These values do not correspond to the true seam peaking h and are not useful for calculating an increase in the probability of failure. Instead, the correct seam peaking h is required, which can be defined in particular as a widening of the pipe material in the area of the weld seam. If the measured values are too far away from the weld seam, they have no direct significance for the seam peaking h, so that they are also not used for its calculation. This leaves the measured values that can be used, which lie in the areas labelled b in FIG. 2. In the lower part of FIG. 2, the values were related to the circular geometry of the inner diameter of the pipe. This shows that the measured values within the areas b lie approximately on a straight line. A linear regression can be used to determine a gradient of the pipe wall within each of these areas b. Linear regression is applicable because, for simplicity's sake, a constant gradient is assumed within the areas b. Based on the correlation in FIG. 3 and the formula shown, the seam peaking height h can be determined from the angle of the gradient . The areas b that are used to determine the seam peaking height can, in particular, be directly adjacent to the area that is therefore not used to determine the seam peaking height, as the weld seam was expected in this area.

[0032] In other words, this relationship can also be seen in the lower part of FIG. 2, where the constant gradients within the areas b are extrapolated linearly, resulting in an intersection point P, and the seam peaking height h is the (perpendicular) distance of the intersection point P from the two straight lines in relation to an ideal circular geometry of the inside of the pipe.

[0033] The seam peaking h can partly be understood as the result of imperfect welding preparation. The sheet material from which the pipe is to be welded must be bent into a tubular shape during the welding preparation. Due to difficulties in the engagement of the bending tools or the transmission of force, the edges to be welded may have a too small curvature, which then forms as a seam peaking h. The above-mentioned linear extrapolation (or the calculation based on FIG. 3) assumes that the sheet material is not bent at all at the welding edge, as a straight extension from the areas b to the seam peaking point P is assumed. With good work preparation, it is to be expected that a seam peaking h determined in this way should always be slightly larger than the actual seam peaking, as at least a partial bending of the material at the weld edge can be assumed. This possibly rather too large value can be understood as a safety factor and makes this method superior to other methods in which a seam peaking h is approximated as accurately as possible, but without being able to sufficiently exclude the possibility that the real seam peaking h may be greater than the calculated value.

[0034] Due to the mobility and rotatability of the sensor carriers 20 described above, the distances measured by the ultrasonic sensors 25 cannot be used directly to calculate the seam peaking height h. This is because the sensor carriers 20 are pressed outwards by the aforementioned springs and therefore the relative distance from the sensors to the inner wall of the pipe is of no particular significance. This also applies because the sensor supports can be inclined if, for example, one of the skids (or both) partially dips into the seam peaking h and/or the other skid rests on an undeformed part of the inner tube geometry. The method described above with the first step of determining the gradients has the initial advantage that the radial position of the sensor carriers 20 (i.e. the absolute values of the measured distances) is not included in the calculation of the seam peaking h, but only the measured gradients of the pipe wall on both sides of the weld seam are used to calculate the seam peaking angle. There is also the further advantage that an inclined position of the sensor support can also be subtracted out, as will be explained below.

[0035] In some cases, an inward seam peaking can occur, i.e. a dimensional deviation of the pipe inwards, whereby the seam peaking has a negative sign, as shown in FIG. 4. In its upper part, an ideal circular geometry is initially shown in grey and radially inwardly displaced measurement results are shown in points adjacent to the weld seam. Analogously to the considerations of the measured value evaluation according to FIG. 2, only the areas b of measurement results are used for the seam peaking calculation. It can be seen from the upper part of this figure that the inferred values lie on an arcuate path, which makes it difficult to extrapolate the values for the formation of an intersection point P, which corresponds to the seam peaking h. A linear extrapolation would systematically lead to too low (negative) seam peaking heights, whereby considerable deviations are also possible. Instead, a kind of mirroring is performed on the ideal circular geometry, resulting in the image in the lower half of FIG. 4. The mirrored values lie largely on a straight line, so that (as already explained) an average gradient can be calculated via a (linear) regression and correspondingly also an intersection point P of the seam peaking h. However, this intersection point must be mirrored again in order to obtain the internal seam peaking. While the mirroring can be calculated in a simplified calculation, e.g. via an offset of the plumbed distances from the circular path, a Kelvin transformation can preferably also be used for this, which promises improved mathematical accuracy, as the Kelvin transformation is true to the angle.

[0036] It also applies that an offset and/or angular deviations of different sensor carriers 20 from each other can be subtracted out, as can be seen below with reference to FIG. 5 and its detail A in FIG. 6. FIG. 5 initially shows several vertical lines between which the sensor carriers 20 are located and which therefore represent the transition line between two sensor carriers 20. Their measurement results are shown as points in the lower half of the diagram, although only the results that are also used to calculate the seam peaking due to their position are shown here. These are six points to the right of the 0 mm position on the welding line, which are largely on a straight line or can be approximated to it without any problems. To the left of the welding line, in a range from 15 mm to 35 mm, there are six measuring points, which are enlarged in detail A of FIG. 6 and are therefore easier to see. The measured values on the right-hand side are recorded exclusively by sensors located on a sensor carrier 20, so that the linear regression can be carried out easily, as shown in FIG. 8. According to FIG. 6, of the measured values on the left-hand side of the seam peaking, the three measured values on the right-hand side also originate from sensors on this sensor carrier 20, as indicated by the vertical boundary line. The three measured values to the left, on the other hand, originate from measured values of a neighbouring sensor carrier 20, whereby a certain inconsistency of the measured values results. This is because there may be a radial offset and a changed inclination between the sensor carriers 20. This inconsistency can be recognised mathematically by determining the gradient of two neighbouring measuring points, as shown in FIG. 9. The six measuring points in FIG. 7 therefore result in the five gradients in FIG. 9. The value of the gradient that lies in the transition from the sensor supports 20 is significantly higher than the others and is to be considered an outlier. The outlier can be eliminated, for example, by determining the median value of the individual gradients when calculating the mean gradient.

[0037] The measured values that are determined within a sensor carrier, on the other hand, have practically no outliers according to FIG. 8, so that their gradients according to FIG. 10 are largely identical.

[0038] In an alternative calculation, the outlier can also be determined first, as explained with the aid of FIG. 9, and only measured values that originate from the same sensor carrier, which lies decisively in the area of the seam peaking, can be used to determine the gradient, as shown with the thick line of the gradient in FIG. 7, which runs through three measuring points on the right-hand side.

[0039] In an alternative calculation, the control unit can also take into account that only measurement results from sensors on the same sensor carrier are used. And if at least two measured values are available on the left and right sides of the weld seam, i.e. the expected seam peaking, this may already be sufficient to calculate the gradients. The use of more measured values increases the accuracy of the determination of the seam peaking.

[0040] Since a large number of measurements are carried out with the pig, which is axially displaced in the pipe, and since it can be assumed that the seam peaking in the pipe axial direction is rather constant or at least changes gradually without jumps, measured values obtained at different axial positions of the pig can also be used to determine the seam peaking. Alternatively, measured values from different axial positions of the pig can be used to determine an average seam peaking angle. Or a seam peaking can be calculated for one axial position of the pig in each case (as described), which is then averaged with other seam peakings for other axial positions of the pig to form an average seam peaking.

[0041] In one embodiment of the invention, the sensors can be evenly distributed around the circumference of the inside of the pipe. This can be designed in such a way that the sensor carriers are practically edge-free. Or the sensor carriers can be designed in, for example, two circumferential rows, with the two rows being offset axially (i.e. in the longitudinal direction of the pipe) and the rows being twisted tangentially so that the inner wall of the pipe is evenly covered with sensors. Alternatively, in an alternative embodiment of the invention, measured values recorded on or adjacent to the weld seam can be ignored. This is harmless, as measured values that are recorded directly on the weld seam do not contain any meaningful information about the seam peaking due to the weld bead. Even if there is no seam peaking in the pipe, the method may still indicate a seam peaking angle >0 and therefore a seam peaking height >0 mm. The method therefore assumes a minimum seam peaking height depending on the pipe diameter and the width of the gap at the longitudinal seam. Geometric considerations can be used to calculate sensitivity curves for the method, from which the minimum values for the seam peaking angle and height can be derived. In the picture below, this was calculated as an example for a 20 pipe and a gap width of 20 mm. In this example, the method is not sensitive to seam peaking angles in the range of approximately +/100 or a seam peaking height of approximately +/1 mm.

REFERENCE SIGNS

[0042] 10 Pig [0043] 12 Base body [0044] 20 Sensor carrier [0045] 22 Arm [0046] 25 Sensor or ultrasonic sensor [0047] 26 Skid [0048] P Sought point of the seam peaking [0049] h Seam peaking, or its height [0050] d Seam peaking width [0051] Half seam peaking angle