EP0572055A1

EP0572055A1 - Method for detecting drillstring washouts

Info

Publication number: EP0572055A1
Application number: EP93201208A
Authority: EP
Inventors: Mohamed Damak; Dominic Patrick Joseph Mccann; David Brian White
Original assignee: Services Petroliers Schlumberger SA; Sedco Forex Technology Inc
Current assignee: Services Petroliers Schlumberger SA; Sedco Forex Technology Inc
Priority date: 1992-05-23
Filing date: 1993-04-27
Publication date: 1993-12-01
Anticipated expiration: 2013-04-27
Also published as: GB2267300B; NO304710B1; NO931863D0; EP0572055B1; GB2267300A; CA2095583C; GB9211048D0; DE69309149D1; CA2095583A1; NO931863L

Abstract

A method of detecting a drillstring washout during an operation involving the addition or removal of pipes from the drillstring such as back reaming, comprising: a) performing at least one such operation and determining by hydraulic coefficient k from the relationship

{P = kQ}^{a}

,where Q is the flow rate, P is the standpipe pressure and a is the flow exponent, from each pipe added or removed, so as to derive a series of values indicating the development of k for said operations, and b) on subsequent operations, determining k and comparing the determined values with the series obtained previously, a drillstring washout being detected when the determined value of k falls substantially below the corresponding value of k in the series. The trend in the development of k can be calculated and used to determine if an anomaly exists in the determined value of k.

Description

The present invention relates to a method for detecting washouts in drillstrings being used to drill a well such as a hydrocarbon well or a geothermal well.
In the rotary drilling of wells, the drillstring used comprises a plurality of pipes linked end to end and having a drill bit located at the lower end thereof. In use a drilling fluid, known as drilling mud, is circulated through the drillstring and out of the bit and returns to the surface via the annulus formed between the drillstring and the borehole wall. Drilling mud contains finely divided solid material which gives the circulating mud abrasive properties. Should damage occur to a pipe in the drillstring the physical stresses of drilling and the action of the drilling mud can cause an opening to form in the drillstring which is known as a washout. If this is not detected, the weakness in the drillstring at that point can cause the string to break due to the torque experienced by the string. This is known as a twist-off and is costly in both time and equipment as it will involve a fishing operation to retrieve the lost end of the drill string before drilling can recommence.
Washouts can be detected by monitoring the flow of drilling mud in the drillstring. The presence of a washout will change the way in which the mud flows in the drillstring and so continuous monitoring of the flow will allow detection. However, in operations such as tripping in or out of hole or reaming, the number of pipes in the drillstring changes in a relatively short time also the flow through the drillstring varies considerably from connection to connection which makes washout detection more difficult. A previously proposed system monitors standpipe pressure for each stand and raises an alarm when anomalous pressures occur. However, this system only monitors the pressure over a single connection and is reset each time a new connection is made. Furthermore, the development of a washout takes a certain period of time and this is often longer than the time between connections in operations such as reaming such that the detection algorithms are not able to detect the anomalies due to the washout. It is an object of the present invention to provide a method by which a drillstring washout can be detected during such operations.
In accordance with the present invention, there is provided a method of detecting a drillstring washout during an operation involving the addition or removal of pipes from the drillstring comprising:

a) performing at least one such operation and determining the hydraulic coefficient k from the relationship ${P = kQ}^{a}$
, where Q is the flow rate, P is the standpipe pressure and a is the flow exponent, for each pipe added or removed, so as to derive a series of values indicating the development of k for said operations, and
b) on subsequent operations, determining k and comparing the determined values with the series obtained previously, a drillstring washout being detected when the determined value of k falls substantially below the corresponding value of k in the series.

In the present method, the flow rate of drilling fluid in the drill string and the standpipe pressure are monitored while the operation in question takes place. For each connection the expected hydration coefficient is calculated from a model obtained from previous connections or operations. This value is compared with the actual value of the hydraulic coefficient measured for the connection. Thus the development of anomalies from connection to connection can be observed and the previous detection algorithms utilised.
Preferably the step b) comprises determining the trend in the development of k and comparing the trend with that calculated for the series.
It is also preferred that the determined values of k are included in the series provided that a washout has not been detected during that operation.
An alternative to determining trends is to estimate a value of k from the series for a given pipe length and comparing this with the current value.
The present invention will now be described by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a schematic view of a rotary drilling rig;
Figure 2 a), b) and c) show hook load, standpipe pressure and flow rate values for a circulation while pulling out of hole operation;
Figure 3 shows the corresponding hydraulic coefficient for the operation shown in Figure 2;
Figure 4 shows a plot of measured hydraulic coefficient (solid line) and calculated trend (dashed line) with drillstring length for a back reaming operation with a drill string washout,
Figure 5 shows the corresponding plot to Figure 4 with no washout.
Figure 6 shows a corresponding plot when running into hole; and
Figures 7 and 8 show flow diagrams for the algorithms used in implementing the present invention

derrick

drill string

hook

travelling block

cable

drill collars

drill bit

pit

drill string

standpipe

bit

The present invention is particularly applicable to situations when the drill string is being rotated and the length is being changed fairly rapidly such as might occur in reaming operations, tripping or stabbing. Under steady state circulation, the flow rate Q is related to the stand pipe pressure by the relationship ${P = kQ}^{a}$
. Since most pressure drop occurs in turbulent flow, the exponent a is very close to 2. The hydraulic coefficient K is a constant which depends on the drill string geometry and on the density of the drilling fluid. Within one circulation period, the drill string length remains constant and so any trend in the hydraulic coefficient is an indication of an abnormal event, for example a decrease in K would indicate a washout. A typical algorithm for detecting anomalies is the Hinkley method (Biometrica 42,6, pp1897-1908, 1971). When this method is applied to monitoring changes in measurements when drilling normally, the detection of a given number of events (typically two or three) within a given time period is used to generate an alarm. The time period might typically be set at 15 minutes which is not suitable for operations such as reaming where the time between connections can be significantly less than this. In the present invention, the hydraulic coefficient calculated for the previous connection or connections is stored and the history of the behaviour of the hydraulic coefficient determined to build up a model of the hydraulic coefficient against depth. The trend from previous operations can also be stored. The hydraulic coefficient for a given stand is compared with the model and trend and the detection algorithm applied to detect anomalies.
In one embodiment, the proposed method does not look only at the current circulation period, but keeps the history of the hydraulic coefficient behaviour from circulation to circulation during the current reaming operation. It also keeps the trend from the last reaming operation provided that the mud and the BHA have not changed. In the case where the mud has changed and there are not enough points to prepare a full model, the fractional change in the hydraulic coefficient k is compared with an empirically derived coefficient. Typically the following relationship is used to confirm that a point can be added to the model:

During the reaming mode the algorithm calculates, for each stand, the hydraulic coefficient value, from the measured flow rate and stand pipe pressure, and builds a linear model of hydraulic coefficient against string length typically using a least squares method. The model is updated as every stand is pulled or added provided a washout has not been detected. By monitoring the changes in the hydraulic coefficient over several stands, the effect of even a gradual washout can be detected.
At the beginning of each reaming operations, when not enough stands have been pulled, the calculated trend in the hydraulic coefficient is compared to that recorded during the last reaming operation.
Once enough stands have been pulled the algorithm compares the hydraulic coefficient of the current stand with the expected value estimated by the linear model built from the previous stands in the same reaming operation. If the measured value is lower than the estimated one by a certain amount typically 1%, a washout is indicated
Figure 2 shows three logs collected whilst circulating and pulling the drillstring out of hole, obtained from the North Sea. The hook load and the stand pipe pressure logs show that a twist off occurred at about 24717 s. The flow channel shows that the circulation periods are often very short (about 6 minutes). Also, that the flow varies from stand to stand between 3800 1/min and 4050/min. As a result it is difficult to detect a washout by simply observing the stand pipe pressure for each connection or even over a number of connections, since it varies from stand to stand considerably because it is proportional to the flow rate squared and the trend is hidden in this variation. It can also be seen that in reaming mode, under the same flow conditions, the pressure generally decreases as the stands are pulled and increases as they are added due to the change in string length and this also masks any trend due to a washout. Figure 3 shows the same event but the third log now show the hydraulic coefficient k. It can be seen that throughout the operation, the k value is dropping gradually until the twist off when it suddenly drops.
Figure 4 shows how the proposed algorithm detects the washout whilst pulling the forth stand, that is, four connections before the twist off. It does this by comparing the calculated trend with that from the previous reaming operation.
Figure 5 shows reaming data in a normal operation (without any washout). It can be seen that the hydraulic coefficient varies linearly with the string length. In this example, all of the measured data fall within the expected range of the model and therefore no alarms are raised.
The algorithms for the method of the present invention are shown in the flow diagrams of Figures 7 and 8.
At the beginning of each reaming operation, when not enough stands have been pulled, the calculated trend, which is the gradient from the least square calculation, is compared to the one from the last remaining operation model. If actual trend >1.65* last trend, a washout alarm is raised. The coefficient 1.65 has been empirically defined by studying sets of field data.
Once four stands have been pulled, the algorithm compares the hydraulic coefficient of the current stand with the expected value estimated by the linear model build from the previous stands in the same reaming operation. At the most sensitive, if the measured value is lower than the estimated value by more than 3 times the standard error of estimate (for sensitivity =9), a washout alarm is generated. By monitoring the changes in the hydraulic coefficient over several stands the effect of even a gradual washout can be detected. To prevent the model including the effect of washout, a point is added to the model only if the two following points do not raise any alarm. To avoid false alarms, the washout alarm is raised only when two successive points lie outside the model.
The implemented algorithm is illustrated by Figure 7. Figure 5 shows a normal evolution of the hydraulic coefficient K during reaming against depth. It can be seen that the hydraulic coefficient varies linearly with the string length and K drops as the stands are pulled. The linear curve models the trend of the hydraulic coefficient from stand to stand. The range of expected values is about ±1% of the estimated value (±3*standard error of estimate).
Figure 5 shows how, once the linear model is built from the previous stand, the algorithm deduces an estimated hydraulic coefficient (+) for the current stand and compares it with the measured value (*). In this case the measures data falls within the expected range and therefore no alarm is raised
Figure 4 shows how the new algorithm detects the washout whilst pulling the fourth stand, that is four connections before the twist off. It does this by comparing the calculated trend with that from the previous reaming operation.
Figure 6 shows running in hole data where the hydraulic coefficient increase as stands are added. At the last point, the pressure is higher than expected because the MWD is switched on. The algorithm checks if the MWD is transmitting data by checking the RPM (Rotation Per Minute) value. The RPM is zero when the MWD is transmitting data. The algorithm currently ignores this point and does not include it in the model.

Claims

A method of detecting a drillstring washout during an operation involving the addition or removal of pipes from the drillstring comprising:
a) performing at least one such operation and determining the hydraulic coefficient k from the relationship ${P = kQ}^{a}$
, where Q is the flow rate, P is the standpipe pressure and a is the flow exponent, from each pipe added or removed, so as to derive a series of values indicating the development of k for said operations, and

b) on subsequent operations, determining k and comparing the determined values with the series obtained previously, a drillstring washout being detected when the determined value of k falls substantially below the corresponding value of k in the series.
A method as claimed in claim 1, wherein the flow rate of drilling fluid in the drill string and the standpipe pressure are monitored.
A method as claimed in any preceding claim, wherein the determined values of k are included in the series provided that a washout has not been detected during that operation.
A method as claimed in any preceding claim, wherein for each addition or removal of a pipe, the expected hydraulic coefficient is calculated from a model obtained from previous connections or operations and is compared with the actual measured value of the hydraulic coefficient for that connection in order to observe the development of anomalies.
A method as claimed in claim 4, wherein step b) comprises determining the trend in the development of k and comparing the trend with that calculated for the series.
A method as claimed in any of claims 1 - 3, comprising estimating a value of k from the series for a given pipe length and comparing this with the current value.
A method as claimed in any of claims 1 - 4, wherein the calculated trend in the hydraulic coefficient is compared to that recorded during a preceding operation of the same type.
A method as claimed in any preceding claim, wherein the operation comprises a reaming operation.
A method as claimed in any preceding claim, further comprising the step of determining if an MWD system is in operation and ignoring any pressure anomalies while said system is in operation.
A method as claimed in claim 9, wherein the step of determining if an MWD system is in operation comprises measuring the rate of rotation of the drill string and detecting when the rate of rotation indicates that the MWD system is in operation.