Non-intrusive subsea leak monitoring.
After withstanding a thorough laboratory demonstration, a set of self-contained, battery-operated subsea leak monitors were set up to monitor the closing of two 20-inch subsea valves between a cross over. These valves had not been operated for 25 years and the client wanted to use the leak monitors to verify whether additional operations were needed.
Before deployment, the leak monitors went through a laboratory demonstration to verify that they would work and quantify the leak in the valves. Leak-detection thresholds depend on valve and pipe geometry, placement of sensor and differential pressure. The prediction model was verified and predicted 95% certainty to detect leakages as low as 42 SCM/h at a 50 bar differential pressure. A small leak was detected into the cross over from one of the valves.
Leaks in subsea installations are detected by several methods. Bubble plumes due to large gas leakages to the environment can be discovered visually. Dye and fluorescent tracers are used for detecting fluid leaks visually or by sensitive cameras or fluorometers. Due to their potential harm to the environment, however, the use of tracers is being phased out in some areas. Chemical analysis of seawater has similar use, based on detection of a substance that has escaped through a leak.
Detection of through-valve and cross-flow types of leaks requires a different set of techniques. Static pressure measurements using block-and-bleed methods seem the most common, but can be impractical and sometimes not sensitive enough to satisfy given specifications.
Acoustic methods have been used to some degree, most commonly by means of hydrophones that pick up ultrasound propagating through water from a leak in the vicinity. Susceptibility to noise and limited sensitivity have been two weak points of such hydrophone measurements.
Another acoustic method for leak detection is performed with ultrasonic sensors in direct contact with the specimen under test. When this method is used, acoustic signals created by turbulent leak flow are picked up directly through the solid structure, eliminating the need for propagation through the strongly attenuating seawater.
Although noise from some sources such as ROV thrusters and manipulators may still disturb the measurements intermittently, the direct mechanical contact enables leak detection and continued monitoring with high degrees of sensitivity and repeatability. Only the differential pressure across the valve is needed for quantifying the leak rate.
Permanent or temporary/retrofit leak monitors can be placed at critical locations, either providing real-time leakage data or logging measurement results to internal memory for extended time periods. The ClampOn DSP Leak Monitor is a completely digital, direct contact acoustic leak monitor. All signal processing is done in the sensor itself, securing the best possible signal-to-noise ratio and thus limiting the influence from other sources.
A number of self-contained, battery-operated subsea leak monitors have been delivered for through-valve leakage assessment on two 20-inch valves. Due to the configuration and dimensions of the line, hydrostatic pressure measurement cannot be used for verifying the desired maximum leakage rate. Acceptance criteria was set to 5 SCM/h, similar to a new valve.
Several options were considered for preventing leaks, e.g. injection of cement or gel sealant. As the non-invasive acoustic leak monitoring solution is deemed more favorable with respect to personnel safety, cost-effectiveness, and future utility of the valves, the leak monitors were deployed first to verify whether additional operations were needed.
ClampOn was selected because of experience in manufacturing acoustic instrumentation for topside, subsea and downhole environments. There was no power or communication available so a data logger and subsea battery pack had to be included in the leak monitoring system. Before deploying the sensors on site, the client had to be convinced that the leak monitor would work and quantify the leak in the valve. A laboratory demonstration was performed using the same type of valve present on the cross over line.
[FIGURE 1 OMITTED]
The subsea leak monitor has a dual-sensor configuration to achieve a combination of small signal sensitivity and acoustic measurement accuracy throughout a wide range of signal levels. Similarly, topside monitors featuring both sensor types were used for the demonstration, giving two sensors upstream and two downstream of the valve.
The demonstration was carried out in three main steps, increasing and decreasing the upstream pressure in steps, and enlarging the size of the leak between each step. Damage was inflicted to the upstream valve seat with a hand file. The acoustic leak monitors were logging continuously throughout the demonstration, as well as indicating acoustic signal levels in real time.
Leakage rates of even less than 1SCM/h were positively detected. Figure 1 shows a sample of the measurement results, indicating the leak monitors with "Type 1" acoustic sensors mounted on the upstream and downstream valve ports. These measurements are from the first pressure cycle, i.e. with the smallest damage to the valve. Considering the constraints of the experimental setup, the measurement results were found to be compatible with the statistical data used for detection threshold and leakage rate estimation.
A total of four ClampOn DSP subsea Leak Monitors, each with a dual sensor configuration, was deployed on the two valves. This dual-sensor configuration is employed to increase the amount of available information and to ensure some redundancy during the closing of the valves. The valves are on a crossover between two lines with different pressure rating.
The client needed to increase the export in the flow line and thereby closing the valves in order to separate the two lines. System setup can be seen in figure 2. The installation is at 200 meters water depth and all operations were ROV operable. The battery pack and data logger allow for monitoring for up to 45 days, using a logging rate of four data points per minute. The leak monitors operated in this case for four days before they were retrieved.
The field program started with the deployment of the four sensors, which were already set up in data logging mode. Pressure gauge and bleeding valve on 4-inch blind flange, as seen in figure 2, were then installed before closing the two valves. Pressure was decreased in the 120 bar pipeline and pressure monitored in the crossover, corresponding to testing of the valve named B in figure 2. The pressure in the crossover was then decreased to below pressure in the 120 bar pipeline, before another monitoring run of the crossover pressure. This implies testing of valve A and B sealings. Equipment was then removed from location.
[FIGURE 2 OMITTED]
Results And Discussion
Detection thresholds and through-valve leakage rate estimates are based on a library of several hundred leakage measurements gathered from valves in refineries, chemical plants, and offshore in the late 1980s.
Predictive equations developed from these statistical data convert acoustic readings to leakage rotes, given the upstream and downstream pressures and some parameters of the flowing fluid and the valve size and design. A total of 155 valves in the data base were carrying gas, for which 54% fell within 1/2 to 2 times the leakage rates estimated by the applicable predictive equation.
The predicative equation does not give a 100% correct answer, but rather the probability that a leak is within a certain range. The accuracy achieved (minimum) is a 70% correct measurement that will be within 50-200% of the leak rate. For example, if the leak rate is 1 liter/min, we can say that it is 70% correct that the leak is within 0.5 and 2 liter/min.
As acoustic leak detection is based on turbulent leakage flow, a certain differential pressure is required across the leak for its detection to be possible. Five bar differential pressure is stated as the minimum requirement for the leak monitors used in the present work. Turbulent flow occurs under conditions of sufficiently high differential pressure and sufficiently low kinematic viscosity, their limits being determined through the Reynolds number.
The acoustic signals generated by turbulent leakage flow generally propagate well in metallic structures, and can be detected at distances up to approximately two meters. However, the detected signal levels are distance-dependent, and a distance correction needs to be made to leakage estimates. Leakages of less than 1 SCM/h were detected in the demonstration, but more conservative estimates for the detection threshold must be used for the subsea conditions.
In the case of subsea, welded-in valves, some acoustic energy escapes through the piping and also to the water, reducing the signal levels that reach the acoustic leak sensors.
There are also maximum quantifiable leak rates, determined by the leak monitor's dynamic range, and maximum detectable leak sizes due to the tendency toward laminar flow in cases of large leak cross-sections. It is evident from the empirical data that accurate knowledge of the upstream and downstream pressure is an important prerequisite for obtaining good estimates of leakage rates and detection thresholds. Whenever possible, varying the upstream pressure during acoustic leak monitoring seems a sound approach for increasing the quality of acoustic leakage data.
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
Figure 3 shows the field results during operation. The noise from ambient sources like ROV activity, closing of valves and bleeding off is evident, and this part of the signal cannot be included in the leak calculation. If we take the raw values from the middle part of the signal (hold pressure and bleed off) and remove the ROV and bleeding off gas activity, we get the leak signature, as seen in figure 4.
It is evident that sensor 3, placed on valve B, has the highest raw values at 50,000-60,000. Lowest values are found on sensor 1 and 2, placed at valve A, at 20,000-30,000. The higher raw values at valve B indicate there is a small leak in the valve. The leak rate is calculated from the raw value level indicated in the figure. At 32 bar differential pressure the leakage rate at valve B was found to be in the range of 8-32 SCM/h. Raw values at sensor 1 and 2 indicate that there is no detectable leak on valve A.
As the test proved the leak on valve B to be larger than the acceptance criteria, the operator decided further investigations and operations were required before being able to decide on increasing the export in the flow line. The plan is to deploy the sensors again, using the same sensor setup.
By Geir Instanes and Sindre Halse Kristiansen, ClampOn AS, Norway
|Printer friendly Cite/link Email Feedback|
|Author:||Instanes, Geir; Kristiansen, Sindre Halse|
|Publication:||Pipeline & Gas Journal|
|Date:||May 1, 2010|
|Previous Article:||Outsourcing's growth in the utility industry.|
|Next Article:||Making your MIG gun consumables last: simple ways to save money, improve weld quality and minimize downtime.|