Final Report Summary - FPSO-INSPECT (Non-Intrusive In-service Inspection Robotic System for Condition Monitoring of Welds inside Floating Production Storage and Offloading (FPSO) Vessels)
The main objective of the FPSO-INSPECT project, was to build a prototype amphibious robot that could carry non-destructive testing (NDT) sensors from an entry port in the top of an FPSO vessel tank, where the NDT sensors can be deployed from a scanner to detect either fatigue cracks in the stiffener to tank shell fillet welds, or corrosion in the shell plates.
At the request of the end-users, it was decided to build a prototype robot vehicle that operated in-air on the floor of the FPSO tank once it had been emptied, but left with sediment that covers the surface.
FPSO vessels are increasingly being used for production and storage of oil from offshore fields. A typical FPSO contains 20 km of internal safety critical welds that require detailed offshore inspection on a five-year cycle. These welds are prone to fatigue cracking due to the drastic increase in loading.
Most common methods of inspection of these welds have major drawbacks, as they require the FPSOs to be dry docked, emptied and cleaned with consequent disruption to production, which means that 90 % of the costs of inspection are associated with the disruption of production and emptying and cleaning the vessel. Also, the inspections are mainly visual and manual and therefore subjective with no hardcopy results.
The amphibious vehicle would have to perform in oil, since according to the end-users it was not the preferred practice in their FPSO vessels to empty the tanks and fill them with water. Therefore, the main objective of the project to the development of an inspection robot that could be used in-the-dry, but which demonstrated its amphibious capability in water was amended.
The main project objectives are the following:
-to demonstrate a robot vehicle with sensors for detecting damage from corrosion and fatigue in the floor plates and stiffeners of FPSO tanks;
- to build an X-Y-Z scanner for the sensors that is mounted on the robot vehicle;
- to develop a positioning and guidance system for the robot;
- to develop sensors and sensor systems for detecting fatigue cracks in the toes of stiffener to floor fillet welds;
- to develop sensors and sensor systems for detecting corrosion on either the internal or external surfaces of the floor plate;
- to develop a man-machine interface that combines robot control with data gathering from the sensors.
The principal aims of the NDT were firstly to detect fatigue cracks emanating from the toes of fillet welds that join stiffeners to the floor of the FPSO tank and secondly to detect corrosion in the tank floors. The sensitivity requirements set by BP and Petronas were 150 mm long for cracks and 10 % loss of wall for corrosion.
With the development of any NDT technique, it was necessary to determine which factors affected the probability of detection and the sizing accuracy. Of particular importance were fillet weld tests pieces into which fatigue cracks were induced. A special method of creating fatigue cracks with a fatigue testing machine at selected positions along the weld toe was developed.
For larger scale trials at the end of the project BP made available a mock-up of a section of FPSO tank.
Five NDT techniques were investigated: phased array, creep wave and plate wave ultrasonics, pulsed eddy-current and ACFM. The phased array technique, although it was able to detect fatigue cracks, as well as other flaws, such as lack of penetration, lack of side wall fusion and porosity, it suffered from the following problems:
Within the sweeping transverse wave scan it was difficult to distinguish cracks signals from pronounced weld cap signals that existed in some of the fillet welds. The test rate was too slow. The sensor was too heavy.
Creep waves are very short range, because a transverse wave is continually leaked to the far surface. Moreover, the transverse wave is always present, giving rise tp multiple reflections that clutter the A-scans with signals.
The important distinction of plate waves from creep waves is that the plate waves can propagate over long distances (> 2m). Also, they are reflected wherever the charge in wall thickness creates a change in acoustic impedance that is sufficient to reflect some of the wave energy. Plate wave would therefore reflect from weld caps (increase in wall thickness) as well as from corrosion (decrease in wall thickness) on either surface of the plate.
The generated eddy-currents contain a range of frequencies, which, because of skin-effect, are able to penetrate to a range of depths in a conductive material. The higher frequency components were restricted to the surface, while the lower frequency components to depths of perhaps several millimeters.
The technique could be made sensitive to loss o metal caused by corrosion in the test plate. A prototype system proved successful in detecting machined holes through thick non-conductive coatings. However, the signal amplitude was proportional to volume loss and not the depth of the corrosion, which would make difficult the detection of large areas of corrosion. The technique was also very slow and it was abandoned in favour of the plate wave ultrasonic technique for detecting corrosion.
The alternating current field measurement (ACFM) technique is a variant of the alternating current potential drop (ACPD) technique used in the laboratory for monitoring fatigue crack growth in mechanical tests. ACPD relies on measuring the increase in resistance of an AC on the surface as it is deflected around a growing surface breaking crack. By using an alternating magnetic field to induce the current in the test surface and magnetic field sensors to measure the strength of any deflection around a crack, ACFM is a non-contact method of detecting surface breaking cracks and measuring their depth. To obtain all the data from a surface crack, ACFM measures the Bx and Bz components of the magnetic field from the scanning sensor.
Each NDT system consists of a sensor, a device for generating the electro-magnetic field or ultrasound pulse and a graphical user interface (GUI) on a lap-top computer for controlling the sensor and displaying results.
As far as the design of the robot vehicle and scanner is concerned, the original proposal was to build a swimming robot vehicle that could swim down in water to the stiffeners at bottom of the tank. Although BP and Petrobras insisted that the robot would have to swim in oil, as it was not their practice to fill the FPSO-tanks with water.
It was decided to press ahead with a swimming robot that would be demonstrated in air and water with only a concept design for operating in oil. The stringent safety requirements when operating in oil would be impossible to meet within the scope of the project.
The scanner movement would work in conjunction with vehicle movement so that the sensor holder would start from the corner between the stiffener and tank-wall, scan one length and stop while the vehicle moved along to a new point to start the next scan. At the end of the stiffener, the vehicle would rotate through a right angle to bring the scanner arm across the weld at the end of the stiffener.
The robot vehicle was designed to swim with a payload of the scanner and sensor to the floor of the tank, where its wheels would allow complete maneuverability inside the spaces between the stiffeners.
In order to control the buoyancy of the robot while swimming, two methods were investigated: constant volume and constant mass. Despite advantages when used in oil (the liquid does not have to be sucked in or blown out), the constant mass type was less stable and more commonly used constant volume type buoyancy control was incorporated with the vehicle.
Other features on the vehicle were four independently powered and steerable wheels, two thrusters and ultrasonic proximity sensorsset around the vehicle for aligning it with the stiffener during scanning.
In order to simulate the global path, a specific design of FPSO tank was selected for the simulation exercise. To guide FPSO-Inspect, a scanning ultrasound sonar would create a sweeping 120 degrees beam under the vehicle. The global path plan would take FPSO-Insect from the entry port in the roof of the tank, follow the ladder down to the floor, then use the tank side to take the robot vehicle to a corner with a bulkhead, where the vehicle could turn though a right angle to cross the tank, following the bulkhead to the other side where the stiffeners to be inspected were lined up.
The man-machine interface integrated controls for the vehicle, scanner and three NDT sensor systems.
At the request of the end-users, it was decided to build a prototype robot vehicle that operated in-air on the floor of the FPSO tank once it had been emptied, but left with sediment that covers the surface.
FPSO vessels are increasingly being used for production and storage of oil from offshore fields. A typical FPSO contains 20 km of internal safety critical welds that require detailed offshore inspection on a five-year cycle. These welds are prone to fatigue cracking due to the drastic increase in loading.
Most common methods of inspection of these welds have major drawbacks, as they require the FPSOs to be dry docked, emptied and cleaned with consequent disruption to production, which means that 90 % of the costs of inspection are associated with the disruption of production and emptying and cleaning the vessel. Also, the inspections are mainly visual and manual and therefore subjective with no hardcopy results.
The amphibious vehicle would have to perform in oil, since according to the end-users it was not the preferred practice in their FPSO vessels to empty the tanks and fill them with water. Therefore, the main objective of the project to the development of an inspection robot that could be used in-the-dry, but which demonstrated its amphibious capability in water was amended.
The main project objectives are the following:
-to demonstrate a robot vehicle with sensors for detecting damage from corrosion and fatigue in the floor plates and stiffeners of FPSO tanks;
- to build an X-Y-Z scanner for the sensors that is mounted on the robot vehicle;
- to develop a positioning and guidance system for the robot;
- to develop sensors and sensor systems for detecting fatigue cracks in the toes of stiffener to floor fillet welds;
- to develop sensors and sensor systems for detecting corrosion on either the internal or external surfaces of the floor plate;
- to develop a man-machine interface that combines robot control with data gathering from the sensors.
The principal aims of the NDT were firstly to detect fatigue cracks emanating from the toes of fillet welds that join stiffeners to the floor of the FPSO tank and secondly to detect corrosion in the tank floors. The sensitivity requirements set by BP and Petronas were 150 mm long for cracks and 10 % loss of wall for corrosion.
With the development of any NDT technique, it was necessary to determine which factors affected the probability of detection and the sizing accuracy. Of particular importance were fillet weld tests pieces into which fatigue cracks were induced. A special method of creating fatigue cracks with a fatigue testing machine at selected positions along the weld toe was developed.
For larger scale trials at the end of the project BP made available a mock-up of a section of FPSO tank.
Five NDT techniques were investigated: phased array, creep wave and plate wave ultrasonics, pulsed eddy-current and ACFM. The phased array technique, although it was able to detect fatigue cracks, as well as other flaws, such as lack of penetration, lack of side wall fusion and porosity, it suffered from the following problems:
Within the sweeping transverse wave scan it was difficult to distinguish cracks signals from pronounced weld cap signals that existed in some of the fillet welds. The test rate was too slow. The sensor was too heavy.
Creep waves are very short range, because a transverse wave is continually leaked to the far surface. Moreover, the transverse wave is always present, giving rise tp multiple reflections that clutter the A-scans with signals.
The important distinction of plate waves from creep waves is that the plate waves can propagate over long distances (> 2m). Also, they are reflected wherever the charge in wall thickness creates a change in acoustic impedance that is sufficient to reflect some of the wave energy. Plate wave would therefore reflect from weld caps (increase in wall thickness) as well as from corrosion (decrease in wall thickness) on either surface of the plate.
The generated eddy-currents contain a range of frequencies, which, because of skin-effect, are able to penetrate to a range of depths in a conductive material. The higher frequency components were restricted to the surface, while the lower frequency components to depths of perhaps several millimeters.
The technique could be made sensitive to loss o metal caused by corrosion in the test plate. A prototype system proved successful in detecting machined holes through thick non-conductive coatings. However, the signal amplitude was proportional to volume loss and not the depth of the corrosion, which would make difficult the detection of large areas of corrosion. The technique was also very slow and it was abandoned in favour of the plate wave ultrasonic technique for detecting corrosion.
The alternating current field measurement (ACFM) technique is a variant of the alternating current potential drop (ACPD) technique used in the laboratory for monitoring fatigue crack growth in mechanical tests. ACPD relies on measuring the increase in resistance of an AC on the surface as it is deflected around a growing surface breaking crack. By using an alternating magnetic field to induce the current in the test surface and magnetic field sensors to measure the strength of any deflection around a crack, ACFM is a non-contact method of detecting surface breaking cracks and measuring their depth. To obtain all the data from a surface crack, ACFM measures the Bx and Bz components of the magnetic field from the scanning sensor.
Each NDT system consists of a sensor, a device for generating the electro-magnetic field or ultrasound pulse and a graphical user interface (GUI) on a lap-top computer for controlling the sensor and displaying results.
As far as the design of the robot vehicle and scanner is concerned, the original proposal was to build a swimming robot vehicle that could swim down in water to the stiffeners at bottom of the tank. Although BP and Petrobras insisted that the robot would have to swim in oil, as it was not their practice to fill the FPSO-tanks with water.
It was decided to press ahead with a swimming robot that would be demonstrated in air and water with only a concept design for operating in oil. The stringent safety requirements when operating in oil would be impossible to meet within the scope of the project.
The scanner movement would work in conjunction with vehicle movement so that the sensor holder would start from the corner between the stiffener and tank-wall, scan one length and stop while the vehicle moved along to a new point to start the next scan. At the end of the stiffener, the vehicle would rotate through a right angle to bring the scanner arm across the weld at the end of the stiffener.
The robot vehicle was designed to swim with a payload of the scanner and sensor to the floor of the tank, where its wheels would allow complete maneuverability inside the spaces between the stiffeners.
In order to control the buoyancy of the robot while swimming, two methods were investigated: constant volume and constant mass. Despite advantages when used in oil (the liquid does not have to be sucked in or blown out), the constant mass type was less stable and more commonly used constant volume type buoyancy control was incorporated with the vehicle.
Other features on the vehicle were four independently powered and steerable wheels, two thrusters and ultrasonic proximity sensorsset around the vehicle for aligning it with the stiffener during scanning.
In order to simulate the global path, a specific design of FPSO tank was selected for the simulation exercise. To guide FPSO-Inspect, a scanning ultrasound sonar would create a sweeping 120 degrees beam under the vehicle. The global path plan would take FPSO-Insect from the entry port in the roof of the tank, follow the ladder down to the floor, then use the tank side to take the robot vehicle to a corner with a bulkhead, where the vehicle could turn though a right angle to cross the tank, following the bulkhead to the other side where the stiffeners to be inspected were lined up.
The man-machine interface integrated controls for the vehicle, scanner and three NDT sensor systems.
