Pipeline Inspection

Early Pipeline Inspection Systems

The next step was internal inspection of natural gas pipelines. A novel system based on line scan cameras and laser illumination in a self contained system was designed. Image data was stored on hard disks and the complete system was powered by Li batteries. This system was mounted on cleaning pig and was named Simplisit. In addition to images customers wanted 3D information as well: What is the depth of that cavity? was typical questions. This lead to the current pipeline inspection system Optopig.

Optopig

Principle of operation

This figure shows a 3D view of the profiling concept. You can see the fan shaped laser beam illuminate the pipe wall. Both an obstacle and a cavity are present. The image formed on the image sensor represents the intersection line between the laser beam and the pipe wall.
In its simplest form the geometry of the pipe can be measured by "counting" pixels in each column of the image sensor which consist of N X M pixels.
Let us assume that we want to find the profile value in column n. We will then check the intensity of all pixels in this column and store the position where we found the maximum intensity. By repeating this process for all N columns we will find the position of all parts of the laser line illuminating the pipe wall.
If we in addition to store the position where we found the maximum intensity, also store the gray level (i.e. intensity) at the position of the maximum value, we will get an image line along the laser line. By moving the laser/camera unit relatively to the pipe, and making a new exposure for (let's say) each millimeter, we can build an image of the pipe by putting these lines together. This is very similar to line scan cameras, scanners and fax machines where the object is scanned line by line.
By putting the profile lines together in a similar fashion, we also get a continuous profile "map" of the pipe wall.

This figure illustrates the principle of optical triangulation. A fan shaped laser beam perpendicular to the pipe wall illuminates a thin stripe on the wall. This stripe is depicted by the the camera and an image is formed on the sensor. Since the angles and distances are constant, the image on the sensor will move depending on the distance between the laser/camera unit and the pipe wall.
The depth of a cavity can be determined by measuring the position of the depicted line on the sensor. A total of 8 cameras are used and by combining those 8 cameras, ovalities of the pipe can be detected and measured.

This figure gives a coarse schematic overview of the Optopic inspection tool and its different parts. The system is based on custom, high speed cameras, storage media, lasers, batteries and complex, custom electronics.
[World wide patent applications.]

The Optopig Hardware

A peek into the hidden inner parts of the 10"-16" Optopig showing the main camera module. In a ring we can see the 8 CMOS image sensors packed in custom chip scale packages. Each sensor has a programmable logic chip that controls the acquisition and processing of images at 500MB/s.

A close-up of the camera system for the 40"-42" Optopig. The right part of the image shows the ring of 8 lasers, while the left part shows the ring of 8 cameras, looking forward at an angle of 45 degrees.

Departure of the 40"-42" Optopig. The compact inspection unit is mounted onto a rugged carrier and inserted into the pig trap. The carrier is wheel based to allow the best possible centring of the camera unit. The sealing disks provides the necessary drive force. This version of Optopig can run at velocities up to 5m/s.

Results from the Use of Optopig

Who wouldn't feel some need for celebration when time finally has come to paint the number of 1000 onto a pipe. The chances of getting "caught" was extremely low before the arrival of Optopig.

2D image of a normal field joint area. To the left we can se the vertical welding seam, with uncoated, dark, pipe surface both before and after. Running through the image horizontally we can see the longitudinal weld. This weld is done before the coating, so we can see that it is covered by coating. We can also see that the weld is ground down towards the field joint.

3D close-up of a small part of a ball valve that is not completely open. The right part of the image shows the glossy inner of the steel ball. The step we can see was measured to 10mm.

2D image of a machined defect in a test pipe. Around the defect we can see some text that tells us that this is a 2.5cm wide and 10% deep defect.

A 3D close-up of the defect. The dark colour indicates that the coating is missing, and possibly that the surface is quite rough.

In a 3D wire frame plot we can clearly see the shape of the defect. By using measurement tools we can determine the size, and depth of the defect.

The Optopig Software

More details on the Optopig system can be found on these pages : optopig.com.

Automatic Anomaly Detection in Pipeline Data