 ·Table of Contents ·Industrial Plants and Structures | Real Time Radioscopy and digital image processing techniques for on line inspection of welds in boiler tubesR. J. Pardikar,Dy. General Manager/NDTL BHEL, Tiruchirappalli-14 Tamil Nadu PIN 620014 INDIA. Contact |
In a boiler industry tubes form a major portion of the steam generating system. Thousands of welded joints are made using automatic and semi-automatic welding machines. In weldments different type of weld flaws occur. These range from porosity to much serious defects like cracks and lack of fusion. Real Time systems using X-rays have been installed at BHEL Tiruchy for on line inspection of butt welds in tubes for boilers welded by MIG welding process on straight tube butt welding machines (STBW). PC based image processing systems are configured by BHEL R & D are interfaced with existing RTR systems at BHEL.
Our experience shows that image processing techniques considerably increased the efficiency in defect detection. This enabled us to substitute Real Time Radioscopy satisfying National and International codes and also increasing the productivity.
Radiography is one of the most versatile NDT method used to identify these types of defect. However, film radiography is a slow and expensive method, particularly for tube inspection where thousands of joints are to be inspected every day. Real Time Radioscopy (RTR) is the alternative to film Radiography with considerable saving in running cost and processing time.
Fig 1: Real time X-ray Imaging system for Boller Tubes.
|
ASME has included real time radiography in Sec. V 1989 and given the guidelines for sensitivity requirements of coded items. In order to meet the sensitivity requirements in par with film radiography it is necessary to interface the RTR system with an image processing system for suitably enhancing the images.
PC based image processing systems configured by BHEL R&D are interfaced with existing RTR systems at BHEL, Tiruchy (Fig.1 shows the configuration of the system).
The straight tube butt welding is a pulsed arc welding process using spray type of metal transfer both at average current levels and low current levels which are very much suitable for out of position welding and thin gauge material welding. In this process the heat for coalescence is produced by an arc between a continuous consumable electrode and the work. Shielding is obtained from a separate externally supplied gas or gas mixture (Argon + CO2 ). The shielding gas protects the weld metal from atmospheric contamination. The materials that can be welded are ferritic steels, stainless steels and their combinations. The gases that are generally used in the welding are Argon, Helium, CO2 or their combinations. The gas flow shall be initiated automatically a few seconds prior to welding and shut off a few seconds after the arc is extinguished.
Each weld shall be subjected to RTR inspection and accepted or rejected based on the following limitations.
Real Time radioscopy is a NDT method in which penetrating radiation (X-rays or g rays) is passed through an object to produce an image on a video monitor and the image is viewed in concurrent irradiation. The arrangement of the source, object and image plane is similar to the conventional film radiography.
X-ray image is converted into a digital image through a long chain, consisting of an X-ray image intensifier, optical lens system between camera and intensifier, video camera and analog to digital converter electronics.
Conventional X-ray image intensifier tube is used to convert the X-ray image to light , then to electrons and back to light on the output screen of the tube. A CCTV camera is focused on the intensifier output screen. The output from the camera is amplified, digitised, stored, enhanced and displayed on TV monitor.
The X-ray image is obviously limited to the image intensifier screen diameter, which is commonly 150 to 400 mm diameter, although larger tubes exists. These type of systems are most widely used with X-ray Kilo voltages in the range of 50 to 300 kV.
X-ray image intensifier tubes have been in continuous development during the past 50 years and tubes are now available with circular input windows up to 400 mm diameters with a primary screen of caesium iodide designed to be efficient with the c-ray energies used in industrial radiography. They tend to loose efficiency at very low kilo voltages, but special thin envelop tubes are available which will operate down to 30 kV. Most X-ray intensifier tubes can provide X2 or X3 electronic magnification, which can increase the image resolution from about 3 lp/mm 5 lp/mm. They have a luminance gain in the range 3000 - 10,000 compared with a standard medical fluoroscopic screen used with 30 kV X-rays, so that the output image is sufficiently bright to be suitable for most type of CCTV cameras. Most equipment use some form of vidicon such as a Plumbicon, Novicon, etc. or a solid state camera. Solid state cameras have a two dimensional light sensitive array with a digital output so that a complete camera consists of only a lens and a sensitive array and no electron scanning beam. However because of simplicity, robustness and reliability of vidicon and solid state cameras are still the most widely used for NDT applications.
The quality of radiographic image can be assessed in terms of three parameters, sharpness, contrast and noise leaving aside minor instrumental problems such as distortion, instability, shading etc.
All radiographic images are unsharp, and the classic way of quantifying this is to measure the unsharpness width of the image of a physically sharp edge. Modulation transfer functions could also be used. In a television fluoroscopic image, there are various causes of unsharpness which need to be measured and combined:
These causes of image unsharpness are inherent in the equipment and in addition to geometric unsharpness due to the focal spot size of the X-ray tube.
As unsharpness is a major factor in radiographic flaw sensitivity, this is, at first sight, a serious limitation to flaw detection capability of an RTR equipment.
One way to reduce the total effective unsharpness is to arrange for the image to be larger than natural size on the primary screen. If the image is M times natural size, the effective total unsharpness is reduced by M, provided of course that in raising M above one.
In film radiography, there is usually a contrast enhancement factor, i.e. the film gradient of about 4 to 6. Television cameras and image intensifier tubes do not increase image contrast, indeed the image conversion contrast is usually slightly less than one, but the use of image processing on a digital image makes the modification of contrast relatively easy. There are a number of contrast enhancement programmes, both for overall enhancement and for local enhancement.
The X-ray image on primary screen is formal by X-ray quanta transmitted through the specimen and absorbed in this screen. The screen usually absorbs only a small proportion of the X-ray quanta incident on it. Even if this primary image is formed from only a few quanta/mm2 /s on the screen, by using intensifiers and television, it can still result in a bright final image because of the amplification process through the system, but at the same time the quantum noise in the primary image will also be amplified. The laws of quantum fluctuations are basic and can not be changed. If there are N quanta/mm2 /s utilised in forming an image, the fluctuation (the noise) is N-1/2 (i.e. N1/2N.100%). If therefore the image is produced from only a few X-ray quanta, the final image will be noisy and the image noise will obscure image detail.
There are three ways of minimising the problem:
In any RTR system, the image is displayed on a television monitor on a line raster. This is completely different nature of image to that on a radiographic film, which usually has a grain structure barely visible without some magnification. Therefore conventional IQI such as wires will not give a complete representation of image quality. In welds, the most serious defect is the small crack, and it is yet not at all clear how well such defect will be imaged on a TV screen, even if shown at several times natural size.
The wire IQI sensitivity is not a satisfactory method of assessing the performance of an RTR system. In particular, it appears possible to have an apparently good IQI sensitivity and at the same time a poor crack sensitivity.
It is therefore recommended that in applications where a high sensitivity is required for planar flaws such as crack, lack of fusion, narrow lack of penetration in welds, that the criteria of satisfactory image quality should be more than a conventional wire or hole IQI sensitivity value. An additional measure of image sharpness is required. This can be provided by the image of a duplex wire IQI such as the type III A in BS 3971:1985.
Theoretical studies of radiographic sensitivity show that wire IQI sensitivity is strongly dependent on contrast parameters, where as crack sensitivity is much more dependent on the image unsharpness which is comparatively large with RTR systems.
Depending on the image quality and flaw sensitivity requirements, a decision must be made on the acceptable total image unsharpness, referred back to the size of the specimen. In case of welds, as cracks might occur, it likely that an unsharpness comparable with that used in film radiography will be needed.
The image processing system converts the analog signal of the sensor in to a digital data stream. The standard TV signal is sampled and converted into a 1 byte deep data word. Typically the images have a size of 512 x 512 = 262144 Pixels, and for a black and white image can be said to have 262144 grey levels. Each grey level representing the intensity corresponding to a pixel. Normally a scale of 0 to 255 is used for the intensity of each pixel, therefore grey level of a pixel will be an integer between 0 to 255. The input signal is digitally integrated over several video frames before further processing.
At BHEL (T) the system employed is a PC based image processor with through memories. The system can accept input either by scanning X-ray film or a radioscopic image system. The processing system can then enhance the image by various means. The signal to noise ratio is improved by image summation or frame averaging. After image summation all succeeding processing techniques serve to improve the presentation of the information contained in the original image.
Most techniques can be performed within few seconds and are used to assist the examiner in the detection of defects.
A typical image processing programme for weld inspection would be:
A digital presentation of the image is provided for archieving.
The purpose of image data analysis is to process a given image so that the result is more suitable than the original image for a specific application. The resulting images are of high contrast and high resolution. The steps involved in image data analysis are:
System Configuration
To avoid time lag between welding and testing, the X-ray real time radioscopy stations are installed in line with the STB welding machine. The tubes after welding are fed to the test station and the results of the weld inspection are conveyed to the welding operator for carrying out necessary corrections in the welding parameters. The handling system for X-ray real time radioscopy receives the tube from welding machine moves the weld joint at the radiation chamber for testing and moves the tube out of the radiation chamber immediately after the testing is completed. The handling system also contains a tube rotating mechanism to segregate good and bad welds after RTR inspection. The handling system is operated from the RTR control room.
X-ray system :
consists of 320 kV constant potential dual focus equipment. Focal size: 3 x 3 mm or 1.2 x 1.2 mm Max. tube current : 10 mA.
| Input screen dia | : | 220 mm |
| Output screen dia | = | 20 mm |
| Resolution | = | 48 line pair/cm |
| Viewing screen | = | P20 (Max. brightness at wavelength = 530 mm). |
Several software programmes have been developed for image improvement and these can be categorised as either point transformations or two dimensional transformations. Point transformations such as grey level manipulations are relatively simple operations. Two dimensional transformation are mathematical procedures that transform an element of the original image into an element of filtered image weighed with its environment. A list of typical software developed for inspection of tubular welds is given below:
Time taken to acquire and enhance the image varies from 20 seconds to 2 minutes, depending on the combination of software used.
The following is the hardware configuration for the image processing
Digital image processing becomes an essential tool for the evaluation of X-ray images in Real Time Radioscopy. Experimental results show that in the absence of image processor the defects are missed in conventional RTR inspection, due to high system unsharpness, noise in the image etc. A reliable defect detection can only be guaranteed if the image input system registers the local contrast sufficiently.
The image processing techniques considerably increased the efficiency in defect detection. The procedures such as averaging and filtering increases the signal to noise ratio and improve the displayed image thereby reducing human errors in interpretation. With the aid of image processor it is possible to obtain high quality images suitable for weld inspection to be comparable with the image qualities required in film radiography. This enable us to substitute Real Time Radioscopy for film radiography satisfying National and International codes and also increasing the productivity.
| © AIPnD , created by NDT.net | |Home| |Top| |