Machine Vision News
Vol. 6, 2000
Previous
Index
Next
 

No freezing of air inlet, no broken glasses, please!
Machine vision makes an attack on old problems

The development of machine vision technology can solve many old problems which are usually handled by means of human vision without any instruments.
 

The air inlet of a gas turbine calls for continuous monitoring

The output power of a gas turbine is regulated by controlling the amount of air flowing into the combustor. The air is fed to the GT through air filters and a compressor. The compressor air inlet (the bellmouth) is equipped with Inlet Guide Vane's (IGV)  functioning as a regulating valve. The IGV's can be turned to suitable positions matching the operating state of the turbine. The air inflow is a typical throttling process, which means that any moisture in air easily condenses and then forms visible ice on the IGV's. The freezing can be observed at temperatures as high as + 5 degrees Centigrade as the phenomenon is dependent of the relative humidity and temperature of compressor inlet air.

The appearance of ice is often a rapid act. As soon as the freezing has been initiated, it may block the bellmouth in a few minutes. It is clearly evident that disturbances in the air intake and bellmouth mean harmful fluctuations of power output and ice formation may lead also to catastrophic consequences for the compressor. In some cases the bellmouth is controlled by an operator through a video monitor. This method consumes much time and is unreliable, especially in cases where several gas turbines are in use. In most cases there are no monitoring facilities available.

CCD Photonics Ltd and Fortum Power and Heat Oy (Vantaa, Finland) have developed an automatic system for monitoring the ice formation at the compressor inlet. The system has a CCD camera installed to look at the air inlet under suitable illumination and a video signal processor for image acquisition. The processor has two image memories: one for a current image and one for a model image obtained at an ice-free moment. The processor makes a continuous comparison between these images and deviations are classified according to their magnitudes. The control algorithm computes a deviation index which can be used as a source of alarm. The user can set the alarm limits. Whenever an alarm condition is met, the multichannel video switches automatically to the appropriate channel to show the state of  the compressor inlet on a video monitor. The alarm signal may also be used to trigger recording of the event.  Says Raimo Mattila from Multitec Finland Oy (Nurmijärvi, Finland) ”We have installed several vision systems at gas turbines and we have found that the turning of  IGV's is the critical moment of operation. The model image will be updated on strict conditions and the heating of air inflow is activated only if necessary – only if there is real freezing on the IGV's”.

A typical case of ice formation is shown in Fig. 1.

Fig.1   The trimming of  the gas turbine camera can be done using a simple PC program. The image at left shows the appearance of the air inlet which has been accepted as a model image. The image at right shows that some freezing is present on four separate IGV blades. The presence of ice is marked using red colour.
 

The optical quality of architectural glass is an important issue

Bent and tempered architectural glass is the construction and facade material of the future and the demand is rapidly growing. New construction applications are springing up around the world and architectural applications  of glass are becoming more sophisticated. Similarly, the size of curved glass surfaces continues to grow, which means that tempered safety glass is needed in larger constructions than before. The applications of cylindrically bent tempered glass are many: wavelike and curved facades, cylindrical curtainwall panels, partition walls inside offices, skylights, staircases, showcases and so on.

The bending and tempering process creates the properties desired for glass, such as strength, using controlled heat processing. Heating the glass to 615 … 650 degrees Centigrade, followed by rapid cooling, creates residual compressive stress on the surface of the glass and a corresponding residual tensile stress in the interior. The resulting stress balance gives glass the desired properties. Tempered, bent glass withstands mechanical strain 4 – 5 times better than annealed glass. This load-bearing ability makes tempered glass suitable for many places that require mechanical strength. Tempered bent glass can also withstand much greater variations in temperature than normal glass. This property makes bent tempered glass more suitable for building facades than laminated glass, for example. Another feature is the increased safety. Tempered glass shatters rather into small pebble-like pieces than larger sharp pieces.

The residual stress state of tempered glass has been traditionally evaluated using photoelastic methods by means of  polarization filters. The basic principle of photoelasticity is the phenomenon of a beam of polarized light passing through a doubly refracting  material wherein it is resolved into mutually perpendicular vibrations which pass through the material, i.e. hardened glass, at different velocities. The amount of phase difference at emergence, when interpreted through a polarizing filter, produces a pattern of light and dark fringes which determine the principal shear stresses through the glass and the maximum tension or compression stress at unloaded surfaces.

CCD Photonics Ltd and Glassrobots Oy (Tampere, Finland) have studied the stress properties of bent, tempered glass using linear and circular polarization methods. Curved glasses are moved between polarization filters under a high resolution vector type CCD camera which records the appearance of  light and dark fringes. The mechanical system is shown in Fig.2. A curved glass moves on rolls after being bent at the temperature of  612 degrees Centigrade. The thickness of the glass is 12 mm, the width is 1000 mm and the ”cylinder” length  is 900 mm. The maximum size of architectural glass prior to bending is 2200 mm * 3500 mm.

Fig.2  Curved tempered glass under stress study at Glassrobots Oy.

An example of dark and light polarization fringes is shown in Fig. 3. The imaging system is capable of distinguishing between various gradients of stress distribution in glass. The leftmost image has been computed using a gradient threshold of  #55 relative units and the middle and rightmost images have been painted using threshold values of  #45 and #35 units, respectively. The imaging system is used to develope the parameter setting properties of the bending/tempering process of  the Rainbowmaker TSF Combi  furnace system manufactured by Glassrobots Oy.

”The goal of our SAFEGLASS project is to develope methods which give an exact measure of  optical quality for architectural glasses”, says Kari Vähä-Antila who is in charge of  technology at Glassrobots Oy.
 

The development work by CCD Photonics was initially funded by  the National Technology Agency (TEKES). The further work is carried out as a part of  the  SAFEGLASS project which belongs to the European Cluster Integrated Machine Vision (EUTIST-IMV). The cluster is a part of the Information Society Technologies Programme (IST) of the Fifth Framework Programme of the European Commission. Other partners of this new Take Up activity are Splintex Glaverbel (Italy) and Satakunta Polytechnic (Pori, Finland).

Fig. 3 Example of stress patterns in bent, tempered architectural glass.
 

Spin-off company

CCD Photonics Ltd  is a spin-off company from Technical Research Centre of Finland
(VTT), founded in 1990 by Dr. Kimmo Simomaa, one of the founding members of  the Vision Club of Finland. The company welcomes any challenging development tasks where modern machine vision methods should be taken into use in old or new designs.
 
 

Contact:

Dr. Kimmo Simomaa

CCD Photonics Ltd  (CCD-Fotoniikka Oy)
Tapiontie 20
02720 Espoo
FIN-02720 Espoo
Phone +358-9-5093417
FAX +358-9-5093417
Mobile +358-50-5260122
email kimmo.simomaa@kolumbus.fi

Previous
Index
Next