st suitable specifications for the piezoelectric element.

It appears that the particle detection is a function of the power of the waves rather than its frequency, with a relation between the size of the detected particle and the power of the wave in logarithm:

where A and B are two constants related to the frequency and the characteristics of the emitting system, and p the size of the detectable particle [22] .

Using high frequency allows for a more focused area of detection and a better discrimination, but it also increases the loss of power inside the melt [23] . Having a more focused area of detection prevents getting the knowledge on a bigger volume of melt. In the past work of Mountford and Sommerville, different particles with known size and shape were aimed to detect by ultrasonic waves. A frequency of 10 MHz was used but the last experiments usually had a frequency of 2.25 MHz. The number of the particles inside molten aluminium was measured for 30 seconds at power level between 76 and 88 dB, under stirred and non-stirred conditions. The test set-up was able to detect particles down to 10 - 15 µm. The results showed promising data both in water and molten aluminium [22] . It should be noted that using lower frequency will allow detecting the concentration of particles in a bigger melt volume, an advantage over other techniques such as LiMCA, and thus improving the knowledge of the cleanliness of the whole melt.

The measurement procedure during the detection can be summarized as following: sending pulses in the melt, usually at a rate of 100 pulses per second at a given frequency, the amplitude is slowly increased until a reflected signal is received. This first signal gives the size of the largest particle in the melt. This means that, to be able to detect smaller particles, the sound system must be able to emit at a high power level, usually around 90 dB [22] . This should be taken into account while building the sound system.

5.1.2. Coupling of the Head and the Buffer Rod

A main issue of the ultrasonic detection is the temperature range at which transducers can be operated. A piezoelectric element cannot be directly inserted in the melt to emit, so the use of a buffer rod is mandatory. One side of the rod is attached to the emitting element and the other side is inserted into the melt. Without cooling, the transducer can reach more than 160˚C due to conduction of the heat in the rod, which is beyond the working temperature of such transducers, thus air-cooling is usually used [3] [9] . A thin layer of thermally conductive material can be placed between the rod and the transducer as seen in Figure 5 to prevent overheating [24] .

5.1.3. Loss in Sensitivity Due to Vibration Noises

The sensitivity of the device is directly related to the Signal-to-Noise Ratio (SNR). It is defined as the ratio between the strength of the desired signal at the end of the probe and the strength of the noise produced inside. This noise is generated due to the mode of vibration and diffraction. It has been shown that the detectable size of the particles depends on the power of the waves rather than the frequency; therefore a higher ratio means better sensitivity of the device as more energy is transferred inside the melt [3] [25] .

The maximal power delivered by the piezoelectric element is of primary importance, but studies also show that using a cladding [3] [25] on the rod, and/or using a tapered rod [3] , improves the signal guidance and increase the SNR. The cladding also prevents the dissipation of the wave power on the sides of the rods, and thus allows more power transmitted to the melt.

The signal to noise ratio is determinant in the ability of the device to differentiate between different particles when the refracted signal arrive at the same time. It is a very important specification of the rod that has to be taken into account.

5.2. Probe & Buffer Rods Definition

The buffer rod has a primordial importance on the transmission of the ultrasonic waves between the piezoelectric element and the melt. Its shape, material and coating have influences on the sensitivity of the system.

The design has to be made between two possible configurations: the “Pulse Echo” configuration, which uses a single rod as both transmitter [26] and receiver, and the “Pitch Catch” configuration, in which two rods are used, one as transmitter, one as receiver, as seen in Figure 6.

The Pulse Echo configuration has the advantage of being cheaper and easier to set up, but suffers from higher noise and echoes inside the rod, which decreases the quality of the detection due to the delay line. The Pitch Catch configuration has a better SNR and allows focusing the detection in one particular area by changing the relative position of the rods, but the set-up is more expensive [3] [27] . Despite that increase in cost, the most commonly used configuration is Pitch Catch, as is also allows for a wider area of detection and a better sensitivity.

5.2.1. Dimensions of the Buffer Rod

Buffer rods are typically between 200 mm and 300 mm in length [3] with a diameter around 15 mm [27] . This allows for the piezoelectric element to be far enough from the melt and for the rod to be submerged to a depth of around 15 mm, which has been seen sufficient to prevent it from going out of the melt, even with surface level variations due to melt movements [22] .

Using a tapered rod is a good solution to prevent echoing inside the buffer rod. Usually an angle around 1.5˚ is enough to see considerable improvement in the SNR [3] . Double-tapered rods were designed as shown in

Figure 5. Schematic of a piezoelectric element coupled to a buffer rod.

Figure 6. Illustration of the pitch-catch principle (left) and the echo-pulse principle (right).

Figure 7 to prevent the diameter difference between the start and the end of the rod, allowing both ends to have a diameter around 15mm and using an angle up to 2˚ [27] [28] .

A focusing lens can be used at the end of the rod to increase the power transmitted to a particular area in the melt [28] as shown in Figure 8. This technique allows for the detection of smaller inclusions [9] but reduce the volume of detection in the melt. As the purpose of the ultrasonic technique is to get a wider and faster knowledge of the particle concentration of the melt, using flat ends for the rods is recommended.

5.2.2. Material Type of the Rod

Material type is especially important from the metallurgical point of view due to aggressive chemistry of aluminium which is not easily combined with many materials. The choice of material for the rod is of primary importance for the life of the probe, as well as for its effectiveness. A good material should meet the following requirements: good wetting ability by the liquid metal under ultrasonic waves, good acoustic conductivity, good resistance to corrosion, good resistance to thermal shocks, low thermal conductivity and high melting point. The durability of the material can also be improved by using coatings and cooling [13] [29] .

Steel rods are mainly used [22] as they provide sufficient acoustic conductivity, but they suffer from corrosion. Good results have also been achieved using Niobium alloys [29] . Other metals can be used, such as silica, tin, alumina or aluminium alloys, but they are not very common [3] [29] .

A stainless steel cladding [27] is almost always applied as although it is not the best choice because of aggressivity of aluminium melts. The cladding is applied by thermal spraying. Ceramic cladding was also used in some works [2] , but its use is less common. Using a cladding protects the sides of the rod against corrosion.

5.2.3. Positioning of the Buffer Rods in the Set-Up

Two types of positioning of the rods are possible: parallel or tilted toward another (V-shape) as shown in Figure 9.

The V-shape setup creates for more focused signals in the melt, thus increasing its sensitivity. The receiving rod receives a higher percentage of the reflected waves and particles down to 10 µm can be detected [22] . However, the detection volume is smaller and the system loses the global knowledge as it gains higher sensitivity.

The parallel shape works on the same principle but the waves are transmitted vertically in the melt, and the receiving rod gets a smaller part of the reflected waves. This allows for a bigger volume in the melt to be analysed, but the amplitude of the initial waves has to be higher to be able to detect smaller particles.

It should be noted that one type of set-up might be preferable depending on the aim of the analysis. If the aim is to detect the smallest particle in the melt, using a V-shape with focusing lens at the end of the rod might work better. If the aim is to get a global knowledge of the particle size and concentration in the melt, using a parallel shape system is recommended.

There has been no comparative research done about the influence of the distance between the two rods on the sensitivity of the system. It should be investigated for the further works either.

Figure 7. Double tapered rod [27] .

Figure 8. Focusing lens at the end of the rod [27] .

Figure 9. Example of a parallel and V-shaped shape set-up [28] .

5.2.4. Contact Problems between Rods and the Melt

The set-up for detection suffers from a few recurring problems at the interface between the rod and the melt. Some corrosion may occur and a cleaning can be needed to increase the lifespan of the rod.

First problem is the coating around the steel rod does not protect the tip from the corrosion. A typical AlFe3 alloy deposit has been found on the tip of the rod in the work of Mountford after a long period of continuous detection [22] .

The second problem, which can be faced in both laboratory and industrial utilizations, comes from the tendency of particle to gather at the tip of the rod [30] . The particles decrease the transmission of the waves by covering the tip and this impacts the efficiency of the detection [3] . A mechanical scratching of the tip is usually performed; either automated or manual solutions have been found [23] . An automated scratching is of premium importance, as it can be done more frequently without human intervention.

5.3. Particle and Melt Movement Effects on Detection Efficiency

Interactions between melt and particles must be also well understood in terms of turbulence, settling and agglomeration. These phenomena affect the detectable particle size range, detectable volume and the reliability of the detection results.

5.3.1. Settling Effect on Detectable Particle Size Range

The modulation of the amplitude of the received signal gives an indication of the site, and the first particles detected are the biggest ones [23] . But some problems may occur due to different settling velocities of particles. The big particles settle rapidly and they move out of the detection zone. Their accumulation on the bottom of the furnace creates noise and impact the detection procedure. This problem can be solved by stirring the melt which carries the settled particles again in the detectable volume, but the stirring should not occur during the detection, rather just before, as stirring can artificially increase the number of detected particles [22] .

In the work of Mountford, pure aluminium at about 730˚C was used for the ultrasonic detection and alumina particles were added and detected for 30 seconds with waves of 2.25 MHz at 72 dB to observe the detecting process. By addition of alumina with known concentration, the number of the detected particles was expected to increase linearly however it was not observed as it is seen in Figure 10. These irregularities in the results could have been caused by the irregular shape of Al2O3 particles [22] . This result brings the question if the detectable size also might be influenced by the shape of the particles.

However, the shape factor has also a direct influence on settling velocity of particles which can cause different problems during the detection [31] [32] . The shape of the particles could have been a reason for unpromising results but more research is needed in this topic.

Figure 10. Variation of the detected particle concentration with added alumina concentration.

5.3.2. Melt Flow Effect on Detection

Ultrasonic detection can potentially be used in launder, holding furnaces and crucible furnaces. Slower melt velocities such as in crucible or holding furnaces make the calibration of over counting particles easier. However, it has been reported by M. Badowski et al. that the melt velocity can disturb the particle movement even in crucible furnaces [32] . But high turbulence such as in the case of launders can make the calibration more complicated but detection can be still performed.

Stirring also affects the melt flow and the detection characteristics since it prevents the settling and keeps the big particles in the flow [22] .

5.3.3. Particle Concentration Effect on Detection

The use of water model is a good way to test theory on the minimal detectable size of the particles. The usual minimal size is around 20 µm [22] , which is also the minimal detectable size by MetalVision. Detection of the particles down to 10 µm is theoretically possible by focusing the zone of detection, using a V-shape system. However, there are some other influences which can contradict the theoretical predictions.

The sensibility of the detection is affected by the concentration of particles in the melt. For high inclusion concentrations, the particle count is often less than the real because some of the small particles are hidden in the “shadow” of the bigger particles. A big percentage of particles are not reachable by the reflected ultrasonic waves [23] . This problem occurs also in lower concentrations but the influence is statistically not important. Stirring the melt is a solution for this, as it will make the particle move out the shadow of big particles, and it is possible to correct the count by calculating the average number of detected particles [22] . But this solution also creates new problems due to a melt movement. In industrial applications, this problem usually doesn’t occur since the particle concentration is very low.

6. Conclusions

The previous research in the years 1985 to 2015 on ultrasonic particle detection in aluminium melts can be summed up as follows:

・ An on-line monitoring of inclusions would allow for faster and more efficient processes.

・ The physics of ultrasonic is well-known and allows creating and arranging models. Existing water models can be used for the first tests to validate the setup.

・ MetalVision is expensive equipment, which has limits when the contrast between concentration of bigger sized particles and smaller sized particles is too high. This could be improved by developing a new on-line system.

・ A set-up for detection by ultrasonic waves might be described in 4 parts: the piezoelectric transducer in charge of producing the waves, the rods which are transmitting the waves to the melt, the interface between the rods and the melt, and finally the interface between the waves inside the melt and the particles.

・ The detection of inclusions in aluminium melts by ultrasonic waves seems promising, and should be experimented more on an industrial level to see whether the change of scale has an influence on the results.

7. Future Prospective

・ This work is a review of different problematics and possibilities associated with the particle detection by ultrasonic waves. Although it is working both theoretically and on a laboratory scale, more tests are surely needed.

・ Future works should focus on combining different futures of ultrasonic detection such as detection of the size of the particles, their position in the melt and the concentration of the melt. The question still stands if it is possible to do all of these together.

・ Another area for the future research would be improving the rod. Finding a compromise between good conductivity and corrosion resistance is very important which would make the detection system cheaper than the current solutions.

Acknowledgements

The research leading to these results has been carried out within the framework of the AMAP (Advanced Metals and Processes) research cluster at RWTH Aachen University, Germany.

Cite this paper

Mertol Gökelma,Damien Latacz,Bernd Friedrich, (2016) A Review on Prerequisites of a Set-Up for Particle Detection by Ultrasonic Waves in Aluminium Melts. Open Journal of Metal,06,13-24. doi: 10.4236/ojmetal.2016.61002

References

  1. 1. Dispinar, D. (2005) Determination of Metal Quality of Aluminium and Its Alloys. University of Birmingham, Birmingham.

  2. 2. Kurban M., Sommerville I.D., Mountford N.D.G. and P.H. Mountford (2005) An Ultrasonic Sensor for the Continuous Online Monitoring of the Cleanliness of Liquid Aluminium. Light Metal 2005, TMS San Francisco February 2005, 945-949.

  3. 3. Poynton, S., Brandt, M. and Grandfield, J. (2009) A Review of Inclusion Detection Methods in Molten Aluminium. In: Light Metals, Vol. 3, John Wiley & Sons, Hoboken, 681-687.

  4. 4. Currie, A.J., Millman, S. and Johnstone, S. (2009) An Ultrasonic Liquid Steel Monitor: Measurement Principle and Probe Design. AIP Conference Proceedings, 1096, 1735-1742.
    http://dx.doi.org/10.1063/1.3114168

  5. 5. ABB—Brochure for PoDFA.

  6. 6. ABB—Brochure for LiMCA.

  7. 7. Smith, D.D., Hixson, B., Mountford, H. and Sommerville, I. (2015) Practical Use of the Metal Vision Ultrasonic Inclusion Analyzer. TMS Orlando.

  8. 8. G&oumlkelma, M., Morscheiser, J., Badowski, M., Dang, T., Le Brun, P. and Tewes, S. (2015) Observation on Inclusion Settling by LiMCA and PoDFA Analysis in Aluminium Melts. International Aluminium Journal, 91, 56-61

  9. 9. Ihara, I., Aso, H. and Burhan, D. (2004) In-Situ Observation of Alumina Particles in Molten Aluminum Using a Focused Ultrasonic Sensor. JSME International Journal Series A, Solid Mechanics and Material Engineering, 47, 280-286

  10. 10. Friedrich, B. and Kr&aumlutlein, C. (2006) Melt Treatment of Copper and Aluminium—The Complex Step Before Casting. IME Process Metallurgy and Metals Recycling, RWTH Aachen, Aachen.

  11. 11. Bao, S. (2011) PhD Thesis “Filtration of Aluminium—Experiments, Wetting, and Modelling”. Norwegian University of Science and Technology, Trondheim.

  12. 12. Mansfield, T.L. (1982) Ultrasonic Technology for Measuring Molten Aluminium Quality. Journal of Metals, 34, 54-57.

  13. 13. Eskin, G.I. and Eskin, D.G. (2015) Ultrasonic Treatment of Light Alloys Melt. CRC Press, Boca Raton.

  14. 14. Auld, B.A. (1990) Acoustic Fields and Waves in Solids. Vol. I & II, 2nd Edition, Krieger Publishing Company, Malabar.

  15. 15. Lecture Notes of the University of Washington. Department of Bioengineering. Seattle.

  16. 16. Metal Vision Manufacturing Canada Ltd., Brochure.

  17. 17. Neff, D.V. (2004) Evaluating Molten Metal Cleanliness for Producing High Integrity Aluminium Die Castings. Die Casting Engineer, 48, 24-30.

  18. 18. Guo, J.F. (2007) The Use of Ultrasonic Techniques in Measuring Inclusions in Al-Si Alloys Melts. Université du Quebec à Chicoutimi, Chicoutimi.

  19. 19. Vadoothker, A.R. (2001) Ultrasound-Assisted filtration System. Patent US6221255 B1.

  20. 20. Eckert, C.E. (1998) Apparatus and Method for Ultrasonic Particle Detection in Molten Metal. Aluminum Company of America, United States Patent US4563895 A.

  21. 21. Puga, H., Barbosa, J., Gabriel, J., Seabra, E., Ribeiro, S. and Prokic, M. (2011) Evaluation of Ultrasonic Aluminium Degassing by Piezoelectric Sensor. Journal of Materials Processing Technology, 211, 1026-1033.
    http://dx.doi.org/10.1016/j.jmatprotec.2011.01.003

  22. 22. Mountford, N.D.G., Simionescu, A. and Sommerville, I.D. (1997) Visualization of the Particle Content of Liquid Aluminium Alloys. Light Metals, 937-943.

  23. 23. Sommerville, I.D., Mountford, N.D.G. and Martins, L.C.B. (2000) Laboratory and Industrial Validation of an Ultrasonic Sensor for Cleanliness Measurement in Liquid Metals. Light Metals, 721-726.

  24. 24. Spicci, L. and Vigna, G. (2014) Heat Drain Device for Ultrasound Imaging Probes. Proceedings of the 2014 COMSOL Conference in Cambridge, Cambridge, 17-19 September 2014.

  25. 25. Jen, C.K., Legoux, J.G. and Parent, L. (2000) Experimental Evaluation of Clad Metallic Buffer Rods for High Temperature Ultrasonic Measurements. NDT & E International, 33, 145-153.
    http://dx.doi.org/10.1016/S0963-8695(99)00042-0

  26. 26. Ono, Y., Moisan, J.F., Zhang, Y., Jen, C.K. and Su, C.Y. (2004) An On-Line Ultrasonic Cleanliness Analyzer for Molten Light Metals. JOM, 56, 59-64.
    http://dx.doi.org/10.1007/s11837-004-0148-9

  27. 27. Ono, Y., Moisan, J. F., Jen, C.K. and Fran&ccedila, D.R. (2002) Development of Ultrasonic Techniques with Buffer Rod in Molten Aluminum. Ultrasonics Symposium, 2002, Proceedings, 2002 IEEE, 1, 805-810.

  28. 28. Ono, Y., Moisan, J.F. and Jen, C.K. (2003) Ultrasonic Techniques for Imaging and Measurements in Molten Aluminium. IEEE Transactions, 50, 1711-1721.

  29. 29. Mansfield, T.L. (1981) Probe for the Ultrasonic Inspection of Molten Aluminium. Reynolds Metals Company, United States Patent US4261197 A.

  30. 30. Gallo, R., Mountford, H. and Sommerville, I. (2003) Ultrasound for On-Line Inclusion Detection in Molten Aluminium Alloys: Technology Assessment. AFS International Conference on Structural Aluminium Castings, Orlando, 2-4 November 2003, 179-194.

  31. 31. G&oumlkelma, M., Le Brun, P., Dang, T., Badowski, M., Morscheiser, J., Friedrich, B. and Tewes, S. (2016) Assessment of Settling Behavior of Particles with Different Shape Factors by LiMCA Data Analysis. Light Metals.

  32. 32. Badowski, M., G&oumlkelma, M., Morscheiser, J., Dang, T., Le Brun, P. and Tewes, S. (2015) Study of Particle Settling and Sedimentation in a Crucible Furnace. Light Metals, 967-972.

Journal Menu >>