|DGZfP-JAHRESTAGUNG 2001 |
|ZfP in Anwendung, Entwicklung und Forschung|
Berlin, 21.-23. Mai 2001 -Berichtsband 75-CD
The aim of this paper is to demonstrate the capability of ultrasonic techniques in the assessment of the structure and properties of ductile cast irons, and also contribute to the database concerning the use of ultrasonic velocity and attenuation measurements for quality control. In the experiments, cast iron specimens with different nodularity percentages (0, 50 and 90%) were cast by the addition of different amounts of alloys to the melt, and then various heat treatments were applied. Ultrasonic velocity and apparent attenuation measurements were carried out on the as-cast and heat-treated specimens. Then, the results of ultrasonic measurements were correlated with the microstructural investigations on the specimens.
Typical applications of nodular cast iron include automotive and diesel-crankshafts, pistons, and cylinder heads, electrical fittings, motor frames and circuit breaker parts, steel mill-work rolls, furnace doors, bearings, tool and die-wrenches, levers, and miscellaneous dies for shaping. Nodular cast iron shows higher strength and ductility than grey cast iron of similar composition. Inoculation of grey cast irons is widely practised and involves adding to the liquid metal shortly before solidification a small amount of substance such as magnesium or cerium. When the residual magnesium content of the melt is sufficient, the graphite grows with a spheroidal shape. Otherwise, various degenerate forms of graphite such as flake graphite may form; in this case, the sharp edges of flakes cause critical stress concentrations that have a more adverse effect on the mechanical properties than compacted forms of graphite with rounded ends. Whether the matrix is fully ferritic or pearlitic or other combinations, variations in the size and shape of the graphite can be produced by variations in chemical composition, casting-section thickness, design, and casting conditions. Heat-treated nodular cast iron usually has more uniform mechanical properties than as-cast one, particularly in castings with wide variations in section thickness. As the matrix phase is varied from ferrite to martensite, strength, hardness and wear resistance increase, but ductility, impact toughness and machinability decrease.
Foundries need to specify the properties of the material delivered to the customers by chemical, mechanical and metallurgical tests. Quality control of nodular cast irons is critical since it is very sensitive to the process variables affecting the microstructure and mechanical properties. The classical procedures characterise the properties in standard states, therefore, they cannot express the actual strength and structural relations in the casting. On the other hand, since these methods are destructive, labour intensive and time-consuming, it is meaningful for both foundries and consumers to have 100% quality testing, which is reliable, rapid and suitable for automatic grading. Thus, developing an ultrasonic technique that could identify the nodularity and the matrix phases non-destructively, would be helpful for the purposes of both quality control and product reliability.
Ultrasonic characterisation takes the advantage of the elastic and an-elastic interactions between the sound wave and the microstructure. Indirect assessment of graphite morphology is possible by ultrasonic velocity measurements that are also dependent on the mechanical properties. Another alternative is the evaluation of the ultrasonic attenuation. Cast irons consist of grains of the matrix phases and graphite particles having remarkably different elastic properties. Although the material shows isotropic properties as a whole due to random orientation of grains, elastic anisotropy and different orientation of crystallographic axes with respect to the axes of neighbour grains makes the elastic modulus different form grain to grain. The grains may also differ in density due to different phases. Thus, sound waves will suffer abrupt changes in their velocities, with corresponding changes in their characteristic impedances, and reflections will take place at grain boundaries or interfaces.
Several studies related to this topic had been published. Emerson and Simmons evaluated the graphite form in ferritic ductile irons by measuring ultrasonic velocity and resonant frequency, . Fuller performed ultrasonic measurements on pearlitic ductile irons with varying amounts of compacted-graphite, and on the ductile irons with different matrix structures having varied nodularity from 40 to 100%. He concluded that ultrasonic velocity and attenuation measurements were not effective when both graphite form and matrix structure varied, [2,3]. Fuller et al. measured ultrasonic velocity and attenuation on ductile irons with varying proportions of ferrite and pearlite in which the graphite form ranged from fully nodular to a mixture of nodular and compacted graphite, . The effects of nodularity and matrix phases on the ultrasonic properties were researched by Lee and Suen, . Cech researched the quality of complex castings of both grey and nodular cast irons having different combinations of ferrite and pearlite in the matrix by using ultrasonic measurement of relative velocity, magnetic and electrical methods, . Orlowicz and Opiekun concluded that the changes in the graphite morphology due to heat treatment could be detected by ultrasonic velocity, . Collins and Alcheikh performed ultrasonic measurements on cast irons with different nodularity and matrix structures. They found a definite relationship between the nodular graphite content and the ultrasonic velocity, but influences of matrix structure on the velocity was not clear. The attenuation increased with decreasing nodularity, however, matrix structures were found to have similar attenuation characteristics except tempered martensite, . Lerner and Brestel measured ultrasonic velocities to predict tensile strength after annealing and normalising of grey cast irons with different graphite morphologies, . Kruger and Charlier developed a model to simulate the backscattered signals from inhomogeneities that are much smaller than the wavelength, .
Three groups of specimens were cast into the sand moulds to produce the slabs having different amount of nodular graphite. Table 1 gives the chemical compositions and the casting parameters. First, specimens having cubic shape (side length of 40mm) were prepared. After grinding and polishing the surfaces of the cubes, ultrasonic measurements were done in the perpendicular directions. After verifying the isotropic behaviour of sound velocity and attenuation, two specimens were prepared by cutting each cube into equal parts. These specimens were machined, ground and polished to obtain the parallel surfaces, and then, a series of specimens were heat-treated. The specimens were heated to 860°C and stayed at this temperature for one hour. Next, the first group was annealed by furnace cooling to 360°C; the second group was normalised by air cooling; and the third group was quenched into oil.
The amounts of the phases and nodular graphite, and the average sizes of nodules were determined at the randomly selected points on the specimens with the Buehler-Omnimet image analyser. Vicker's hardness tests were performed by Heckert hardness device. The average values were calculated by considering five independent measurements on each specimen. The densities of the specimens were also measured.
Ultrasonic measurements were carried out using the KrautKramer-Branson USIP-15 and a 2 MHz straight-beam probe. A uniform force was exerted on the probe to keep the thickness of oil film constant at the interface. The unknown velocities in the specimens were determined by comparative measurements on the reference block K2 whose ultrasonic velocity is accurately known.
|V= t Vref / ( k T )||(1)|
where t is the thickness of the specimen, Vref is the velocity of longitudinal wave in the reference block, k is a measure representing one scale on the time-base range, and T is the scale value of the back-wall peak. The mean values were calculated by repeating the measurement three times on each specimen. For the velocity measurements in the reference block, the time of flight between two back-wall echoes was measured using Panametrics analyser 5052UAX50 combined with Philips oscilloscope PM 3365A, and then, the ultrasonic velocity was calculated by dividing the twice the thickness of the specimen with time.
Since absolute attenuation measurements require specialized ultrasonic equipment and techniques, apparent attenuation that is the observed loss in the ultrasonic energy was determined. In addition to the true loss, it also includes losses attributable to instrumentation, specimen configuration, beam divergence, and measurement procedure. This measurement procedure is readily adaptable for the determination of relative attenuation between materials. Apparent attenuation was determined using Equation-2 by considering first and second back-wall echoes,
|a = [ 20 log (A1 / A2) ] / 2 t||(2)|
where A1 and A2 are the amplitudes of the reflections in dB, and t is the thickness of the specimen. Three measurements were done on each specimen, and the average value was calculated.
The longitudinal ultrasonic velocity can be related to Poisson's ratio (m), density (r) and elastic modulus (E) according to the Equation-3.
|VL2 = (E / r) [ ( 1-m ) / ( 1+m )( 1-2m )]||(3)|
It has been reported that mis independent of the phases in the matrix of the cast irons since the velocity ratio of shear and compression waves is constant irrespective of the matrices, . Since the density values of the specimens are close to each other (between 7.02 and 7.06), the elastic modulus is the main parameter in the variation of ultrasonic velocities. Fig.1 shows that there is a direct correlation between the longitudinal ultrasonic velocity and the amount of nodular graphite, and the values obtained are in agreement with the literature. Specimens with the highest content of nodular graphite have the highest velocity, and ultrasonic velocity decreases with decreasing nodular graphite percent.
|I||Matrix||65%F + 35%P||93-100% F||82-90% P||100% M|
|Graphite size (mm)||27||23||26||31|
|II||Matrix||80%F + 20%P||95-100% F||90% P||M + few RA|
|Graphite size (mm)||24||24||22||25|
|III||Matrix||5%F + 95%P||100% F||95-100% P||100% P|
|F: ferrite, P: pearlite, M: martensite, RA: retained austenite|
|Group I :||3.7%C, 1.4%Si, 0.2%Mn, 0.017%P, 0.011%S, 0.033%Cr, 0.03%Ni,
0.001%Mo, 0.286%Cu, 0.006%Sn, 0.035%Mg|
Casting temp. = 1430°C, Inoculation = 2 minutes.
|Group II :||
3.1%C, 2.6%Si, 0.22%Mn, 0.016%P, 0.017%S, 0.029%Cr, 0.015%Ni,
0.002%Mo, 0.275%Cu, 0.018%Sn, 0.022%Mg|
Casting temperature = 1430°C, Inoculation = 8,5 minutes.
|Group III :||3.6%C, 1.11%Si, 0.34%Mn, 0.077%P, 0.074%S, 0.022%Cr, 0.04%Ni,
0.002%Mo, 0.27%Cu, 0.015%Sn|
Casting temp. = 1360°C, no inoculation.
|Table 1: Data about casting procedures, chemical compositions and microstructures|
Fig.2 gives the variation of relative amplitude with respect to the number of back-wall echoes. As the amount of nodular graphite decreases, the number of back-wall echoes observed on the screen also decreases. Since the scattering of ultrasonic waves by flake graphite is much more readily than nodular graphite, a decrease in nodularity resulted in higher attenuation in terms of echo amplitude. The results given in Fig.3 support the statement that the apparent attenuation of the ultrasonic waves is sensitive to the variations in the amount of nodular graphite.
|Abb 1: Relationship between nodular graphite content and ultrasonic velocity in the as-cast specimens.|
|Abb 2: Comparison of screen heights of the back-wall echoes.|
|Abb 3: Variation of apparent attenuation with respect to amount of nodular graphite.|
From Fig.4 it is seen that an increase in the amount of the nodular graphite causes an increase in both ultrasonic velocity and hardness. Indeed, the correlation between ultrasonic velocity and mechanical properties is difficult since the matrix phases also influence the mechanical properties. For the known amount of nodular graphite, hardness increased in the order of ferrite, pearlite, and martensite. However, for a given amount of nodular graphite, the specimens having different matrix phases did not show any variation in ultrasonic velocity. Thus, it was concluded that the elastic modulus of ductile iron is decided mainly by the amount of nodular graphite.
|Abb 4: Relationship between longitudinal ultrasonic velocity and hardness.|
The sensitivity of ultrasonic velocity and apparent attenuation against the changes in the nodular graphite content allows grading of nodular cast irons. It is also possible to detect local unnodularized zones in the castings. In the specimens having high amount of nodular graphite, the longitudinal wave propagates faster. The differences in density and Poisson's ratio among the specimens are negligible, then, the main reason for the velocity change is the modulus of elasticity. Variation in the amount of nodular graphite also affects the sound attenuation. When the nodular graphite content decreases, the attenuation increases due to the increased scattering effect of the non-spherical graphites. The matrix phases also affect the mechanical properties, however, this influence could not been detected by velocity measurements. Although apparent attenuation measurements are more sensitive, the reliable utilisation of it requires further research. When the matrix phases are known, ultrasonic measurements can be used for determining graphite morphology that is also an indicator of the strength properties. If parallel surfaces exist and side-wall effect is negligible, these measurements provide a quick tool for quality control.
The authors gratefully acknowledge Erkunt and Dökta companies, and the staff of the Welding Technology and NDT Centre of Middle East Technical University.