Noises Analysis used for defects identification on pieces

Prof.dr.eng. Vasile LUCA, Prof.dr.eng. Sergiu FĂTU, Prof.dr.eng. Cornel SERBAN Prof.dr.eng.Rodica POPESCU

Transilvania University of Braşov, Romania

Abstract

Samples from steel and cast iron with defects (crashes, retasures, inter-crystal corrosion, different forms for graphite, silicon inclusions, oxides) were manufactured.

The samplex, tested in the elastic field have noises in the audio and high frequences, which can be captured with a piezo-electric traductor. The interpretation of information made with the PC lead to the possibility of establishing a connection between the noises characteristics and the type of defects.

Worldwide the casting of parts of nodular cast irons with spherical graphite has been recording a continuous increase, this material replacing some steels as well as grey cast irons. Both cases yield economic advantages, by either the reduced specific energy consumptions or the diminished part weight.

Ensuring an adequate quality to raw cast parts of nodular cast irons with spherical graphite implies however more rigorous technological restrictions than those for the smelting and casting of steels or grey cast irons. Consequently a more thorough quality control is required, leading to the control of each cast raw part, graphite nodulisation taking place in the form.

Since 1979, the Tractor Manufacturing Company of Braşov has implemented in addition to the classical determinations of the chemical structure, hardness, mechanical strength, breaking elongation and metallographic analysis, also nodulisation control carried out directly on the parts by determination of the ultrasound propagation velocity. Besides its obvious advantages the method also implied some disadvantages: a relatively long control time (gauging of the defectoscope, preparation of the control surface, thickness measurement with the slide gauge, measurement of the relative thickness with the defectoscope, computation of the ultrasound propagation velocity, results assessment on a nomogram), a relatively rapid wear of the defectoscope sensor.

In time it could be observed, that all cast raw parts with an adequate degree of nodulisation generate upon tapping a sound of certain characteristics, significantly distinct from the raw parts with incomplete nodulisation. Hence the idea of quality control by sound emission analysis of the raw parts was generated, yielding a simpler and more rapid method, describing the quality of the entire volume of the part.

1. Analysis methods of the sound emission

Electromagnetic excitation of the raw part

Figure 1 presents the diagram of an installation of audio-frequency used in the assessment of part or raw part quality. The frequency generator (1) emits a sinus or impulse type signal, the frequency being adjustable from 0 to 20 kHz.


Figure 1. Installation for the analysis of sound emission with electromagnetic excitation of the raw part: 1 – frequency generator; 2 – amplifier; 3 – exciter; 4 – raw part; 5 – receiving transducer; 6 – amplifier; 7 – oscilloscope; 8 –frequency-meter; 9 – voltmeter.

The signal is amplified by amplifier (2) and transmitted to the electromagnetic exciter (3) which generates longitudinal, transversal and superficial waves in the raw part of a complex geometry (4). Whilst for simple geometry test pieces wave propagation can be described by physical – mathematical methods, in complex geometry parts unpredictable reflections, refractions and interferences are generated, eliciting in each point of the part different reactions.

The receiving transducer (5) senses the reactions of the excited part, and the amplified signal is transmitted to oscilloscope (7), frequency meter (8) and voltmeter (9). BY varying the frequency of the excitation current, depending on the propagation mode of the waves through the part, the voltmeter will record a higher or lesser signal, allowing for plotting a diagram of the received oscillation amplitudes versus frequency. The maximum points can be established with univocity with the oscilloscope, by the method of the Lissajous figures, and the frequency corresponding to these maximums can be recorded with high accuracy with the frequency meter.

Mechanical impulse excitation of the raw part

Figure 2 presents the diagram of a sorting installation of cast raw parts, achieved by analysis of the sound emission consequent to a mechanical impulse. The  10 mm diameter hardened steel ball (1) falls freely through the galls tube (2) of length 200 … 300 mm, set experimentally for each type of tested raw parts. Following the impact on the part, the ball rises back into the tube and is withheld there by the magnetic field of the permanent magnet (3), thus avoiding a second fall and mechanical impulse, respectively. The impulse generates in part (40 a wave field characterised mainly by a certain frequency, that is the eigen-frequency of the tested object.

The signal sensed by the receiving transducer (5) is amplified by amplifier (6) and can be visualised on the memory or high luminescence inertia oscilloscope (7). Figure 3 presents the time related evolution of the received signal.


Figure 2. Installation for sound analysis with electromagnetic excitation. 1 – steel ball; 2 – glass tube; 3 – permanent magnet; 4 – raw part; 5 – receiving transducer; 6 – amplifier; 7 – oscilloscope; 8 – frequency meter.


Figure 3. Characteristics of the sound emission generated by mechanical impulse: A – maximum amplitude; t I – initiation time;  - damping; ps – sensibility threshold of frequency meter; tt – total recording time of frequency.

Experimentally it could be established, that the dimensional deviations of the cast raw parts to not influence significantly the time related evolution of the aspect of the oscillogram. This is however strongly influenced by the shape and dispersion of the graphite inclusions, by the presence of micro-cracks, as well as by the presence of shrink holes and micro shrink holes. When the level of the received and amplified signals exceeds a certain value – ps (sensibility threshold), the frequency meter (8) will display the frequency of the oscillations, or, operating a s a counter, it will record the number of perceived oscillations during the total time Tt. The aspect of the oscillogram, the frequency or the number of oscillations will indicate the quality of the raw part.

Analysis of the sound emission of raw parts subjected to mechanical strain in the elastic field

The method is presented in figure 4. The mandrel 91) adapted to a hydraulic machine for mechanical tests will de displaced with a constant velocity, the straining force on the raw part (2) increasing progressively, but only up to a value which will not generate stress exceeding the elasticity limit. During elastic deformation the part will emit noise with a frequency of 50 kHz … 1.5 MHz, which are transmitted to the pressing tool and are received by the piezoelectric transducer (4).


Figure 4. Diagram of the sound emission analysis of the mechanically strained raw parts: 1 – mandrel; 2 – raw part; 3 – rubber buffers; 4 – piezoelectric transducer; 5 – signal analyser.

In order to avoid the recording of noise generated by the friction between the part and the clamping device, rubber buffers (3) are placed between these. During straining signals of the type represented in figure 3 are recorded, but of significantly higher frequency, received by the analyser (5).

Possible signal sources are: the friction between the crystal grains, development of micro-cracks, cracking of the networks, needles or cementite plates, as well as silicate inclusions. Thus the method lends itself excellently to highlight the presence of lamella graphite, of free cementite and of shrinking holes.

The “Locon” device made by PAC of Princeton, N.J. can be used for a signal analyser, which can stock and automatically analyse all these noises. An ultrasound defectoscope can be used for laboratory research, which operates on reception, coupled by an interface to a process analysing computer.

2. Experimental results

By applying the electromagnetic excitation to the raw parts, determinations were carried out on several bearing covers, coming from different charges, some corresponding, others with an uncorresponding nodulisation. The obtained resonance curves are presented in figures 5 and 6.

Figure 5a presents the resonance curve of a cast bearing cover with a corresponding microstructure, confirmed by the measurement of the propagation velocity of ultrasounds. Figure 5b presents the resonance curve of a raw part having a correct structure too, but being slightly thicker consequently to the casting form not closing completely, and with a small burr in the separation plane.


Figure 5. Resonance curves of some raw parts

This raw part is by 147 g heavier than the first one. Comparing the two resonance curves significant differences can be noticed, due mainly to the differences in dimensions and weight

Figure 5c presents a cast part of correct dimensions, but of uncorresponding nodulisation highlighted also by ultrasound. A sample was taken from this part, and following microscope analysis about 30% lamella graphite could be identified. This is consistent with Rm = 38.4 daN/mm2 and A = 3.7% The aspect of the resonance curve is significantly different from those of the first two parts.


Figure 6. Resonance curves of some finished parts

Thus the conclusion emerges, that for assessing material quality the dimensional deviations have to be minimised. Figures 6a, b and c present the resonance curves of finished covers, all three parts coming from charges of corresponding characteristics. The differences between these resonance curves are very small, thus confirming the applicability of this method in the evaluation of material quality for parts of constant dimensions and weights.

The method of sound emission analysis based on mechanical impulse was tested on raw parts of larger dimensions, i.e. on clutch plates cast also of nodular cast iron with spherical graphite. Of the batch of raw parts received in ultrasound control, 20 plates were analysed by the method presented in figure 2. The sensibility threshold of the frequency meter was set at 1 V, and it was switched to impulse counting operation mode. For all plates accepted by ultrasound control, the number of recorded impulses varied between 3922 and 4206.

Determinations were carried out also on five plates which had not been accepted by ultrasound control of the nodulisation. For these frequencies of a much smaller number of impulses were recorded, i.e. 641 … 2100.

Thus the idea of implementing this control method is generated, as being simpler, more rapid, requiring less costly equipment and less qualified personnel. The method can be used for a first selection of raw parts, carrying out ultrasound control only on the rejected parts, for confirmation or invalidation of the results.

The method of sound emission analysis of raw parts in the elastic field is under research, the authors having still struggling with difficulties related to equipment as well as data assessment.

REFERENCES

  1. Spanner J.:Journal of acoustic Emission, nr.6, 1987, Vol.2.
  2. Theisk I.: Die Anwendung der Schllemissionanalyse zur Riss prufurg in Gusstucken. Giessereitechnik, nr. 2, 1990.

1