Compacted Graphite Iron and Ultrasonic Testing
John A. Griffin
Robin D. Griffin

INTRODUCTION
The University of Alabama at Birmingham was involved in a project to determine the feasibility of using compacted graphite iron (CGI) in a commercial part. One of the project goals was to develop an ultrasonic velocity method to separate acceptable from unacceptable microstructures as a production quality control procedure.

Ultrasonic velocity is commonly used as a quality control technique in the production of cast iron. It is very sensitive to graphite shape, and can be determined inexpensively and quickly. Ultrasonic velocity is a function primarily of the modulus, and density of the material. The porosity size and distribution also affects ultrasonic velocity due to their effects on density, modulus and pulse wave shape. The velocity of a longitudinal ultrasonic wave through a material can be expressed as (Krautkrämer, 1983):

                               Equation 1

where νl is the longitudinal ultrasonic velocity, E is Young’s modulus, ρ is density, and μ is Poisson’s ratio. Young’s modulus increases as the square of the density, i.e., as porosity decreases (Klima and Baaklini, 1984):

E = E₀ exp(-bP)                                                                                                   Equation 2

Where E0 is Young’s modulus for the nonporous material, P is volume fraction porosity, and b is a factor related to the pore size, shape and location. Poisson’s ratio generally increases with increasing density, but the effect is usually small.

The microstructure of cast iron consists primarily of ferrite, pearlite and graphite. The density difference between the iron matrix and graphite in cast iron produces a large impedance at the iron/graphite interface. Therefore, an ultrasonic wave traveling through the iron will react to the graphite in a manner similar to a pore. The matrix microstructure will influence ultrasonic velocity but to a much smaller degree than the graphite shape and number.

PROCEDURE
Effects of section shape, transducer frequency, contact method, and microstructure on velocity needed to be understood to provide a robust and accurate procedure. A variety of cast irons, primarily CGI, were collected from three production foundries to develop a set of standards for experimental and production control. Three shapes were produced from the supplied cast iron as illustrated in Figure 1. The Oak Box standard was designed to eliminate any interference from sidewall reflections to the ultrasonic wave. This design is used to produce the cleanest signal possible. Unfortunately, most production castings are unlikely to have a location that matches the relatively large diameter of the Oak Box standard so a smaller diameter standard was produced called a Working standard. Finally, a Lens shaped standard was produced to test for any effects caused by a work piece with a curved surface. The standards had surface finishes smoother than 50Rq.

Figure 1. Ultrasonic Standard Designs.

The most common transducer frequencies used for castings range from 1-10 MHz so two transducer frequencies were tested: 2 and 5 MHz. Most production facilities use a non-contact technique to speed production and compensate for rough or irregular surfaces. In some situations, contact measurements are also made so in this experiment, velocities were calculated using both non-contact and contact methods. Figure 2 illustrates the tank and transducer arrangement used for the non-contact calculations. For the contact calculations, a delay line was used to remove any rattle caused by the main bang of the transducer, as illustrated in Figure 3. The delay line was made of wrought 6061 aluminum with a thickness of 0.5 in. (12.2 mm) and a diameter of 1 in. (25.4 mm).

Both setups used a PanametricsÒ 5800 pulser/receiver and a TektronixÒ TDS210 digital oscilloscope. The two transducers were aligned on the same axis and powered by the pulser/receiver. Ultrasound was sent by one transducer and received by the other in a pitch/catch technique. The received ultrasonic signal was recorded on a digital oscilloscope for subsequent analysis.

Figure 2. Apparatus for Measuring Ultrasonic Time-of-Flight Non-Contact on CGI Specimens.

Figure 3. Apparatus for Measuring Ultrasonic Time-of-Flight Contact on CGI Specimens.

Time of flight was measured using both the first break and second peak of the ultrasonic signal as illustrated in Figure 4. First break is often used if the material being tested distorts the peaks such that an accurate measurement is difficult.

Figure 4. Typical Waveform of Ultrasonic Signal.

The ultrasonic velocity was calculated by dividing the specimen thickness by the time required for the signal to travel through the specimen. A reference signal was first recorded with the delay line between the two transducers. The test specimen and delay line were then put between the transducers and the signal captured. The transit time through the test specimen was determined by comparing the reference signal obtained with the delay line to the signal obtained with the delay line and specimen between the transducers. Multiple measurements were made to assure accuracy. A constant coupling pressure was maintained on the transducer-delay line-specimen system to assure that consistent signals were obtained.

A representative sample of each iron was polished using a procedure developed at UAB. Twenty-five images at 200X of unetched microstructure were recorded for each sample for graphite analysis. The microstructural images were analyzed using an image processing software package that measured a variety of characteristics including nodule count, surface area of graphite/volume of material (SV), shape factor, and percent nodularity. Shape factor was calculated using the following formula:

Shape factor = (4p*Area)/(perimeter)2 Equation 3

Graphite particles with a shape factor greater than 0.65 were considered nodular (a perfect circle has a shape factor of 1).

No attempt was made to control either the chemistry or pearlite/ferrite content of the standard materials. All irons were CG with varying levels of success with one exception. One ductile iron sample was included to provide graphite shape extremes for the study.

The range of microstructures used in this study are presented in Figures 5 through 10 in the un-etched and etched condition. The graphite shape varied considerable over these samples as did the amount of pearlite/ferrite in the matrix.

Figure 5. Microstructure of Foundry A CG Iron Standard #1962.
(A) Unetched. (B). Nital Etch.

Figure 6. Microstructure of Foundry A CG Iron Standard #2051.
(A) Unetched. (B). Nital Etch.

Figure 7. Microstructure of Foundry C CG Iron Standard #2489.

Figure 8. Microstructure of Foundry C CG Iron Standard #X72.

RESULTS AND DISCUSSION
Effect of First Break Vs. Second Peak Velocity Determination

Both first break and second peak velocity measurements were made and typical comparisons are illustrated in Figure 11. The analysis showed that the first break velocity is about 0.005 in/μsec higher at velocities above 0.19 in/μsec and about 0.10 in/μsec higher at velocities below 0.19 in/μsec. These results indicate that both measurement methods performed well but if comparisons between castings are going to be made, ultrasonic velocity should be determined using the same time of flight criteria.

Figure 9. Microstructure of Foundry C CG Iron Standard #W72.

Figure 10. Microstructure of Foundry B Ductile Iron Standard.

Effect of Standard Shape
The effect of standard shape on ultrasonic velocity is illustrated in Figure 12. Due to material constraints, we had fewer Lens-shaped and Oak Box standards than Working standards. However, varying the standard shape within these limits did not significantly change the velocity response to microstructure. Side wall reflections did not influence or interfere with the thru velocity technique used in this study but may pose a problem if using multiple reflections in a thru-back method. The curved surface of the Len-shaped standards did not increase or decrease the pulse strength or shape.

Effect of Transducer Frequency
The effect of transducer frequency for contact and non-contact set-ups are illustrated in Figures 13 and 14 respectively on the working standards. In both cases, the velocity shifted upward as the transducer frequency was increased. For the contact measurements, the shift was less than 0.05 in/usec while for the non-contact measurements, the velocity shifted up between 0.05 and 0.001 in/usec with an increase in transducer frequency from 2 to 5 MHz. The shapes of the curve were unaffected. Velocity is independent of frequency in homogeneous materials as shown in Equation 1. Most materials however contain an inhomogeneous microstructure which will absorb and scatter the ultrasonic wave depending on the wave frequency. This scattering causes velocity to be dependent on frequency.

Figure 11. Effect of Velocity Determination Method on Velocity.

Effect of Transducer Coupling Method
The effect of non-contact versus contact transducer coupling methods are illustrated in Figure 15. The non-contact velocities were lower than the contact velocities and the differences were greater at higher velocities. The differences in velocity could come from a number of sources. Velocity calculations for contact measurements are straight forward with no other variables. Non-contact calculations require not only the thickness of the sample but the distance between

the transducers and the ultrasonic velocity of the water. A water temperature variation of a few degrees can measurably change its velocity. The location of the transducers can also make accurate measurement of their separation difficult. Any error in these measurements can produce an offset from the contact velocity calculations.

Figure 12. Effect of Standard Shape on Velocity.

Figure 13. Effect of Transducer Frequency on Velocity-Contact Measurement.

Figure 14. Effect of Transducer Frequency on Velocity – Non-Contact Measurement.

Figure 15. Effect of Transducer Setup on Velocity

Correlation of Microstructure Measurement to Velocity
Percent graphite nodularity is typically specified to a level by the customer and charts exist that allow the foundry to quickly rate a microstructure. Image analysis methods also exist for characterizing graphite morphology. Careful image analysis of the graphite can provide less subjective results but this method is much more time consuming. Nodularity can be calculated using image analysis programs by selecting a shape factor value below which a graphite shape becomes non-nodular, such as was performed here. Other microstructural features can simultaneously be measured using image analysis including the average shape factor and the graphite surface per unit volume. Excellent correlations between graphite surface per unit volume (SV) and ultrasonic velocity have been made in the past (Patterson, 1979). However, this descriptor is also influenced by volume fraction and number density which do not affect ultrasonic velocity as strongly as shape.

Figures 16, 17 and 18 show the relationship between velocity and nodularity, shape factor and graphite SV. Figure 16 plots ultrasonic velocity at 2 MHz versus percent nodularity for all standards using non-contact. The correlation between nodularity and velocity was above 90% for an exponential fit and these results are excellent for irons that varied widely in chemistry and matrix structure. These results indicate that the velocity measurement should provide assistance in separating acceptable and unacceptable graphite structures in CG iron parts. The results were similar for the average shape factor as illustrated in Figure 17 with a correlation above 90% for an exponential fit. The relationship between graphite SV and velocity was not as good and these results are illustrated in Figure 19. Surface per unit volume appeared to contain two populations and is obviously affected by other unaccounted for microstructural features such as variations in the volume fraction or number density of graphite particles.

SUMMARY AND CONCLUSIONS
The University of Alabama at Birmingham was involved in a project to determine the feasibility of using compacted graphite iron (CGI) in a commercial part. One of the project goals was to develop an ultrasonic velocity method to separate acceptable from unacceptable microstructures as a production quality control procedure. Effects of section shape, transducer frequency, contact method, and microstructure on velocity needed to be understood to provide a robust and accurate procedure. A variety of cast irons, primarily CGI, were collected from three production foundries to develop a set of standards for experimental and production control. Three shapes were produced from the supplied cast iron.

Velocity measurements were made using both 2 and 5 MHz transducer frequency, contact and non-contact methods and different size and shaped standards. Varying the standard shape within the limits of this study did not significantly change the velocity response to microstructure. Side wall reflections did not influence or interfere with the thru velocity technique used in this study but may pose a problem if using multiple reflections in a thru-back method. The curved surface of the Len-shaped standards did not increase or decrease the pulse strength or shape.

Figure 16. Relationship Between Graphite Nodularity and Velocity.

Figure 17. Relationship Between Graphite Shape Factor and Velocity.

Figure 18. Relationship Between Graphite SV and Velocity.

Small but significant shifts in velocity were found with changes in transducer frequency. For the contact measurements, the shift was less than 0.05 in/usec while for the non-contact measurements, the velocity shifted up between 0.05 and 0.001 in/usec with an increase in transducer frequency from 2 to 5 MHz. The shapes of the curve were unaffected.

Similar to transducer frequency, small differences in velocity were observed with differing transducer coupling methods. The non-contact velocities were lower than the contact velocities and the differences were greater at higher velocities. The differences in velocity could come from a number of sources. Velocity calculations for contact measurements are straight forward with no other variables. Non-contact calculations require not only the thickness of the sample but the distance between the transducers and the ultrasonic velocity of the water. A water temperature variation of a few degrees can measurably change its velocity. The location of the transducers can also make accurate measurement of their separation difficult. Any error in these measurements can produce an offset from the contact velocity calculations.

The correlation between nodularity and velocity was above 90% for an exponential fit and these results are excellent for irons that varied widely in chemistry and matrix structure. These results indicate that the velocity measurement should provide assistance in separating acceptable and unacceptable graphite structures in CG iron parts. The results were similar for the average shape factor with a correlation above 90% for an exponential fit. The relationship between graphite SV and velocity was not as good. Surface per unit volume appeared to contain two populations and is obviously affected by other unaccounted for microstructural features such as variations in the volume fraction or number density of graphite particles. These results indicate that ultrasonic velocity should provide a valuable tool for separating acceptable and unacceptable graphite structures in CG iron.

Back to top