Crack growth performance of tire compounds.

Besides performance parameters such as rolling resistance, wet traction and abrasion, the resistance of the various tire compounds to crack growth is one of the most crucial requirements of a tire, especially with respect to safety. In order to guide the compounder, several laboratory tests have been developed to assess the crack growth resistance of tire compounds. The Monsanto Fatigue to Failure Tester (MFTF), or the De Mattia test machine are still frequently used in the rubber industry for laboratory screening measurements of the resistance of rubber to dynamic crack growth. Even though many of the machines have been modified, e.g., to perform temperature controlled MFTF-experiments [ref. 1] or to measure the crack length in sample [refs. 2-4], the correlation of their results with tire machine tests, or the performance of tire compounds in the service, is often unsatisfactory.

Until the development of more elaborate test methods in recent years, these were the only tools of the tire compounder. These more elaborate test methods often use the pure shear sample (a long horizontal test piece with a long cut to avoid edge effects) which is dynamically loaded by a servo-hydraulic drive [refs. 5-7]. These machines are generally limited in productivity having only one sample position, and have therefore become rather a research instrument than a common tool for tire compound development.

Time-consuming and expensive tire tests are still the most common tool in the tire industry to determine the crack-growth performance of the various rubber components of a tire. These tests, because of the time and expense, limit developments to two or three candidate compounds, often missing potential breakthroughs in technology.

In this article, a new machine called the tear analyzer is described. This machine, when operated under particular conditions, has been found to be capable of predicting (in the laboratory) the same crack growth resistance of various rubber parts of the tire such as sidewall, tread and breaker compounds as found in tire machine tests.

As well as the correlation for tire compounds a correlation between service performance and tear analyzer results has also been established for a number of non tire products such as multi-rib-belt compounds.

Experimental

Tear analyzer operation

The tear analyzer measures the crack propagation rate in rubber and records the values together with the mechanical parameters during dynamic straining. A servo-hydraulic drive dynamically powers a piston with sample clamp attachments. The servo-hydraulic drive is controlled by two frequency sweep generators which can provide any time-dependent load in the frequency range of 0.1 to 100 Hz.

The samples (four on the first machine, 10 on the second) are subjected to a dynamic load in a temperature-controlled chamber that can be adjusted between -20[degrees]C and 150[degrees]C. The ambient conditions can be either ozone-free air, nitrogen atmosphere or air/nitrogen with enhanced ozone concentration. In addition to control of the oven temperature the sample surface temperature is also monitored by an infrared pyrometer.

The signal generated by the frequency generators determines the time dependence of the strain. The corresponding time-dependent stress in the sample is measured by fast load cells.

Rubber strips 60x15x1.5 nun, with a lateral incision of 1 mm in depth, are used as test samples. Four or ten samples are measured simultaneously. It is reported in literature [ref. 8] that crack-growth measurements, performed with the tensile strip sample, give the same results as tests with the pure shear sample.

Each load cell, and its corresponding sample clamp attachment, is connected to a computer-controlled stepping motor to maintain a constant prestress, i.e., to ensure constant minimum stress during the whole time of the experiment. This continuous control of the minimum strain is extremely important for investigations of rubber samples with high tension set or a comparison of materials with different tension set values. By recording the stepper motor displacements during the experiment it is possible to obtain a measure of the dynamic tension set of each sample.

The mechanical data, which are recorded on a PC, include maximum and minimum stress and strain; total, elastic and dissipated energy densities; tearing energy; stepper motor displacements; and the surface temperature of the sample in the vicinity of the crack.

A CCD-video camera system, mounted on a stepper motor-driven x-y manipulator, is used to determine the crack contour length c.

At a certain piston displacement a picture of the sample is taken. It is then transferred into a frame grabber and simultaneously stored in analog format on videotape. The digitized picture from the frame grabber is then evaluated by the software which localizes the crack position and determines the contour length by following the black-white boundary line of the crack. This procedure is independent of how the crack propagates through the sample. The size of the crack contour length c, and the number of cycles n, are also recorded on the computer.

From the crack growth data, the logarithmic crack growth velocity, or the crack growth rate d(1nc)/dn=1/c(dc/dn), is calculated and can be plotted against the stress or strain amplitude, any energy density (total, elastic, dissipated) or the tearing energy. The detailed data evaluation procedure has been discussed in the literature [ref. 9].

When a car is driven and the tire rotates, each small rubber element is subject to a time-dependent load as a point on the tire tread enters and leaves the contact patch. The period that the rubber element is in the contact patch, compared to the time between, is relatively small. This time-dependent load is simulated by using a pulse waveform (haversine) produced by two frequency generators to drive the servo valve. One generator supplied the signal for the fast piston movement thus mimicking the passage through the contact patch. The other generator which triggered the first one, furnished the time between two signals to simulate the tire revolution time. Figure 1 illustrates the type of waveform that was used.

Tire tests

Before tire tests could be carried out it was necessary first to make tires. Tires for the sidewall tests were made either with Bayer materials or some compounds under contract. Tires for testing the breaker, cushion gum and X-ply truck tread compounds were made with proprietary materials by tire companies.

There are different tire tests used by tire companies to determine the crack growth of the different tire components. The difficulty is in finding the conditions that test only the component in which the company is interested. Many of these tests involve testing the tire under specific conditions of speed, load and atmosphere on a fixed diameter drum that is driven. Other tests are on vehicles under specially set conditions. With all these accelerated tests it is hoped, once again, that the correlations with service performance remain consistent.

Sidewall compounds

There are three common tests for sidewalls. One test involves testing tires with pre-cuts through the sidewall at an angle of 45[degrees]. The tires are run on a drum at a fixed load and pressure. The average cut length is determined every 1,200 miles. Another test is to run the tires without pre-cuts in the sidewall but under an ozone atmosphere. A third test is to run the tires with a large overload and at low pressure and speed so that the sidewalls are severely deflected.

Breakers

In one of these tests the development of the temperature at the breaker edge was measured for the different compounds after the tires were run on a track at 130 km/h at 100% load test for up to 100,000 km. In another test car, fires were subjected to both a wheel test (cyclic slip angle [+ or -]3[degrees]) at 100% load, 2.4 bar, 38[degrees]C, 4,800 km and a controlled road test (48,000 km, 100% load, 1.8 bar). In both cases, the separation resistance was determined. Besides the road test, for truck tires an extended endurance wheel test was performed (step-up load test, load from 66% to 230% over 170 hours, 55 km/h)

Cushion gum

These results have been obtained from tires that have been retreaded using the pre-cure method and then run for up to 100,000 miles. The extent of the cracking at the shoulder, between the tire and the pre-cured retread, is measured each 10,000 miles.

X-ply truck treads

The rating of these compounds came from a series of extended tests on loaded trucks, operating under controlled arduous conditions, in a tropical climate. The extent of groove cracking at the base of the grooves is measured.

Cut/chip resistance of radial truck tire treads

The cut chip wear resistance was measured after the tires have been subjected to severe service on a gravel and caliche track at speeds around 50 km/h. After the test the worn treads are compared to a set of standard tread compounds and a rating is given to them.

De Mattia

The De Mattia cut growth results shown, as a comparison with the tear analyzer, were obtained from precut samples according to ASTM D 813-87 and ISO 133 1983 (E).

Results and discussion

The components of the tire that have been investigated included sidewall, breaker, X-ply truck tread and inner liner compounds. Cushion gum compounds, for retreaded truck tires, have also been investigated.

Sidewall compounds

In addition to the most commonly used natural rubber/butadiene rubber (NR/BR) sidewall compounds, natural rubber/ethylenepropylene-diene rubber (NR/EPDM) systems have proved to provide an ozone-resistant alternative with excellent appearance to standard NR/BR sidewalls [ref. 10]. As an example, table 1 gives an experimental NR/EPDM compound formulation (L94) which was tested against a standard NR/BR sidewall compound (L92). These two compounds came from the first set of tires built to test the performance of the NR/EPDM system.

  Table 1 - NR/EPDM and NR/BR control
              compound

Sample               L92         L94
SMR 5                 50          60
EPDM XG003(a)          -          40
Buna CB 10            50           -
N660                  50           -
N375                   -          50
Oil                   10          20
ZnO                    3           3
Stearic acid           2           2
Wax                    2           -
TMQ                    2           -
6PPD                   2           -
Sulfur               2.19       2.19
MBS                  1.00          -
CBS                     -        0.5

(a) EEPDM XG003 is an experimental type.

Several pre-cut De Mattia cut-growth tests, performed in different laboratories, gave the result that the NR/EPDM system should exhibit a better performance than the NR/BR vulcanizate. The same ranking was obtained in crack growth measurements with a travelling microscope, at 20 Hz sinusoidal load in pure shear, using a mechanical energy resolver (MER), for strains between 20% and 60% [ref. 11].

The tear analyzer, when operated at the similar low frequency sinusoidal load conditions as the De Mattia test confirmed the result that the NR/EPDM was best. In the whole strain amplitude region investigated 25%-55%), at 4 Hz sinusoidal load, it was found that L92 showed higher crackgrowth rates than L94 when plotted against the strain amplitude or the elastically stored energy density, [W.sub.e1].

Unfortunately, these results were exactly opposite to those shown in tire machine tests. In figure 2, the results of the precut tire machine test are shown for the two compounds from table 1. Obviously, the NR/EPDM compound L94 exhibits a much worse crack-growth resistance than the NR/BR vulcanizate L92 which shows only minimal crack growth.

The tests were then repeated with the tear analyzer in the pulsed mode. In this pulsed mode, with the waveform as shown in figure 1, a completely different result was obtained. The results suddenly corresponded almost exactly with the tire test results. It was known that pulse testing was sometimes used [refs. 5-7] but not that the results would change so dramatically compared to a sinusoidial load. Figure 3 shows results of a pulsed load experiment, on the same compounds, at 15% strain and T = 40[degrees]C.

The actual pulse conditions were a piston displacement (with a 25 Hz pulse to simulate the passage of the contact patch) followed by a 1/7 s recovery time, the aim of which was to approximately mimic the tire revolution time. Due to limitations of the servo-hydraulics, this (25/7 Hz) was the closest fit to the waveform of a small element of a tire (as it passes successively through the contact patch) which could be obtained with the existing equipment. Under this pulsed load, this particular NR/EPDM compound cracks much faster than the NR/BR system and even the qualitative shape of the crack-growth curves are similar to those from the tire tests. To reduce the testing time, different pulsing conditions have been tried. At 30% strain and 15/4 Hz pulse conditions, the measuring time has been reduced by a factor of three, but the same ranking of the cut-growth rate of the compounds, as in the tire test, has been found.

It should be noted that these two compounds were chosen from the first tire trial carried out in 1990. The results were relatively poor for the NR/EPDM but, on the other hand, were a good example to use to find a good predictive test.

The ranking reversal which occurs through the change of the load conditions, from a low frequency sinusoidal load to a high frequency pulsed load, is, thought to be caused by the different relaxation time spectra of these specific NR/EPDM and NR/BR formulations. To determine this effect, the stress-relaxation cycle was measured by repetitive straining and relaxation with a tensile test machine, and the amplitude dependence of the complex modulus of elasticity was measured with a MER. It was found that the NR/EPDM compound (L94) exhibits a larger tan [delta] and a higher fraction of dissipated energy (this was also found in the tear analyzer measurements) than the NR/BR vulcanizate L92.

The complete relaxation between two strain cycles, only possible in the pulsed mode gives a higher crack-growth rate for L94 and a lower one for L92. However, the observed ranking reversal is still not completely understood and no reference to this effect has been found in the literature. Most machines use a cycle closer to the sinusoidal load conditions than the pulsed type and do not obtain a good correlation with machine tests.

Numerous sidewall compounds have been investigated with the tear analyzer. Three series of tires have been specially built and tested on both wheel tests (by different tire manufacturers) and on the tear analyzer in pulsed mode (for formulation, see table 2). Figure 4 displays the tear analyzer results in comparison With the tire test results from the second tire trial. Smoothed lines have been drawn through the crack-growth data. Obviously, the tear analyzer reproduces the tire test results very well, as can be seen from figure 5, where a correlation between the tire test and the tear analyzer results is shown. It can also be seen that there are two NR/EPDM compounds which show much better results than in the first test.

[TABULAR DATA OMITTED]

This same set of tires was tested on an ozone wheel test of a different tire manufacturer. In this case the crack-growth rate was calculated and compared to the rating results. The same ranking was found for the crack-growth rate as for this tire test rating. From previous experiments it is known that the NR/EPDM compounds tested are virtually ozone resistant [ref. 10], which means that the ozone wheel test represents a fatigue test only.

A third set of tires, with improved NR/EPDM sidewall formulations, was built and tested at another tire customer's plant. These sidewalls refused to crack throughout the test. The crack-growth resistance of this new NR/EPDM compound was at least as high as the NR/BR compound (L92) which was confirmed by the tear analyzer.

Further confirmation of the predictive capability of the tear analyzer (in pulsed mode) for sidewall compounds has been obtained by many tire companies.

Cushion gum compounds

Truck tires are retreaded sometimes three times in their life. One retreading procedure, known as pre-cure retreading, involves grinding the remainder of the old tread to form an accurately shaped base on which a pre-vulcanized replacement tread is bonded. The link between the new tread and the carcass is provided by a thin cushion gum layer, the edge of which is visible in the wing area of the tire. At the edges of this cushion gum, cracking problems can be observed after long periods of service.

Experiments with the tear analyzer were aimed to develop a predictive testing procedure for the crack-growth resistance of three proprietary cushion gum compounds in service. Pulsed mode experiments at T = 80[degrees]C, 30/4 Hz and 10%-25% strain have been performed. In figure 6 the elastic energy density are shown. It be seen that R is greater than O, and both are higher than H. When the crack growth rates were plotted against the strai amplitude, or the elastic energy density, the cushion gum H was found to be better than cushion gum R, which in turn was better than cushion gum O. In service, by contrast, it had been found that cushion gum R was better than H, and again both were better than O.

The tests, however, also showed that the moduli of the three cushion gum compounds were quite different. Now when the crack growth rate is plotted against the stress, the same ranking is found as in service. This indicates that a cushion gum operates under conditions of constant stress rather than constant strain. This ranking reversal occurs because of the very different energy densities of compound R and H at constant strain and at constant stress. At constant strain R stores more energy than H which is the opposite to constant stress.

Breaker compounds

Breaker compounds have also to be considered at constant stress rather than constant strain load conditions during service [ref. 6]. Four proprietary breaker formulations have been tested for a tire manufacturer. In figure 7, the ranking of the crack growth rate is shown. This ranking was confirmed by the tire manufacturer to be the same as found in service. Since the moduli of all four samples were close, the same ranking could be obtained at constant stress as well as at constant strain.

Car and truck tire breaker compounds were tested for another tire manufacturer and for both compounds a correlation between the tire test rating (for separation resistance) and the crack growth rate could be found.

Biased truck tire tread compounds (groove cracking)

Four compounds with known service performance were supplied by a tire company. They were tested at T = 100[degrees]C in the pulsed mode (30/4 Hz) at strains between 15% and 30%. The compounds had quite different moduli and total energy density. Nevertheless, the elastic energy densities were close as a result of different fractions of dissipated energy. Quite different crack-growth rates were found and the ranking at constant strain was that compound 4 was better than 6 which was better than 2 which in turn was better than 3 (figure 8). This was later confirmed to be the groove-cracking performance ranking found in tire service tests. If the quantitative rating, as provided by the tire manufacturer, is correlated with the crack-growth rate, 1/c(dc/dn), on a semilog scale, a linear relationship is found.

At constant stress, a different ranking was found which indicates that for groove cracking problems the compounds have to be considered under constant strain.

Radial truck tire tread compounds

The cut/chip wear resistance of truck tire tread compounds as determined by a gravel road test is compared to the crack growth rate as measured with the tear analyzer at 80[degrees]C, 30/4 Hz and 20% strain in figure 9. Considering the complexity of the cut/chip wear process, a satisfactory correlation of the tear analyzer data with the results of the tire test was still found.

Inner liners

Inner liner cracking is complex problem since liner failure is often associated with problems relating to tire constructions, such as finer cord penetration and liner flow, cracks which start in the vicinity of the carcass cords and near the vent lines on the inside, of the tire. If the cracking problems are not associated with the of the above-mentioned problems but rather with the compound formulation, the tear analyzer can also predict liner fatigue performance. In figure 10, the crack-growth rates of two truck tire liner compounds from different tire manufacturers are displayed. Compound A exhibits a significantly better crack-growth resistance than compound B - the same observation which has been made by retread companies.

Conclusion

A new laboratory test method for the assessment of the dynamic crack-growth resistance has been established. Excellent correlations between tear analyzer results and the service performance of various tire components, such as sidewall, breaker, tread and inner liner compounds, have been obtained. The machine can handle up to ten samples per run, thus providing a powerful and fast predictive tool for testing the crack-growth resistance of tire compounds. It is suggested that this type of machine will become more common in the future, being used to determine the optimum compound formulations, and tire testing will then only be used to confirm results.

References

[1.] H. Wesche, Kautsch., Gummi, Kunstst. 33, 103 (1980). [2.] D.G. Young, E.N. Kresge, A.J. Wallace, Rubber Chem. Technol 55, 428 (1982). [3.] M.-P. Lee and A. Moet, Rubber Chem. Technol 66, 304 (1993). [4.] R. Clamroth and U. Eisele, Kautsch., Gummi, Kunstst. 28, 217 (1975). [5.] D. G. Young, Rubber Chem. Technol 59, 809 (1986). [6.] D. G. Young, Rubber Chem. Technol 63, 567 (1990). [7.] D. G. Young, Rubber World 4, 30 (1991). [8.] G.J. Lake and P.B. Lindley, J. Appl. Polym. Sci. 8, 707 (1964). [9.] U. Esele, S. Kelbch and H. W. Engels, Kautsch., Gummi, Kunstst. 45, 1064 (1992). [10.] A.J.M. Sumner and H. Fries, Kautsch., Gummi, Kunstst. 45, 558 (1992). [11.] D. Wall, Polysar internal technical memorandom (1991).