You are on page 1of 8

Fatigue Testing

Laboratory 7 Report November 5, 2013

Mechanical Behavior of Engineering Materials (ME108) Fall 2013 University of California at Berkeley

Group 7 Lab Section 104, 1st Rotation

Joey Johnescu Kevin Kung Nick Renda Jon Saltz

!"#$%&'(&)'*+%*+,& -./&& 5./&& 4./&& >./&& C./& D./& E./& F./& H./& 012!30)!&.......................................................................................................................................&4& 67!389:)!687&.............................................................................................................................&4& !;<83=&............................................................................................................................................&4& <?@<36A<7!0B&A<!;892&........................................................................................................&>& 3<2:B!2&...........................................................................................................................................&C& 962):22687&.....................................................................................................................................&D& )87)B:26872& .................................................................................................................................&E& 3<G<3<7)<2&...................................................................................................................................&E& 0@@<796?&........................................................................................................................................&F&

"

1.0

ABSTRACT In this lab we observed the fatigue behavior of steels. Three 1045 steel samples were put

under high-speed cyclic loading in an R. R. Moore testing machine using different applied weights until failure. After failure, the fatigue life in cycles was recorded and compared for each case to our calculated expectations. Our results showed a lot of variation and indicated that fatigue behavior is unpredictable and heavily dependent on specific sample properties. The broken samples were then viewed under an optical microscope. From these images the crack propagation could be observed, and a micro-scale analysis of the failure modes could be done. 2.0 INTRODUCTION Safety has become a fundamental concern and limiting factor as engineers continue to push the boundaries of mechanical design. One possible cause of material failure occurs when materials undergo fatigue, a type of failure caused by cyclic loading at stress levels below the ultimate strength of the material. Fatigue occurs in the stages: crack initiation (formation of microcracks containing stress raisers), crack growth (crack propagation), and fracture (sudden failure). For this lab, we will perform stress-control fatigue tests according to the ISO standard code, examine the fatigue properties of steel specimens under different stress amplitudes, and use optical microscopy to study how crack growth was affected by the loading conditions [1]. Fatigue tests will be performed with straight shank specimens consisting of AISI 1045 steel. 3.0 THEORY Fatigue is a type of failure which results from cyclic loading at stress levels below the ultimate strength (i.e. the point at which a statically loaded sample would fracture). Fatigue damage is the cumulative effect of many small plastic deformations that occur during cyclic loading [2]. It is generally thought of as having three stages. Crack initiation is when microcracks form in areas with stress raisers or defects, like notches or rough surface features. Crack growth is when these cracks propagate and finally, fracture is when a dominant crack leads to sudden material failure. Fatigue Life (Nf; [cycles]) is defined as the number of stress cycles or strain reversals that a material experiences prior to fracture, and is simply the sum of the periods for the 3 stages mentioned, but since fracture is nearly instantaneous, it is usually omitted. We are left with Nf = Ni + Ng, where Ni corresponds to crack initiation and Ng corresponds to crack growth. For brittle materials, this is further simplified by approximating that Nf ! Ni since fatigue life is dominated by crack initiation. Likewise for ductile materials, we say that Nf ! Ng [3].

Fatigue can be analyzed in three different ways, with two macroscopic approaches that look at the relationship between Nf and global parameters, namely stress amplitude ("#; [MPa or ksi]) and strain amplitude ("!; [no units]), and a microscopic approach which relates Nf to the local driving force of fatigue, stress intensity ("K). In this lab, we only deal with the stress-based macroscopic approach, in which we use the nomenclature that S="# and N=Nf. The S-N approach is only valid for high-cycle fatigue (e.g. Nf > 103), where the stress is mostly within the elastic region. When S and N are plotted on a semi-log plot, one can see two distinct regions of finite and infinite fatigue life. In the first region (~103 < Nf < ~106), the S-N curve follows the power law relation S=A(Nf)b where A and b are material-specific fatigue parameters with A = S10002/Se and b = (1/3)log(S1000/Se). S1000 is the stress amplitude corresponding to a thousand cycles and Se is the endurance limit, or the stress amplitude that will provide an infinite fatigue life. Empirical testing has found that S1000 ! 0.9SU and Se ! 0.5SU, which leads to a general equation of S ! 1.62SU(Nf-0.085) [ksi]. In the second region (Nf > ~106), S=Se by definition [3]. In this lab we also used an optical microscope to observe our samples after fracture, so as to gain some insight into the differences between our experimental and calculated results. We expect to see smooth bands in the region where gradual crack propagation occurred, which usually corresponds to a scratch or defect in the materials surface. Likewise, a rough fracture surface indicates rapid propagation, which occurs during material failure [2]. 4.0 EXPERIMENTAL METHODS For this experiment, we used three straight shank specimens of AISI 1045 steel that were machined in accord to the standards set by the ISO code 1143 [1]. The test was preformed with an Instron R.R. Moor high-speed rotating beam fatigue testing apparatus with a loading capacity in the range of 9 to 101 lb and a maximum rotational speed of approximately 10,000 rpm. The specimen was mounted in two holders with loading harness and weight pan assemblies used to applied specified load. Upon specimen failure, the machine is equipped with a safety-switch that automatically stops the motor to prevent damage to the machine. For the experiment, the following steps were taken for each test specimen: 1. Measure and record the dimensions of the fatigue specimen 2. Load the specimen in the machine 3. Set up the loading hardness and weight pan assemblies with a minimum of 9 lb of weight 4. Set the rotational speed to 50% using the speed control knob and turn on the power entry module $

5. Turn on motor 6. Once at steady state, add additional weight to reach specified load and reset the rotation counter immediately once the weight is applied 7. Once the specimen fails, record the number of revolutions, turn off the power entry module, unload the eight, and dismount the specimen The steps were repeated with weight values of 12, 13, and 14 lbs. 5.0 RESULTS Our primary results are shown in the charts and tables below, as well as in Appendix A: Load W (lb) Moment arm L (in.) Specimen minimum diameter D (in.) Stress amplitude S (ksi) (Eq 1) Fatigue Life N Calculated (Eq 2) Experimental

12* 4 0.1440 81.87 45228 379,553 13 4 0.1440 88.69 16635 49,473 14 4 0.1440 95.51 6590 57,562 [Table 1]: Summary of the data obtained note the difference between the experimental and calculated values of NF. *Signifies test was ended early, as the sample wouldnt break. !!"# ! !"!" !! !
! !!!!"

!!"# !! ! !"#

[Equations 1 and 2]&


*'& *)& 2+I%,,&0JK$L+MN%&OP,LQ& *$& *"& *(& ''& ')& '$& '"& '(& (& !(((((& "(((((& #(((((& 7MJ#%I&'(&)RS$%,& $(((((& +,-./-,012& 3451678190,-&

2&T,&7G&

[Figure 2]: A plot comparing the difference between the calculated and experimental values of NF. %

6.0

DISCUSSION From the graph in Figure 2 we can see a marked decrease in fatigue with increasing stress

amplitude, which agrees with the theoretical assumptions made. Crack propagation is directly proportional to stress amplitude, and increases with higher stress amplitudes resulting in a shorter fatigue life. Fatigue cycles follow a power-law relationship (shown in a slightly altered form in Eqn 2) as long as the loads occur in the materials elastic range. When comparing our results to the mathematical power-law model, the observed material fatigue performance of the samples does not quite agree with the predicted behavior. This difference between experimental and theoretical values can be attributed to sample variation, as well as the statistical bias that occurs from only testing three samples. Both the 12 and 14 lb samples performed much better than anticipated: the sample tested with 14lbs of weight exhibited more than nine times the expected cycles to failure, and the sample tested with 12lbs was broken by adding extra weight because it showed no signs of fatigue and there wasnt enough time to continue the experiment indefinitely. Its likely that these samples varied significantly, perhaps from a different density, dimension, or surface finish, which resulted an anomalous fatigue life. The experiment could be vastly improved by increasing the number of samples tested, as well as testing a wider variety of weights for more data points. In addition to using quantifiable measurements, we can use microscopic analysis on our samples to verify our theoretical assumptions. These photographs, showed in Appendix A, demonstrate several theoretical aspects of fatigue: crack propagation speed can be roughly approximated based on the surface topology at the cross section. Comparing the three cases also shows that the 14 lb loaded sample had a much rougher surface than the 13 or the 12 lb samples, which substantiates the idea that materials with shorter fatigue lives have rougher surface textures. The smoother surfaces in the 12 lb sample show a slower crack propagation speed, and measured a corresponding higher number of cycles before failure. In all three samples, we observed that the edges in general had rougher surfaces than other parts of the cross-section. This suggests that fracture began before cracks could completely propagate through the material, which indicates the fracture strength of the material was reached before the crack was completed. The rougher edges also indicate the presence of stress raisers, which are more likely to initiate cracks. The surface finish of the outer edge of one sample is shown in the last set of photos in Appendix A: it demonstrates that a surface smooth to the human eye is in fact full of imperfections, and can be a contributing cause to crack formation. )

7.0

CONCLUSIONS Our results largely did not correspond to theoretical fatigue life predictions. While we

expected the highest load to result in a shorter fatigue life, this wasn't the case. Also, the 12 lb load case resulted in a much higher fatigue life than expected. These discrepancies can be attributed to machining tolerances, or perhaps material defects. It should be noted that in our lab only three tests were performed. Standard fatigue life curves are derived from thousands of samples; in other words, our tests have no statistical relevance. Indeed, the fatigue failure by its nature is a stochastic process, one that relies heavily on probability. The results of this lab suggest that parts that undergo many cycles of loading should be designed with large factors of safety. 8.0 1. 2. 3. REFERENCES N. E. Dowling, Mechanical Behavior of Materials, 3rd edn (Upper Saddle River, NJ: Prentice Hall, 2007). K. Komvopoulos, Mechanical Testing of Engineering Materials, 2nd edn (San Diego, CA: University Readers, 2011). International standard, ISO 1143-1975 (E): Metals-Rotating Bar Bending Fatigue Testing.

9.0 9.1

APPENDIX Appendix A: Fractured Surfaces of Tested Specimens

12 lb weight, Side 1

12 lb weight, Side 2

13 lb weight, Side 1

13 lb weight, Side 2

14 lb weight, Side 1

14 lb weight, Side 2

'

14 lb weight, angled view Side 1

14 lb weight, angled view Side 2

You might also like