K. E. Goeck-e is a Chief Mechanical Technician, University of I)avton. Research Iinstitute, 300 College Park, Dayton, OH 45469-0128. Original manuscript ...
A Technique to Measure Fatigue Crack Growth Threshold by K. E. Goecke and M. A. Moshier ABSTRACT-An experimental technique for determining fatigue crack growth threshoid is presented. This experimental technique uses an increasing AK step loading procedure to determine threshold going from a no-growth to growth status. Stress relief annealing of the Ti-6A1-4V test specimens eliminates load history effects normally associated with the precrack, providing a measurement equivalent to what is achieved by a standard ASTM load shed test. In addition to measuring load history free thresholds, this increasing AK technique can be used to investigate different load history effects on threshold by using the threshold step measurement with different precrack histories and without the subsequent annealing process. Verification of the threshold step measurement is demonstrated by comparing measurements with standard ATSM load shed testing results. KEY WORDS-fatigue crack growth, threshold, titanium, experimental testing Equipment
The fatigue crack growth threshold AK step test requires the following test equipment: fatigue loading frame, load cell, servo-hydraulic controller, constant current DC power supply, voltmieter with micro-volt capability, resistance temnperature detector (RTD), optical microscopes, and a comI1-
puter for control of the load signals and data collection. A s'hematic of the equipment setup is shown in Fig. I anid will be discussed further in the experimental setup section. Material and Specimens
The Ti-6-4 aIloy used in this study was supplied in the solution-treated overaged (STOA) condition. Plate material was forged from a parent bar producced in accordan ce with AMS 4928. After forging, the miaterial was preheated at 938 0 C (± 11C) for a minimum of 30 minutes, solution treated to 932°C (±14'C) for 75 minutes, and fan air cooled. The plate was then mill annealed in vacuuLm at 704 0 C (±14'C) for two hours, and fan cooled in argon. The restalt was a titanium alloy with an alpha-beta microstructure of approximately 60% primary alpha with the rema'inder transformed beta. t The mechanic'al properties of the Ti-6-4 plate are oy = 930 MPa and GuTs = 978 MPa. M. A. Moshier, AFRAIMLLM, Wright-Patterson AFB, OlH 45433, is curren tly a Sen ior Research Scientist, RHAMM 'kchnologies,Beavercreek, OH 45434. K. E. Goeck-e is a Chief Mechanical Technician, University of I)avton Research Iinstitute, 300 College Park, Dayton, OH 45469-0128. Original manuscript suibmitted: Decemnber 12, 2000. Final manuscript received: I)ecember 17, 2001.
182 e Vol. 42, No. 2, June 2002
Standard ASTM compact tension C(T) specimens were used. The dimensions for W, B. and A, are 40 mm, 10 mm,
and 7.2 mm respectively. To facilitate optical crack lerngth neasurements on the surface, the C(T) specimen surfaces were mechanically polished to a mirror-like finish by progressively decreasing the abrasive size used for polishing. A onie micron diamond paste was used for the final polishing. Experimental Setup
The experimental setup is shown in Fig. 2(A). The AK step test for d etermining threshold uses an experimental setup that is similar to that used for a standard ASTM load shed threshold test. Fatigue crack growth threshold is determined using a block loading procedure.2 Direct current potential difference (DCPD) is used as the measurenvent method for determining the onset of fatigue crack growth from a nongrowth status. The basic concept of DCPD is to monitor the voltage difference across the crack created by an applied constant curreit to the specimen. Crack extension is then indicated by an inc'rease in the voltage difference across the crack caused by the increase in resistance to current flow. Crack extension is also verified visugally with the aid of traveling optical microscopes. In order to properly set up the test frame for DCPD, tooling for the testing of C(T) specimens is installed in the test frame with attention to isolating the test specimen electrically from the test frame. The use of strengthened zirconia pins with mica washers between the specimen and grips can be used to provide an added level of electrical isolation (Fig. 2(B)). Isolating, the test specimen from the test frame ensures that the specimen will be the only current path, thus changes in voltage readings are due solely to crack growth. Wire leads are attached directly to the specimen for both the current input and the voltage pickup leads. Four threaded holes are machined on the face of the specimen and brass screws inserted into the holes to provide ease in the attachment of the leads (Figs. 2(A) and (13)). In addition to monitoring the voltage difference across the crack, a resistance temperature detector (RTD) is attached to the specimiien for moniitoring the specimen temperature. A rise in the specimen temperature causes an inc'rease in the specimeni resistan ce and a resulting increase in the measured voltage and can therefore erroneously indicate crack extension. To maintain the specimen electrical isolation, the use of thermocouples for monitoring specimen temperature is not recommended.
The step loading is specified in terms of a constant K for each loadinig block. The computer calculates the load
Fig. 1-Schematic of test setup
from the appropriate K solution using the geometry, crack length, and specified K. The load signal is then generated by computer and used as the command signal for the servohydraulic controller. Load, temperature, and DCPD signals are acquired and the DCPD and temperature data plotted so that crack extension can then be determined. Prior to measuring the long crack threshold using the constant R, AK step test, each C(T) specimen is precracked, to provide a crack of sufficient size and straightness to eliminate the effect of the machined starter notch and to eliminate the effect of any load history. Optical microscopes are installed on the load frame to monitor crack length during the precrack portions of the test. It was found that precracking less than 0.50 mm influenced the subsequent threshold measturement. Therefore, a conservative precrack value of 0.75 mm was used for establishing each specific load history. The load history is established at a constant AK and R by optically measuring the crack in specified intervals so that the load can be calculated and adjusted by the computer. In order for the stress intensity factor solutions to be accurate, the difference in crack lengths from side to side should not be greater than 20% of B (specimen thickness) and steady crack growth should be observed on both sides of the test specimen. Once crack extension is detected the test is stopped. After crack extension is detected a new precrack can be established with a new 0.75 mm of crack growth at a specified K. After the new precrack is completed, the threshold is measured in
Iieie B Fig. 2-Photo of test setup (view A), photo showing load pins, mica washers, and DCPD attachment screws (view B)
the same manner as before. Precracking and threshold testing in this manner allows for investigating many load histories. Multiple threshold tests can be conducted on each C(T) specimen. The specimen geometry used in this study allowed for up to 16 threshold tests. Constant R Increasing AK Step Loading The step load procedure is used to determine at what AK crack extension occurs. The method involves subjecting a specimen to 200000 cycles at a constant R and AK below which crack extension is anticipated to occur. If the crack extension is not detected within that block of 200000 cycles, Kmnax is increased 0.2 MPa mi/ 2 and R held constant and the test continued until the crack extension occurs. The computer does not adjust the loads during each loading block. The resolution of the system used in this technique to detect Experimental Mechanics * 183
C4 i. -1.1- I......
-11
-
crack extension is sufticient to detect crack extension increments of less than 2 [tm of growth, based on finite difference calculations and the voltage resolution of the system. Using a constant threshold growth rate as defined by ASTM of 10-10 n cycle, the 2 .m resolution would require 2000(0 cycles to detect any crack extension. To ensure that crack growth is detected in a block of loadinig, 200000 cycles are used in each loading block. At worst, the crack growth rate of the measured threshold is 1(-10 m/cycle, and could be as low as 10-1i mJcycle. The threshold is defined to be the average of the K levels where no crack extension occurs and where 'rack extension first occurs. Because the increm
