VERIFICATION OF A TECHNIQUE FOR HOLOGRAPHIC RESIDUAL STRESS MEASUREMENT Michael Steinzig* Gregory J. Hayman* Michael B. Prime* *HYTEC Incorporated, 110 Eastgate, Los Alamos, NM 87544 ABSTRACT This paper describes a holographic hole-drilling system for measuring residual stresses. The technique uses optically measured surface displacements near a drilled hole to obtain a measure of the specimen’s residual stress. Unlike manual fringe counting methods, the system described here uses automated software to both acquire displacement data and convert it to residual stress. We describe a verification of the software implementation where the results from the holographic system are compared to a “known” state of stress in a specially prepared ring and plug specimen. The verification shows that the software implementation can give residual stress results that compare well with predicted stress levels over the full range of planar stress states: from equi-biaxial to uniaxial to opposed-biaxial (one component tensile, one compressive). INTRODUCTION The hole drilling method is a widely used technique for experimentally determining residual stress. Typically, a strain-gage rosette is used to measure surface strains that result when subsurface stress is relaxed due to the drilling of a small hole. These strains are then related to the state of stress in the hole region prior to the stress relaxation. In a similar manner, one can relate surface displacements to sub-surface stress states. The use of optical methods for measuring the surface displacements encountered during hole-drilling was developed during the 1980’s. The driving force for this development was the inherent disadvantages of the strain gage method for measuring residual stresses; that is, the strain gages must be applied to a plane and smooth surface, the hole must be drilled very precisely in relation to the strain gage, and significant time and expense are required to apply the gages. In (McDonach, et al., 1983), the authors developed the application of moiré interferometry in conjunction with hole drilling. Holographic hole drilling was explored independently by (Antonov, 1983), (Bass, et al., 1986), and (Nelson and McCrickerd, 1986) during the mid 1980’s. In these approaches, an interference fringe pattern relates the displacements that have occurred as a result of the hole-drilling to the sub-surface residual stress. Numerous authors published variations of the holographic technique throughout the 1990’s, (Hung and Hovanesian, 1990), (Lin, et al.,1994), (Zhang and Chong, 1998). More recently, (Diaz, et al., 2000) incorporated an in-plane sensitive electronic speckle pattern interferometer (ESPI) system, and automated fringe analysis for rapid stress calculation. INTERFEROMETRY OVERVIEW The following is an introduction of the topic of ESPI. For a more thorough review of the subject matter, refer to (Jones and Wykes, 1989) or (Rastogi, et al., 1997). In a classical interferometer, two beams of coherent light, henceforth referred to as the illumination and reference beams, are combined and the resulting wavefront is passed to a detector. In speckle pattern interferometery, one or both beams exhibit speckle patterns due the reflection of said beam(s) off of a diffusely reflecting surface. When these two beams form an image on a charge-coupled device (CCD), an interferogram results. The CCD registers the intensity of the speckle pattern interferogram as: I(x, y ) = A 2I (x, y ) + A 2R (x, y ) + 2A I (x, y )A R (x, y )cos[φ (x, y )] (1) where the (x,y) pair denotes a location on the test specimen as imaged onto a particular sensor in the CCD array. I is the interference intensity at the CCD array, AI and AR are the amplitudes of the illumination and reference beams, respectively, and φ is the effective phase angle of the interference. As the path length of either light beam changes, the phase angle, φ, changes. For a frequency-doubled Nd:YAG laser a 2π change in the phase angle corresponds to 532 nanometers of overall path length change. For a typical ESPI arrangement, a single coherent laser beam is split to generate a reference beam and an illumination beam. The illumination beam is reflected off of the test surface and is then imaged onto a detector. The reference beam is sent directly to the detector, as in Figure 1. Published in Residual Stress Measurement and General Nondestructive Evaluation, PVP-Vol 429, The ASME Pressure Vessels and Piping Conference; Atlanta, GA July 23-26, 2001. Ed. Don Bray. Observation Direction Zoom Illumination Direction CCD Fiber Test Surface LASER PZT Framegrabber PC Monitor Figure 1. Typical ESPI system layout This interferometer can be used to measure shape change of the test surface because surface displacements result in a path length change in the illumination beam. The resulting interferogram can be modeled as I' (x, y ) = A 2I + A 2R + 2A I A R cos[φ + ∆P ] (2) where the (x,y) notation has been dropped to simplify the equations. In order to quantify the surface displacements, the optical path length change ∆P must be solved for in Equation (2). The optical path length change must then be related to the mechanical motion of the test surface. One method of solving for ∆P is to introduce known path length changes in the reference beam using a phase shifter, and to acquire four interferograms before and after surface displacement. The equations (before and after surface displacements, respectively) are: I i = A 2I + A 2R + 2A I A R cos[φ + ∆S i ] , (3) I i' = A 2I + A 2R + 2A I A R cos[φ + ∆S i + ∆P ] ∆Si is the known phase shift and has values of 0, π 2 , π , and 3π 2 for i = 1, 2, 3, and 4, respectively. The path length change due to the surface displacements can now be solved for using ( ( )( )( )( )( )( )( I' − I' I − I + I' − I' I − I 1 3 1 2 4 2 4 3 ∆P = tan −1 ' ' ' ' I 2 − I 4 I1 − I 3 − I1 − I 3 I 2 − I 4 ) ) (4) This result must be passed through a phase unwrapping algorithm to remove the modulo 2π effect caused by the arctangent function. The path length change is related to the mechanical surface displacements by ∆P = 4π λ K •u = 4π λ (K xux + K yu y + K zu z ) (5) where K is the sensitivity direction as defined by the angle bisector of the illumination and observation directions, shown in Figure 1. RESIDUAL STRESS MEASUREMENT SYSTEM In the current work, a single-illumination-beam phase-shifting ESPI has been developed for the measurement of residual stress. The system’s software and related algorithms quickly and easily transform surface displacement measurements into residual stress values. System Configuration The residual stress measurement system consists of a laser unit, optical fibers, a video head, an illumination head, and a personal computer (PC). The laser unit houses a single frequency 20mW Nd:YAG laser with its associated power supplies and controllers, beam splitting optics, laser-to-fiber couplers, and a phase shifting device. Its purpose is to generate the coherent laser light, split this light into a reference beam and an illumination beam, and then deliver these two beams to the video head and illumination heads, respectively, via fiber optic cables. The illumination head collimates the light arriving from the laser unit and is used to illuminate the test specimen’s surface. The video head images the test surface and the reference beam onto a CCD camera. The PC is used to control the phase shifting of the reference beam, acquire video images from the CCD, save this data Published in Residual Stress Measurement and General Nondestructive Evaluation, PVP-Vol 429, The ASME Pressure Vessels and Piping Conference; Atlanta, GA July 23-26, 2001. Ed. Don Bray. to the hard drive, and then analyze the data to obtain a residual stress measurement. Stress Analysis The stress analysis algorithms are based on the work of (Nelson, et al., 1997), but have been adapted to work well with an ESPI system. A set of interferograms are acquired and saved to memory. Then, the drill is placed in front of the test specimen and a small hole is drilled to a given depth. Next the drill is removed and more interferograms are acquired. These interferograms provide the change in the illumination beam’s path length, ∆P, which corresponds to the surface displacements caused by the stress relaxation. This path length change is related to the residual stress through the following equation, ∆P1 C11 ∆P 2 = C 21 ∆P 3 C 31 C12 C 22 C 32 C13 σ xx C 23 σ yy C 33 τ xy (6) in which Cij contain constitutive properties, geometric properties, holographic sensitivity factors, and finite element derived coefficients, and ∆Pi = ∆P (ri , θ i ) − ∆P (ri , θ i + π ) (7) in which ∆Pi is the phase (path length) difference between two surface points which are diametrically opposed to one another but at the same radial location from the hole center. to a solution heat treatment, cold working to improve strength, followed by natural aging to ensure stable properties. The T351 designation refers to plate that has been stress relieved after processing by uniaxial stretching to 1.5-3.0% strain. After this stretching, residual stresses on the order of 1-2 ksi (6.8-13.6 MPa) are expected to exist in the material (Prime, et al., 1999). The modulus used for this material was E=73.08 GPa, Poisson’s Ratio ν=0.33, and yield strength of 290 MPa. The specimen was nominally a 4” (101.6 mm) diameter ring, with a 2” (50.8 mm) diameter hole in the center and a corresponding 2” plug. The ring/plug diametrical interference was approximately 0.005” (0.127 mm). The components were 0.5” (12.7 mm) thick. For assembly, the plug was cooled to liquid nitrogen temperature (-160 °C) leaving approximately 0.0038” (0.0965 mm) clearance with the room temperature (20 °C) ring. Grease was used to lubricate the interface and minimize any friction between the surfaces. Prior to assembly, eight strain gages were mounted on the ring and used to precisely determine the stress state after the assembly. Figure 2 shows the locations of the gages, six of which measured the radial strain component and two of which measured hoop strains. The gages were placed on the ring near the plug interface where the strains are the highest. One gage would be sufficient to uniquely determine the stress state, however, using multiple gages allow one to confirm that the stress state is axisymmetric, and to establish uncertainty bounds on the stresses. strain gages Therefore, three holographic pairs of data are sufficient to determine the plane-stress state of the material being drilled into. However, in ESPI, the CCD typically consists of 256,000 pixel elements. Therefore, thousands of data points around the hole are available to determine the stress state. Currently the software samples thousands of phase change triads, with each triad providing an independent stress calculation. The individual results are averaged to provide the reported stress value. RING AND PLUG SPECIMEN An aluminum shrink-fit ring and plug was chosen as the known residual stress specimen for several reasons. First, this specimen has a closed form solution for the residual stresses, and the stress distribution is relatively simple: the stresses are constant in the plug, and only a function of radial position in the ring. Second, the ring and plug specimen has appeared routinely in the literature as a test specimen for evaluating residual stress measurement techniques, e.g., (Gnaupel-Herold et al., 2000). Third, this type of specimen provides the full range of biaxial stress states for testing. In the ring near the interface, σθ is positive and σr is negative. Near the outer edge of the ring the stress state is nearly uniaxial since σr goes to zero. In the plug, the stress state is equi-biaxial, σθ = σr. Therefore, with one specimen it is possible to demonstrate the ability of the holographic system on three stress states. Aluminum 2024-T351 was chosen as the material for the ring and plug. The 2024 alloy is readily available, well characterized, and has good yield strength characteristics. The T3 temper refers Figure 2. Strain gages locations on ring and plug The stress state of the ring and plug can be calculated (Shigley and Mitchell, 1998) by first calculating the pressure p= ( Eδ Ro2 − Ri2 2 Ri Ro2 ) (8) which is taken as positive, and δ is the radial interference between the ring and plug. In the plug the stress is equal to the pressure in the radial and angular directions. In the ring, the stress is: Published in Residual Stress Measurement and General Nondestructive Evaluation, PVP-Vol 429, The ASME Pressure Vessels and Piping Conference; Atlanta, GA July 23-26, 2001. Ed. Don Bray. 2 1 − R o Ro2 − Ri2 r 2 pR 2 R2 σ θ = 2 i 2 1 + 2o Ro − Ri r σr = pRi2 (9) was added to the pressure calculated from the strain gage readings to give the “known” pressure of 9.52 ksi, (65.64 MPa) and calculate the residual stresses plotted in Figure 4. EXPERIMENTAL SETUP AND PROCEDURE Equation 10 can be combined with the usual elastic relations to give ε (p,r), which can be inverted to give p(ε,r) and thus the strain gages applied to the surface of the sample can be used to calculate the pressure. The eight measured strains gave an average value for the interference pressure of p=8.27±0.45 ksi (57.0 MPa ± 3.1). The measured strains also confirmed that the stresses were axisymmetric to within the precision of the measurements. Further confirmation of the accuracy of the strain gage readings is provided by the measured interference between the ring and the plug before assembly of 0.0023” ± 0.0003” (0.058 mm ± .0076 mm), which gives a calculated value of p=9.19 ksi ± 1.16 (63.4 Mpa ± 8.0). Also, the change in outer diameter of the ring after assembly was measured as 0.0019" ± 0.0006" (0.048 mm ± 0.015 mm), which gives a calculated value of p=7.47 ksi ±2 (51.5 Mpa ±13.8). A pressure of 8.27 ksi was used for comparison to the holography system measurements, since that value is expected to be more accurate than the dimensional measurements. Several measurements of the as-received 2024-T351 plate using the holography system indicated that the residual stresses near the surface of the plate were 1.25 ± 0.42 ksi (8.6 Mpa ± 2.9), consistent with the reported values for residual stress in 2024-T351 aluminum, as reported by ALCOA. The initial stress of 1.25 ksi Ring & Plug Specimen For these tests, a 3-foot by 2-foot granite table was used as the workspace. The video head, illumination head, ring and plug specimen, and drill assembly were each secured to a 2-foot by 2foot optical breadboard, which in turn was placed onto the granite table. The laser unit was placed directly behind the optical breadboard, as shown in Figure 3. The optical breadboard allowed for easy alignment of the video head and drilling assembly relative to the ring and plug specimen. The drilling axis and the video head were placed perpendicular to the surface of the test specimen. The illumination head was oriented such that the angle between the test surface and the illumination beam was 25.5°. The drilling assembly consists of a small air-driven turbine drill that is mounted to a translator table. The drill operates between 30,000 and 40,000 rpm. Such drills have been shown to drill a hole without introducing significant stresses (Flaman and Herring, 1985). The drilling depth is manually controlled via a micrometer screw that has a resolution of 0.01 mm per increment. The bit used was a 1/16” (1.59 mm) diameter square-edged end mill with 2 flutes. Holes were spaced at least five hole diameters apart, which means that neighboring holes would change the stress distribution by less than 1%. Holes were drilled variously to depths of 0.5 and 1.2 times the hole diameter to confirm that different hole depths did not produce different results. Laser Unit Illumination Head Drill Video Head Figure 3. Experimental setup of residual stress measurement system and the test specimen Published in Residual Stress Measurement and General Nondestructive Evaluation, PVP-Vol 429, The ASME Pressure Vessels and Piping Conference; Atlanta, GA July 23-26, 2001. Ed. Don Bray. There are several other possibilities for errors in this setup. Two ratios of hole depth to diameter were used, 0.5 and 1.2. Since the end mill had a nominal diameter of 0.0625” (1.588 mm), this means the hole depths were nominally 0.03125” and 0.075” (.79 and 1.91 mm) deep. In practice, it is difficult to achieve this depth exactly, and so each depth was measured. It was found that in most cases the holes were within 0.001” (0.0254 mm) of the desired depth, with a few holes being as far as 0.003” from the expected depth. In addition, the hole diameters were measured, and found to be consistently larger than the nominal diameter by about 0.008” (.212 mm). This difference between nominal and actual measurements can have a significant impact on the results, but since the software is “hard wired” for only specific h/D ratios, the results in Figure 4 are based on the nominal h/D values The larger than nominal hole sizes may indicate an eccentric alignment of the bit head with the drilling axis, or that the drilling axis is not exactly perpendicular to the axis of the ring and plug. Either of these situations will contribute to errors. It might be expected that the error would be completely fixed, but in practice these two parameters (the eccentricity and perpendicularity) would probably change every time the drill bit or the part position was changed. The issue of the drill being not perpendicular to the part may have error significance of its own, since the original equations were derived assuming complete perpendicularity. RESULTS AND DISCUSSION In order to test the holographic system’s ability to measure various residual stress states, holes were drilled in many locations on the surface of the ring and plug. The results of the measurements with the holographic system and the strain gage results are plotted in Figure 4. The plug is in a state of equalbiaxial compression, shown by the horizontal line from a distance of 0 to 1” (0-25.4 mm). The ring has opposed biaxial stress, compressive in the radial direction and tensile in the hoop direction. It should be noted that the radial stress must go to zero at the outer edge of the ring, and therefore the initial 1.25 ksi stress added to the calculated pressure was reduced exponentially to zero as a function of radius. Although no error bars have been shown in Figure 4, we can see that the values measured with the holography system span the expected value fairly well. The worst discrepancy seems to occur near the ring/plug interface, especially in the hoop stresses. We speculate that this is caused by the gradient of stress as a function of radial position. The original equations are developed for a stress state with no gradient, and have not been adjusted to account for this. Qualitatively, it makes sense that this would have the largest effect on the hoop stresses in the ring near the interface, since both the gradient and the magnitude are largest there. Ring and Plug Holographic and "Known" stresses 20.0 18.0 Calculated-strain gage 16.0 14.0 H/D=1.2 12.0 H/D=0.5 Stress [ksi] 10.0 Plug 8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0 -8.0 -10.0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Distance from plug center to hole center [in.] Figure 4. Holographic hole-drilling results compared to strain gage analysis of residual stress Published in Residual Stress Measurement and General Nondestructive Evaluation, PVP-Vol 429, The ASME Pressure Vessels and Piping Conference; Atlanta, GA July 23-26, 2001. Ed. Don Bray. CONCLUSIONS In general, residual stress measurement with any technique available today can be expected to have uncertainties of at least ± 2-3 ksi (6.9-20.7 MPa). (Lu, 1996) The data from the holographic instrument described here certainly seems to be within that range, and actually seems better than that, although the “known” stress used in this validation also has an uncertainty of at least ±0.5 ksi (3.4 Mpa), making quantification of the accuracy of the holographic method difficult. We have manually adjusted the analysis to account for the actual h/D values for 4 different holes (drilled in the ring near the interface) and found that in each case both the hoop and radial stresses dropped by about 10%. In future work, the software will be modified to allow the actual h/D diameter as an input, to account for variations in hole depth and diameter. The effect of stress gradient and non-perpendicular holes will also be investigated. In particular, it should be possible to utilize the holographic displacement information to back-calculate a value for the stress gradient. Direct comparisons with other methods of stress measurement, (x-ray diffraction and neutron scattering) are also planned for future studies. ACKNOWLEDGEMENTS We would like to thank Professor Drew Nelson for his input during the software development. Jack Hanlon contributed to the optical development of the PRISM system during the time this project was a Co-operative Research and Development Agreement with Los Alamos National Laboratory. REFERENCES Antonov, A. A., 1983, “Inspecting the level of residual stresses in welded joints by laser interferometry”. Weld. Prod. Vol. 30, pp. 29-31. Bass, J. K., Schmitt, D., and Ahrens, T.J., 1986, “Holographic in situ stress measurements”. Geophys. J. R. Astr. Soc. Vol. 85, pp. 13-41. Diaz F.V., Kaufmann G.H., and Galizzi G.E., 2000, “Determination of residual stresses using hole drilling and digital speckle pattern interferometry with automated data analysis”, Optics and Lasers in Engineering, Vol. 33, pp. 39-48. Flaman, M. T., and Herring, J. A., 1985, “Comparison of Four Hole-Producing Techniques for the Center-Hole Residual Stress Measurement Method,” Experimental Techniques, Vol. 9, pp. 3032. Gnaupel-Herold, T., Prask, H.J., Clark, A.V., Hehman, C.S., Nuygen, T.N., 2000, “A comparison of neutron and ultrasonic determinations of residual stress”, Measurement Science and Technology, Vol. 11, pp. 436-444. Jones, R., and Wykes C., 1989, Holographic and Speckle Interferometry: A discussion of the theory, practice and application of the techniques, Cambridge University Press, Cambridge. Lin S.T., Hsieh C.T., and Hu, C.P., 1994, “Two holographic blind-hole methods for measuring residual stresses”, Experimenal Mechanics, Vol. 34, pp. 141-147. Hung YY, and Hovanesian JD, 1990, “Fast detection of residual stresses in an industrial environment by thermoplastic-base shearography”, Proceedings of the SEM Spring Conference on Experimental Mechanics, Albuquerque, NM, pp. 769-775. Lu, J., 1996, Handbook of Measurement of Residual Stresses, Fairmont Press, Lilburn, GA. McDonach, A., McKelvie, J., MacKenzie, P.M., and Walker, C.A., 1983, “Improved moiré interferometry and applications in fracture mechanics, residual stress and damaged composites”, Experimental Techniques, Vol. 7, pp. 20-24. Nelson, D. V., and McCrickerd, J. T., 1986, “Residual-stress determination through combined use of holographic interferometry and blind hole drilling”, Experimental Mechanics, Vol. 26, pp. 371378. Nelson, D. V., Makino, A., and Fuchs, E. A., 1997, “The Holographic-hole drilling method for residual stress determination”, Optics and Lasers in Engineering, Vol. 27, pp. 323. Prime, M.B., Jacobson, L., Pacheco, M. 1999, “Residual Stresses Measured Before and After Stress Relief in Rolled Aluminum Plate,” Book of Abstracts, 1999 ASME Mechanics and Materials Conference, R. C. Batra and E. G. Henneke, ed., Blacksburg, Virginia, pp. 241-242. Rastogi, P.K., ed., 1997, Optical Measurement Techniques and Applications, Artech House, Inc., Norwood, MA. Shigley, J. E., and Mitchell, L. D., 1998, Mechanical Engineering Design, McGraw Hill, pp. 76-77. Zhang J., and Chong T.C., 1998, “Fiber electronic speckle pattern interferometry and its applications in residual stress measurements”, Applied Optics, Vol 37, No. 28, pp. 6707-6715. Published in Residual Stress Measurement and General Nondestructive Evaluation, PVP-Vol 429, The ASME Pressure Vessels and Piping Conference; Atlanta, GA July 23-26, 2001. Ed. Don Bray.