Ghostscript wrapper for D:\Digitalizacja\MTS84_t22z1_4_PDF_artyku³y\mts84_t22z1_2.pdf M E CH AN I KA TEORETYCZNA 1 STOSOWANA 1/ 2, 22 (1984) SOME PRACTICAL PROBLEMS OF DISPLACEMENT AND STRAIN MEASU REMEN T BY IN COH EREN T SU PERPOSITION OF INTERFEROGRAMS PIOTR W E S O Ł O W S K I Institute of Physics T echnical University Budapest I n this paper th e suitability of using the regulated pathlengh interferometer due to autom atization of 3- D displacem en t an d surface strain measurement as well as problems faced in dealing with t h e accuracy of th e m easurem ents are examined. D at a needed for determ in in g all th ree com pon en ts of displacement vector are stored on single plate. The displacement an d deform ation of surface poin t of a disc subjected to diametral compression was m easured to investigate th e effectiveness of this technique. Introduction The quan titative investigation of displacement vector fields by holographic interfe- rom etry requires recordin g in at least three different interferograms. The conventional m ultiple h ologram m eth od [3] gives different perspectives for views through each of the plates. C onsidering t h e autom atization of the evaluation of holograms (the first problem- autom atization of fringe counting) th e im portan t problem is, what kin d of interferometer should be used. I n this experim ent such an interferometer was chosen, which records in depen den t in terferogram s simultaneously by incoherent superposition of holograms [6]. Th e m ost im port an t property from a poin t of view of autom atic fringe inspection is an accurate identification of correspon din g object poin ts on each of interferograms; this is because of th e sam e observation al direction for all the interferograms. R egardin g th e m eth ods of evaluation of the displacement vector, the authors [2] exam ining th e H ain es an d H ildebran d m ethod, th e Bronch- Bruevich single hologram m eth od as well as the im proved version of the latter (D hir- Sikora m ethod), have arrived at a conclusion t h a t by com parin g theoretical an d experimental results, the most suitable an d with the smallest error is th e D hir- Sikora's. This m ethod uses at least four observation poin ts t o obtain an overdeterm ined set of linear simultaneous equations relating th e fringe shifts with th ree un kn own com pon en ts of displacement and obtain s the result by th e least square principle [11]. The discrete values of displacement com pon en ts are sm oothed ou t by cubics splines, after th is strain s are com puted. 70 P . WE SO Ł O WSK I The most important properties of regulated pathlength interferometer The recording of different interferograms on a single plate for evaluation of 3- D di- splacements is made simultaneously. Independency of interferograms is a result of the absence of correlation between beams n ot belonging together. The uncorrelation can be realized by producing an optical pathlength difference among corresponding beams, which is larger than the coherence length of the laser used in the experiment. I n order to be able to reconstruct interferograms separately there must be present so much reference beams as illumination beams. H olographic displacement measurement Let us investigate the problem in D escartes orthogonal coordin ates: y Observation F ig. 1 A displacement vector: L = L x i+L y j+L z k. ( I ) P 3 is a point on the object corresponding to a space vector Rp, for the object illuminated from a point source defined by R t , the reconstructed image can be observed from a point described by R 2 . Let us define the illumination vector K t and the observation vector K 2 : Ł . = x p i+y p j+z p _k, (2) tr — i * * - * I ) 2 + (y P - y t ) 2 (x 2 - x p )i+ (y 2 - y P \ Rz k- K u A ~ 2 where K t , K 2 are the unit illumination and observation vectors, respectively, and \ St\ - \ Et\ - k - - jp (3) (4) (5) PROBLEMS OF MEASUREMENT 71 is the magnitude of these vectors, with the wavelength X of the laser light. The phase shift due to observations of the viitual image from different directions can be written as: Ó = KL = 2nN. (6) where 2nN = Q (7) is fringe locus function. K matrix consist of m different sensitivity vectors: K = K1 K2 K3 where (8) Km = KT-KT, and tn 2: 3, means number of observations. Independent on how carefully and how many observations are made, there will be always some errors E. F — K• T —O f 91 The purpose of the analysis is to mininalize all of the errors squared, that is: In order to minimize the square of error the partial derivates of Eq. (10) must be zero: r — I > UfC" — Km\ • T — Q m l 2 l — 0 n n where i =» x, y, z m = 1,2, 3, ...,/• is number of all observations. The solution of Eq. (11) in matrix form is: L = (Kr • £ ) ] • KT N2n (12) where KT is transposed matrix of K. Experiment As a model there was used a plexiglass disc with diameter 90 mm, thickness /; = 10 mm (E = 3.2 • 104 N/m 2 , fi = 0.385) subjected to diametral compression [4]. The optical elements used in experiment: Mmirrors, BSbeamsplitters, COLL.colli mators, PLmicroscope lens with pinholes, Hholographic plate (Agfa Geavert 8E75), Oobject, LHeNe laser with output power of 50 mW. In the measurement the „zerofringe" method was applied by using an elastic strip 72 P . WE SO Ł O WSK I d M4 which was stuck between the loading framework and the holographic slab. Two collimators (Carl Zeiss Jena make) were used for producing two plane wave references ( 0 50 mm). The basic principle of building of holographic interferometers is to maximalise it's sensitivity on. the smallest displacement component and to decrease it's sensitivity on the largest displacement component. The illumination points (P I LI , P2L2) were chosen accordingly to sensitivity of the L z component and L x , L y components, respectively. The difference of the optical pathlengths between corresponding beams was Al = 1820 mm. Evaluation of interferograms Interferograms were recorded in double- exposure method. They were evaluated along horizontal as well as vertical diameter. In the reconstruction process the holographic plate was illuminated by original reference beams; the two independent interferograms were photographed through six observational points (Fig. 4) Typical fringe pattern belonging to different references are shown on Fig. 5. The numbers of the fringe order were determined semiautomatically on the basis of photographs with interference patterns. In each of investigated points twelve sets of fringe order data were given (six different observational point belong to one reference). The determination of the number of the fringe order looks as follows: the observer marks the geometrical centre on each fringe (black and white) along the investigated line and then with the aid of a drawing digitizer puts the geometrical data and integer fringe number to a computer connected via interface card. The program based on the least square principle chooses an appropriate polynomial and evaluates fractional fringe order numbers at the required points. Displacement components along the horizontal diameter were determined using the Eq. 12. Displacement components: L x ,L y ,L z along the horizontal diameter were computed in 17 points and are shown on Fig. 6. Displacement components along the verical diameter are shown on Fig. 7. F or further derivatives e, = *") / 7" \ „ " only the central range of y e ay PROBLEMS OF MEASUREMENT 73 KT'MtfB Fig. 5 G (—28.125; 28.125) is considered because of boundary disturbances due to applied forces. From the Fig. 7 can be seen, that point ^4(0; — 45 ;0) practically didn't] move so the orthogonal coordinates can be reduced at that point. The rigid body motions are expressed in rotations ©x,, 0r, @z,. i i i i i i i i I—l—i i—i—i—i—i [ I 1 1 I I I L_J Fig. 6 From the plots of Fig. 6 and Fig. 7: 6X, = 3.67 • 10 5(rad), Qy, = 6.1 • 10" 6 (rad), <92, = = 3.1510 5 (rad). To obtain the value of deformation the values of Lx(x, y = const), Ly{x, y = const), Lx{x — const, y), Ly{x = const, y) at discrete points of the surface are needed; afterwards there is a need to smooth the data by fitting an apprioprate curve to it. Smoothed spline functions [1], [8] are appropriate choice for fitting the displacement data as the objective of subsequent differentation. They are attractive for three reasons: their definition is based on the theory of mechanical deformation, second derivates are understood a priori, and their application to interferometrie determination of strain has been studied [13]. The values ex evaluated along horizontal diameter are shown on Fig. 8 and the values of ey are shown on Fig. 9. *28.125 45 \ y [mm} Fig. 7 1 1 1 1 1 > /s • 1 1 1. 1 1 1 1 < " / * — i i ê 7 ir 5 3 2 1 .I . J i l l [1O"S] . - - * ._ / \ \ 1 I 1, I 1. 1 1 - - — i 1 7/8 6/8 5/8 4/8 3/8 2/8 1/8 0 1/8 2/8 3/8 4/8 5/8 6/8 7/8 x/R Fig. 8 1 / 1 5/i 1 j | i4/8 1 / 3/8 1 2/8 i 1/8 - I 0 -7 - 8 - 9 1/8 1 • » - | 2/8 1 <̂ | Vertical D. 3/8 4/8 5/8 1 1 1 • • .̂ experim. Y^stheory V s i i i y/R 1 I X 1 Fig. 9 [74] P R OBLE M S O F M EASU REM EN T 75 D iscussion about accuracy of the measurement There are two main error sources: a) errors associated with measuring the system geometry b) innacurate determination of fringe order values. Two inequalities corresponding to these errors are as follows (8): where cond ® = \ \ g\ \ • \ \ gr% (15) is the value characteristic of th e sensitivity of interferometer, and 11jp I = (max • eigenvalue of Ę T • ^ 2 , (16) where KT is transpose matrix of K. F or quick determination of the error, which is found by evaluation of displacement vector components an other way was proposed [12]: where
OPMALi;Hfł n P H H H K O rE P E H T H O H C yn E P n O3H Li;H K[ H H T E P O E P O r P AM M
HHcbopiwainm o MexaHiwecKHX xaparcrepHCTHKax oSieKTOB B 3KcnepyMeiiTaJiŁnoJi M examroe n o n y-
«aeTCH H3 3KcnepHMeHT0B n o onpefl&Jieinno nojieft fledropiwamni H HanpHHceriHH. flJia TaKHX H ccn eao-
BaHHH MoryT 6tix& Hcnonb3OBaHbi MeTOflH roJiorpadpn^ecKoft H H Tepi})epoMeTprai. B DTOH paSoTe n oKa-
3aHo onpe^eJieH H e n o jia nepeiwemeHHH ToueK noBepxHOCTH oSieKTa H cnoJit3yH H H iepdiepOMeTp c p e r y-
HHpoBaHoii OHTHKOH njrreM, KOTopbrił # aeT BO3MO>KHOCTB aBTOAiaTHqecKoft Hncnei