Published in: International Journal of Material Forming. Document Version: Peer reviewed version

Transcription

1 A Vitual Inspection Technique fo Assessing the Dimensional Accuacy of Foged Compesso Blades Using FE Modeling and CMM Inspection Makem, J., Ou, H., Amstong, C., Rennie, A., & Nikov, S. (2008). A Vitual Inspection Technique fo Assessing the Dimensional Accuacy of Foged Compesso Blades Using FE Modeling and CMM Inspection. Intenational Jounal of Mateial Foming, 1(1), DOI: /s Published in: Intenational Jounal of Mateial Foming Document Vesion: Pee eviewed vesion Queens Univesity Belfast - Reseach Potal: Link to publication ecod in Queens Univesity Belfast Reseach Potal Geneal ights Copyight fo the publications made accessible via the Queens Univesity Belfast Reseach Potal is etained by the autho(s) and / o othe copyight ownes and it is a condition of accessing these publications that uses ecognise and abide by the legal equiements associated with these ights. Take down policy The Reseach Potal is Queens institutional epositoy that povides access to Queens eseach output. Evey effot has been made to ensue that content in the Reseach Potal does not infinge any pesons ights, o applicable UK laws. If you discove content in the Reseach Potal that you believe beaches copyight o violates any law, please contact openaccess@qub.ac.uk. Download date:15. Feb. 2017

2 A Vitual Inspection Technique fo Assessing the Dimensional Accuacy of Foged Compesso Blades Using FE Modeling and CMM Inspection J. Makem 1, H. Ou 1, C.G. Amstong 1, A. Rennie 2, S. Nikov 2. 1 School of Mechanical & Aeospace Engineeing, Queen s Univesity, Belfast, Ashby Building, Stanmillis Road, Belfast, UK. BT95AH. URL: jmakem01@qub.ac.uk; h.ou@qub.ac.uk 2 Rolls-Royce plc, UK. URL: ABSTRACT: This pape pesents eseach fo developing a vitual inspection system that evaluates the dimensional toleance of foged aeofoil blades fomed using the finite element (FE) method. Conventional algoithms adopted by moden coodinate measuement pocesses have been incopoated with the latest feefom suface evaluation techniques to povide a obust famewok fo the dimensional inspection of FE aeofoil models. The accuacy of the appoach had been veified with a stong coelation obtained between the vitual inspection data and coodinate measuement data fom coesponding aeofoil components. Key wods: Foging, Finite Element, Coodinate Measuement, Dimensional Inspection. 1 INTRODUCTION The high demand fo manufactuing quality and optimum pefomance of components within the aeospace industy equies pats to adhee to the elevant geometical and dimensional specifications. Consequently, components ae inspected to evaluate thei dimensional toleance. In paticula, the dimensional inspection of complex pats compising fee-fom geomety, such as aeo-engine blades, is becoming inceasingly impotant due to the equiement fo highe pecision and efficiency. Geneally, the objective behind the majoity of inspection techniques is to detemine if the pat meets a set of specific design citeia by assessing the deviation between the manufactued pat and the nominal model. Fo quality assuance puposes, high pecision dimensional measuement techniques ae employed to evaluate the dimensional toleance of compesso blades. The contact inspection pocess of coodinate measuement is one of the most effective measuement techniques, poviding both high accuacy and epeatability. The pocess employs a compute contolled coodinate measuement machine (CMM) to inspect the pat automatically by moving a tactile pobe along the wok-piece suface, to measue the coodinates of individual contact points. Compesso blades ae manufactued using the closed die hot foging pocess. The majo contibutos to dimensional and shape eos in foged aeofoil components ae pess and die elasticity duing foging, mateial sping back and themal distotion at the cooling stage [1]. Due to the complexity of the hot foging pocess, foging design pactice is lagely based on the tial and eo appoach [2]. These foging tials employ a diect compensation appoach which involves modifying the die shape by a faction of the measued foging eo until the blade dimensions ae within toleance. The finite element (FE) method has been used successfully to simulate the foging of compesso blades [3]. By accuately pedicting mateial defomation, stess/stain ate distibution and tempeatue change in hot foging, the FE method may be used to quantify the foging eos and optimise the die shape, emoving the need fo expeimental testing and theeby educe the oveall development timescales and costs. Consequently, as

3 the FE eo govens die shape modification, it s necessay to assess the accuacy of the foging eos poduced in simulation by compaison with coodinate measuement data fom actual aeofoil sections. Thus, to gain an accuate compaison between the FE geneated eo and that quantified using the CMM, the FE eo must be evaluated using a simila appoach to that employed by the coodinate measuement pocess. This pape pesents a vitual inspection appoach used fo the validation of an FE blade foming simulation. The developed system povides a fully automated tool fo eo quantification in a vitual envionment. A obust famewok has been developed incopoating vaious iteative algoithms to egiste the FE model in elation to a CAD nominal and geneate inspection data fo compaison with CMM measuements on a eal pat. It will be shown that a stong coelation exists between simulated and eal pats, veifying the validity of the appoach. 2 BLADE INSPECTION The pocess of aeofoil inspection via CMM compises thee main stages, including; pat localisation, blade section inspection and finally blade paametes analysis [4]. A unique coodinate system can be established by defining six datum points on the blade. The blade sections ae then measued using a two stage pocess involving an initial ough measuement of each section followed by a moe accuate assessment of the pofile in fine measuement mode. Finally, afte the measuement data fo each blade section has been coectly aligned with the espective nominal pofile, blade paametes and pofile toleances may be evaluated. 2.1Pat localisation The pocess of localisation, also known as egistation, mathematically locates the pat pio to inspection by detemining a igid body 3D coodinate tansfomation between the design coodinate system (DCS) and the measuement coodinate system (MCS). Fo aeofoil inspection, the method o six point pinciple is commonly used to detemine the MCS [5]. Fistly, the pimay datum plane is detemined by thee points on the fee-fom concave suface of the blade, including K 1 and K 5 on the oot section and K 3 on the tip section. The seconday datum plane is constucted using the cental axis of cylindical featues at each end of the blade, defined by P 1 and P 2. Finally, the tetiay datum plane, othogonal to both datum planes is detemined using the last datum point, K P, located on the oot block. Finally, both coodinate systems ae then aligned via igid body tansfomation. The position of the datum points on a nominal blade model is shown in figue 1. Fig. 1 Datum point locations on nominal blade model. 2.2 Blade measuement The coodinate measuement pocess employs a contou measuement mode, whee by each blade pofile was measued at a constant height (z coodinate). CMM data is ecoded at thee sections along the blade fo both concave and convex sufaces. Each section is defined by six datum points (K points) aound the pofile of the component. A two stage pocess involving a ough and fine measuement mode is used to evaluate each section. 2.3 Blade paametes analysis Once the pofiles of each section have been obtained, a toleance evaluation algoithm is implemented wheeby the convex and concave pofiles fo each section ae positioned with espect to the coesponding nominal pofile. The shape toleance is then epesented in the fom of a plot compaing the nominal and measued pofiles. Also, the thickness, δ 1, bow, δ 2 and twist, δ 3 deviation ae evaluated at the elevant K point stations, as shown in figue 2. (a) (b) Fig 2. Measuement of (a) section thickness & (b) twist eo.

4 The thickness and bow deviation may be calculated using equations 2 and 3. t ( x x ) + ( y y ) 2 Kcc KCV 2 KCC KCV δ = (1) δ1 = δ t δ t (2) δ = y y (3) 2 K 3 K 3 whee δ t = measued thickness and δ t = nominal thickness. 3 VIRTUAL INSPECTION SYSTEM The main pogamming languages employed in the development of this softwae wee Python and MATLAB. Figue 3 shows the basic stuctue of the vitual inspection famewok. Fig 3. Stuctue of vitual inspection system. 3.1 Component egistation To initiate localisation, datum points wee fistly identified on the blade fomed fom the FE foging simulation. As the initial position of the blade was in oughly close poximity to the nominal coodinate system due to constaints applied in the FE simulation, no peliminay tansfomation pocess was equied. An initial estimate of the position of datum points K 1, K 3 & K 5 shown in figue 1 wee found via intepolation fom pofiles geneated by intesecting the FE mesh at the tip and oot sections. Subsequently, using the ules of othogonality and sequence fo datum fame constuction [6], the seconday datum was established by estimating the location of P 1 and P 2. This was achieved by fitting the nodal coodinates situated on each cylindical suface accoding to the least squaes pincipal. Consequently, afte identifying the espective point clouds using a shape ecognition algoithm, each dataset was submitted to the least squaes calculation which yielded the coodinates of the datum points on the espective cylinde axes. Afte locating the position of five datum points, the nomal vectos, p,s and t of the espective datum planes in the MCS wee detemined using equations 4 6. p = ( K5 K1) ( K3 K1) / ( K5 K1) ( K3 K1) (4) s = ( P2 P1 ) p / ( P2 P1 ) p (5) t = p s / p s (6) ( ) Accoding to the sequence ule in datum setting, the fist stage of the localisation involves aligning the components pimay datum plane with that of the nominal. This was achieved by aligning p with p, the diection cosine of the pimay datum in the DCS. This fist tansfomation was decided by a tanslation and two otations in sequence. The second phase involved alignment of the seconday datum planes by otation about p. Finally, to align the tetiay datum planes, a single tanslation, T, between the two coodinate fames was applied. The entie tansfomation is epesented by: ( β ) R( θ, ) X T X = R α + 1 m (7) whee R ( θ,α ) α to align p, R ( β ) to align s and = otational matix fo angles θ and =otational matix fo angle β =point on unegisteed pat. X m Afte the initial egistation of the component, the localisation pocess was epeated until the discepancy between the datum points on the blade and the coesponding points on the nominal wee within a desied toleance. 3.2 Compaative analysis and pat validation Once the pat was successfully egisteed, measuement data was geneated by intesecting the aeofoil suface at the espective K point sections using the method descibed. The section thickness was ecoded fo each of the nine sets of K points. Also, the bow deviation was ecoded at the mid section, as well as the twist deviation at the mid and tip sections. To evaluate the pofile toleance, an additional alignment pocedue based on an iteative closest point algoithm [7] was equied to fit the elative pofile data to the espective nominal cuve.

5 4 RESULTS and DISCUSSION A contou plot displaying the esults of the localisation pocess is shown in figue 4. (a) Befoe Localisation (b) Afte Localisation Fig. 4 Deviation between Nominal & Foged Component The egistation algoithm effectively locates the blade using the iteative six point appoach. The deviation between the nominal and the FE model has been damatically educed in this aea. Deviation Deviation (a) Hoizontal Axis (b) Hoizontal Axis Fig. 5 Compaison between (a) FE and (b) CMM Mid Convex Pofile Eo A stong coelation exists between the FE esults measued using the vitual system and the actual fom measuements. In paticula, the data ecoded at the mid convex section displays the stongest coelation with a discepancy of less than 18% between the maximum pofile deviations, as shown in figue 5. Geneally, only small discepancies between the vitual and actual CMM esults fo both twist and bow deviation wee appaent. The vitual measuement of bow deviation diffeed by appoximately 0.01mm in compaison the CMM esult, wheeas a maximum diffeence of 0.2 in twist deviation was ecoded at the tip section. In tems of the thickness deviation measuements, values ecoded at the oot and mid sections ae easonably consistent with CMM values. Howeve, slightly lage discepancies ae appaent at the tip which may be attibuted to the inability of the FE method to accuately simulate the defomation in this egion. 5 CONCLUSIONS The vitual inspection system povides a complete, obust pocedue fo the dimensional inspection of foged aeofoil blade models fomed using the finite element method. The localisation pocess used by the system accuately egistes the component with the nominal model. The inspection and paametes analysis modules incopoate vaious iteative algoithms and the latest evaluation techniques to successfully quantify the foging toleance to within a high degee of accuacy. The stong coelation between measuements geneated fom this system and actual CMM measuement data, validates the appoach as an effective means to quantify the foging toleance using the coodinate measuement pocess in a vitual envionment. ACKNOWLODGENTS The authos wish to thank the Engineeing and Physical Sciences Reseach Council ESPRC and Rolls-Royce plc fo funding and suppot of the pesent eseach (EP/C004140/1). REFERENCES 1. Ou H., Lan J., Amstong C.G., Pice M.A., Walloe S.J., Wad M.J., Reduction in post foging eos fo aeofoil foging using finite element simulation and optimisation. Modelling Simul.Mate.Sci.Eng, 14, (2006) Ou H., Amstong C.G., Pice M.A., Die shape optimisation in foging of aeofoil sections. Jounal of Mateials Pocessing Technology, 132, (2003) Hatley P., Pillinge I., Numeical simulation of the foging pocess. Comput Methods APPL Mech Eng, 195, (2006) Hsu T., Lai J., Ueng W.E., On the development of aifoil section inspection and analysis technique. Int. Jounal of Adv Manuf Technol, 30, (2006) Hsu T., Lai J., Ueng W.E., Hwang J., An iteative coodinate setup algoithm fo aifoil blades inspection. Int J Adv Manuf Technol, 26, (2005) Huang X., Gu P., CAD-model based inspection of sculptued sufaces with datums. Int Jounal of Poduction Reseach, 36, (1998) Li Y., Gu P., Fee-fom suface inspection techniques state of the at eview. Jounal of Compute-Aided Design, 36, (2004)