You are on page 1of 6
Two-Plane Balancing of a Rotor System Without Phase Response Louis J. Everett Assistant Poessor, Department of Mectaricl Enginoeing, Texas A&M Unwersty Cateye Staton, TK 7843 Measurements ‘This paper presents, and experimentally verifies, « two-plane balancing technique ‘for rigid rotors and possibly flexible rotors operating at a constant speed. The Technique, based upon influence coefficients, extends the single-plane four-run balancing procedure to two planes. Like the four-run method, this technique is most easily performed graphically and does not require response phase measurement. Despite the additional runs required to obtain data, its simplicity and applicability to ‘a wide range of equipment renders it more useful, in some cases, than the standard ‘wo-plane influence coefficient method. Introduction ‘Unbalance related vibrations account for approximately 80 percent of all vibration problems in rotating equipment. Un- balance can be the cause of not only excessive vibration, but also short beating life and noise. This helps explain why rotor balancing has received considerable interest, and why the literature containg many contributions (as detailed in this ppaper’s references) addressed specifically to understanding and correcting unbalance, ‘There is a major trend in industry today for balancing rotors insitu rather than in a shop. One reason for this trend is the expenditure required to obtain and operate the diverse ‘equipment necessary to balance the variety of machinery in a large plant. Another reason for the preference of in situ bal- ancing is that it often reduces downtime introduced by trans- porting the machine to the shop. In addition, in situ balancing Often produces higher quality balance. This is due in part to the fact that the rotor is balanced in its own bearings and housing. The industry has responded to this trend, and more recent equipment often allows access to balance planes and in- corporates slots for convenient weight addition. Of course, there is equipment for which in situ balancing is not popular; for example, high speed, critical equipment such as large com= pressors and turbines. Such equipment is usually balanced at its operating speed by the manufacturer or at a specialized ‘overhaul shop. ‘Over the years, several balancing techniques have evolved, ‘each favorably suited for a certain application, operating speed, or class of machinery. Most methods are essentially a variation of two major balancing philosophies; influence cocf- ficients, and modal balancing. The application of modal balancing has been limited due partially to the fact that the rotor must be run at a known mode shape. ‘The standard two-plane influence coefficient method is possibly the most widely used technique in industry today for in itu balancing. It is commonly available on many program “Conte by the Technical Commitee on Vbraon and Sound for pulcaton nthe Joona or Viton, Acousrcs, Stes, RSLS |B Dance. Manwsctip received a ASME Heedqurts, December 23, 1986 162/ Vol. 109, APRIL 1967 ‘able calculators and microcomputers. Unfortunately, the ‘method requires an accurate measurement of vibration phase angle in order to produce useful results. Accurate phase measurement is generally difficult to obtain on general pur- ‘pose machinery partially because vibration phase is sensitive {o speed variations (especially near a resonance) and operating speed is difficult to control on equipment running unloaded as, uring the balancing process. In addition, the presence of other equipment operating near the transducer can affect ‘measurement accuracy. Furthermore, it is not uncommon for ‘machines to be running at @ speed which is too low to trigger the key phase instrumentation. The influence coefficient ‘method is also sensitive to other system parameters which can result in anillconditioned influence coefficient matrix thereby producing erroneous corrections weights (Darlow, 1982). ‘There is clearly a need for a technique applicable to in situ balancing, and free of the anomalies experienced by the stan- dard influence coefficient method. The technique should be easy to use, should not require excessive operator interven ion, nor sophisticated measuring equipment. This paper presents a method, similar to the four-run single-plane method, which has demonstrated the preferred qualities on two occasions. ‘The remainder of this paper is organized as follows. First, the standard two-plane influence coefficient balancing method is presented. Following the standard method is a discussion of balancing without response phase measurements and a ‘mathematical presentation of the technique. Following the presentation, a step-by-step procedure is included. The paper concludes after presenting two experimental verifications. ‘The Standard Two-Plane Influence Coefficient Balane- ing Method ‘The influence coefficient method for two-plane balancing discussed by Kroon (1944) will be briefly presented since itis the basis of the procedure, According to influence coefficient theory, vibration resulting from an unbalance is proportional to the unbalance. The constant of porportionality is generally ‘Transactions of the ASME represented as « complex number, or vector, and is called an influence coefficient. Mathematically, the unbalance in a rigid rotor can be modeled as two distinct rotating masses, each ly- ing a unit distance from the rotation axis in distinct planes along the rotor axis. The ewo planes which are imagined 10 contain these masses can theoretically be chosen arbitrarily. Usually planes are chosen which allow convenient placement for removal of correction mass, Jn the standard method, vibra- tion is measured at another two planes. These measuring planes are not necessarily identical to the unbalance planes. For the development of the procedure, the four planes dis- ‘cussed will be called unbalance planes one and two, and ‘measurement planes one and two. The numbering is arbitrary. ‘The standard balancing procedure consists of four steps. First, the unbalanced rotor is run and the resulting vibration, ‘magnitude and phase, is measured. Mathematically the vibra: tion transmitted to the two measuring planes is expressed as, vivt fun Tala] [ Moms _ [ o vent Tashy anh Us, matrix expression, and throughout this paper, capital laters denote salars (real numbers) and lower-ese bold let- ters denote unit vectors (Complex numbers with a magnitode ‘of one). represents influence coeficens, is the vibration, and Ufs unbalance. The subscript on U denotes the unbalance Plane, the subscript on V denotes the measuring plane. The Supercrpt on V denotes that this isthe initial run ofthe rotor The influence coefficients are generally complicated functions of many parameters inloding rotor geometry, dynamis pro: Peri, stiffness, damping, running speed, and: choice of measuring and unbalance planes, Tn the second sep of thestandard balancing method, ial correction mas is placed on imbalance plane | at an arbitrary radius and angular postion. The chosen position of the tal ‘ass on the rotor must be noted, The zero angular poston is defined relative to the postion ofthe key phaser. With the ‘mass aixe, the rotors run and vibration readings are taken. Inthe third sep, the otor is stopped, the ral mass on plane one is replaced by a mass on unbalance plan two. The ris again spun andthe vibration measured: The resuling ‘ibrations can be expressed as [an J uty Tata ] [Um eitsw geerce bee na [ na ] fut Take) [ Gm a peere | beers Here 17 denotes the product of the il weight and its radial distance fom the rotation athe subset on denotes the Unbalance plane. Step of the method tase () and ©) fo determine the influence coefficients as tain = (Hive) am m @ fate = (Viiv) Journal of Vibration, Acoustics, Stress, and Reliability in Design fats = (Vere) Hw Taka = (H- eM) ® Having the influence coefficients, equation (1) is solved for the unbalance in each plane by inverting the influence coef cient matrix. If the matrix is ill-conditioned (nearly singular) ey smal eors in measurement can produce late ert ter the matrix inversion. The unbalance can now be theoretically corrected by applying masses 180 deg away from ch computed unbalance, the radial postion foreach correc- tio is chosen such thatthe product of distance and mas {denial tothe magnitude of the computed unbalance Balancing Without Response Phase Measurements ‘The four-run method (see Den Hartog (1956), Blake and Mitchel! (1972), and Carlson (1979)] is a method for achieving single-lane balance ina rotor. The four-run method is unique to other popular methods in that it is a graphical solution Which requires measurement of vibration amplitudes only and not vibration phase angle. After making vibration amplitude ‘measurements on four constant speed runs of the rotor, the correction mass is easy to find. Since the method does not re- {quire response phase measurement, it has certain advantages. For example, measurement errors are insignificant [see (Bake and Mitchell, 1972)] as long as the error is consistent, the method is simple and quick, works well even for slow speed rotors, and measurements can be taken at any convenient plane such as the supports. ‘There currently does not exist in the literature a technique for performing true two-plane balancing. without response pphase angle measurements. Blake and Mitchell (1972), hhowever, applied a simple extension ofthe four run method to balance a rotor in two planes. The results were satisfactory; however, based on the correction masses calculated, the actual amount ‘of initial dynamic unbalance is questionable. In another work performed by Gunter, Springer, and Humphris, (1982), a variation of the four-run method was used in com- bination with modal information, to balance a flexible rotor. ‘Their method was applied to balance the rotor at its critical speeds. ‘What follows is the presentation of a method applicable to ‘worplane balancing of rotors. The method is an extension of the single-plane four-run method. Its mostly a graphical solu- ion and does not require response phase measurement. It is also similar to the standard two-plane influence coefficient ‘method being a graphical solution to the same governing equa- tons. The method is verified via two experiments. The method is proposed as an alternative to standard influence coefficient techniques, nota replacement, ‘Two-Plane Balancing Without Response Phase To apply the method, one first spins the rotor and records the vibration magnitudes in the two measuring planes. Next, one applies atrial unbalance in unbalance plane | of the rotor, ‘The angular position of the trial mass on the rotor must be noted. The Zero-angle positon is arbitrary but constant throughout the procedure. The positive sense of angles is also arbitrary. After placing the mass, one agtin spins the rotor, and records the vibration magnitudes. For two more trials, fone replaces the previous trial unbalance with an identical ‘magnitude of unbalance ata different angular position on un- balance plane 1, spins the rotor, and records the vibration ‘magnitudes. The six vibration measurements are given by ‘APRIL 1987, Vol. 109/163 ‘Table Table fr colecting data ‘ial Unbalance Migstede; —___ aw "Wbsion Nsrataate Bhi | Tage aie 7 imam | in aegee Pasel Pare? smb | diane | Magutnde | Magaie fn ee 1 |, wy 3 | =" ” | m2 ‘ 5 | “¥ ‘ =v | | vive wet [ Fata thy | ie aa o tun tata | L tem for k= 1,23 Using (1) and (4 the following equations are obtained VEG GHIn tn Meme fork = 12,3andj= 1,2 o ‘The left-hand side of equations () represents two vectors with measured magnitudes and unknown directions, the right- hhand-side vector has unknown magnitude and direction. "To obtain a more useful form of the equations, first rotate cach equation by the known angle of — w,, and by the Unknown angle ~ v9. The equations are now Wy ta Wo we © [Now rotate through 180 dey and reflect about the real axis to obtain V5() We) 4 yo w, 2G a mY in fork = 1,2,3 and, 12 o By introducing a unit vector, n, directed along the negative real axis, equation (7) can be written as BOG) 19 wat, Wy (82) 4 Th fork = 1,2,3 andj = 1,2 ® ‘The left-hand side of equation (8) represents a vector of own length and orientation to which a vector of known Jengthis added. The loci of such a vector sum is esy to con stret being circle of known radius and center. Equation (@) 164/ Vol. 109, APRIL 1987 is solved in the following manner. First, with j set to 1 (con- sidering only vibrations measured on plane 1) a circle is drawn representing the left-hand side of () for one ofthe trial masses (for one value of #). On the same diagram, wo more circles are constructed for the two remaining plane 1 measurements. All thee circles will be distinct since the trial unbalances were Tocated at distinct angles. Also, since all three circles represent the loci of the same vector (recall the trial mass magnitude is the same for every run) they should intersect and therefore ‘uniquely define the unknown vector on the right. The distance between the intersection point and the origin of the figure represents /,, Wg. The angle between the negative real axis, ‘because of the introduction of vector m in equation (8), and a line through the origin and intersection point determines the angle of the unit vector i,/¥9. By setting j to 2 (considering vibrations measured on plane 2), a similar construction ean be ‘made. This determines the unknown vector I; Wy (lay/¥8)- BY removing tial masses from unbalance plane one and pia fone trial mass at atime for three runs on the second plane, the six equations (9) can be generated from which two more con- structions are made. HOI iy mera me ( 5,6 and, ° ‘To make use of the information obtained from the construc- tions, equation (1) must be rotated to obtain ou a, Usa, J Using the graphical solutions to (8) and (9), equations (10) ‘can be solved for the unbalance in each of the two unbalance planes. ‘Two-Plane Balancing Procedure To use this method for balancing, one will need to make seven runs of the rotor and 14 vibration magnitude measurements. Table 1 can be used to record the necessary in- formation, The procedure is performed using the following steps 1 Run the unbalanced rotor and measure vibration ides (Vf, V9) at two planes. Locate & trial mass (JV w,) on plane 1, run the rotor and measure vibration magnitudes (V}, VJ) on'two planes. 3 Change the position ofthe trial mass om plane 1 to (W w,), run and measure vibrations (V3, ¥4). 4 Perform Step 3 again for W’ws, and Vj, V2. 5 Perform Steps 2, 3 and 4 using trial masses Ww, 1 ‘Ws, and Ww located on plane 2. Vibration measurements are HD, Hi, VD, and (H, Y9)- 6 Draw’ three circles. Circle one is centered at V2. w; with radius of Vj. Circe to is centered at V4 w with radius (of V4. The third ie centered at V3 w, with radius of V}. 7 Measure the distance (di) from the origin to the ap- proximate intersection and the angle (8) from the real axis to the line between the origin and intersection. 8 Divide d by trial mass magnitude (1) to obtain f, Substract@ from 180 deg to obtain the angle of i,,/¥. ‘9. Draw throe more circles as in step 6 using centers and radii of (V3 w,, V4), (V2 Wa, V3), (V3 Wa» V3). Use the circles to compute Jy; and the angle of f,/¥f a8 was done in Steps 7 and 8. Transactions of the ASME | Disk Disk 2 Sleeve epee Probe Probe LJ U Fig. Rig used inthe examples Table 2 Data taken trom the ist experiment case Vibetion Messrements Baleacing Speed: $150 rpm ‘Tal Unbalace: 08 Gram TST ages] — Wits sae] we agen | et tae Sian] SERS |g me agatet[ae | as || as” || ve fom fe | a fa ve | 8 ATS 65 | 86 22 | fa rea mee | "| “abie'S Results ofthe fist experiment using the new method Method of This Paper Using Data From Table 2 Plane I Pine Magnitude | Phase | Magnitude | Phase (Grams) | (Deg) | (Grams) | (Des) azo jisiz | oso | 15 07 | 190 | 089 | 5 46 I" 01 Journal of Vibration, Acoustics, Stress, and Reliability in Design Table 4 Results ofthe frst experiment using the CASE 1 Reralts From The Standard Method sing Dats From Table 2 Data ie | Magaitude | Phase | Magnitsce | Phase Taken | (Gram) | (Deg) | (Grams) | (Dee) 14 sor | 2 | aon | ape el 205 | ms | oats | 9 ue | as | oa | 26 oss | oo | ze | 129 os | am | se | an | a a6 | 1m | am] us | asa oa | one fan | ae | 0 as | oss | am | ase | se 0, 3,6 1.91 53, 349 | ager 10 Again draw three circles as in Step 6 using centers and radit of (1 wa, V2), (A ws VD, (MY Wo, V7). Use the circles to compute fj; and the angle of iy/v) as was done in Steps 7 and 8. LI Repeat Step 10 drawing circles with centers and radii of (VE wa, V3), (VE we, V5), (V9 we, VO) 10 determine Ja; and the angle of 3/98. 12 Solve equation (10) for unbalance. Experimental Veri ‘The method was verified using a rotor with two disks, see Fig. 1. Both disks had a radius of 30.5 mm. Disk one had a ‘mass of 608 grams, disk two was 822 erams. Two proximity probes were located near the two sleeve bearings as indicated. The signal outputs were processed using a commercially available digital vector filter, which output vibration phase and amplitude. The critical speed of the rig was measured to be 1160 rpm. The collected data for the first experiment is shown in Table 2 ‘The method of this paper was programmed on a portable microprocessor with a minimum hardware configuration. This demonstrates the feasibility of using it for in situ balancing. ‘The graphic solution was determined via computer as follows. ‘After reading the input data, the program draws on the CRT, fone set of the three circles represented by equations (8) and (©). A small dot appears on the screen which can be driven around the screen using the keyboard. When the dot is placed at the intersection point, the processor digtizes the position of| the dot thereby knowing the location of the circle intersection. Once all four figures are digitized in this manner, the pro- cessor proceeds to compute the correction masses. ‘The results obtained using the method of this paper are displayed in Table 3. Also shown in Table 3 are the actual cor- rection weights applied, and the residual vibration after cor- rection. The output from a standard two plane influence coef- ficient method is shown in Table 4. Note that forthe standard method, one needs only vibration measurements from three APRIL 1987, Vol. 109/165 Table §- Data taken tom the eecond experiment ar smamber case 2 Vibration Mestrements Balancing Speed: 3000 sn ‘ial Unbalnce: 08 Gram Tage oi ‘Wbration Nesraremente Te degree Plane 1 Plane 2 and pine | Magaitede | Phaze | Magnitude | Phase (is) (ie) | (Deg) 0 | to ones 16 | me nea | 2 135 a1 | m6 Plane 1 | wo | oa j= | ° 266 | Plane 2 Resuls of Wrst experiment using the now method ASE 2 enulte From The Mathod of This Paper Using Data From Table 5 eee ee esa | ai [os ee fet tet ea) fl x ees | cee (age el teeta is = ! ‘runs; hence Table 2 presents more data than necessary. Table 4 therefore presents all possible outcomes from the standard ‘method by utilizing all permutations of the data in Table 2. In this experiment, all permutations of the data produced atleast ‘one Jarge and impractical correction weight. Table 5 shows another test case with a different running speed. Shown in Table 6 are the results obtained from the pre- sent method; again it improved the state of balance of the rotor. Table 7 shows results forall permutations ofthe data in ‘Table 5. Notice that using a trial mass on plane I at zero dee produced, in all cases ofthe standard method, a gross error in the correction mass. This implies that one should avoid plac- ing trial mass at this location. Also note that the last three 166 / Vol. 109, APRIL 1987 Table 7 Results of he second experiment using the standard method CASE 2 Revulte rom The fandard Method Using Data From Table 5 ten Tat ‘ier ass | a au | ae a0 | os rows of Table 7 are very close to the values computed via the present method; hence the operator would have obtained satisfactory results if he knew to place the mass properly. Un- fortunately, such a priori knowledge of the proper placement Of trial mass is not available. Note also that the present method produced acceptable results in both experiments without any @ priori knowiedge. Conclusions ‘was specifically developed for field balancing but is applicable to a wide spectrum of balancing problems. The method ex- tends the four-run single plane method and has greatest utility ‘when the two-plane influence coefficient methods fails to ob- tain the desired state of balance quality due to inaccuracies phase measurement, While the two cases presented in this Daper are not exhaustive, they clearly illustrate the ability of the present technique to handle rotors simply and consistently. Although the method of obtaining two-plane balance presented inthis paper does not require measurement of vibra- tion phase angles, it does require more runs than the standard technique. This can prove to be an advantage, however, since ‘these extra runs can help overcome problems with improperly located trial masses. The two examples performed demonstrated both problems with the standard method and ‘the success ofthe present method. Since this method is based upon influence coefficient theory, it has a sound theoretical foundation. Furthermore, because it does not require response phase measurement, it aiso has several advantages and can easily be applied to Held balancing problems. Acknowledgments ‘The author wishes to thank John Vance currently with ‘Texas A&M University for suggesting the topic of this paper. ‘Also a thank you is due Fouad Zeidan who performed the ex- periments in expert fashion. ‘Transactions of the ASME References Bshop, RFD, and Gavel G. ML 199, “The Vibration and Bane lng of an Untalasat Feibie Rotor founal af Mechanical Engineering Stone, Vols Nos 1 6. Bisop, RED ad arkinson, AG, 1969, “On te lian of Modes in the Balancing of Fell Sbat Procoins of the Ittaion of Mechanical ines, Vol. 17, No. 16». 407 ‘Bishop, K. ED, ad Parkin, AG May 1972, “Oa te Use of Balac ingMachines for Pcuble Rtas,” ASME ourel of ngncring or ad, lke, M.P., and Michi, W.S., 172, Varox ney Acoomi Measons: sear Hanon, Spartan Bosh, ‘Caron, P.O. Ly 197%, “Four Run Balncing Without Phase” Pro cosings, echnary Vibration Montorin and Anat Seminar ond ecg New Oreang, Ua ‘Chur, A. Hand Plunket, R, November 1961, “lanl Fexbe Rotor,” ASME Jounal of Engineering Jor Indy, Vl. 83, Seis 8, No 8, 9.3853. Darl, MS 18, “The tdenfication and Elimination of Nonadeen- ent Baloe Planes i Ines Coes Dalsneng, ASME Puper No, marae. ‘en arog, J.P, 1986, Maciuoseas Vinarons MeGraw-fll New Yok Gunter, E Speinget, Hand Hamp, Re Rey 962, *Balnsing ‘Matias Flexible Rotor Bering Sysem Wihout Phase Measurement” Energia Blt, Vol 3, No 0 “Keenburer, W196, “Bancng Fle Rotors on Two Generally Fe be Bearings." Brown Borer Rett, Vol 4, No- 9, pp 6617, ‘eleaburger, W., May 1972, "Should Flexte Rotor Be Balanced in Wo (2) Panes?” AME fural of Engincring for Indy p38 Kroon, RP 194, “Balancag of tang Apparat, Par :” ASME tourna of ppt Mecha, Vl 1, 1944, 9. A. “LeGrow, IY 197, "Mullane Buencig of File Rotors A Method of Ceaiaing Creaton Welt” ASME Paper 7 Vir-5 197 Lindley, AG. nd Bishop, KE. D. 196, “Some Resnt Research 00 the Balicing™ of Lge Peubie Rotrs* Proceedings af the Itiation of “Mechel Expncrs, Vo. 177, NO. 30, 9.8 ‘Moors, L. 8, and Dodd E.G. 196%, "Mie Halaning of Large Feible ‘Rotors Genera Electric Compan Jour, Vl. 3, No. Parkinson, A.C, and Bishop, RED, 1965, "Reds Vibraon in Modal Balancing,” Jour of Mechanic Eneleerig Scene, Vo. NoIe 9-33, “Tang. TM, and Tromplr, P. Re, 196k "Dynaaee of Synchronous Procesing Turborotors With Paulas Refecace 10 Bueaing: Pa t= Ap. plleatlon," ASME Journal of Applied Sacha, Val 90, Sere Ep. 25.20 ‘esr, J.M., and Bade. Re H., May 193, “Espermentl Evan ofthe Ext Point Speed nd Least Sgaees Procluer for Flext Rotor Baltcing by the Inflence Coetficeat Method,” ASME Journal of Enenco Ig for tet. 9 8, ‘heal, EL 1936, “Dynamic Balancing of Rotating Machinery inthe id” ASME Vol 36, pp, METS. APPENDIX ‘This appendix contains a BASIC program which performs the balancing procedure discussed in the paper. The program runs under MS DOS. 30 Em 10(2,8) TOC, ¥C2. 2,2) 22,2) Y42,39,0002),012), "YG. 10.2).18.2) 092.040), 20 am mo¢a.a),ro).19022,9) oe Be S Seat omE Z)oneocant- HOH CheotNi foo "anrry cbster ‘tl mee sieute °, 710 0" "PaDHT Ser Socetion ot "OROMC2)*" mee co LISS Heo uecd pitutoneer ae TN Ov7 e028 60" “GRunt "enter vibration ong. on “PLHK(D)¢ GF jams sith as Eaal aus 489 ron sto ae Set To Son tet wy 2 0 "naa? Sisration oe SPLa4Gh)™ plane wth % 30 PRISE WOMS(D)#= mane on vtRACS™ puane ", Bo en rap zinta ra Set Bo" Na senste Be gh Het BD 258,298, 2,2)NEKT AGHEXT JvREEE Journal of Vibration, Acoustics, Stress, and Reliability in Design So ac ae t4, 325 QRCC 2-2) NO Bo ENP R ap olay nd SB iss ho veo muy nceenoa inion: Ho Ibo mm wanamG/y) So nie Be nts ee ae So Wot ns ‘bows way ma EB PE BEeeeieastt me ERRRU Soe nese Rea Be Mctelmea/cat) Seana sees S60 Fhunr sme SoPUnet)o" pane eavzee iF se sux Beata aynibe o/s sasenesa Baa eee Bo EEle. wnrnunmonnenneonm 83 BEER curs He weasecrn) sar Teo GoTo 1260 a a a eee Ft "ans (oh) akoAt tao LOL Folnncar. way oupanco Tum oL2AT-0 Se Frscs.riocérnerany ‘APRIL 1987, Vol. 109/167

You might also like