PASSING THROUGH A CRITICAL AT CONSTANT SPEED? Zlaten Rocke Supervisor Operating Performance ‘Sienane Power Corporation Bradenton, Floride Croatia and inthe United States since 1970. Mr. Racic graduated from SOE in Milwaukee, Wisconsin with a BSMET’ and with MGA’ degree from” Wova University in Fare Lauderdale, Florida. His ong career started as a merchant marine engineer, and currently he is fenployed at Sionens Power Corporation. In his line of work he has resolved unerous rotordymanie. and Vibration problems on practically any OEM rotating equipment. rnRcoucriON This case history on a machine lecated in the Far East deals with a vibration problen en a conbustion turbine driving an ~100 NVA generator. The preblen observed was a "thermal vector"(at the CT exhaust end) which was changing wich change in. oad. Several consultants had tried to correct the condition by balancing, but all they were getting was a shifting of the beginning and the end point Of the "thermal vector". {THE APPROACH A first look at the Polar plot showing the run-up and loading to -70 Wd revealed a Very low, lee Figid mode eritical at ~1100 RPM, a more pronounced and rigid node critical at 2070 RPM and very good Balance condition at operating speed of 3000 RPA (Fig.1). hon the generator was synchronized, and Loaded, ‘the vibration amplitude and phase continued to charge in a form very mich” receebling a ‘resonance. The only difference was that th happening at @ constant speed (Fig. 1,3). Wien the Unit was partly unloaded, the vibration vector followed the load-up curve up to a point of trip. During deceleration, the batance condition ‘changed, affecting also the and rigid node response at the turbine exhaust end (Fig. 2,3). Subsequent runs confirned the repeatability of the ats With the assumption that if it looks Like a critical and snells Like a eritfeal, ft mst be a eritteal, @ trial weight was placed ‘in the middle span of the turbine rotor at the angle based on critical speed location estinated from the polar plot (Fig. 1). The result was amazing. (Fig.4,6). Not only the "Loop" was gone, but the total “length of the thermal vector" had shrunk! A cubsequent shut down showed also a recuced “critical at ruming speed’, a further ‘snall taprovenent of 2nd rigid critical and slightly worsening and angular shift of ‘the ist rigid critical. LYSIS When dealing with a Mclean thermal vector, the Fesuttant seen on the polar plot is the vectorial ‘su of the residual “unbalance vector and” the changing thermal. vector. Balancing above cennot influence the total "vector spread" but only shift ite location (Fig.7)-. But in this case balancing has reduced the total ‘wector spread, i-e., balancing had reduced the resultant anplitudes over a range of ~180", at constant speed, similar to the effect of reducing @ typical resonshce response. Because the balance weights vere placed at 210°. (90" behind the estinated peak of the displacenent vector), the balance respense sens to be (ike correcting a rescnant response amplitude, LUsION The polar plot cFig.1) shows the start-up curve before belancing and the rather odd vector change during the Increase of load, until achieving steadystate condition at 70 Md. It is alnost a circular Course from .27 ir/s pk at 200 to «17 in/s pk at Slo. The initial ettonpes in balancing the end poini of the steady state did not yield favorable Pesulte during load transient. The reascn behind fe is that ony correction weight which would recuce Vibrations at steady” state will presumably eteriorate the transient vibration according £0 the Laws of Msuperinposition". hen untoading the turbine, the vector did_not Fottow back the path completely. The state of balance of the turbine, therefore, differs from the initial one. This fact explains’ sufficiently the chservation that the vibration in the critical ‘Speed renge of the 2nd rigid moce differs from that fof the start-up. Leaving aut the above mentioned strange (etreular) course of the veetor change, the phenonenon, up to ew, is. understandable. There is 3 residuat Trbatance which causes vibrations in the critical speed range hen starting up, there is a transient Veetor change hen heating up the turbine shaft hich goes temporarily in direction 180* and cones, Back after ahile and there Ts 2 repeatable vector ‘dus to loed temperature which follows these parameters (see the difference between the 70 HW Beint and the TRIP point). The "superimposition" Of these three components results’ in the total ‘eetor change depicted in Figure 8. A correction weight at 210° (at the high point of the velocity vector") was installed. Obviously, the vibratione in the critical speed range have been reduced by this shot for the stert-up as well ‘as for the shutdown. That is understandable since the vectors of both, et start-up and shutdown had been in the saxe quadrant before balancing. ious, something happened, the physical exptanat fon of shich requires more sophisticated theories: The Yector spread beteen the transient saxinun point land the steady state point at 70 Hl has shrunk by the installation of weights. AS mentioned above, this could not have gecurred ag ong as the simple aus of linear” superiaposition of the single conponents (independent fran each ether) apply. then attempting balancing transient vibration CThernel vector®) at canstant speed, two cases are chserveds 1. The transient maxima is reduced, the veetor change renaining the sane is only parallel shifted. 1, The trensient maximum is reduced, the Vector ehange has becone snaller or almost disappears. case I. follons the simple Laws of superinpesition. Case it. can only be explained if more complex physical love apply. One of then iz the non-Linear Spring constant of the supporting systen. ‘he critical speed in the non-Linear system varies. With the magnitude of vibration. The case 1s, imaginable that 9 systen is more sensitive to Unbatance with higher vibrations then with smal Ler Vibrations. the change of the state of balance, ue to heat ing up the turbine shaft, my alvays be the sone, But {f the transient maxinun ean be Teduced by field balancing, the sensitivity to lwelance con decrease and the turbine ansuors Co the sone change of the state of balence with & snaller vector spread. The phenomenon can be Conbined with the chenges of the phase engle. Mon- inear spring characteristic can occur in reat Lifes. Each joint between 2 machine conponents 1s a potential ease of non-linearity. Many parts of the machine are fitted with clearances, bearing shells in bearing casings. The other possibility of case I. is ¢ simple change in syoten stiffness Influenced by the systes heatirg.. ‘As the machine is heated up, the systen support becomes eotter and the eritical speed agrates to the Lefe (FIg.9)- As depicted in Figure 9, after the unit reaches operating speed, if operates with the amplitude of Vibratfon an the left slope of the solid Lino Meritieal™ at corresponding phase. ‘As often in our Lives we do sonething because it orks, even if we hove no csplete explanation why fete 's0., If pore similar cases are studied in the future, hopefully we will be able to find the Bol — rs Shrone fats tan oe