Dynamic response of large ACC fan systems

Transcription

1 response of large ACC fan systems 2017 ACCUG A C L V Jacques Muiyser, Ochse Lombard, Johan van der Spuy, Danie Els Stellenbosch University Albert Zapke Enexio 1 of 42

2 Outline of 42

3 3 of 42

4 Test Case A air-cooled condenser (ACC) fans At the Test Case A coal-fired power sta ons, steam is condensed in an air-cooled condenser (ACC) by forcing ambient air through inclined heat exchangers with an array (288) of large Ø9 m, 270 kw axial flow fans situated at a height of 50 m.

5

6 Fan blade loading Each fan is suspended from a fan bridge. Distorted inlet air flow condi ons due to winds and other fans 1 as well as the downstream flow obstruc on (bridge) cause varying aerodynamic loads. 1 Van der Spuy, S.J., Von Backström, T.W. and Kröger, D.G. (2009). An evalua on of simplified methods to model the perormance of axial flow fan arrays. R & D Journal of the South African Ins tu on of Mechanical Engineering, vol. 26, pp of 42

7 7 of 42

8 2 2 Muiyser, J., Els, D.N.J., Van der Spuy, S.J. and Zapke. A. (2014). Measurement of air flow and blade loading at a large-scale cooling system fan. R & D Journal of the South African Ins tu on of Mechanical Engineering, vol 30, pp of 42

9 Test Case A 9 of 42

10 10 of 42

11 Fast Fourier transform (FFT) showed peaks at Ω, 2Ω and 3Ω f. 11 of 42

12 The response of a single degree of freedom system to periodic excita on Consider single degree of freedom system in the figure mx(t) + cx(t) + kx(t) = kf(t) with periodic func on f(t) = a + Re A e where A = f(t)e dt The steady state response is then with x(t) = a + Re A G e G = 1 1 p k m c x(t) c, ω = k/m, ζ = + i2ζp 2mω kf(t) 12 of 42

13 Formula on Non-linear least-squares op miza on is used to fit a Fourier series to a periodic measured response, r(t), using P harmonic terms where the variables are: a, a, b, a, b, a, b, ω. Q(t) = r(t) a + Re a + ib e (1) Then for the op mum curve fit a + ib A G or A a + ib G (2) The reconstructed excita on, f (t), is then given by f (t) = a + Re A e (3) 13 of 42

14 Test Case A: Full scale flapwise bending force results 14 of 42

15 Full scale results Results show similar excita on for both cases with a lower peak on the windward side during decreased winds. Higher wind speed Lower wind speed Bending load [kn.m] Bending load [kn.m] Reconstructed excitation [kn.m] Rotational position [degrees] Reconstructed excitation [kn.m] Rotational position [degrees]

16 Analysis Test Case X No resonance! Measurements recorded at a different plant, Test Case X, where f 3Ω did not include large blade vibra ons. Higher wind speed Lower wind speed Amplitude [N.m] Flapwise bending moment [kn.m] Azimuth angle [degrees] Frequency [Hz] Amplitude [N.m] Flapwise bending moment [kn.m] Azimuth angle [degrees] Frequency [Hz]

17 Test Case X Effect of surrounding fans 17 of 42

18 measurements: Analysis 18 of 42

19 Analysis of vibra on source The effect of the fan bridge on fan blade vibra on3 Strain gauges were a ached to a flat plate fan blade to determine the effect of bridge solidity and distance from the rotor. MicroStrain SG-Link Strain gauges 3 Work performed in conjunc on with the final-year project of Nico R. Basson.

20 Analysis of vibra on source The effect of the fan bridge on fan blade vibra on It was found that the amplitude of vibra on increases with: Increasing flow rate Decreased distance between fan rotor and bridge Increasing bridge solidity Load amplitude Reference curve Grid at 150 mm Plate at 150 mm Plate at 250 mm Flow rate [m 3 /s] 20 of 42

21 Analysis of vibra on source The effect of the pla orm height 21 of 42

22 Analysis of vibra on source The effect of the pla orm height Increase of the measured vibra on as the pla orm height is reduced from 4.5D to 2.5D A slight decrease 1.5D, change in excita on mode 22 of 42

23 23 of 42

24 Test Case A 24 of 42

25 Instrumenta on Full bridge torque strain gauge Full bridge bending moment strain gauge 2 Speed sensor Low speed sha 25 of 42

26 Startup Torque Shaft torque [N.m] Shaft speed [rpm] Time [s] 26 of 42

27 Startup torque low iner a fan blade comparison: Test Case B Test Case B New genera on low iner a fan blade that was tested under full scale condi ons. 27 of 42

28 Startup Power Power [kw] Electric Fan shaft Time [s] 28 of 42

29 Shutdown 20 Staft speed [rpm] Staft torque [kn.m] Time [s] 29 of 42

30 Sha loads Shutdown Spectogram Wind effects Vibra on frequencies Excita on reconstruc on of vibra on sources Pla orm height Sha loads Simula on of fan system 30 of 42

31 Low speed sha bending Total force length (moment) on output sha rela ve to gearbox Wind speed 25 m/s (a) wind direc on 66 (b) wind direc on 92 (c) wind direc on of 42

32 FFT 100 Magnitude [db] ω 2ω 3ω Frequency [Hz] Rota on speed 1ω (2.11 Hz) Walkway load pulse 2ω (4.23 Hz) Blade first bending mode 3ω (6.35 Hz) 32 of 42

33 Low speed sha bending wind direc on of 42

34 Low speed sha bending wind direc on of 42

35 Low speed sha bending wind direc on of 42

36 36 of 42

37 MSC ADAMS used for the dynamic simula on of the fan system Flexible bodies used Fan blade proper es selected to be the same as the finite element fan blade to provide a blade with similar vibra onal characteris cs as the full-scale fan blade. 37 of 42

38 Vibra on characteris cs Imprac cal to solve eigenvalue problem Virtual experiment by applying a force to a model with sweep or a chirp force func on and measuring the response at a point. Linear sweep 38 of 42

39 ADAMS vibra on characteris cs 39 of 42

40 System vibra on characteris cs 40 of 42

41 recommenda ons 41 of 42

42