Experimental dynamic characterization of machine tool spindle.

Zapciu, Miron ; K'nevez, Jean Yves ; Cahuc, Olivier 等

1. INTRODUCTION

The modern approach on the dynamic behavior of machine tools is to measure specific parameters for its operation in order to have an intervention only when parameters indicate. This method is called conditional maintenance (predictive). For the conditional maintenance to be effective, we must make reliable and accurate measurements of machines. A number of variables machines can be used as indicators: temperature, oil pressure etc. However, experience has shown that the most reliable parameter that gives the earliest and the best way the deteriorated condition of a rotating machine is vibration. By measuring and monitoring the level of vibration produced by a machine, we obtain an ideal indicator on his condition. If the increased vibration of the machine can detect a defect, the analysis of vibration characteristics of the machine helps to identify the cause. Usually it's not possible to make an assumption about the causes and excitatory vibration. This information could be obtained by making a frequency analysis.

The vibration diagnosis is based on identification of mechanical phenomenon, the frequency of vibration it generates. To do this we must represent the vibrations in the space of frequencies, this operation is carried out by making a mathematically transform FFT using signal time vibration. Each component of the spectrum is a well-defined characteristic frequency (imbalance, resonance or misalignment). The frequency analysis is generally when the vibration level of the machine is considered higher than the permissible threshold.

The measurements systems used in this paper are Vibroport 41 (Schenck) and LMS Pimento module (Zapciu & Paraschiv, 2007), (Bisu, 2007). These devices are very efficient and help to develop a database, but do not allow the collection with an expert system. Recently, the MPPIV module uses a new measuring device built into a program CM400 expert, to establish ways of surveying and exploiting measurement results.

2. LEVELS OF PREDICTIVE MAINTENANCE

Predictive monitoring programs have three main objectives: warning earlier possible about the potential defects on equipment equipped with continuous monitoring systems; monitoring the dynamic condition of general purpose machinery; primary warning of defects that may remain hidden within complex characteristic, for example rolling-element bearing flaws.

[FIGURE 1 OMITTED]

Generally, most of effective predictive-monitoring programs will incorporate a wide variety of parameters to accurately characterize condition and provide earliest warning of significant changes. The process of typical predictive monitoring program consists in three levels starting with first baseline overall analysis and finishing with a detailed vibration signature on each piece of monitored equipment (Fig.1).

If the component at running frequency dominates the spectrum no greater than 20 ^ 25 % of the amplitude at running frequency, then monitoring overall amplitude is two or three broad bands is a appropriate method for early detection on rolling element bearings and other mechanical rotating parts of machines and equipments in industrial mechanical engineering.

Experienced organizations report that a fully implemented of regularly predictive monitoring program will eliminate unexpected failures and reduce the number of machines in questionable marginal condition to less then 6-8 % of the total.

Fourier discreet function has a very important role in many applications of digital signal processing, such as linear filtering, spectral analysis and estimation. The main reason of its importance lies in the existence of efficient algorithms for calculation of the function using formula:

[X.sub.k] = [N-1.summation over (n=0)] [x.sub.n] exp(-2[pi]i/N nk) k=0, ... N-1 (1)

where [x.sub.0] and [x.sub.N-1] numbers are complex.

Unlike discreet Fourier transformation, which requires [N.sup.2] operations for a set of N points, Fast Fourier transformation uses only N x log x 2(N) operations, which constitutes a significant improvement in performance, especially for large sets of data.

Bearing machine tool spindle frequencies are best identified in the envelope spectrum. Studies of the envelope time history may be valuable in determining when and how many significant events take place.

Bearing frequencies (monitoring defect formulas) assuming pure-rolling motion are the following:

Outer race defect:

[f.sub.OR] = n/2 [f.sub.r] (1 - BD/PD cos [beta]) [Hz] (2)

Inner race defect:

[f.sub.IR] = n/2 [f.sub.r] (1 + BD/PD cos [beta]) [Hz] (3)

Ball defect:

[f.sub.B] = PD/BD [f.sub.r] (1 - [(BD/PD cos [beta]).sup.2]) [Hz] (4)

where:

n=number of balls or rollers; [f.sub.r]=relative rev/s between inner and outer races; PD=Pitch Diameter; BD=Ball Diameter; [beta] = contact angle.

The formulae apply to the shown deep groove bearing and only in the case of pure rolling operation. It is recommended to use Bearing Frequency Calculators supplied by bearing manufacturers (e.g. www.skf.com).

3. MEASUREMENT OF SPINDLE VIBRATIONS APPLICATION--FOR MILLING MACHINE

In order to know the proper speed domain for a High Speed Milling Machine tool the authors experiment the vibration of the HSM 600U spindle using Tracking analysis module of Vibroport 41 (Rigal et al., 2005), (Sutter & Molinari, 2005).

Considering Fast Fourier Time spectrum filtered out of operational domain of the machine tool, one of the most important frequencies was 780 Hz (46800 rpm) (Fig. 2).

Eigen frequencies of the assembly spindle-bearings (without cutting process) obtained experimentally (run-down spindle by 36000 rpm to 7000 rpm were: 29130 rpm (first harmonic of 242.5 Hz), 14550 rpm (242.5 Hz) and 12.600 rpm (210 Hz).

[FIGURE 2 OMITTED]

In the figures 3 and 4 are presented the tracking signals of speed vibration acquired for two speed domains (Zapciu et al., 2007). The maximum level of vibrations speed was 0,23 mm/s rms (level of vibrations is very acceptable).

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

Speed vibration of the spindle for another machine tool (PC MILL 100 by EMCO) is presented in the figure 5.

For this milling machine was obtained the following speed, recommended to be avoided in the work domain: 2280 rpm (Fig. 6) and 4000 rpm. The level of speed vibration was 0,3 mm/s; the threshold specific for future attention is about 1,2-1,4 mm/s rms for this machine tool.

4. CONCLUSIONS

This research paper proposes a maintenance survey for the machine tool spindle using vibration level (Tracking vs Speed). The second important objective was to separate the dynamics of the spindle of the machine tool in order to have a good control for the cutting processes. In this context, the actual subject is important and it can help to elaborate a proper model for predictive maintenance based on vibration.

This work was validated by the experimental results based on the measuring of the level of vibration of the spindle of two milling machines. Future work will include analyzing and filtering signal in order to better diagnose the bearing condition of the machine tool spindle. An effective predictive monitoring program will be designed to accurately recognize the earliest significant changes, and use the simplest and least costly means to separate the machine that have problems from the large number of machines in a good dynamic condition.

5. REFERENCES

Bisu, C. (2007). Study of Self-Maintained Vibration in 3D Cutting--A new modelling applied on turning, PhD Thesis, IMST faculty, University POLITEHNICA of Bucharest.

Rigal, J.F.; Zapciu, M.; Mabrouki, T.; Belhadi, S. (2006). Sawtooth chip formation in hard turning and the approach to separate process and machine vibration frequencies. ICMaS, Bucharest, ISSN 1842-3183, p.133-136.

Sutter, G.; Molinari, A. (2005). Analysis of the Cutting Force Components and Friction in High Speed Machining. Journal of Manufactu'ing Science and Engineering, Trans. of the ASME, Vol.127, pp. 245-250.

Zapciu, M.; K'nevez, J.Y.; Gerard, A. (2007).Using tracking analysis of predictive maintenance concept in order to obtain dynamics of machine tool spindles. International Conference ICAMaT 2007, Sibiu, Romania.

Zapciu, M.; Paraschiv, M. (2007). Predictive maintenance and use of tracking concept to analyze dynamics of machine tool spindle. International Conference--TMCR'2007, pp.512-516, May 31th-June 3th, 2007, Chisinau.