35.5 Chatter Detection

Back

Regenerative chatter is easily detected by an operator due to the loud, high-pitched noise it produces

and the distinctive “chatter marks” it leaves on the part surface. However, automatic detection is

required for intelligent manufacturing (Cho and Ehmann, 1988; Delio et al., 1992). At the onset of

chatter, process signals (e.g., force, vibration) contain significant energy at the chatter frequency. It is

a well-known fact that the chatter frequency will be close to a dominant structural frequency. The

most common method to detect the presence of chatter is to threshold the frequency signal of a

process signal. To analyze the frequency content of a signal, a Fourier transform, or fast Fourier

transform, is performed. If the frequency content of the resulting signal near a dominant chatter

frequency is above a threshold value, then chatter is determined to be present. It should be noted that

machining process signals also contain significant energy at the tooth-passing frequency. If the

dominant structural frequencies and tooth-passing frequency are sufficiently separated, then the

tooth-passing frequency may be ignored when determining the presence of chatter. If the dominant

structural frequencies and tooth-passing frequency are close, then the signal must be filtered at the

tooth-passing frequency using a notch filter. Also, forced vibrations, such as those resulting from

the impact between the cutting tool and part, must not be allowed to falsely trigger the chatter

0 0.01 0.02 0.03 −4

−2

0

2

4

Time (s)

x (mm) Fx (kN)

y (mm) Fy (kN)

0 0.01 0.02 0.03 −4

−3

−2

−1

0

Time (s)

0 0.01 0.02 0.03

−0.1

−0.05

0

0.05

0.1

Time (s)

0 0.01 0.02 0.03

−0.3

−0.2

−0.1

0

0.1

Time (s)

FIGURE 35.17 Time-domain simulation for Example 4 with Ns ¼ 32,000 rpm and d ¼ 3.795 mm.

This section presented the technique of time-domain simulation as an alternative means to analyze

regenerative chatter. Time-domain simulations are the direct numerical simulations of the force

process and structural vibrations. A process parameter is changed iteratively from simulation to

simulation to determine the critical value at which chatter occurs. A sufficiently small time step

must be utilized in the numerical integrations to account for the small system time constants

associated with the large structural frequencies.

35-18 Vibration and Shock Handbook

© 2005 by Taylor & Francis Group, LLC

detection algorithm. These thresholding algorithms all suffer from the lack of an analytical method of

selecting a threshold value. This value is typically selected empirically and will not be valid over a

wide range of cutting conditions and machining operations.

35.5.1 Example 5

An experimental face-milling operation, a complete

description of which is given in Landers

(1997), is conducted with a spindle speed of

1500 rpm and a tool with four teeth. The

dominant structural frequencies are 334, 414,

653, and 716 Hz. The machining force Fz is

sampled at a frequency of 2000 Hz, and the timedomain

signal is transformed into the frequency

domain via a Fourier transform using 80 points

(i.e., one spindle revolution). The power spectral

density of the force signal is shown for depths-ofcut

of 1.0 and 1.5 mm in Figure 35.18 and

Figure 35.19, respectively. In Figure 35.18, there

is significant energy at 100 Hz, which is the toothpassing

frequency. There is also significant energy

at 750 Hz due to structural vibrations; however,

the system did not chatter, as evidenced by the lack

of chatter marks on the part and a high-pitched

sound during machining. In Figure 35.19, there is

significant energy at 665 Hz as well as 100 Hz.

Chatter was evidenced by the chatter marks left on

the part surface and the high-pitched sound

during machining. The results demonstrate that

the chatter frequency is 665 Hz, which is near the

dominant structural frequency of 653 Hz. Note

that the power spectral density at the frequency of

0 Hz is ignored in Figure 35.18 and Figure 35.19.

This component is stronger than the components

at all other frequencies since the machining force

Fz fluctuates about a static, nonzero value. In this

application, a thresholding algorithm may ignore

the low frequencies where the tooth-passing

frequency is strong; however, if the operation

were to be performed at a higher spindle speed, say 7500 rpm, or the number of teeth were increased

from 4 to 20, the tooth-passing frequency would be 500 Hz, close to the structural frequencies. In this

case, the force signal would have to be filtered at the tooth-passing frequency.

0 100 200 300 400 500

Frequency (Hz)

Power spectral density (N2S2)

600 700 800 900 1000

0

1

2

3

4

5

6

× 10−5

FIGURE 35.18 Power spectral density of Fz in a facemilling

operation with d ¼ 1.0 mm.

0 100 200 300 400 500

Frequency (Hz)

Power spectral density (N2S2)

600 700 800 900 1000

0

0.5

1

1.5

2

2.5

3

3.5

4

× 10−4

FIGURE 35.19 Power spectral density of Fz in a facemilling

operation with d ¼ 1.5 mm.

This section presented techniques to detect the occurrence of regenerative chatter. The phenomenon

of regenerative chatter is easily detected by an operator due to the loud, high-pitched noise it

produces and the distinctive “chatter marks” it leaves on the part surface. The most common

method to detect the presence of chatter is to threshold the frequency signal of a process signal. In

this case, one must be careful to separate out the spindle rotation and tooth-passing frequencies.

Regenerative Chatter in Machine Tools 35-19

© 2005 by Taylor & Francis Group, LLC

Навигация