Page 244 - Sensors and Control Systems in Manufacturing

204
Cha p te r
F o u r

downtime. Also, in the new machine software, it was found faults
occurred because the software had not been properly matched to the
number of tool positions physically present on the tool magazine.
Once, such a fault actually caused a major collision within the machin-
ing volume.
4.7.8.2 Detecting Tool Failure
An important element in automated process control is real-time
detection of cutting tool failure, including both wear and fracture
mechanisms. The ability to detect such failures online allows reme-
dial action to be undertaken in a timely fashion, thus ensuring consis-
tently high product quality and preventing potential damage to the
process machinery. The preliminary results from a study to investi-
gate the possibility of using vibration signals generated during face
milling to detect both progressive (wear) and catastrophic (breakage)
tool failure are discussed next.

4.7.8.2.1 Experimental Technique The experimental studies were car-
ried out using a 3-hp vertical milling machine. The cutting tool was a
381-mm-diameter face mill employing three Carboloy TPC-322E
grade 370 tungsten carbide cutting inserts. The standard workpiece
was a mild steel plate with a length of 305 mm, a height of 152 mm,
and a width of 13 mm. While cutting, the mill traversed the length of
the workpiece, performing an interrupted symmetric cut. The sensor
sensed the vibration generated during the milling process on the
workpiece clamp. The vibration signals were recorded for analysis.
Inserts with various magnitudes of wear and fracture (ranging from
0.13 mm to 0.78 mm) were used in the experiments.

4.7.8.2.2 The Manufacturing Status of Parts Figure 4.6 shows typical
acceleration versus time histories. Figure 4.7a is the acceleration for
three sharp inserts. Note that the engagement of each insert in the
workpiece is clearly evident and that all engagements share similar
characteristics, although they are by no means identical.
Figure 4.7b shows the acceleration for the combination of two
sharp inserts and one insert with a 0.39-mm fracture. The sharp
inserts produce signals consistent with those shown in Fig. 4.6, while
the fractured insert produces a significantly different output.
The reduced output level for the fractured insert is a result of the
much smaller depth of cut associated with it. It would seem from the
time-domain data that use of either an envelope detection or a thresh-
old crossing scheme would provide the ability to automate the detec-
tion of tool fracture in a multi-insert milling operation.
Figure 4.7 shows typical frequency spectra for various tool condi-
tions. It is immediately apparent that, in general, fracture phenomena
are indicated by an increase in the level of spectra components within