Monitoring of cutting tool wear …
1
MONITORING OF CUTTING TOOL WEAR BY USING CONTROL
SYSTEM SIGNAL
Toma Udiljak
Tihomir Mulc
Doc.dr.sc. Toma Udiljak, University of Zagreb, FSB, I. Lučića 5, 10000 Zagreb
Tihomir Mulc, B.Sc., SAS Zadar, M. Oreškovića 1, 23000 Zadar
Keywords: tool condition monitoring, open control system, signal sensitivity, face turning
ABSTRACT: Safety and reliability of operation of industrial manufacturing processes is a
very important prerequisite of the economic productivity. Sudden process disturbances
such as collision, overload, breakdown and tools wear are not fully understandable, and
cause production system failures. In order to prevent the effects of excess behaviour
regarding wear or eventually tool breakdown, modern technological systems pay particular
attention to predicting tool condition. Numerous theories of monitoring are trying to classify
and explain tools wear, but none have given any satisfactory results as yet, at the same
time insuring flexible, simple and price-regarding acceptable process control. Open
structure of modern digital control opens up new possibilities and prospects in this
respect. In many cases, the combination of digital plant signals and internal data of the
machine control system, along with advanced methods of signal analysis can replace the
external control systems. The integration of process control software module into the
machine control system allows fast reactions should there be any process disturbances,
without any additional hardware expansion. This paper studies the sensitivity of signals
contained in the control system to the cutting tools wear processes in face turning.
1. INTRODUCTION
Machine tools and production systems are generators of developing new production
equipment i.e. machine is the technical structure, collection of many technologies,
designed with the aim of reshaping raw materials into functional units useful to men. Over
the recent years, machine tools and production systems have gone through dramatic
changes caused to the greatest extent by the development of information technology and
flexible automation. The shift from classical towards sophisticated, fast flexible highefficiency machining cells is obvious. In the field of particle separation machining over the
recent years, high-speed machining has become a standard. It has changed the attitudes
towards machining, cutting tools and machine tools shaping. In order to achieve highspeed machining, the development of dynamic machines of light structure and low mass,
compact construction and high rigidity is required. In such circumstances, the installation
of engine-spindle of high rotation frequencies, liquid-cooled, with automatic tool clamping
system (HSK – nesting), feed gears equipped with digital drive systems and fast guides
has become a standard.
Control of high-speed machines is a very demanding task which requires powerful and
efficient systems of process monitoring and diagnostics. Basic conditions for good
management of machining monitoring include knowledge about the process state and
undertaking of adequate actions. Since machining process is an open system and is not
2
Toma Udiljak, Tihomir Mulc
uniquely defined, there may come to process disturbances that are not quite
understandable, nor predictable. The main parameter generating unexpected
disturbances in the machining system during machining is the process of cutting tools
wear, caused by interaction between tools, workpiece and machining conditions. The
diversity of input parameters, constant development of new materials, geometry and new
tool materials, as well as higher machining speeds, with simultaneous setting of
increasingly strict standards regarding safety, complicate the control process monitoring,
so that process monitoring remains one of the most demanding tasks in further
development of machining devices.
2. MACHINING PROCESS MONITORING SYSTEMS
In particle separation machining different methods for monitoring and controlling of
processes are applied, but the main monitoring source is the signal obtained from the
sensor. Sensor converts one physical value into another (force, sound emission,
vibrations into electrical signal). Built-in sensors have to be supported by additional
equipment for analysis and reception of signals with adequate assessment system.
Additional equipment needs to be adapted to the particular machine and machining
operations. From technical aspect, linking of external systems and the machine control
system is always related to certain difficulties, so that some of the disadvantages may be
highlighted in using the conventional monitoring systems:
 additional devices and sensors are necessary, and they need to be adapted to the
machine,
 external systems provide good exploitation results only after good preparation,
 available information contained in the control is not used,
 maintaining the monitoring system and defining of parameters is often demanding and
complicated.
It is possible to identify the system condition when the failure shows in the
measuring signal. The influence of failure on the measuring signal should not be only
theoretical, but should act on the signal flow with the possibility of reproduction (wear –
forces, vibrations, sound emission, etc.). Unfortunately, it is not always possible to
establish a simple link between the condition of the system and the signal, but signal
changes can result due to various causes, so that the interpretation of failures significantly
influences the efficiency and reliability of the monitoring system. The development of
sensory methods and systems is led by the tendency to realise maximal reliability in most
of the machining conditions, and the improvement of sensitivity on the observed
phenomenon. Regarding the required reliability of the monitoring system, sensors have to
satisfy various needs with regard to detection of the condition. On the one hand, failures
need to be detected very quickly, and on the other hand, the decisions have to be
trustworthy so as to eliminate losses due to false alarms. The problems of noise analyses
and often contradictory information of senses in signal analysis represent the focus of
research, since even the most successful strategy of decision-making is limited if the input
information is not sufficiently extensive and reliable.
Based on the previous experiences, it seems that the methods of analysing
particular signals cannot provide any major improvements in the monitoring system, so
that latest research are directed to the development of multi-sensory systems with the aim
of obtaining better, more reliable and safer information on the condition of the monitored
process or system. The application of advanced technologies of reception and analysis of
signals such as assessment of parameters, neural network, pattern recognition, fuzzy
logic, represent possible tools regarding the need to process ambiguous signals and
noises. Thus also the modern open CNC control system offers some possibilities for
establishing simple, inexpensive and easy-to-manage monitoring systems.
Monitoring of cutting tool wear …
3
3. STRUCTURE OF OPEN CONTROL SYSTEMS
Controller affect significantly the capabilities of machining systems. The trends in
developing control systems are directed towards establishing intelligent systems with
integrated modules for adaptation to the dynamic environmental changes, possibilities of
integration of new users’ applications, and learning possibilities from the process. The
definition of the open CNC system can vary, depending on the equipment manufacturers
and CNC machine tools users. A typical CNC open system has standard functions that
are used generally for all the machine tools, Figure 1.
MMC
Man Machine
Communication
Operator area
Machine
Parameter
Putting into
operation
Supervision
NC-core
Interpreter
Recognize order
Preparing
Processing data
Interpolation
Reactions supervision
Servo-axis
Correction signal
processing
Communication:
data, orders
System data:
supervision signal
and data
Modul for
monitoring
Machine data:
configuration data
Figure 1. Structure of the open control system ()
Kinematic configuration of processing machines are diverse. Depending on the machine
kinematics and its specific characteristics, some properties can have different operation
algorithms, although the general CNC structure is the same. CNC system is formed by
selecting the software modules from the standard library and their automatic linking. There
is the possibility of developing the missing functions and their adding to the standard
library. Thus, standard functions library can be supplemented by specific modules for tool
monitoring in order to provide the users with new possibilities in the field of “on-line”
process monitoring with regard to avoiding collision, breakdown, overload and monitoring
of tool wear. Software module installed in the control system also provides the fastest
reaction in case of a known process disturbance. Hence, modern open control systems
allow integration of additional software modules that use flow or momentum digital signals
for monitoring. However, for every concrete case the sensitivity and applicability of such
systems in various processing conditions need to be checked, and the supervision
strategies need to be adapted accordingly.
4. DESCRIPTION OF THE EXPERIMENT PLANNING
The aim of the experiment is to determine the sensitivity of drive system
parameters to tool edge wear in the process of face turning. For the purpose of studying
sensitivity of the monitoring signals to tool wear, a numerically controlled face head was
applied, used for flywheel machining. Digital Siemens motors were used for the main and
feed gears. The face turning unit was fitted within the unit of a special machine controlled
by Siemens digital control system, Sinumerik 840D, Figure 2.
The procedure of determining the sensitivity of control signals to tools wear is divided into
two parts: determining the level of tool edge wear, and gathering the data during the
4
Toma Udiljak, Tihomir Mulc
machining process and its further analysis. In order to shorten the procedure of testing
sensitivity of control signals to tools wear, grinding wear was done according to the
experiences gained in work (1), where it was established that topographical similarity of
grinding wear matches process wear after 30 to 60 seconds of machining.
Figure 2. Unit for face and peripheral turning (SAS-Zadar )
Plates were ground on the jig grinder Moor, along the whole plate circumference thus
reducing the radius of the tool nose, Figure 3. This means that with flank wear, if uniform
wear distribution is assumed, the radius of the tool nose is reduced, i.e. theoretical
roughness is increased according to the expression:
RtVB 
f2
8rVB
(1)
Monitoring of cutting tool wear …
5
where the new radius of the tool nose is determined by:
rVB  r  VBtg
(2)
r
CUTTING A-A
VB
SECTION A-A
A
A
VB
r

Figure 3. Geometry of the cutting edge in grinding
Due to geometry tolerance of the plate itself, completely even grinding of the flank is
difficult to achieve. Ground values of the flank wear band are given in Table 1.
Experiments were carried out without stopping the machine operation process, and the
conditions under which the experiments were carried out are contained in Table 2.
Table 1. Flank wear area obtained by grinding
Cutting tool flank wear
VB1
0.22 mm
VB2
0.42 mm
VB3
0.55 mm
Table 2. Face turning conditions
Workipiece material
16MnCr5
Starting diameter d 0  mm
 109.7
End diameter d1 mm
 154
Number of revolutions n okr / min
600-400
Cutting speed vc  m / min
218
Feed rate speed vc  m / min
82.5
Feed rate f  mm / okr 
0.15
Cutting length l mm
22.15
Cutting time t s
18.6
Type of machining
Dry machining
Insert type (Sumitomo)
CC MT 09 T304 NFP-T 110A
6
Toma Udiljak, Tihomir Mulc
In order to insure good topographical similarity of the ground and the process wear, 4
workpieces were machined prior to every ground plate, so that the adaptation time of
ground wear amounted to about 75 s, which is in accordance with the results obtained in
().
During the machining process, current data on electricity, velocity and position were
gathered from NC users’ core. Data were gathered by means of PLC (Simatic 5). Control
system data were gathered by means of a software written in programming language Step
5 (in PLC), Figure 5. Data were stored into the appropriate data block (DB).
NC - CORE
.......
SINUMERIK
%_N_PROG_2_MPF
N10 G71 G18 G90 CDON G54
N20 G0 G90 X56 Z170
N30 Z-0.7 M3 S550
N40 G96 S218 F0.15 D1
N45 M25
N50 G1 X78.362
N55 G1 X=78.699-R1/2 Z-0.265
N60 G97 S600
N70 G0 Z1.2
N80 G1 Z10 F0.15
N90 G0 Z170 M5
N100 M98
N110 M99
N115 G0 X56
N120 M30
..........
..........
..........
PLC
SIMATIC S5
User memory
Input / Output modules
840D
Data filling for X-axis
"Kanal1".M
Dyn(25)
M 121.2
S
CMP>=I
MW124
IN1
4000
IN2
M 121.2
A
JC
L
T
JU
Q
R
Data filling for S-axis
"Kanal1".M
Dyn(25)
M 121.3
S
CMP>=I
MW124
IN1
4000
IN2
R
Q
M 121.3
A
JC
L
T
JU
//
M006: L
L
<=I
ON
JC
SLW
LAR
//
OPN
L
T
L
T
L
T
L
T
L
T
//
L
+
T
MOO5:NOP
M
121.3
M006
0
MW 126
M005
4000
MW 126
M
33.0
M005
3
1
//
M003: L
L
<=I
ON
JC
SLW
LAR
//
OPN
L
T
L
T
L
T
L
T
L
T
//
L
+
T
MOO4:NOP
"Tiho OS-S"
"TIME OF DAY"
DBD (AR1,P#0.0)
"S-SPEED"
DBD (AR1,P#4.0)
"S-CURRENT"
DBD (AR1,P#8.0)
"S-POSITION"
DBD (AR1,P#12.0)
"MARPOSS-CORECTION"
DBD (AR1,P#16.0)
MW
20
MW
0
M
121.2
M003
0
MW 124
M004
4000
MW 124
M
33.0
M004
3
1
"Tiho OS-Z"
"TIME OF DAY"
DBD (AR1,P#0.0)
"Z-SPEED"
DBD (AR1,P#4.0)
"Z-CURRENT"
DBD (AR1,P#8.0)
"Z-POS"
DBD (AR1,P#12.0)
"MARPOS-CORECTION"
DBD (AR1,P#16.0)
MW
20
MW
0
124
124
126
126
Figure 5 – Structure of the signal flow
The start for recording the data was activated from the users’ NC program, using M
function (in this case auxiliary function M25 was selected). Data were recorded through
input modules into the data block (DB) in PLC tact. Thus recorded data could be read or
further analysed. If the data is further analysed, they need to be read prior to the
beginning of the next processing cycle, since they become replaced by new data.
5. RESULTS ANALYSIS
By analysing the obtained data it may be noticed that an interesting value is certainly the
electricity signal, i.e. area below the electricity curve, which could have been predicted.
Simultaneously, data for feed and main servo axis were recorded. The diagrams of
electricity momenta for the main engine and feed axes, for different levels of edge wear
are presented as follows in Figure 7 a, b, c, d, e.
Monitoring of cutting tool wear …
7
a) idle stroke
b) flank wear VB1
c) flank wear VB2,
d) flank wear VB3
Figure 7.
Curves of drive signals for the main and feed axis
8
Toma Udiljak, Tihomir Mulc
Table 3 gives an overview of area behaviour below electricity curve for certain axes
depending on the level of wear of the cutting tool edge.
Table 3 Area below electricity curve depending on the level of cutting tool wear
Area below
electricity curve
(%)
Vreteno S
Prigon X
Idle
stroke Po
(%)
452.35
736.5
Wear
PVB1
VB1
1325
739.2
Difference Wear
PVB1-Po
PVB2
VB1
VB2
874.6
1347.9
2.6
668.6
Difference Wear
PVB2-Po
PVB3
VB2
VB3
895.6
1771.5
-67.9
744.2
Difference
PVB3-Po
VB3
1319
7.7
It is noted that the tool wear has the greatest influence on the power, i.e. electricity of the
main engine, whereas it has almost no influence on the feed axis. In order to eliminate
higher frequencies of the electricity signal, the area below the curve has been taken into
consideration. The increase of the value of the area below the electricity curve caused by
the increase of flank wear from VB1 to VB3 is 33%, which is a significant increase and a
clear indicator of tool edge wear. The experiment confirms that there is a clear correlation
between the flank wear and the area below the main engine electricity curve, and that the
system in this case is sufficiently sensitive for monitoring the tool wear. These results
suggest that the area below the main engine electricity signal can be used to detect the
tool wear, confirmed by some authors who even give the linear expression showing the
dependence of the main engine electricity on the flank wear band during turning process
(6):
P  C  VB  P0
(3)
Linear dependence of main engine electricity in this experiment was not confirmed, figure
8.
2000
1800
Power / area (%)
1600
1400
1200
1000
800
600
400
200
0
1
2
3
Flank wear VB
Figure 8.
Dependence of the area below main engine electricity curve on the level of
flank wear.
Monitoring of cutting tool wear …
9
6. FURTHER RESEARCH IN IMPROVING THE INFORMATION BASE
Software module integrated into control offers, contrary to external sensory system, an
economical solution. The drive system does not act directly on the executive part of the
tool, but is connected with the machining process through mechanical components. These
components have to be modelled adequately with relation to the process disturbances,
since the basis of reliable process monitoring lies in the high quality information obtained
from basic signals. Main disturbance influences of transmitting systems include:
 resting and sliding friction of the drive chain, with non-linear behaviour depending on
the movement velocity, and the state of rest,
 acceleration electricity which changes with system load,
 clearance caused by the change of direction in the drive chain.
Considering the values of disturbance, these need to be analysed and separated from the
basic signal during signal processing, so that only the processing signals remain. The
following effects have to be taken into account:
 acceleration effect through inertia in acceleration process,
 friction effects in moving axes, spindle, guides or rotation friction,
 holding effects in standstill and clearance in change of direction.
Figure 9 offers an overview of corrective magnitudes that are eliminated already in the
prototype from the disturbance signal influence, thus generating good signal, Figure 9.
Figure 9 Dependence of the main engine electricity and the level of wear
Further processing of thus obtained valuable signals can be carried out by sophisticated
technologies of artificial intelligence, neural networks, pattern recognition. Since no clearly
defined algorithms nor theory are necessary for the operation of neural networks, because
they have the possibility to acquire knowledge through a series of examples, they are very
suitable for working with data on tools wear and prediction about their remaining reliability.
The possibility of neural networks to create reliable indicators of tools wear depends
strictly on the network structure, as well as on the conditions of the learning possibilities of
the network. On the other hand, pattern recognition can be defined as categorisation of
input data into recognisable classes by isolating significant properties or attributes from a
mass of minor details in the data (13). Results achieved by the application of the pattern
recognition method show that it has sense in the monitoring of tools wear (). In order to
accelerate the learning process in such systems, the application of grinding wear method
seems suitable, since it allows fast, economic, sufficiently big and reliable learning
sample.
7. CONCLUSION
Open control with digital gears open up new possibilities and prospects in “on-line”
monitoring of machining systems. In many cases the tools monitoring through control
system can replace the conventional external monitoring systems. By combination of
digital drive systems with additional information from the control system, methods of
isolating characteristic features from the signal and sophisticated data processing
technologies, high reliability and safety of signal analysis is achieved. Also, supervising
the machining process through software integrated modules in NC core allows fast
reactions to known processing disturbances, with no additional hardware restrictions on
the system. In this way, practical sets of processing monitoring modules can be
10
Toma Udiljak, Tihomir Mulc
developed, hardware-independent and open i.e. reconfigurative. The applicability of such
systems is mainly limited by sensitivity in relation to the observed phenomenon, which has
to be pre-defined. The control system platform with a developed minimal number of
additional monitoring functions, simplifies the man-machine connection, and makes it
acceptable by the operator. Further development of such systems, and the method of
isolating characteristic features, at the same time applying the technologies of artificial
intelligence, present a significant step towards realising a simple, reliable, user friendly
way of monitoring of cutting tools and processes.
8. LITERATURE
1. Udiljak T., Doprinos razvoju metoda za ispitivanje i praćenje stupnja istrošenosti oštrice
reznog alata, dissertation, Sveučilište u Zagrebu 1996.
2. Isermann R., Uberwachtung und Fehlerdiagnose, VDI-Verlag, Dusseldorf 1994.
3. Shultz H., Hochgeschwindigkeitsbearbeitung, Carl Hanser Verlag, Munich Vienna
1996.
4. Stute G., Regelung an Werkzeugmaschinen, Carl Hanser Verlag Munchen Wien 1981.
5. Koning W.,Klocke F., Fertigungsverfahren 1, Springer-Verlag Berlin Heidelberg New
York 1997.
6. Cuppini D., D'Errico G., Rutelli G., Tool wear monitoring based on cutting power
measurement, Wear, 139(1990) 303-311
7. Šavar Š., Obrada odvajanjem čestica I i II dio, Školska knjiga Zagreb, Zagreb 1990.
8. Cebalo R., Fleksibilni obradni sustavi, Vlastito izdanje, Zagreb 1995.
9. Zdenković R., Obrada metala skidanjem, Sveučilište u Zagrebu, FSB 1965.
10.Novaković B., Majetić D., Široki M., Umjetne neuronske mreže, Školska knjiga Zagreb,
Zagreb 1997.
11.Zimmermann H.J., Neuro+Fuzzy, VDI-Verlag, Dusseldorf 1995.
12.ISO International Standard: Tool Life Testing with Single Point Turning Tools,
Stockholm, 1997.
13.Damodarasamy S., Raman S., An inexpensive system for classifying tool wear states
using pattern recognition, Wear, 170(1993) pp.149-160