You are on page 1of 21

CUTTING TOOL

RELIABILITY AND RUL


PREDICTION
Eng./ AHMED MEDHAT

Supervised by:
Prof./ YASSER SHABAN
Dr./ RAMY M. KHALFA
INTRODUCTION – WHY AUTOMATION
AND MONITORING ?

 INDUSTRIAL REVOLUTIONS
 MONITIRING SYSTEMS
 DATA ANALYSIS
 TIME-SEREIS DATA
 CATASTROPIC FAILURE PREVENTION
 DOWNTIME REDUCTION
CUTTING TOOL MONITIORING

 TOOL CONDITION MONITORING SYSTEM


(TCMS)
 SENSOR MECHANISM – ONLINE
MONITORING
 STATE OF THE CUTTING TOOL
 DAMAGE LIMITATIONS OF THE CUTTING
TOOL
 CRITICAL MANUFACTURING
 WORKPIECE QUALITY, SCRAPS RATE
COVARIATES OF CUTTING TOOL

 CUTTING FORCES (Fx, Fy & Fz)


 CUTTING SPEED (v) y

 FEED RATE (f)


x
CASE STUDY – CUTTING TOOL & WORK-
SPECIMEN

 Cutting Tool: Seco TH1000-coated carbide grade is


used, coating structure of the cutting tool includes a
nanolaminate physical vapor deposition (PVD) top layer
for maximum toughness and high chipping resistance.
 Work-Specimen: cylindrical bar of Titanium (Ti-
6A1-4V) alloy matrix reinforced with 10%-12% of TIC
ceramic particles is used for turning.
CASE SYUDY-DATA DESCRIPTION
CASE SYUDY-DATA DESCRIPTION CON.

 five covariates speed, feed rate and cutting forces


acting on the cutting tool.
 speed and feed rate values, it is shuffled through
experiment within 3 values for each covariate
 25 identical cutting tool will be used for machining
Ti-6A1-4V specimen
 The same values of speed (v) and feed rate (f) are
repeated 5 times on different 5 cutting tools
 The cutting forces are recorded for each experiment
from beginning to failure .
 Failure of cutting tool is specified by the value of
wear, a threshold value of wear 0.2mm
CASE SYUDY-EXAKT SOFTWARE

 Exact software is a predictive maintenance tool built


on statistical modeling, mainlyon Proportional
Hazard Model (PHM).
 Simple Weibull Model
 Hazard Function h(t,Z).
 Reliability Function R(t,z).
 Remaining Useful Life (RUL).
CASE SYUDY-EXAKT SOFTWARE

 β (shape parameter) – Item’s age


 η (scale parameter) MTTF (Scales Time axes)
 Zi(t) Covariates
 γ1, γ2 are covariate-related parameters
 Tr – Observation time
EXAKT-BASE MODEL & SUB-MODELS

 Built the base models with all covariates & and


working age. “ALL”
 Subtract sub-models from base one with a lower
number of covariates.
 Test each sub-model to reach to most significant
model with high accuracy in presenting all
covariates.
SUB-MODEL (V-F)

 Significant Model
 Working age is a covariate (β > 1)
 Wald Test Value
 P-value
MODEL FITTING TESTS

 Kolmogrov-Smirnov Test (K-S TEST): it


is a statistical test that tests our model to show
how fit it is with the data set.
 P-Value: this is an indication of the probability
of our model to be far from failure representation
– it’s good to be a small value.
MODEL FITTING TESTS

 RESIDUALS
 The average residual value must equal 1
 points are randomly scattered about the horizontal
line y = 1 means that model fits the data well.
 about 90% (at the least) of the residuals are
expected to fall within these limits for the model
to be accepted.
HAZARD FUNCTION h(t,Z)

 Using MATLAB we will plot the hazard function


for an example of speed (v) = 80 and feed rate (f)
= 0.15.
 Note that the hazard and area under the curve
which represent the failure probability are
increasing with working age at covariates values
applied.
HAZARD FUNCTION h(t,Z)

 hazard function is plotted for different three values of


cutting tool speed (v) at the same value of feed rate (f), we
can note that as w increases speed the probability of
reaching failure also Highly increases and the rate of hazard
is increase
RELIABILITY FUNCTION R(t,Z)
RUL PREDICTION ACCORDING TO
COVARIATES

 (v) [40:50] and feed rate [0.2:0.25] at current


working age of 200.
 RUL is about 156.969 sec, that mean the Total
Useful Life (TUL) for cutting tool will be 200 +
159.96985 = 360 sec. but we note that the value of
calculated standard deviation is relatively high
which about 128.7579. that make RUL value
predicted is less reliable.
OPTIMAL POLICY

 The graph shows the minimal expected


replacement cost of $0.184127/h ( = 0.184127 x
$1/h).

 The expected cost of the strategy of replacement


only at failure = $0.272949/h.
OPTIMAL POLICY

 Cost per unit time (of both preventive and failure


replacement combined),

 - Cost per unit time due to preventive replacements,

 - Cost per unit time due to failure replacements,

 - Percentages of preventive and failure


replacements,

 - The expected time between replacements


(including both the preventive and failure
replacements).
REFFERENCES

 1. Dimla, D.E., Multivariate tool condition monitoring in a metal cutting operation using neural networks. 1998.
 2. Dimla Sr, D. and P.M. Lister, On-line metal cutting tool condition monitoring.: I: force and vibration analyses.
International Journal of Machine Tools and Manufacture, 2000. 40(5): p. 739-768.
 3. Shaban, Y. and S. Yacout, Predicting the remaining useful life of a cutting tool during turning titanium metal matrix
composites. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, 2018.
232(4): p. 681-689.
 4. Shaban, Y., S. Yacout, and M. Balazinski, Tool wear monitoring and alarm system based on pattern recognition with
logical analysis of data. Journal of manufacturing science and engineering, 2015. 137(4): p. 041004.
 5. Swanson, L., Linking maintenance strategies to performance. International journal of production economics, 2001.
70(3): p. 237-244.
 6. Pejić Bach, M., et al., Predictive Maintenance in Industry 4.0 for the SMEs: A Decision Support System Case Study
Using Open-Source Software. Designs, 2023. 7(4): p. 98.
THANK YOU

You might also like