Professional Documents
Culture Documents
Weibull Analysis To A Practical Case
Weibull Analysis To A Practical Case
Module M13
RELIABILITY, MAINAINABILITY & RISK
MAJOR ASSIGNMENT
Abstract
Application of Weibull technique to the bottle neck of Ternium Siderar
This study is about the Weibull analysis technique which characterizes the
reliability problem and determines the cause root of each failure mode for the
Torch Cutting Machine. This equipment represents the main availability problem
of the Continuous Casting Machine (CCM), the Ternium Siderars bottle neck.
At the beginning of this report a briefly description of Ternium Siderar process is
introduced. After this the CMMS records are analysed and the CCM main
availability equipment, component and failure mode are detected. Then the
Weibull analysis is applied, the cause reason are identify and possible solutions
are presented. In addition to this, its financial benefits are evaluated.
Finally there hand graphical result are contrasted against the ReliaSoft Weibull
++7 Software. Similar answers are achieved for both alternatives.
INDEX
I.
RELIABILITY INVESTIGATION......................................................................................................3
A.
1.
INTRODUCTION ...................................................................................................................................3
Ternium Siderar Steel Process .......................................................................................................3
a)
2.
a)
3.
a)
3.
4.
2.
1.
Figure 6: Torch........................................................................................................................................ 8
Figure 7: Nozzle....................................................................................................................................... 8
Table 3: Analysis of Incomplete Sample of Toch-Nozzle Times to Failure .................................................. 9
D.
Torch.............................................................................................................................................8
a)
b)
c)
1.
C.
Figure 3: ABC Pareto Failures CCM, Source: Ternium Siderar Maintenance Site .................................. 5
2.
CONCLUSION ....................................................................................................................................16
CONTRASTING THE RESULT WITH WEIBULL SOFTWARE .......................................................................16
1.
Input data ....................................................................................................................................17
2.
Software Answer - Weibull ++7 ...................................................................................................17
a)
b)
c)
d)
e)
f)
g)
G.
UNAVAILABILITY ANALYSIS................................................................................................................5
2.
E.
F.
B.
REFERENCES .....................................................................................................................................27
I.
Reliability Investigation
A.
Introduction
a)
a)
Plant
Dealy Capacity
[Ton]
Minimum Dealy
Capacity [Ton]
Blust Furnace
10500
7900
Continuous
Casting
8000
9000
Today, the Blast Furnace N1 and 2 are working at their minimum technical
capacity. They couldnt produce less pick iron than 7900 Ton/day, otherwise
they become unsteady (permeability- thermal -chemical process problems).
What is more, the Hot Street Mill has an average capacity1 of 9000 Ton/day.
Nowadays, the Hot Street Mill has 96 free hours a month. These hours have not
been program because there are no available slabs to feed it up.
Furthermore the company cant import slabs from another company since there
is no profit margin. (The market cost and sells price make it not possible to
import slabs in order to transform into coils.)
In addition to this, the Argentinean market is demanding more downstream
products that do not have been produced because of the slabs limitation.
To conclude, the actual scenario generate the CCM bottle neck with excess of
pick iron (it is stocking as a solid pick iron) and free hours in the Hot Street Mill.
Every slab that is not produced at Siderar must be import as a final product
from other companies. In this cases Siderar act as a trader. To sum up, it is
crucial to improve the CCM availability (downtime and low speed failures)
Slab Thickness=200 mm
Its depends of the production mix (chemical composition, thickness and width). The same considerations
apply for the CCM.
Ladle
Tundish
a)
B.
Mould
Segments
Torch Cutting
Unavailability Analysis
This analysis only takes into account the unavailability due to equipment
failures.
Considering the CMMS records2, as we can see in the Figure 3, the Torch
Cutting is the main availability problem for the Continuous Casting Machine.
(For the last two years)
Hours
Hydraulic
Marking
Tundish
Mold
a)
Downtime
Figure 3: ABC Pareto Failures CCM, Source: Ternium Siderar Maintenance Site
Ternium Siderar has selected SAP system as a CMMS. What is more, we have an Intranet Site that consolidate
this information
It is important to highlight that the Torch Cutting Machine have not produced
downtime (unless in the last years). But if we also considerer the speed
reduction, the Torch Cutting becomes the main availability problem.
iii
v
The media used for autogenous equipment are heating gas and oxygen, each
extremely hazardous when handled inappropriately.
b) User requirements
Cut the strand of 200 mm of thickness and 1650 mm of width, at a casting speed
which goes from 0,8 m/min up to 2 m/min, at a pre-selected speed piece length +/- 30
mm, in a maximum time of 2 minutes and 30 seconds, with a certain quality (see cut
quality chart, figure 1, appendix 1).
-
As the Continuous Casting Machine is the bottle neck of the company, there is no
of failures
failure
33
Actuator
8352
1.044
9,6E-04
Wire/Connections
8352
4.176
2,4E-04
Drive
8352
2.784
3,6E-04
Clutch
8352
4.176
2,4E-04
8352
8.352
1,2E-04
Motor CA
8352
1.392
7,2E-04
10
8352
835
1,2E-03
8352
8.352
1,2E-04
Sensor
Software
Mechanical
a)
Number
50
Torch
22
8352
380
2,6E-03
12
8352
696
1,4E-03
Filter
8352
1.670
6,0E-04
Brake
8352
8.352
1,2E-04
Guide
8352
4.176
2,4E-04
Transmision
8352
1.670
6,0E-04
Valves
8352
2.784
3,6E-04
For the torch cutting machine all the elements are related by OR gates in
the fault tree function Cutting the Slab. Every component that fail, makes
the Cutting the Slab function failed. For the Torch element there is no
redundancy component in the original design. The fault three is not
represented since I consider that is no necessary because it has all OR
gates.
3,0E-03
2,5E-03
2,0E-03
1,5E-03
1,0E-03
5,0E-04
H
ol
es
an
d
To
on r ch
ex
io
n
Se s
ns
o
A
ct r
ua
to
M
r
ot
or
C
A
Fi
Tr
a n lte
sm r
isi
on
Dr
W
ive
ire
V
/C
a
on lve
ne
s
ct
io
ns
M
C
ot
l
ut
or
ch
s
C
on Gu
tro
id
lC e
en
So ter
ftw
ar
e
Br
ak
e
0,0E+00
Components
b)
4. Torch
The torch is a heavy duty, water cooled, component that feed gas and oxygen.
The torch finishes in the nozzle which is the element that mixes the gases in
order to create a large heating effect.
a) Figure 6: Torch
b) Figure 7: Nozzle
For process and quality requirements, the nozzle must be replaced twice a week
during the production windows. This work is performed by fix daily shift. For
practical reason, this preventive task is programme for every Tuesday (4 nozzles
work for 48 hours) and Thursday (4 nozzles work for 120 hours). For this reason
we can see in the Table N3 that four nozzles are suspended at 48 hours and 4
nozzles are suspended at 120 hours.
T1L
T1R
T2L
T2R
T1L
T1R
T2L
Week TTF
Week TTF
TTF
TTF
TTF
TTF
TTF
(h) f / s (h) f / s (h) f /s (h) f /s
(h) f /s (h) f /s (h) f /s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
1
27
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
2
28
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
3
29
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
4
30
103 f
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
5
31
100 f
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
6
32
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
12 f
12 f
12 f
7
33
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
8
34
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
9
35
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
10
36
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
11
37
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
12
38
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
13
39
120 s
120 s
120 s
120 s
34 f
34 f
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
14
40
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
15
41
120 s
120 s
120 s
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
16
42
120 s
120 s
120 s
120 s
83 f
83 f
83 f
48 s
48 s
48 s
48 s
48 s
48 s
48 s
17
43
120 s
120 s
120 s
120 s
0,5 f
120 s
103 f
48 s
48 s
48 s
48 s
25 f
25 f
25 f
18
44
120 s
120 s
120 s
120 s
1 f
1 f
48 s
48 s
48 s
48 s
120 s
120 s
120 s
19
45
120 s
27 f
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
120 s
120 s
120 s
20
46
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
120 s
120 s
120 s
21
47
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
120 s
120 s
120 s
22
48
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
120 s
120 s
120 s
23
49
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
120 s
120 s
120 s
24
50
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
120 s
120 s
120 s
25
51
120 s
120 s
120 s
120 s
48 s
48 s
48 s
48 s
48 s
48 s
48 s
120 s
120 s
120 s
26
52
120 s
120 s
120 s
120 s
48 s
48 s
48 s
120 s
120 s
120 s
T2R
|
48
120
48
120
48
120
48
120
48
120
48
120
12
120
48
120
48
120
48
120
48
120
48
120
48
120
48
120
48
120
48
83
48
120
25
f /s
s
s
s
s
s
s
s
s
s
s
s
s
f
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
f
s
s
f
120
47
120
48
120
48
120
48
120
48
120
48
120
48
120
48
120
s
f
s
s
s
s
s
s
s
s
s
s
s
s
s
s
s
During the last year 418 Nozzles had been changed. N=418
C.
Weibull Analysis
1. The Cumulative Hazard Method
199
197
a)
Observed
Hazard
(1/n)
Cumulative
CDF
hazard
F(t)=1-e{-H(t)}
H(t)
F(t)
%
0,5
418
0,00239
0,0024
0,0024
0,24%
417
0,00240
0,0048
0,0048
0,48%
416
0,00240
0,0072
0,0072
0,71%
12
415
0,00241
0,0096
0,0095
0,95%
12
414
0,00242
0,0120
0,0119
1,19%
12
413
0,00242
0,0144
0,0143
1,42%
12
412
0,00243
0,0168
0,0167
1,66%
25
411
0,00243
0,0193
0,0191
1,89%
25
410
0,00244
0,0217
0,0215
2,12%
25
409
0,00244
0,0241
0,0238
2,36%
25
408
0,00245
0,0266
0,0262
2,59%
27
407
0,00246
0,0290
0,0286
2,82%
34
406
0,00246
0,0315
0,0310
3,05%
34
405
0,00247
0,0339
0,0334
3,28%
47
404
0,00248
0,0364
0,0358
3,51%
48
403 to 205
83
204
0,00490
0,0413
0,0405
4,05%
83
203
0,00493
0,0462
0,0452
4,52%
83
202
0,00495
0,0512
0,0499
4,99%
83
201
0,00498
0,0562
0,0546
5,46%
100
200
0,00500
0,0612
0,0593
5,93%
103
199
0,00503
0,0662
0,0641
6,41%
103
198
0,00505
0,0712
0,0688
6,88%
120
197 to 1
Analysing the times to failure, it is considered t0=0. Taking into account 418
elements and plotting them in the Weibull paper (see Plot 1), we can obtain:
10
N = 418 elements
= 0,725
b)
11
b)
TTF
Mode of
f/s
(h)
Failure
CDF
F(t)=1-e{H(t)}
F(t)
%
0,5
mounting
407
0,00246
0,00246
0,00245
0,25%
mounting
406
0,00246
0,00492
0,00491
0,49%
mounting
405
0,00247
0,00739
0,00736
0,74%
12
mounting
404
0,00248
0,00986
0,00982
0,98%
12
mounting
403
0,00248
0,01235
0,01227
1,23%
12
mounting
402
0,00249
0,01483
0,01472
1,47%
12
mounting
401
0,00249
0,01733
0,01718
1,72%
25
mounting
400
0,00250
0,01983
0,01963
1,96%
25
mounting
399
0,00251
0,02233
0,02209
2,21%
25
mounting
398
0,00251
0,02485
0,02454
2,45%
25
mounting
397
0,00252
0,02736
0,02699
2,70%
199
48
197
120
12
199
197
TTF
Mode of
f/s
(h)
Failure
CDF
F(t)=1-e{H(t)}
F(t)
%
27
slag
407
0,00246
0,0025
0,0025
0,25%
34
slag
406
0,00246
0,0049
0,0049
0,49%
34
slag
405
0,00247
0,0074
0,0074
0,73%
47
slag
404
0,00248
0,0099
0,0098
0,98%
48
83
slag
204
0,00490
0,0148
0,0147
1,47%
83
slag
203
0,00493
0,0197
0,0195
1,95%
83
slag
202
0,00495
0,0246
0,0243
2,43%
83
slag
201
0,00498
0,0296
0,0292
2,92%
100
slag
200
0,00500
0,0346
0,0340
3,40%
103
slag
199
0,00503
0,0396
0,0389
3,89%
103
120
f
s
slag
-
198
-
0,00505
-
0,0447
-
0,0437
-
4,37%
-
d)
e)
13
The graphical Weibull competing failures mode, show the following results:
Mounting Failure Mode graph
= 2,8 x 104 hours (characteristic life), time at which 63% of the nozzles
mounting
= 550 hours (characteristic life), time at which 63% of the nozzles are
expected to failed.
D.
mounting
running in failure mode. That result confirms that the mounting problem is a
childhood problem and it is necessary to work in two ways:
a) Improve the nozzle mounting process.
14
Slag
failure mode. That result confirms that the slag problem is not an inherent
15
nozzle failure mode. The context is affecting and making the nozzle to fail. It is
necessary to work in the slag formation cause reason in order to improve the
nozzle life.
For this porpoise I will recommend:
Consult to the supplier about if it exist edge detector system for this
application
Consult about edge detector system to another CCD around the world.
E.
Conclusion
F.
16
1. Input data
Following the Weibull ++7 instructions, I feed up the system with the following
data (the same data that I used for Weibull Analysis of competing modes of
failure):
mounting
17
This answer is quite similar that I could obtain by the manual graphical method.
In addition to this, I consider that the software give more accurate answer since
every result is mathematically calculated. In the graph method there are some
manual mistakes that affects to the reliability answer. Apart from that, the most
important part is the concept and the analysis that are associated with the
result. It is the same
mounting
mounting
mounting
represents a childhood problem and this is the key part of the analysis.
a)
18
b)
c)
19
d)
e)
20
f)
g)
21
G. References
ii
GeGa Lotz GMBH (1997), Operation and Maintenance Manual for Torch Cutting
Machine.
iii
ii
http://www.gega.de/files/duesenheft_duesendaten_stand_juli_2009_1.pdf
Section 4.2 Analysis of data with many censored items The Cumulative Hazard
Method, M13-Reliability, Maintainability and Risk
xi
27