Analysis of possible causes of cracks initiation on barking drum

Peter Trebun a
a,
, Frantiek Trebun a
b,
, Frantiek imck
b
a
Technical University of Koice, Faculty of Mechanical Engineering, Department of Industrial Engineering and Management, Nemcovej 32, Koice 04200, Slovakia
b
Technical University of Koice, Faculty of Mechanical Engineering, Department of Applied Mechanics and Mechatronics, Letn 9, Koice 04200, Slovakia
a r t i c l e i n f o
Article history:
Received 13 February 2014
Received in revised form 19 May 2014
Accepted 25 June 2014
Available online 5 July 2014
Keywords:
Barking drum
Crack
Residual stress analysis
Numerical analysis
a b s t r a c t
During operation of barking drum equipment the cracks were identied on the input part
of drums cylindrical shell. On the basis of operator request the analysis of reasons of pos-
sible causes of crack initiation on drum shell has been realized. The analysis includes
numerical computations of stress elds in the locations of stress concentrators resulting
from loading of drum during operation as well as inuence of vertical offset of drives to
the stress state in drum shell. The residual stresses in selected locations of drum shell were
determined by the hole-drilling method. On the basis of analysis new concept of drum shell
strengthening has been proposed in order to ensure its further safe operation.
2014 Elsevier Ltd. All rights reserved.
1. Introduction
The aim of the paper is to provide a solution to the problems that arose during operation of the barking drum. On the shell
of the drum, a number of cracks were detected and the operator of the equipment was concerned about its condition. The
analysis of stresses and states was carried out by the numerical and experimental methods of mechanics.
The debarking process has been used in the pulp and paper mills for over a century. The pile of logs fed into the drum
is put into crosswise tumbling or rolling motion by rotating the drum, which is tted with log prisms welded to inner side
of the drum shell. While tumbling and rolling inside the drum, the logs rub and hit against each other and bark loosens
from the log surface. The drum shell has a number of longitudinal loop-holes (bark slots) through which the loose bark
should come out of the drum. In principle, the barking drum is a combination of a debarker and a bark separation unit
[1,2].
The method of debarking has evolved from a batch debarking to a continuous debarking process. In batch debarking, the
debarking drum is lled to a high degree while keeping the discharge gate closed and rotating the drum. The second method
uses continuous charging and discharging of the barking drum, and this type of equipment is described in the paper.
2. Structure of barking drum, localization of cracks on drum shell
The barking drum (Fig. 1) serves for debarking of logs with the length varying from 2.5 m to 6.0 m [3]. The apparatus con-
sists of a drum body, driving and supporting tyres, a discharge gate, housing and exhausting system. In the drum, wood logs
with diameter up to 700 mm can be processed, maximum capacity of the equipment is 320 m
3
per hour, and angular velocity
of the drum during debarking is about 5 rpm [4].
http://dx.doi.org/10.1016/j.engfailanal.2014.06.012
1350-6307/ 2014 Elsevier Ltd. All rights reserved.

Corresponding authors. Tel.: +421 55 6023235; fax: +421 55 6334738 (P. Trebun a). Tel.: +421 55 6022462; fax: +421 55 6334738 (F. Trebuna).
E-mail addresses: peter.trebuna@tuke.sk (P. Trebuna), frantisek.trebuna@tuke.sk (F. Trebun a).
Engineering Failure Analysis 45 (2014) 106117
Contents lists available at ScienceDirect
Engineering Failure Analysis
j our nal homepage: www. el sevi er . com/ l ocat e/ engf ai l anal
The body of the drum is designed as a cylindrical shell and consists of two parts (front and rear), which are driven by the
system of tyres (Fig. 2) with drives controlled by frequency changers. Rotation of the drum leads to debarking of logs inside
the drum. In most cases, one block of drives consists of 16 tyres located on the circumferential and longitudinal directions of
the drum cylinder (Fig. 1).
On the inner sides of both drum cylinders, 18 prisms are welded on their perimeters in the axial direction (Fig. 3). They
serve for better debarking of logs during their rotation inside the drum. In the shell of the drum, loop-holes are created for
bark removal (Fig. 3).
The body of the drum was made by welding of sixteen cylindrical shells where the last circumferential weld on the front
(charging) part consisting of 10 cylindrical shells was done after positioning of the drum on the tyres, so that the inner
prisms in the locations of the last circumferential weld of the front part of the drum were interrupted (Fig. 4).
Fig. 1. Barking drum (from leaet of producer) [3].
Fig. 2. System of rubber tyres for actuation of cylindrical drum shell.
Fig. 3. Longitudinal prisms welded to internal wall of drum shell.
P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117 107
After several years of the drum operation, cracks were found in the front (charging) part of the drum shell, especially, in
the locations of the interrupted prisms (Fig. 5). The cracks were repaired by welds, but after a short time of operation they
were detected again.
Because of the recurring cracks, short steel logs were welded on the interior side of the drum shell, Fig. 6a. The outer side
of the shell was reinforced with stiffening rings from both sides of the circumferential weld on the front part of the drum,
Fig. 6b.
At the time of the reinforcement of the drum by short steel logs, new cracks were detected on six long prisms in the front
part of the drum, Fig. 7a.
As Fig. 7 shows, the cracks on the prisms lay at the plane perpendicular to the drum axis and they covered almost the
whole cross-section areas of the prisms. During the reparation works also newly detected cracks were welded. The problems
with cracks led to the conclusion that it was necessary to provide thorough analysis of stress and deformation in the barking
drum [5,6].
3. Analysis of stresses in drum shell by the nite element method
For the assessment of causes of crack initiation in the drum shell the nite element method was used [79]. When
carrying out the computation, several types of simulations of interactions of wood logs and prisms during debarking were
performed, especially, loading of logs by concentrated and distributed force.
Fig. 4. Discontinuity of prisms in location of last weld on the front part of drum.
Fig. 5. Cracks identied on end of prism.
108 P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117
Fig. 8a shows a computational model for loading of the prism by concentrated force, Fig. 8b and c shows elds of principal
stresses r
1
and their directions for this load case.
Fig. 9a gives a computational model for loading of the prism by distributed loading, Fig. 9b and c shows elds of principal
stresses r
1
and their directions for the given load case.
Fig. 6. Reinforcement of drum in location of cracks. (a) Short prisms and (b) outer ringlet (view from internal side of drum).
Fig. 7. Cracks on prisms in front part of drum (a) location of cracks on prisms and (b) illustration of crack shapes.
P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117 109
Fig. 10 shows typical cracks on the drum shell. From comparison of Figs. 9 and 10 it is obvious that the cracks occurred in
the area of stress concentration between the loop-hole and the prism.
In the process of numerical analysis, simulation of loading of the perforated rings of the shell drum was also performed.
This simulation of shear loading documented transmission of the torque from individual drives in case of loading of the drum
by wood logs. The simulation was performed for all types of perforated rings.
Fig. 11a shows the computational model of loading of the perforated rings by shear stresses (they represent various tor-
sion moments), Fig. 11b and c gives elds of principal stresses r
1
and their directions, respectively, for one type of the per-
forated shell.
In the last stage of the drum shell analysis, attention was paid to the causes of cracks in the prisms, Fig. 7.
Since the cracks lay in one plane and were perpendicular to the prism axes, they could only be caused by tensile stress.
The cracks inspired authors to modeling various possible loadings that could cause their formation. The computational
model was loaded by self-weight of the drumas well as by the weight of wood logs determined on the basis of the processing
velocity of the barking drum. The simplied model had no loop-holes for bark discharging, i.e. the corresponding stress con-
centrators were not considered at this stage of computation. A simplied model of the drum that was positioned on the sup-
ported and driven tyres is given in Fig. 12.
With respect to the fact that cracks in the prisms (Fig. 7) could be caused only by bending moments acting in the planes
perpendicular to the drum axis, the inuence of the vertical setup of one block of the drum drive tyres on the stress in the
prisms was examined, (Fig. 12).
Fig. 8. Loading of prism by concentrated force (a) computational model, (b) eld of principal stresses r
1
and (c) directions of principal stresses r
1
.
110 P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117
Fig. 13a gives elds of von Mises stresses for the correct vertical positioning of drive rubber tyres of all blocks in the plane
parallel to the drum axis. Fig. 13b shows elds of von Mises stresses for one block of drive rubber tyres shifted vertically by
10 mm.
As Fig. 13b shows, the vertical shifting of the tyre block led to the stress concentrators that were located in the circum-
ferential direction of the cylindrical shell that could be a reason of crack initiation.
4. Experimental determination of residual stresses in shell of barking drum
Taking into consideration the fact that the cracks in the barking drum shell were located in the area of the circumferential
weld that was performed during positioning of the drum shell on the tyres, residual stresses were analyzed by the
hole-drilling method [10].
Fig. 9. Loading of prism by distributed force (a) computational model, (b) eld of principal stresses r
1
and (c) directions of principal stresses r
1
.
P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117 111
Fig. 14 shows positions of strain-gages for the measurement of residual stresses by hole-drilling method. Fig. 15 illus-
trates locations of the drilled holes of strain-gage rosettes 1, 2 and 5. The process of drilling in location of strain-gage rosette
3 is documented in Fig. 16.
The drilling of holes was carried out by apparatus RS 200 using strain-gage rosettes RY 21-3/120. The process was per-
formed in 10 steps by drilling 0.5 mm at a time and the resulting hole depth was 5 mm. The hole diameter was 3.2 mm. The
released strains were registered in each step of the measurement by strain-gage equipment P3 [11].
The magnitudes of principal residual normal stresses and their directions were computed from the measured released
strains in the Restress program using ASTM E 837-01, Integral and Kockelmann methods [12,13].
Table 1 gives magnitudes of the principal residual stresses and their directions for individual locations measured accord-
ing to ASTM E 837-01. Angle u represents declination of the principal stress vector r
1
from the axis of drum rotation.
5. Discussion of results and conclusion
The existing cracks (even if they were near the welds) could not be considered as a result of low strength of llet welds. It
was documented by the strength of the unit length of the weld which was satisfactory and in no point the resultant force
acting inside the weld unit length exceeded the proposed nominal value.
On the other hand, the identied cracks documented and examined by the authors pointed out possible reasons of their
formation due to stress concentration and fatigue failure [14,15].
Assessment of the structures by fatigue is based on the assumption that nominal stresses lie in elastic area and their
amplitudes do not exceed 1.5 multiple of the yield point for normal and 1.5/

3
p
multiple of the yield point for shear. If
we assume that the number of loading cycles can be determined directly from the number of drums rotations, then e.g.
for 5.5 rpm during 20 h of operation per day during the year (350 days), we have 2,310,000 cycles requiring the assessment
of fatigue life. If we also consider that prisms are welded to the drum, then according to the common standards for the design
of steel structures, we can choose the detail category (notch) 90, which according to the standard [16] requires the use of a
chamfer angle smaller than 45 and the length of the chamfer at least 6 times bigger than the thickness of the welded rib.
Fig. 10. Typical cracks initiated during operation on the shell of drum.
112 P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117
This design principle was not followed and accordingly the reinforcements caused sudden change of stiffness in those loca-
tions, which caused formation of cracks.
When repairing the drum, the chamfer angle on the ends of the welded prisms with the length of 1900 mmwas decreased
to 30, therewith one of the two conditions given in the standard was fullled.
With respect to the number of loading cycles from the beginning of the operation to the present day, it was necessary to
assess fatigue life of the equipment. If we take into account that the differences in loading of variable loading of the struc-
tural detail (drum shell) were not important, then it can be assumed that the loading of structure led to constant stress
amplitude and fatigue can be treated as fatigue with the amplitude of nominal or actually shape stress. The curves of the
fatigue strength for individual categories of details are given in standards for the design of steel structure [16]. Fig. 17 gives
the curve of fatigue life with two slopes also for the detail category 90 that belongs to the examined structure.
Fig. 11. Loading of perforated shell by shear stresses (they represent torque loading) (a) computational model, (b) eld of principal stresses r
1
and (c) eld
of directions of principal stresses r
1
.
Fig. 12. Model of input part of drum supported by tyres.
P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117 113
For category 90, the stress amplitude for N
MKD
reaches 379 MPa, which is approximately 1.5 times of the yield point, for
N
D
= 5 10
6
cycles it reaches 66 MPa, and for number of cycles N
L
= 10
8
this value is 36 MPa.
The magnitudes of partial fatigue damages can be determined according to the above mentioned standard by comparison
of the whole damage D
d
with the allowed limit value of damage D
L
.
Reliability condition is expressed by relation
D
d

X
D
i

X
n
i
N
i
6 D
L
If there are no more accurate values determined, value D
L
= 1,0 can be considered.
Fig. 13. Fields of von Mises stresses (a) correct setup of driving tyres and (b) vertical overhang of driving tyres by 10 mm.
Fig. 14. Positions of strain-gages in locations 15 for measurement of residual stresses.
114 P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117
Fig. 15. Locations of strain-gage rosettes 1, 2 and 5.
Fig. 16. Drilling in location of strain-gage rosette 3.
P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117 115
For the known stress level determined, e.g., by numerical simulation using the nite element method, if the stress does
not exceed 66 MPa, but is bigger than 36 MPa in the boundaries of the stress amplitude, the technical lifetime is limited by
the interval 5 10
6
to 10
8
loading cycles (rotations of the drum). During previous operation of the drum, the number of load-
ing cycles exceeded 8.6 10
6
, i.e. for stress amplitude 59 MPa the cracks were initiated in the critical locations of the prisms.
This very fact requires us to know the stress eld for the dened support and loading.
Taking into account the levels of residual stresses that reach maximum 132.6 MPa in the locations of measurements and
taking these values as middle stresses, the stress amplitude for the computation of fatigue life can be decreased by approx-
imately 17%, i.e. with the amplitude 49 MPa. These are also the reasons for premature damage of the supporting elements
either prisms or the drum shell.
On the basis of the analysis it can be stated that:
(1) Measurement of the residual stresses in ve locations showed that the levels of residual stresses were very high. The
stresses had the character of tensile stresses and they were oriented mostly in the direction of the drum axis. High
levels of the residual stresses decreased the stress amplitude and that consequently caused failures in the prisms
and welds.
(2) The simulations conrmed that crack directions corresponded to the directions that were perpendicular to the direc-
tions of the principal tensile normal stresses.
(3) The newly detected cracks in the prisms could result from inaccurate positioning of the supporting and drive units.
(4) Measures proposed by the supplier of the equipment could not be considered as sufcient for prevention of crack ini-
tiation, because the reasons for crack initiations were not eliminated. The ring reinforcement on the outer side of the
drum did not solve the problem of the elimination of axial stresses. Longitudinal prisms with smaller chamfer angle
decreased the inuence of stress concentration, but the length of the chamfer did not t the values of standards.
(5) Welding of the 1900 mm long 18 prisms could be considered as only partial removal of the critical location in the area
of the most frequent occurrence of cracks, but they did not solve the problem of the location of new crack initiations
which the authors pointed out in their report.
Acknowledgement
This work was supported by Projects VEGA 1/0937/12 and VEGA 1/0102/11.
Table 1
Magnitudes of principal residual stresses and their directions determined according to ASTM E 837-01.
Measured location r
1
(MPa) r
2
(MPa) u
1 124.13 5.61 11.45
2 40.94 9.43 3.83
3 132.63 3.40 79.30
4 122.43 2.34 71.05
5 50.59 29.27 74.75
Fig. 17. Curve of fatigue life with two slopes for individual detail categories.
116 P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117
References
[1] Jalonen A. GentleBarking a log debarking method for higher yield From innovation to reality. Metso Paper Pori Oy, Pori, Finland. <http://
www.eucalyptus.com.br/icep02/antti_jalonen.pdf>.
[2] Barorth R. Literature review of the latest development of wood debarking. Report A No 27. University of Oulu, Control Engineering Laboratory; 2005. p.
29. ISSN 1238-9404.
[3] <www.metso.com>.
[4] EasyTyre BARKING DRUm BDET. Maintenance instructions. 4915724. pp. 20.
[5] Trebun a F et al. Analysis of crack initiation in the press frame and innovation of the frame to ensure its further operation. Eng Failure Anal
2011;18(1):24455.
[6] Trebun a F et al. Identication of causes of radial fan failure. Eng Failure Anal 2009;16(7):205465.
[7] csi L, leszts P. Moving toward a more realistic material model of a ductile material with failure mode transition. Materialwissenschaft und
Werkstofftechnik 2012;43(5):37987. ISSN 0933-5137.
[8] Vavro J, Vavro Jr J, Krdla M, Vavrov A. Optimization of the design of the thin shell mechanical Structures. Univ Rev 2010;4(1):3540. ISSN 1337-6047.
[9] Saha A et al. Investigation of probable cause of damage of steam drum of naphtha cracking furnace. Eng Failure Anal 2009;16:138796.
[10] Trebun a F, imck F. Kvantikcia zvykovy ch napt tenzometricky mi metdami. Preov: Grafotlac; 2005, ISBN 80-8073-227-2.
[11] Trebun a F, imck F. Prrucka experimentlnej mechaniky. Koice: Typopress; 2007. 970-80-8073-816-7.
[12] ASTM-E 837-01. Determining residual stresses by the hole-drilling strain-gage method. Philadelphia: American Society for Testing and Materials;
2001.
[13] Tech Note TN-503-6. Measurement of residual stresses by the hole-drilling strain gage method. Raleigh, NC: Measurements Group Inc.
[14] Trebun a F, Burk M. Medzn stavy lomy, Grafotlac Preov; 2002. ISBN 80-7165-362-4.
[15] Trebun a F, imck F. Odolnost

prvkov mechanicky ch sstav. Koice: EMILENA; 2004, ISBN 80-8073-148-9.

[16] Juhas P. a kol. Navrhovanie ocel

ovy ch kontrukci. STN 73 1401.

P. Trebuna et al. / Engineering Failure Analysis 45 (2014) 106117 117