Original Article

Proc IMechE Part B:

J Engineering Manufacture
2017, Vol. 231(10) 1812–1819
Effect of operating parameters on Ó IMechE 2015
Reprints and permissions:
plate bending by laser line heating sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0954405415612678
journals.sagepub.com/home/pib

Biplab Das and Pankaj Biswas

Abstract
Laser line heating is a plate bending technique. It is produced due to the heating effect of a laser beam when irradiated
over a suitable heating path. In this work, the effect on angular deformation under different operating parameters, such
as energy (in terms of laser power), scanning speed and number of passes along with the thickness of the substrate
material, was studied under straight line scanning schemes. In this experiment, CO2 laser has been used and the sub-
strate material used is the mild steel sheet. The effect of each operating parameter was taken into account to find the
optimum process parameter. The design of experiment has been applied with the help of Taguchi method, and the
results were obtained and analysed with the help of analysis of variance and Taguchi analysis (signal-to-noise ratio), for
the determination of the optimized values of the operating parameters with their effect towards angular deformation in
the process. A regression analysis also has also been performed to obtain a suitable co-relationship between bending
angle and other operating parameters.

Keywords
Multi-pass, angular deformation, Taguchi design of experiment analysis, regression analysis, line heating

Date received: 5 September 2014; accepted: 22 September 2015

Introduction For the occurrence of TGM, the value of the Fourier

number (F0) has to be lesser than unity.2 The Fourier
The process of line heating assisted with laser is found number (F0) can be expressed as follows
to be one of the most versatile techniques for various
surface development activities. The process is based on cd
localized heating without increasing the temperature F0 =
t2 v
beyond recrystallization point. The mechanism of the
process is dependent on various process parameters, where C, d, t and v are thermal diffusivity, laser beam
the shape of the substrate and the material properties diameter, sheet thickness and scanning velocity,
to be used for the substrate. Here, in this case the tem- respectively).
perature gradient mechanism (TGM) has been incorpo- According to Lawrence et al.,3 for TGM to occur,
rated. In this case, bending is induced by thermal laser spot diameter is somehow equal to that of the
gradient between surfaces of the material to be pro- thickness of the sheet. The effect of bending angle with
cessed. Temperature gradient is the reason for bending reference to the change in laser parameters has been
of thick plates; various analytical methods are pro- proposed by Shichun and Jinsong.4 They proposed that
posed for the determination of temperature field which the increase in angular deformation occurs with the rise
is required for the simulation of the line heating in laser power, number of scans, thickness of the mate-
technique. rial and thermal effect index and decrease with higher
Zhang et al.1 presented the effect of line heating fac- spot diameter, thickness of the sheet and scanning
tors including heat input, heating line position from
plate edge and plate size on angular deformation which Department of Mechanical Engineering, Indian Institute of Technology
were explored with the numerical simulation. The main Guwahati, Guwahati, India
motto is to develop a relation between the heating
Corresponding author:
parameters and the angular deformation, which are Pankaj Biswas, Department of Mechanical Engineering, Indian Institute of
required for fabrication of ship hull by the process of Technology Guwahati, Guwahati 781039, India.
line heating. Email: das.biplab@iitg.ernet.in
Das and Biswas 1813

speed. The analysis based on TGM was proposed by

Shen et al.5 by taking the cross section into consider-
ation. An equation was proposed for the prediction of
angular deformation with the variation in laser power
and scan speed.
Vollertsen6 has developed a relation for the determi-
nation of angular deformation. The expression was
obtained on the basis of TGM. The expression for the
bending angle was obtained on the basis of process
parameters and material properties. Similar formula-
tion was also given by Yau et al.,7 where the expression
of bend angle is obtained based on two-layer model for
TGM along with addition of counter bending effect for
Figure 1. Experimental set-up for straight line bending.
some of the pure elastic strains, which resulted into
dual equations, for obtaining both counter bending
and bending angle. Mucha et al.8 have derived an
on TGM. The output results obtained were suitably
expression for obtaining the bending angle; the expres-
analysed under Taguchi analysis and analysis of var-
sion obtained is based on TGM. The output obtained
iance (ANOVA), and with the development of relation-
reveals that the coefficient of thermal expansion and
ship between the operating parameters, suitable
heat capacity has a significant effect as material para-
optimum parameters were obtained.
meters with reference to that of laser forming process.
Kyrsanidi et al.9 have developed a parametric mathe-
matical model where plastic bending is considered at Experimental details
the time of heating. It is also found that the developed
model took very less computing time in comparison to A CO2 gas laser machine (Orion 3015, LVD make) was
the existing model. used for the experiments. Rectangular mild steel blanks
Shen et al.10 have derived a formula for obtaining of dimension of 200 mm 3 200 mm and thicknesses of
angular deformation by laser forming process. A model 3 and 4 mm were taken as the working specimens. For
was developed on the basis of conventional equilibrium the experiment to be started, the specimens were put on
and compatibility conditions. Mucha11 has developed a the machine work table with the help of clamp, with
model for the process of plate bending on the basis of job holding attachment as shown in Figure 1; about
TGM. The model provides solution for longitudinal 5–6 mm gap was maintained between the work table
and transversal angular deformations. Shi et al.12 pres- and the specimen so that there should be no constraints
ent the study of TGM for the forming process. Under for the specimen while undergoing counter bending.
processing condition of TGM, the plate bends in the Numerous numbers of experiments were carried out by
direction of both x- and y-axes. The model has been altering the operating parameters. The angular defor-
developed for estimating the angular deformation mation of the plate was determined with the help of
about y-axis, which has been obtained based on the coordinate measuring machine (CMM) for individual
heat transfer theories and mechanics of elasto-plasti- set of experiments.
city. Shen et al.13 also proposed a work related to the For calculating the stand-off distance, initially the
forming of plates under the action of laser as a heat width of the zone (w) affected by moving heat source is
source under simultaneous scanning of laser in parallel taken into account along with the uniform velocity (v).
directions. The outcome reveals that the angular defor- It is estimated to be equal to the summation of the dia-
mations were obtained for both single and multiple meter of the laser beam (d) and the thermal diffusion
scanning under the same line, keeping less gap between length (l ). The thermal diffusion length (l ) can be
the two scanning paths. Casalino and Ludovico14 have calculated as
proposed a feed forward neural network with back pro- rffiffiffiffiffiffiffiffiffiffiffi
pagation technique for evaluating bending angle and  c3d
l =2 ð1Þ
for selecting the TGM and buckling mechanism (BM) v
laser forming parameters. The result reveals that neural
where ‘c’ is the thermal diffusivity, which can be again
network provides fast and precise outcome in compari-
expressed as
son to finite element analysis and it is easier to use
them than the multivariate regression analysis. k
In this article, an experimental study was conducted c= ð2Þ
r3C
for obtaining the optimized values of operating para-
meters which include scanning speed, laser power and where ‘k’ is the thermal conductivity of the material, ‘r’
the number of scans on the angular deformation with is the density of the material and ‘C’ is the specific heat
the change in thickness of the substrate material based capacity of the material.
1814 Proc IMechE Part B: J Engineering Manufacture 231(10)

Therefore, the width of the zone (w) can be expressed Table 1. Operating parameters and their selected levels (under
as Taguchi DOE L-16 array).
rffiffiffiffiffiffiffiffiffiffiffi! rffiffiffiffiffiffiffiffiffiffiffi!
c3d c3d Experiment Traverse Power Number of Sheet
w=d+2 2 =d+4 ð3Þ No. speed (W) passes thickness
v v (mm/min) (mm)

From the above expression, if the velocity (v) is 1 500 300 1 3

taken to be very large, then the width of the zone (w) 2 500 400 2 3
can be considered to be the measure of beam diameter 3 500 500 3 4
4 500 600 4 4
(d). On the basis of this theory, it is assumed that the 5 400 300 2 4
heating impression is obtained with the help of moving 6 400 400 1 4
laser beam at high speed (probably be the highest speed 7 400 500 4 3
possible, speed attainable with the work station they 8 400 600 3 3
worked with) to be the beam diameter (d) at a particu- 9 300 300 3 3
10 300 400 4 3
lar stand-off distance between the tip of the nozzle and 11 300 500 1 4
the workpiece surface. As in this case we need to main- 12 300 600 2 4
tain the TGM, the laser beam diameter is to be equal 13 200 300 4 4
to that of the thickness of the substrate (workpiece). 14 200 400 3 4
For this reason, suitable stand-off distance was main- 15 200 500 2 3
16 200 600 1 3
tained. The rectangular plates were heated along the
centre line of the top surface of the plates.

Taguchi method of design of experiment uniform for all, then the mixed level design is to be
taken into consideration. By considering both the val-
The design of experiment (DOE) is a method where the
ues (number of parameters and levels), the suitable OA
experiments were performed according to a fixed sys-
has to be chosen from the manuals or handbook. The
tem or plan. This is necessary for solving any problem
design should be extracted from the listed designs and
related to engineering in principles, techniques and data
it is to be used.
collection stage, ensuring in getting some logical and
Here, the experimental design for multi-pass line
justifiable conclusions. It also enables the designs for
heating was taken into consideration by selecting four
the determination of individual and interactive effects
laser operating parameters; they are as follows: traverse
of many factors, which could affect the output results
speed, laser power, plate thickness and number of
in any design. Here in this case, Taguchi method of
scans. All the parameters were varied in four levels
DOE is taken into consideration before performing the
except the plate thickness which is varied in two levels.
experiments and analysis.
L-16 OA was incorporated for the DOE as shown in
Taguchi method15 has been designed for obtaining
Table 1, and with the help of parameter and output
improved quality processes and outputs where the per-
signal-to-noise (S/N) ratio and S/N analysis were per-
formance of the output depends on several factors. It
formed. As in this case the main aim is obtaining large
can be thought that for carrying out any test and devel-
bending with the change in operating parameters,
opment, simple logic is sufficient for establishment of
higher values of S/N ratio give better result correspond-
possible combination of several factors along with the
ing to optimal process parameters.
ranges to be taken into consideration. Taguchi has
developed a set of designs for applications. The special
features of this type of designs are the usage of ortho- Results and discussion
gonal arrays (OAs). These arrays help in performing
minimum number of experiments required for a given
Single-pass line heating
set of parameters. Here, we have used Taguchi DOE A series of experiments on single pass (run) were per-
for optimizing the operating parameters. The experi- formed for obtaining the angular deformation of mild
ment design must satisfy the following two conditions: steel plates of dimension 200 mm 3 200 mm 3 3 mm.
The experiments were carried out by altering the laser
1. The number of trial runs must be determined. power and traverse speed, keeping the laser diameter
2. Trial conditions are to be specified. constant, that is, 3 mm. The peak temperatures
obtained from the experimental study at different loca-
tions (i.e. as shown in Figure 2) are shown in Table 2
Selection of the level of parameters which shows the effect of operating parameters on peak
For the selection of the level of parameters, the initial temperatures at different locations for single-pass line
selection of the number of operating parameters heating process. The results are shown in Table 3 which
involves in the process and then the level is to be shows the effect of angular deformation for single-pass
defined. If it happens that the number of levels is not line heating process.
Das and Biswas 1815

Table 2. Peak temperature in thermocouple locations under single-pass line heating.

Exp No. Power (W) Traverse Temperature at Temperature at Temperature at Temperature at

speed (mm/min) position ‘A’ (°C) position ‘B’ (°C) position ‘C’ (°C) position ‘D’ (°C)

1 400 500 142.027 97.266 99.665 146.963

2 400 400 164.923 115.136 121.176 168.714
3 400 300 207.066 154.467 157.59 215.56
4 400 200 262.974 201.42 223.14 300.519
5 500 500 220.825 49.419 139.485 224.79
6 500 400 226.86 59.374 182.369 343.546
7 600 500 147.455 103.694 146.77 380.806
8 600 200 372.536 238.252 309.375 523.17

Table 3. Results for experiments under single-pass line heating.

Experiment Power Traverse speed Angular

No. (W) (mm/min) deformation
for single pass (°)

1 400 500 0.2502

2 400 400 0.8084
3 400 300 0.9666
4 400 200 1.026
5 500 500 0.81328
6 500 400 1.1901
7 500 300 1.2799
8 500 200 1.3999
9 600 500 1.1133
10 600 400 1.3667
11 600 300 1.1496
12 600 200 1.652

Figure 3. Traverse speed versus bending angle for different

laser powers.

Figure 3 shows the curve between the bend angles

versus traverse speeds for mild steel plate of 3 mm thick-
ness. The values obtained were under different sets of
power and traverse speed. It has been observed that
constant power and increase in traverse speed lead to
the reduction in angular deformation (bend angle).
Another study has been performed based on the trend
line which has been obtained based on the values of
bend angle obtained under different heat inputs per unit
length, as shown in Figure 4, for single-pass line heating
process. Here, the heat inputs per unit length have been
calculated using laser power (P) and traverse speed (v),
that is, P/v.
From Figure 3, it is also revealed that with uniform
increment of laser power under constant traverse
speed, the angular deformation is not uniform. With
the increase in power, the behaviour follows non-uni-
formly. From Figure 4, it has been observed that with
the increase in line heat input, that is, heat input per
unit length, the angular deformation increases. But the
rate of increase in angular deformation is more with
lower values of line heat input. After a certain line heat
input (i.e. approximately 150 J/mm), the rate of increase
Figure 2. Positions of thermocouples. in angular deformation is very low. So from Figure 4, it
1816 Proc IMechE Part B: J Engineering Manufacture 231(10)

can be concluded that the line heating should be carried input in the irradiated area from the first pass could
out near the line heat input value of 150 J/mm. So in serve for the additional forming by reducing the
this way for different thicknesses of plates, the opti- temperature-dependent flow stress of the material, in
mum line heat input can be calculated. that a hot plate is easier to form than a cold one.
Figure 5 shows the results obtained under multiple
scanning schemes under different heat inputs per unit
Multi-pass line heating length for samples of 3- and 4-mm-thick mild steel
We have developed the design of experiment on the plates.
Taguchi DOE, L-16 array, on the basis of that experi-
mentation has been done and the corresponding results
of bending angle are tabulated in Table 4. Figure 5 rep- Analysis of DOE results
resents the heat input per unit length versus bend angle The obtained experimental results were investigated
under different scanning schemes for mild steel plate of under Taguchi method of analysis. Taguchi analysis
3 and 4 mm thicknesses. incorporates the basic three objectives, which are as fol-
With the increase in material thickness and strength, lows: (1) for the determination of the optimum condi-
more amount of energy is required for obtaining the tion, (2) factors contributing to the results and
same angular deformation. Here in this technique as determination of percentage contribution of the operat-
multiple passes are involved, it has been observed that ing parameters to the output results and (3) the
with the increase in the number of passes, the angular expected results at the optimum condition.
deformation increases. This is because the retained heat Identification of the optimum condition can be
obtained by simple arithmetic calculations.15 On deter-
mination of optimum condition, trial run must be
accompanied so as to know the difference in the output
in the experiment conducted in optimum and non-
optimum condition. Optimum may not be the set of
experiment that has been obtained under OA.

S/N and its significance. The ‘signal’ depicts the useful

information and the ‘noise’ represents the value which
is undesirable.16 The expression for the S/N ratio has
been formulated in such a way that the higher factor
level settings are being utilized for obtaining the opti-
mized quality characteristics of the experiment.
From the Taguchi analysis, we obtain the plot for S/
N ratio in Figure 6. The values of the graph indicate the
optimal parameters based on the bending angle as an
Figure 4. Heat input per unit length versus bending angle. output for the process.

Table 4. Experimental results on the basis of Taguchi L-16 array DOE.

Experiment Traverse speed Power Number Plate thickness Resultant angular

No. (mm/min) (W) of passes (mm) deformation (°)

1 500 300 1 3 0.65

2 500 400 2 3 1.49010
3 500 500 3 4 1.42505
4 500 600 4 4 3.00825
5 400 300 2 4 0.04000
6 400 400 1 4 0.17920
7 400 500 4 3 4.59330
8 400 600 3 3 3.61996
9 300 300 3 3 2.69666
10 300 400 4 3 4.24840
11 300 500 1 4 0.64720
12 300 600 2 4 1.30280
13 200 300 4 4 2.52780
14 200 400 3 4 2.54715
15 200 500 2 3 2.58338
16 200 600 1 3 0.77000
Das and Biswas 1817

Figure 7. Percentage contribution of individual operating

parameters.

Figure 5. Heat input per unit length versus bend angle under and 86.85%, respectively, which indicate the wellness
different scanning schemes for 3 and 4 mm thicknesses of plates. of data to fit into a statistical model.
From Figure 7, it can be observed that the bending
angle has a significant influence on sheet thickness and
number of passes (scans) than the other parameters. It
is also observed from the values of percentage contribu-
tion that both scanning speed and the number of passes
are dominant with respect to all other parameters.

Regression analysis. A multivariable regression analysis

has been done for obtaining a relationship between the
angular deformation and the input parameters (i.e. laser
power, traverse speed, sheet thickness and number of
passes). The equation obtained based on the regression
analysis is shown in equation (4)

˘ = 2:87  0:00151v + 0:00229P + 1:03N  1:12t

Figure 6. Mean S/N ratio for the input parameters. ð4Þ

From Figure 6, it has been observed that the usage where ˘ is the bend angle (°), v is the traverse speed
of the values of the parameter having negative mean S/ (mm/min), P is the laser power (W), N is the number of
N ratio value will lead in obtaining poor quality of passes and t is the sheet thickness (mm).
output (bend angle). For obtaining the optimum bend Here, R2 acts as a measure of accuracy of the result
angle, the parameters should be selected in the positive to be obtained as output in relation to that of the input
region of vertical axis for all parameters. It is observed parameters with the experimental results. Higher value
from Figure 5 that optimum bend angle can be of R2 indicates that the empirical relation obtained, that
obtained using laser power = 600 W, traverse speed = is, equation (4), above has a better predicting capability
200 mm/min, number of passes = 4 and sheet of the output (bend angle), and there will be less mis-
thickness = 3 mm. match between the experimental output and the output
obtained from equation (4). As per Table 6, which is
the regression table for equation (4), the value of R2 is
ANOVA. It is a technique which has been used for pro- found to be 92.1%, which shows that the model can be
viding measure of confidence. This technique has been used with sufficient accuracy.
applied to the experimental results for the determina- A test run has been performed based on the para-
tion of the effectiveness of the parameters that are meter tabulated in Table 7 and it was observed that the
responsible for the required output. This technique experimental output obtained validates well with that
helps in providing confidence on the basis of variance of the output obtained from the regression equation
of the data.16 (i.e. equation (4)). The results are also plotted in
From the above analysis as shown in Table 5, the R2 Figure 8. The results obtained are of great accuracy
and the adjustable R2 are found to be about 95.62% with minimal amount of percentage of error. Thus, this
1818 Proc IMechE Part B: J Engineering Manufacture 231(10)

Table 5. ANOVA table for bending angle.

Source DF Seq. SS Adj. SS Adj. MS F p Percentage contribution

Traverse speed 3 0.7946 0.7946 0.2649 1 0.465 2.629723889

Laser power 3 1.647 1.647 0.549 2.07 0.223 5.450736528
Number of passes 3 21.4155 21.4155 7.1385 26.94 0.002 70.87446759
Sheet thickness 1 5.0341 5.0341 5.0341 19 0.007 16.66032347
Residual error 5 1.3248 1.3248 0.265 – – 4.384417579
Total 15 30.2161 – – – – –

DF: degree of freedom; Seq. SS: sequential sum of square; Adj. SS: adjusted sum of square; Adj. MS: adjusted mean of square.

Table 6. Regression table.

Predictor Coef. SE coef. T p

Constant 2.867 1.05 2.73 0.02

Traverse speed 20.00151 0.001044 21.44 0.177
Laser power 0.002286 0.001044 2.19 0.051
Number of passes 1.0316 0.1044 9.88 0
Sheet thickness 21.1218 0.2334 24.81 0.001
S = 0.466887, R2 = 92.1%, R2(adj) = 89.2%

SE: standard error.

Table 7. Values of the bend angle obtained from the regression equation.

Experiment Traverse Laser power No. of pass Plate thickness Bend angle from Bend angle from Percentage
No. speed (W) (mm) experiment regression of error
(mm/min) analysis equation

1 400 400 1 3 0.80 0.852 6.103

2 300 400 1 3 0.96 1.003 4.287
3 400 700 1 4 0.45 0.419 7.398
4 500 600 1 4 0.037 0.039 5.128
5 600 450 2 4 0.523 0.5746 8.859
6 200 600 4 3 5.10 4.702 8.464

model can be used for predicting the bend angle by

altering the input parameters within the range.

Conclusion
In this article, bending of mild steel sheet is carried out
with the process of laser line heating with the input
process parameters considered as traverse speed, laser
power, number of passes and sheet thickness, and the
response obtained is bend angle as the output.

 The line heating process parameters are optimized

with respect to multiple performances in order to
achieve best quality of angle of bend.
 Optimization of the process parameters was carried
out using Taguchi method of analysis and a statisti-
Figure 8. Comparative plot between the bending angles cal technique called ANOVA.
obtained from experiment and regression equation.  It was identified that laser power of 600 W, traverse
speed of 200 mm/min and number of passes of four
Das and Biswas 1819

for sheet thickness of 3 mm are the optimal combi- 5. Shen H, Shi Y, Yao Z, et al. An analytical formula for
nations of laser line heating process parameters that estimating the bending angle by laser forming. Proc
produce an output of 5.1° as an angle of bend. IMechE, Part C: J Mechanical Engineering Science 2007;
 ANOVA statistics revealed that the number of 220(9): 993–997.
passes is the most influencing parameter in achiev- 6. Vollertsen F. An analytical model for laser bending.
ing good results followed by sheet thickness, laser Laser Eng 1994; 2: 261–276.
7. Yau CL, Chan KC and Lee WB. A new analytical model
power and traverse speed.
for laser tube bending. In: Geiger M and Vollertsen F
 A suitable co-relationship has been obtained (eds) Proceedings of the LANE laser assisted net shape
between the laser power, traverse speed, sheet engineering 2, 1997; 2: 357–366.
thickness and number of passes with bend angle 8. Mucha Z, Hoffman J, Kalita W, et al. Laser forming of
using multiple regression analysis. The expression thick free plates. In: Geiger M and Vollertsen F (eds)
validates well with the experimental results. Proceedings of the LANE laser assisted net shape engi-
 Thus, it is concluded that the procedure proposed neering 2, 1997; 2: 383–392.
in this article for optimization of laser line heating 9. Kyrsanidi AK, Kermanidis TB and Pantelakis SG. An
process parameters will significantly improve the analytical model for the prediction of distortions caused
angle of bend for mild steel sheets and plates. by the laser forming process. J Mater Process Tech 2000;
104(1–2): 94–102.
10. Shen H, Yao ZQ, Shi YJ, et al. An analytical formula for
Declaration of conflicting interests estimating the bending angle by laser forming. Proc
The author(s) declared no potential conflicts of interest IMechE, Part C: J Mechanical Engineering Science 2006;
with respect to the research, authorship and/or publica- 220(2): 243–247.
tion of this article. 11. Mucha Z. Deformations and stresses induced in materials
by moving beam of CO2 laser. Proc SPIE 2007; 6598:
65980M-1–65980M-9.
Funding 12. Shi Y, Shen H, Yao Z, et al. Temperature gradient
The author(s) received no financial support for the mechanism in laser forming of thin plates. Opt Laser
Technol 2007; 39(4): 858–863.
research, authorship and/or publication of this article.
13. Shen H, Shi YJ and Yao ZQ. Numerical simulation of
the laser forming of plates using two simultaneous scans.
References Comp Mater Sci 2006; 37(3): 239–245.
1. Zhang X, Ji Z, Liu Y, et al. The effect of line heating fac- 14. Casalino G and Ludovico AD. Parameter selection by an
tors on angular deformation generated by line heating. J artificial neural network for a laser bending process. Proc
Shanghai Jiatong Univ 2006; 40(6): 1010–1014. IMechE, Part B: J Mechanical Engineering Science 2002;
2. Li W and Lawrence Y. Numerical and experimental 216(11): 1517–1520.
investigation of convex laser forming process. J Manuf 15. Roy R. A primer on the Taguchi method. Dearborn, MI:
Process 2001; 3(2): 73–81. Society of Manufacturing Engineers, 1990.
3. Lawrence J, Schmidt MJ and Li L. The forming of mild 16. Kolahan F, Manoochehri M and Hosseini A. Applica-
steel plates with a 2.5 KW high power diode laser. Int J tion of Taguchi method and ANOVA analysis for simul-
Mach Tool Manu 2001; 41(7): 967–977. taneous optimization of machining parameters and tool
4. Shichun W and Jinsong Z. An experimental study of geometry in turning 2011. World Acad Sci Eng Technol
laser bending for sheet metals. J Mater Process Tech 2011; 74: 82–85.
2001; 110(2): 160–163.