Reliability Improvements in Repair Welding of High-strength Steels` IIW Doc IX-2002-01

Dr.sc. Ivan Samardžić*

Dr. Thomas Siewert **
* Dr.sc. Ivan Samardžić, Mechanical Engineering Faculty in Slavonski Brod (CROATIA)
** Dr. Thomas Siewert, National Institute for Standards and Technology, Boulder (Colorado,USA)

Key words: degradation, high-strength steels, reliability improvement, repair welding

Summary: Engineers often have to deal with repair of weldments after they have been put into
service. Here, the concern is that the service environment may have changed the structure,
independent of whether the problems originated during the initial construction (cold and hot
cracks, inclusions, porosity, etc.) or during subsequent service (wear, damage, stress-corrosion
cracking, etc.). In the case where the material characteristics have changed, the welding
procedures used during the original construction may no longer be appropriate. Thus, it is
necessary to create a new welding procedure qualification to define the specific details for an
appropriate repair. In this paper, emphasis is given to repair welding of railway tanks
manufactured from high-strength steels.
1. Introduction
The weld joint typically has a higher risk of failure than the rest of the structure (e.g. a weld
degradation factor “V” of less than 1 is often used in calculations) [1-6]. One reason for this is
that one or more properties of the base material near the weld could have degraded by the
process of welding, leading to a failure through a dominant failure mode (the failure that has the
greatest probability of appearing, such as stress-corrosion cracking). The failure risk might
further increase if repair welding is performed. Therefore, the welding, quality control, and
quality assurance technologies need to be developed to more stringent requirements if we want to
avoid the conditions for failure, and so achieve higher reliability for welded construction. We
are often faced with dilemmas during repair welding:
• Do the reliability requirements for the welded construction even permit repair welding?
• How many times can repair welds be performed at one place?
• Which technology should be used to produce higher reliability and less degradation at the
welding joints?
In some cases, there are standards that provide answers, or inspectors (with appropriate
authority) who can offer an opinion. In other cases there are not.
Different kinds of welding flaws can appear at repair welds, but the most serious are cracks.
They can be characterized by location, orientation, type, size, etc.
This paper deals with possible problems that could occur during repair welding of TStE 420,
a high-strength steel (HSS) that is used to manufacture pressure vessels for storage and transport
of liquefied gases. The following sections describe different aspects of the problem: why repair
welds have a greater variation in cooling rate (than production welds), why this variation in
cooling rate leads to a greater range in hardness, and why this range in hardness leads to poorer
properties in aggressive environments, such as cathodic polarization. The starting point of the
discussion (and key to the problem) is the complex temperature field that occurs during repair
welding, and which can cause significant degradation of the welded-joint zone (such as a
decrease in the reliability of the zone of the repair weld).
________________________________
`Contribution of NIST, not subject to copyright in U.S. Some tradenames are included for
information purposes without endorsement or criticism.

1
2. Degradation in properties due to the complex temperature field during welding

The thermal cycles associated with the joining of two plates by welding can degrade the
properties of the material adjacent to the weld. The degree of degradation is dependent on many
factors, and there are several reasons why repair welds may exhibit greater degradation [7-9].
For example, repair welds are usually short, so their temperature fields are more complex than
long welds (the simple 2-D temperature models apply only to semi-infinite welds). The
degradation is most severe at weld starts and stops, but can also occur at other locations along the
weld and for different stress conditions (welding of hybrid materials, uneven thicknesses, near
the plate end, etc.). The choice of the welding process also determines the temperature field, due
to the heat input into the material from the moving heat source. Accordingly, we may find a very
complex temperature field inside the material. Also, there will be transition areas where the
temperature fields may respond in a manner that is intermediate between the 2-D and 3-D results,
depending on the boundary conditions or the start conditions. A stable temperature gradient is
established after a short time for suitable dimensions around a moving heat source in a long
weld. In transition areas, the cooling rate varies, so the mechanical properties can be significantly
different from the areas with 2-D heat flows (areas that are quite distant from weld starts and
stops). Because the temperature fields in the transition regions are so complex and because the
effects are more extreme near the ends, this report will concentrate on the weld starts and stops.
The degradation in properties at weld starts and stops occurs for both long and short arc
welds. It is of greater concern in short welds, because the proportion of the weld that
experiences the 3-D heat-flow conditions is larger. This is shown schematically in Figure 1,
which presents the cooling rate variation (w, K/s) and cooling time in welded joints (over the
temperature interval from 800 ºC to 500 ºC). In the transition area of the weld joint (some area
betweeen the ends and the middle), the heat flow (and so cooling rate) is intermediate. At the
end of the weld (after a length of 2-D heat flow near the middle of the weld), the 3-D
temperature field is present again. Since the cooling time and cooling rate of the welded joint
influence the mechanical properties (such as the hardness and strength) of the welded joint, they
should be kept within required margins so that any serious degradation of the mechanical
properties in the transition areas at weld starts and weld stops can be avoided.
The length of the transition regions depends mainly on the heat flow of the heat source and
its effective energy input, but depends also on material thickness, physical properties of material,
preheating temperature, temperature between phases, etc..
In some cases, this problem can be avoided by the use of weld-on/weld-off plates that are
mounted before welding and removed after welding. However, in most repair welds, mounting
of weld-on/weld-off plates is not possible, so the transition or non-stationary (weld start and
weld stop) areas remain in the structure.

2
 1a)

w, K/s Cooling rate (w, K/s)

A... maximal hardness mean

value at weld bead start
length “a”
t 8/5 , s B... mean hardness value at
normal bead length “b”
C... maximal hardness mean
Cooling time (t , s) value at weld bead stop
8/5
length “c”
D... mean hardness value in base
Weld bead length, mm metal
E... hardness scattering of
1b)
normal bead length “b”
HV F... hardness scattering of base
metal
NSF...non-stationary (3-D) heat
flow
E C QSF.. quasi-stationary (2-D) heat
flow
A Q ... quality
a b c
F
B
D

Weld bead length, mm

Weld bead start length Normal weld bead length Weld bead stop length

Heat Flow NSF QSF NSF

Quality Q degradation Standard Q Q degradation

a b c

Total weld bead length, mm

Figure 1. Definitions - theoretical diagrams of variations of influencing factors on weld quality along the bead length
for: cooling rate w, K/s (1a), cooling time t8/5, s (1a) and hardness value HV (1b).

2.1. Determination of hardness differences in the heat affected zone (HAZ) along the welded
joint
A series of welds was produced to investigate the effect of hardness gradients on the
expected mechanical properties of the welds. The welds were produced by a number of processes
(GMAW, SAW, and surfacing) on HSS with yield strengths up to 500 MPa. After test plates

3
were welded and surfaced by controlled welding parameters, specimens were prepared for
hardness measurements (HV1, HV5, HV10), microstructural analysis, and corrosion testing. A
schematic view of the specimen preparation after welding of HSS steel TStE 420 is presented in
Figure 2. Two characteristic diagrams were drawn from the data showing the distribution of
hardness HV5 along the HAZ near the fusion line after surfacing of a plate with a thickness of 15
mm (Figures 3 and 4). The hardness was measured by the contour method [10].

10 v

Weld bead start Weld bead stop

25
WM
15 BM HAZ "Fusion line"

10
Figure 2. Schematic of specimen preparation after welding of HSS steel TStE 420.
450

HV5

400
Welding proces: Submerged Arc Welding
Base metal: HSS TStE 420, thickness 15 mm,
Preheating: To=19 ( C)
Welding speed: v=350 (mm/min)=5,8 (mm/s)
Current: I=460-480 (A)
Voltage: U=29-31 (V)
Heat input: Eef=2430 (J/mm)
wire diameter 4 (mm):
350

300

250
0 100 200 300 400

Distance from weld bead start to weld bead end, mm

Figure 3. Hardness HV5 in the heat affected zone along the fusion line from weld start to weld end. The solid line
shows the average. Welding process: submerged arc welding, base metal: TStE 420, δ=15 mm, filler metal wire
diameter: ∅4 mm.

4
 480

HV5

440

400

360

320

0 50 100 150 200

Distance from weld bead start to weld bead end, mm
Figure 4. Hardness HV5 in the heat-affected zone along the fusion line from weld start to weld end. Welding
process: manual arc welding, base metal: TStE 420, δ=15mm, filler electrode: Fox 2,5Ni, diameter: 3,25
mm, To=20 oC, I=125.3 A (σ=15.6), U=22.9 V (σ=4.76), v=3,7 mm/s. Specimen G2.

Figures 3 and 4 show an increase in hardness at the weld start and stop regions, compared to
the middle of the weld.

Higher values of hardness were always noticed at the end of the weld compared to those at the
start. A similar distribution of hardness in the HAZ along the fusion line was noticed on all
samples (30 samples), but the hardness values varied due to different welding parameters, such
as heat input. Variation in the heat input changed the cooling rate, i.e. the time to cool from 800
to 500 ºC.
Based on these investigations, the influence of cooling rate in the weld joint could be
analysed for the example of an experimental arc strike shown in Figure 5. An arc strike was
produced with an SMAW electrode (simulation of an improper procedure) on a fine-grain steel
TStE 420. Hardness measurements in the weld region showed an increased hardness over that of
the base material, which is very important because this material is used for pressure vessels that
are used in aggressive environments (subject to stress corrosion).

5
 HAZ Mark Location HV5
WM
1 HAZ 473
11
2 HAZ 466
BM 8
1 3 HAZ 473
2
5 4 HAZ 473
10 5 mm
4 5 HAZ 466
3
6
6 BM 190
7
9 7 BM 195
8 BM 185
15 mm 9 WM 391
10 WM 380
11 WM 396
Figure 5. Hardness HV5 results in HAZ, along the fusion line, in WM and BM after producing an “arc strike” in
steel TStE 420 (simulation of improper procedures).

At locations with higher hardness (weld starts and stops, arc strikes, ...), a stronger
inclination to cracking (cold cracks) and stress corrosion is expected.
2.2. Experimental investigation of degradation at weld starts and stops
We investigated the degradation of properties at weld starts and stops, for comparison to the
weld center and to the base material. This continues earlier laboratory investigations of the
sensitivity to stress corrosion of weld joints and base material made from high-strength steel
Nioval 50. Those investigations concluded that the parameters that determine the severity of
stress corrosion are: maximum hardness in the weld joint, aggressive media, and the strength
level. [11, 12, 13] In the present experiments, the base material TStE 420 was studied. The
composition of steel TStE is presented in Table 1, while the principal mechanical properties are
presented in Tables 2 and 3.
Table 1 Composition of base material [14].
Steel - TStE 420 C Si Mn P S Ni N Al V
Our measurements 0,15 0,38 1,46 0,017 0,005 0,22 0,018 0,023 0,13
Steel producer data 0,15 0,36 1,52 0,013 0,006 0,21 0,016 0,016 0,14

Table 2 Basic mechanical properties, from manufacturer's documentation [14].

Steel Yield point Rt (MPa) for different Strength Elongation, Bending,
TStE 420 plate thicknesses (mm) Rm A5, % α=180o
≤10 11-15 16-25 ≥25 ≤35 MPa Long. Trans.
Our data - 503 - - 604 A5=27 - -
Z=61,9%
Data from 420 420 410 410 530-680 17 3a 4a
the steel mill

Table 3 Basic toughness (J) [14].

Non aged
Steel Rolling Temperature, oC
Direction +20 0 -20 -40 -50
TStE 420 Longitudinal 63 55 47 31 27
Transversal 39 31 27 20 16

6
 Results of laboratory investigation of Charpy V at temperatures 20, 0, -20, and -40 ºC are as
follows. [14]
Temperature 20 oC: arithmetic mean KV=203 J (individual values: 209 J, 195 J and 205 J)
Temperature 0 oC: arithmetic mean KV=190 J (individual values: 183 J, 196 J and 193 J)
Temperature -20 oC: arithmetic mean KV=191 J (individual values: 196 J, 182 J and 196 J)
Temperature -40 oC: arithmetic mean KV=163 J (individual values: 168 J, 156 J and 167 J)

2.3 Design of experiments

The surfacing of three plates (dimensions 500 mm x 250 mm x 15 mm) was predicted
through a design of experiments. After weld surfacing (and recording the welding parameters
and preheating temperatures), three specimens were produced for testing of hydrogen
embrittlement. As shown in Figure 6, one specimen had the weld-pass start in the middle of the
test section (P), one had the weld centered in the test section (S), and one had the weld stop in the
middle of the test section (K). To show the tendency of the base material to stress corrosion,
three specimens were also made from plate that was not surfaced.
One specimen was taken out of every four groups of specimen types (base material, weld-
pass start, half of weld pass, weld-pass end) as a reference. These specimens were not exposed to
hydrogen.
500
180 16

75 P N7 K
d=15

130
250 P K
180
r 30 K
P
25
70
350
weld beads
P... weld bead start
K... weld bead stop
Figure 6. Testing plates after welding (bead on plate). Two plates with 3 passes each, and one plate without welded
beads – only base metal.

2.4 Surfacing
Surfacing was performed using an MAW basic electrode, Fox 2.5 Ni, diameter 3.25 mm. The
temperature of the plates before surfacing was 15 ºC. Welding parameters were recorded by an
on-line monitor. Average values of the surfacing parameters are listed in Table 4.
Specimens for hydrogen brittleness testing were machined from the plates. The specimens
were identified according to bead location, as listed in Table 4.

7
Table 4. Arithmetic mean for welding parameters

Plate 1 Plate 2
Bead mark 3 4 5 6 7 8
Current, A 23 21,4 22,4 22,3 22,1 21,8
Voltage, V 140 138 134 136 133 140
Bead length, li 185 182 180 175 180 180
Welding time, s 64 63 60 55 61 60
Welding speed, mm/s 2,9 2,9 3 3,2 3 3
Heat input, J/mm 1110 1022 1000 954 996 1017

2.5 Exposing the specimens to hydrogen and fracture by tensile test of the specimens.
To evaluate the tendency of the base material, weld start, weld end, and middle section of the
weld to hydrogen embrittlement, the cathodic polarization technique was used. As described in
references [15], [16], [17] and [18], the presence of atomic hydrogen in a material leads to
decreased plasticity. Due to differences in microstructure between the weld zones, different
levels of embrittlement are expected, as a function of the base material and the weld location.
Cathodic polarization was performed with a current density of 40 mA/cm2 for 24 hours, without
any stress on the specimens. The solution was 0,5 mol H2SO4 + 10 mg As2O3 per dm3, deaerated
by nitrogen for 30 minutes before cathodic polarization. Hydrogen-charged specimens were
pulled on a tensile test machine with a deformation speed of 0,24 mm/s. After testing, the IK
index was calculated by the formula:

Z − Z (H )
F= ⋅100% ....................................................................................................... (1),
Z
where:
Z ... reduction in area of uncharged specimen of base metal
(used as a reference value), and
Z(H) ... reduction in area of a hydrogen-charged specimen.

2.6 Results
The mechanical properties of the hydrogen-charged specimens are presented in Table 5.

Table 5. Mechanical properties of hydrogen charged (CP) specimens

Specimen Cathodic Re Rm A5 Z IK Remark
polarization (CP) MPa MPa % %
1 Base Material (BM) 503 604 27,0 61,9 Base metal
2 BM+CP 519 613 23,0 29,2 52,8 Base metal, hydrogen charged
3 Weld Middle (S) 493* 653* 14,4 28,5 Middle of bead
4 S+CP 531* 582* 8,8 20,5 66,9 Middle of bead, hydrogen
charged
5 Weld Stop (K) 522 608 17,7 60,1 Bead stop
6 K+CP 490 534 18,0 12,2 80,3 Bead stop, hydrogen charged
7 Weld start (P) 477 599 18,0 61,7 Bead start
8 P+CP 497 575 9,7 16,6 73,1 Bead start, hydrogen charged
Based on square specimen area.

8
 Specimens not charged with hydrogen Specimens charged with hydrogen
1 2
Weld Start Weld Stop Weld Start Weld Stop
3 4
Weld Start Weld Start
Weld Stop Weld Stop
5 6
Weld Start Weld Stop Weld Start Weld Stop
7 8

... Failure location after tensile test

Figure 7. Failure locations after tensile testing

Figure 7 shows the location of fractures in the specimens. In specimens that had been treated
with hydrogen, the fracture occured at the weld starts and stops; for the specimens not treated
with hydrogen, the fracture occured at approximately 25 mm from the weld stop, i.e. in the base
material.
Values of the brittleness index (IK) were obtained from formula 1. Comparison of the
different values of IK shows that the smallest value of IK occurs for the base material. The
middle of the weld has a lower value of IK index than the weld starts and stops. Therefore, the
middle of the weld is more resistant to hydrogen embrittlement than the weld starts or stops.
Comparison of the weld starts with the stops shows that the weld stop is more sensitive to
hydrogen.
We were also able to develop some relationships between the IK values and the maximum
hardness in the HAZ adjacent to the fusion line. Regions with higher hardness have higher
inclination to hydrogen enbrittlement, i.e., higher degradation of base material due to welding.
This investigation confirms our thesis of an embrittlement of the weld starts and stops. A
tendency to greater stress corrosion with an increase in maximum hardness values in the weld
was also noted.

3. A short review of repair welding and repair welding techniques

Welding is used to repair manufacturing errors during construction, as well as for field repair
of damaged welded structures that have been in service. However, a discussion of the concerns
of repair welds is seldom found in the literature. This problem is most serious for pressure
vessels working in aggressive media. Experience on railway tanks and spherical tanks shows that
a large number of stress corrosion cracks have occured on poor-quality repair welds [11].
Therefore, the problem needs more investigation and analysis than ever before.
A half-bead technique is usually proposed for high-strength steels. The authors' experience
suggests that this technique cannot be easily applied to weld repairs produced with SMAW
because the distance from the annealing bead to the melting line must be 1 mm to 3 mm (Figure
8, detail “A”). But, when it is applied, then it is important to implement this technique on both
the weld starts and stops. Correct implementation of this technique requires adequately trained
welders. The maximum hardness can be decreased by increasing the preheating and reheating
temperature, which could be useful for weld starts and stops. The results of experimental repair
welding depend on the way in which weld starts and stops are placed in the melting zone of a
previous weld.

9
 elektrode 2,5 mm

DETAIL A 1-3mm
1-3 mm

1 mm

DETAIL A
(without grinding)

d/2
gringing

-0,5 mm

DETAil A
(with grinding)
electrode 3,25 ili 4mm
1 mm

Figure 8. Weld bead sequences during repair welding using “half bead technique” [19].

4. Experimental repair welding

Repair welding was performed on plates of 15 mm thickness, with the parameters and
welding techniques used during the welding of 10 railway tanks of capacity 110 m3. Figure 9 is
a schematic representation of the groove preparation. The sequences of the passes are presented
in figures 10 and 11.

15 o
60+- 3
Inner side
4+ 1

Outer side
Attachment weld by the SMAW process
Figure 9. Preparation of a weld joint in production railway tanks.

10
 4
3
Inner side

1 Outer side
Grooving from outside
Figure 10. SAW from inside of tank, followed by grooving from outside

8 7
Inside

6
5

Outside
Figure 11. SAW from outside of tank, after grooving.

Table 6. Welding parameters for SAW process for each pass.

Pass Current Voltage Wire feed Preheating Heat input
I, A U, V v, mm/s temp., oC E , J/mm
1 400-460 20-26 420 80 1413
2 440-460 28-30 420 80 1780
3 440-470 27-30 420 85 1853
4 440-470 27-30 420 80 1853
5 460-490 24-27 400 60-70 1817
6 440-470 25-28 440 80 1644
7 440-480 25-29 450 100 1625
8 440-470 26-30 450 105 1699

To evaluate the repair procedure, a longitidunal crack was assumed to be present in the HAZ
along the melting line at a depth of 8 mm and for a length of 100 mm. A repair groove was cut
with a carbon electrode and a “V” groove joint was prepared by grinding (groove angle of 60 º,
radius in groove root was approximately 6 mm).
To achieve lower values of the hardness in the repair welds, the weld starts and stops were
placed in the melted zone of the previous weld.
(figure 12).

11
 Section A-A

A A

Section B-B

B B

Figure 12. Repair welding - weld pass start and weld pass stop placed in the previous weld metal.

Repair welding was performed by SMAW with a Fox 2.5 Ni electrode, 3.25 mm in diameter.
The preheating and interpass temperatures were kept between 150 and 200 ºC. The average heat
input was 1000 J/mm. After cooling of the test plate, specimens were cut from the weld start,
middle and weld end to measure the hardness (HV5). The hardnesses for the weld start and stop
regions are presented in the next figure.

SMAW passes in repair weld S M A W pas s es in repa ir w eld

244

268 232 265

265
257 268
277 251 274
260
286 280 277
269 293
274 257

232 257 210

230 262 212

Fusion line of original SAW passes Fus ion li ne o f ori gina l S A W pas s es

Figure 13. Maximum hardness value at TStE 420 steel repair weld start (a) stop (b).

These experimental investigations confirmed our expectation that the hardness would not
increase significantly. The hardness values at the weld starts and stops were not statistically
different from those for the middle of the welds.

5. Conclusion

It is important to consider the problems of differing temperature fields when approving

welding procedures, especially for the weld starts and stops. The number of weld starts and stops
should be minimized, and the remaining starts and stops should be designed and implemented in
ways that the embrittlement is minimized. To assess the overall reliability of a structure, the
properties of all weld repair regions (start, middle, and stop) must be determined. Only with such
a complete knowledge of the properties can the goal of increased reliabilty of welded structures

12
be reached. Beside general conclusions on weld starts and stops, this paper provides specific data
for the repair welding of HSS TStE 420.

References:
1. Lukačević, Z., Samardžić, I. Weldability analysis based on probabilistic approach to reliability and
weakening of welded joints. 3rd International Seminar "Numerical Analysis of Weldability", Graz-
Seggau, Austria, 25-27 September 1995.
2. Lukačević, Z. Probabilistic Weld Efficiency and Reliability Approach for Pressure Vessels. IIW Doc.
XI-407-82.
3. Lukačević, Z. Weld Efficiency, Reliability and Classes of Weld Joints (Philosophy on Weld
Efficiencies). IIW-XI-374-80.
4. Lukačević, Z. Quantification of weldability, joint efficiency and reliability. Proceedings of the
International Conference on the Joining of Materials - JOM-7. The European Institute for the Joining
of Materials. Helsing∅r, Denmark, 1995.
5. Lukačević, Z. Reliability Assessment of Welded Products. Proceedings of the International Welding
Conference. CMES. Hangzhou, China, 1984.
6. Lukačević, Z., Samardžić, I. Approach to weldments reliability modeling. The International Welding
Conference, Japan-Slovak Welding Symposium, Kozice (Slovak Republic), 5-6 March 1996.
7. Samardžić, I. Temperature fields application in weld repair technique. Proceedings of the 19 th
International Welding Congress, Bratislava, 1994, 311-321.
8. Samardžić, I. Stoić, A. Weakening due to temperature field variation during welding. The International
Welding Conference, Japan-Slovak Welding Symposium, Kozice (Slovak Republic), 5-6 March 1996.
9. Samardžić, I., Galović, A., Galović, M. Nonstationary temperature fields at welding. Croatian Welding
Journal “Zavarivanje” 36, 6, 1993, 211-217.
10. Mraz, L. Hardness measurement in underbead HAZ of single pass welds; Doc IIW IX-1609-90; 1-39.
11. Belić, A., Lukačević, Z., Samardžić, I. Accelerated tests for stress corrosion of welded joints.
Croatian Welding Journal “Zavarivanje” 36, 4-5, 1993, 155-163.
12. Gayk, W. Empfehlungen zur Werkstoffauswahl und zu den Schweissdaten fur NH3- Lagerbehalter
und NH3- Kesselwagen, Mat.-wiss u. Werkstofftech. 19, 1988, 126-132.
13. Crackwell, A. Stress corrosion cracking of steels in ammonia, The Institute of Refrigeration, London
1988.
14. Gojić, M. Malina, J. Ivaniš, K. Evaluation of hydrogen embrittlement resistance of duplex stainless
steel for the oil industry. Oil (Nafta) 43 (6) 313-321, 1992, (in Croatian).
15. Base and filler metal producer certificates. Quality control department of factory PHTTV, Đuro
Đaković d.o.o., Slavonski Brod, 1995, (in Croatian).
16. Chavane, A. Habashi, M. G. M. Pressouyre and I. Galand, Corrosion 42 1986, 54.
17. Oriani, R. A. Hirth, J. P. Smailowski, M. Hydrogen Degradation of Ferrous Metals. Noyes
Publications, New Jersey, 1985, 822-852.
18. Bernstein, I. M. Thomas, A. W. Hydrogen Effects in Metals. Publication of the Metallurgical Society of
AIME, Warrendale, 1983, 527-666.
19. Deljusa, I. Repair welding technik “half bead technik”. Croatian Welding Journal Zavarivanje 33, 1,
1990, 17-21

Acknowledgement. This publication is based on work sponsored by Croatian-American Joint

Fund in cooperation with the Ministry of Science and Technology of Croatia and the National
Institute of Standards and Technology (NIST) in the U.S.A. under project JF 137.

13