Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/343190330

Galvanizing Bath Management During Galvanize to Galvanneal and Galvanneal


to Galvanize Product Transitions

Conference Paper · April 2004

CITATIONS READS

2 672

3 authors, including:

Joseph R. McDermid Frank E Goodwin


McMaster University International Zinc Association
211 PUBLICATIONS 2,565 CITATIONS 657 PUBLICATIONS 2,616 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Frank E Goodwin on 24 July 2020.

The user has requested enhancement of the downloaded file.


Galvanizing Bath Management During Galvanize to Galvanneal and Galvanneal to
Galvanize Product Transitions

Joseph R. McDermid
Department of Mechanical Engineering, McMaster University
1280 Main St. W., Hamilton, ON, Canada L8S 4L7
tel: 905-525-9140 x27476
e-mail: mcdermid@mcmaster.ca

Éric Baril
Noranda Inc. – Product Development
2250 boul. Alfred Nobel, Suite 300,Ville St-Laurent, QC, Canada H4S 2C9
e-mail: eric.baril@montreal.norfalc.com

Frank E. Goodwin
International Lead/Zinc Research Organization
PO Box 12036, Research Triangle Park, Durham, NC, USA 27709-2036
e-mail: fgoodwin@ilzro.org

Key words: galvanizing, galvannealing, phase diagram, product transition, fluid flow modeling, bath
intermetallic transformation, bath management

INTRODUCTION

Although the continuous galvanizing process has been in existence in (more or less) its present form for more
than half a century, it is only in the last ten to fifteen years that significant progress has been made in
understanding the phenomenon occurring in the liquid zinc bath. Much of this progress is owed to concerted
efforts by the steel and zinc industries both independently and through the ILZRO consortium and academia.
These efforts have resulted in a much deeper understanding of the flow patterns in the CGL bath (both in
thermal equilibrium and the transient thermal conditions during ingot melting), a reasonably accurate
description of the Zn-rich corner of the Zn-Al-Fe system and work on the chemistry and kinetics of processes
occurring during the galvanneal (GA) to galvanize (GI) and the GI Æ GA product transitions. Further progress
on the determination of bath intermetallic precipitation zones during ingot charging is expected in the near
future as transient thermal bath flow solutions and bath chemistry thermodynamic data are coupled. Progress
on the modelling of short-term transient flows in the CGL bath is also expected.

Much of the original impetus for undertaking the above work was to enable galvanizers to understand and
control bath chemistry and dross formation and movement during the GA Æ GI and GI Æ GA product
transitions. It is on this subject that the present contribution will focus.

Galvatech '04 Conference Proceedings 855


EFFECTIVE AL AND THE ZN-RICH CORNER OF THE ZN-AL-FE PHASE DIAGRAM

A reasonably accurate description of the Zn-rich corner of the Zn-Al-Fe phase diagram is an essential step in
understanding the processes occurring during product transitions. The relevant portion of this system, at low Al
and Fe contents, is now reasonably well described due to the recent efforts of Tang [1], McDermid and
Thompson [2] and Giorgi et. al. [3]. These first two researchers have conducted independent studies of the
solubility of Fe in dilute Zn-Al baths and have come to the conclusion that the Zn-rich corner with Al contents
of less than 0.50wt% contains three intermetallic phases in order of increasing bath Al content: ζ-FeZn13, δ-
FeZn7 (also sometimes referred to as δ-FeZn10) and η-Fe2Al5ZnX. Nominal compositions for these
intermetallics are shown in Table I. It should be noted that the solubility of Fe in these alloys is quite low and
that there is significant variation of the Fe solubility for each one of the intermetallics shown in Table I.

Table I: Chemical composition of CGL intermetallics as per Figure 2 [2, 4 - 6].


wt%Al wt%Fe wt%Zn
η-Fe2Al5ZnX
37 – 46 31 – 37 18 – 25
(top dross particle)
δ-FeZn7
1.5 – 3.5 2.2 – 9.5 87 – 93
(bottom dross particle)
ζ-FeZn13 0.7 – 1.0 5.8 – 6.1 93.2

As can be seen from this Table, the intermetallic compounds are not line compounds and, particularly in the
case of the η particle, can have a wide variation of composition due to the presence of soluble constituents.
However, it is convenient to represent them by the line compound compositions for the limited compositions of
interest to galvanizers. This latter approach was taken in the above work on determining the solubility limits of
Fe and Al in the liquid baths. As per Table I, it should be noted that the η particle is the floating “top dross”
particle usually encountered in GI operation and the δ particle is the sinking “bottom dross” usually seen during
GA.

In the case of the studies undertaken by Tang [1] and McDermid and Thompson [2], general agreement was
derived on the equation for the solubility of Al and Fe for the case where η-Fe2Al5ZnX is the dominant bath
intermetallic, where Tang’s equation is given by [1]:

33066
ln [ Fe]2 [ Al ]5 = 28.1 − (1)
T

where [Fe] and [Al] are the soluble Fe and Al contents for a given Zn-Al-Fe bath in wt% and T is the bath
temperature in degrees K. McDermid and Thompson’s equation, determined from an extensive experimental
program of solubility experiments, is given by [2]:

30391 ± 3106
ln [ Fe]2 [ Al ]5 = 24.247 ± 4.221 − (2)
T

where the symbols in the equation have the same meaning as per equation (1) and the confidence limits are at
the 95% level. An examination of equations (1) and (2) will show that they agree within the experimental error

856 Galvatech '04 Conference Proceedings


of the latter study. No confidence limits on the coefficients of equation (1) were quoted for the study of Tang
[1]. The solubility limit of the δ-FeZn7 field is given by McDermid and Thompson in the equation [2]:

11746 ± 1199
ln [ Fe][ Al ]0.30 = 11.794 ± 1.556 − (3)
T

where the symbols in the equation are as per equations (1) and (2). It should be pointed out that the Fe
solubility curve for the ζ-FeZn13 region will not be defined here as operation in this phase field is not commonly
encountered in CGL operations.

The solubility curves of McDermid and Thompson [2] for the region of interest for galvanizing and
galvannealing are shown below in Figure 1 and will be used extensively throughout the subsequent discussion.
This diagram has been referred to as the “metastable” diagram due to the fact that some discussion continues as
the exact equilibrium phase configuration for this region of the Zn-Al-Fe system [2, 5, 6]. As is well known, GI
tends to operate wholly within the η+L phase field whereas GA tends to operate either within the three phase
η+δ+L phase field or the two phase δ+L phase field. It has been advocated that it is preferable to operate GA
within the three phase field to prevent or reduce the formation and accumulation of bottom dross particles [7, 8].
It should be noted, however, that less reactive grades of steel (e.g. re-P IF) may require that the GA bath be
operated in the two-phase δ+L region so that they can be galvannealed at a reasonable power input [9].

0.050

ζ η “composition”
0.045

+
0.040 δ+L
η+δ+L
L
0.035

0.030
δ “composition”

invariant “knee point”


0.025
wt%Fe

0.020 η+L

0.015 liquid

0.010

0.005 Fe/Al solubility curve

0.000

0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30

wt%Al

Figure 1: Phase diagram of the Zn-rich corner of the Zn-Al-Fe system at 460°C as per McDermid and
Thompson [2]. Please note the exaggerated scale for Fe versus Al and that the Al scale starts at 0.08wt%.

It should be pointed out that the total Al and Fe content of the galvanizing bath almost always exceeds the
solubility limits pictured in Figure 1 due to higher metastable Fe solubility in the vicinity of the strip/bath

Galvatech '04 Conference Proceedings 857


interface. Thus, δ or η intermetallic particles are almost always present to a greater or lesser extent in the CGL
bath depending on its Fe content. The presence of these particles results in the dissolved or “effective” Al
content of the bath being significantly different from the total Al commonly returned from chemical analyses of
bulk bath samples, particularly in the case of samples taken during GI operation in the η+L phase field due to
the high Al content of this intermetallic phase (Table I). These particles are also the origin of galvanizing
drosses. As effective Al is the determining parameter in the formation of the desired coating microstructure and
properties for continuously galvanized products, it is desirable to control the process based on effective Al [7, 8,
10]. For the two phase fields in the diagram, this process essentially consists of plotting the bath analysis result
on the phase diagram for the temperature of the bath at the time of sampling and drawing the appropriate tie-
line (parallel to the intermetallic “composition” lines shown in Figure 1) through the analysis point and
determining the effective Al from the intersection of this line and the appropriate solubility limit curve [10].
For the three phase regions in Figure 1, the Gibbs phase rule dictates that the effective Al is invariant for a given
temperature and the effective Al is given by the intersection of the solubility curves and intermetallic
composition lines. In the case of the δ+η+L phase field, the effective Al is given by the “knee point” shown in
Figure 1. Further details of this calculation for each of the relevant phase fields can be found elsewhere [10].
However, use of multiple solubility curves is somewhat impractical and, fortunately, computer tools to
determine the effective Al and bath intermetallic speciation in the CGL bath for the range of bath chemistries
commonly encountered in galvanizing and galvannealing have been developed (i.e. Noranda’s NEAC™ and
Teck-Cominco’s DEAL™). These programs use the total bath Al, total Fe obtained from bath samples and bath
temperatures as inputs, returning the bath effective Al.

A further examination of the phase diagram in Figure 1 will show that, in fact, the GA Æ GI and GI Æ GA
product transitions involves both a change in bath effective Al and the bath intermetallic compound in
equilibrium with the surrounding bath. In the case of the former transition, the equilibrium intermetallic
changes from either δ-FeZn7 to η-Fe2Al5ZnX (or a mixture of δ and η to entirely η) whereas the latter transition
will result in either the partial or complete conversion of η to δ. The implications of this conversion of bath
intermetallics will be discussed in further detail below.

MATERIAL FLOW IN THE GALVANIZING BATH

Extensive work has been carried out on the development of flow patterns in the continuous galvanizing bath
under both steady-state and transient thermal conditions, the main efforts on which have been undertaken by
ILZRO [11-16], Noranda [17-21] with contributions from other authors [22-24]. The overall objective of this
work has been to control the formation of dross particles in the CGL bath and manage their movement such that
they have little or no impact on the quality of the galvanized product. These efforts have shown that strong
steady-state flow patterns develop in the galvanizing bath, affected mainly by the moving strip, hardware
geometry and (to a lesser extent) flow from the output jet of the heating inductors. Typically, the magnitude of
the flow velocities is a strong function of strip width and strip velocity or line speed. It has also been shown
that the mixing of Al into the pot is quite uniform in the steady-state case [11, 12, 18, 25], with the transient
mixing behaviour of the pot being a combination of plug flow and continuously stirred reactors [12]. This last
result was based on earlier water modelling work [18], which showed that the composition of the bath becomes
quite uniform some 20 min. following complete melting of the ingot.

More recent works have concentrated on the development of non-isothermal flow patterns during the melting of
a typical CGG ingot at the back of the bath [13-15, 21]. These authors have shown that the buoyancy effects
from the cooler (and denser) melting ingot, coupled with the full-fire flow from the inductors used to maintain
the temperature of the bath, have a significant effect on the overall flow patterns in the CGL bath. These
models have shown that the plume of cooler, higher Al zinc descends to the bottom of the pot, where it is forced

858 Galvatech '04 Conference Proceedings


to the front of the bath via the flow imposed by the moving strip and returns to the back of the bath via return
flow along the side walls of the pot. In turn, the hotter. lower density Zn from the inductors is forced to the top
of the bath, where it is recirculated into the bulk by cooling and coupled flow from the descending ingot plume.
The interaction of these flows results in strong recirculation flows being developed at the back of the bath [13-
15, 21]. This general flow pattern of descending cool, high Al alloy was also found for the melting and
dispersion of submerged 5wt% brightener bars [23].

These latter flow patterns have profound implications for the control of dross during both steady-state and
product transition operation of the CGL. As can be seen from Figure 1, the Fe solubility in Zn-Al-Fe baths
decreases sharply as the bath Al increases in the η+L phase field. As the most commonly used Al addition
materials have Al contents significantly higher (i.e. 0.50wt% and greater) than the bulk bath Al concentrations
commonly used in CGL operations, the local Fe solubility is significantly exceeded in the descending Zn plume
from the melting ingot, resulting in a zone in the bath in which the solubility limit is exceeded and dross
particles can nucleate and grow [16, 22-24]. This results in significant localized dross formation, the particles
of which are subsequently dispersed throughout the bath by the general flow. Dross can be controlled by
minimising these Al gradients through modified feeding practices [22, 24] or by the insertion of baffles in the
bath [21] such that Al gradients in the portion of the bath through which the strip passes are minimised. A fuller
summation of all of the implications of bath flows for dross control are given elsewhere [26]. In terms of the
GA Æ GI product transition, the descending Al-rich plume also functions to transport Al to the accumulated
bottom dross, providing additional driving force for the δ Æ η intermetallic conversion reaction [27].

GENERAL PRINICPLES OF CGL BATH CHEMISTRY CONTROL

CGL bath chemistry control is a subject that has received some significant attention over the last decade and
work on the subject is available in the open literature for the interested reader [7, 8, 10, 27-30]. A fundamental
principle of all bath chemistry control approaches is that they are based upon the control of effective Al as
defined by the phase diagram, as per Figure 1, although there are some differences between authors on the
recommended methodologies. Some authors advocate a model-based control method based on the results of
fundamental studies derived from a mixture of laboratory and industrial data [7, 8, 30] whereas other authors
advocate an approach based on custom empirical models tailored for individual CGLs [10, 27-29], of which the
latter approach has been successfully demonstrated on industrial baths in the open literature. All of the
approaches employed essentially use a mass balance on effective Al and Zn in which Al and Zn inputs from
CGG ingots or brightener bars are balanced against the Al and Zn leaving the CGL bath in the coating and via
the dross phase. The latter two quantities are usually calculated based on models built using the preferred
approach. The difference in the mass of target bath effective Al versus the actual mass of effective Al (obtained
from bath samples) is computed and the appropriate bath addition recommended [7, 8, 10, 27-30]. The concept
of the mass balance approach to bath chemistry control will be used in the subsequent discussions. However, it
should be pointed out that these models are usually constructed under steady-state GI or GA operation, whereas
the transient bath chemistry conditions during the GA Æ GI and GI Æ GA pose unique challenges to the
application of mass balance models for CGL bath chemistry control.

BATH CONTROL DURING GA Æ GI PRODUCT TRANSITIONS

For most single pot lines, the GA to GI transition consists of adding a large amount of Al to the galvanizing
bath in a single or closely spaced number of additions such that the Al content of the bath is increased quickly.
This addition usually takes the form of high Al containing materials to minimise the impact on the bath Zn
level, either high Al jumbos or brightener bars containing 5 to 10wt% Al. This large “spike” addition takes the

Galvatech '04 Conference Proceedings 859


bath composition from the δ+L or δ+η+L phase fields into the η+L phase field without significantly changing
the bath Fe content, as illustrated schematically in Figure 2. Typically, most galvanizers raise their GI bath Al
considerably above the knee point to “clean” the bath of dross particles. As can be seen from Figure 2, the δ
intermetallic is no longer thermodynamically stable in the CGL bath at these higher bath Al concentrations and
it undergoes a transformation to the η-phase according to the reaction:

2 δ-FeZn7 + 5 Al Æ η-Fe2Al5ZnX + (7 – X) Zn (4)

As can be seen from equation (4) the transformation of the bath intermetallics consumes Al, the sole source of
which is the bath. Given that significant quantities of δ phase can accumulate in the bath as bottom dross
during the preceding GA campaign, bottom dross particle conversion can consume large quantities of bath Al
and will tend to decrease the bath effective Al below target levels unless properly compensated for [10, 27],
presenting a challenge for bath control.

0.050

0.045
A B C
0.040

0.035
Al addition gap –
0.030
total vs. effective
wt%Fe

0.025

0.020
additional dross
0.015

0.010

0.005

0.000
0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30

wt%Al
Figure 2: Schematic representation of the GA Æ GI transition process using the Zn-Al-Fe phase diagram [2].
The transition starts at point A and proceeds to point B or C. In this case, the target bath chemistry is
0.20wt%Al. The difference between points B and C is the use of total Al (B) or effective Al (C) as the
measurement of bath Al. In this case, calculations based on total Al would take the bath composition to point B
whereas calculations based on effective Al would take the bath composition to point C.

Figure 2 also illustrates a further challenge to bath chemistry control during the GA Æ GI transition in that the
Fe/Al solubility at the target GI composition (points B or C) is frequently significantly lower than that at the GA
composition (point A), resulting in two effects: (i) additional bath Al is consumed in the precipitation of η-
Fe2Al5ZnX dross particles [10, 30] and (ii) the effective Al is very different from the total Al in the bath sample,
as illustrated for points B and C in Figure 2. The only solution for the former problem would be to increase the
GI bath Al content to slightly greater than the knee point, but many galvanizers to not want to do this as they
wish to operate with the cleaner baths seen at higher Al contents [31]. The solution for the latter is to use only
effective Al for bath control, as illustrated in Figure 2. In this case, a total Al reading of 0.20wt% for point B
will result in an effective Al of 0.167wt%Al, caused by the high bath Fe of 0.040wt%. Correcting this by using

860 Galvatech '04 Conference Proceedings


effective Al in the mass balance calculations will result in the amount of Al used in the transition addition being
raised to bring the bath to a total Al of 0.24wt% to give an effective Al of 0.20wt%, as per point C in Figure 2.

There is some debate as to the exact mechanism of the δ-FeZn7 Æ η-Fe2Al5ZnX transformation, with some
authors claiming that the η nucleates and grows on the δ surface while the δ particle dissolves into the melt,
eventually leaving a free η particle [32, 33] whereas other authors claim that the δ particles dissolve with new η
particles nucleating and growing form the bulk bath [34, 35]. The growth mechanism of the η particles then
proceeds via Ostwald ripening controlled by the diffusion of Al and Fe from the bulk bath [36]. Regardless of
the exact mechanism, there is general agreement that the δ Æ η transformation is quite rapid [32, 34, 35] with
the end result being fine η particles which slowly grow in the high Al bath [36, 37] and gradually rise to the
surface to be skimmed. It should be pointed out, however, that the time taken for the η particles to rise to the
bath surface is a strong function of the particle size [26] (i.e. 1×10-7m/s for 5µm particles to 16×10-4m/s for
200µm particles, assuming Stoke’s Law behaviour) and can be significant affected by the higher flow velocities
commonly encountered in the bulk bath. For example, it has been shown by Liu et. al. [33] that top dross
particles during a GA Æ GI transition were approximately 5µm in diameter for the first 3 hours after the
transition. These particle then gradually coarsen to larger particles of approximately 20µm to 60µm as the
transition proceeded to steady-state operation [4, 26, 33]. Thus, there is usually a significant delay between the
transition addition and the skimming off of the converted top dross particles due to the particles being “trapped”
in the general flow field, which can vary from 10 to 24 hours [10, 27, 33]. It has been generally perceived that
these dross particles subsequently impact the strip, resulting in so-called “dross pick-up” defects. However,
recent studies have shown that some “dross” related defects are, in fact, due to oxide entrapment on the coating
surface or substrate defects [33, 38].

An additional factor contributing to the difficulty of bath Al control during GA Æ GI transitions is the use of
10wt% Al brightener bars for the high Al addition material. Unlike lower Al content materials where the
melting plume descends into the bath [13-15, 19, 23], this material floats on the surface of the CGL bath.
Dispersion of the contained Al into the bath is via surface flow and diffusion, which tends to be rather slow [7,
27]. Furthermore, this Al is subject to oxidation and the local formation of dross particles [7, 27] resulting in
significant losses to products not “useful” to the transition process, namely increasing the bath effective Al and
converting the δ intermetallic to η. There is general agreement that the efficiency of 10%Al brightener bars is
quite low, with Liu et. al. [30] maintaining an efficiency of approximately 70% whereas McDermid and Dewey
[27] support an efficiency ranging from 50 to 75% based on a mass balance of the CGL bath. By performing
GA Æ GI transitions on the same line as the above 10%Al brightener bar work, however, McDermid and
Dewey [27] found that the efficiency of 5%Al brightener bars was close to 100% and that the time for the
transition could be reduced by 33% versus using 10%Al brightener bars. These differences were attributed to
the fact that 5%Al brightener bars (and their melting liquid) sink in the CGL bath [23, 27], avoiding oxidation
of the contained Al and delivering it quickly to the zone of δ intermetallic conversion at the bottom of the bath,
increasing the local Al level and speeding up the conversion kinetics by providing larger local driving force for
the transformation reaction. Furthermore, the bottom of the bath is a zone of higher Zn flow velocity where the
Al will be more quickly incorporated into the bulk bath, particularly if the descending Zn plume is present due
to the melting of a CGG ingot at the back of the bath, as was the case for these tests [27]. This concept has been
successfully extended to the use of 4.5%Al CGG jumbos for GA Æ GI transitions [39]. Thus, it is strongly
recommended that materials used for the spike addition contain 5wt%Al or slightly less.

As mentioned previously, all of the above factors tend to conspire to make CGL bath control very difficult for
the period of the transition, arbitrarily defined as the time required for the dross rate to return to its nominal GI
value (n.b. the dross rate in this case is defined as the mass of dross produced per unit area of sheet coated, as
determined by the mass of a dross ingot divided by the strip area throughput during the time the dross ingot was

Galvatech '04 Conference Proceedings 861


produced) [27]. This problem is illustrated in Figure 3, which shows one of a series of GA Æ GI transitions
carried out at ProTec’s CGL#1 using 5wt%Al brightener bars for the Al “spike” addition with subsequent
feeding of the bath being solely by the addition of 0.45wt%Al CGG jumbo ingots [27]. It should be noted in
Figure 3 that “mfeffAl” indicates the mass fraction of effective Al in the bath. This latter material is used in
steady-state GI operation in a fixed ratio with SHG jumbos to maintain the target bath Al level. Variations in
dross rate are also shown in the Figure and it can be seen that a large spike is seen in the dross rate at
approximately 20 hours following the original addition, due to the converted bottom dross particles reaching
that top of the bath to be skimmed as discussed previously. It should be noted that the continuous symbols in
the diagram are predicted effective Al contents computed by a mass balance model, corrected appropriately for
bath samples for cases where the prediction fell outside the ±5% analytical error.

GA-GI Transition test - mfeffAl prediction corrected for bath samples after bbar additions, with dross rate vs. time

0.002400 0.004000
100 bbar addition 90 bbar addition
400 bbar addition

0.002200 0.003500

0.002000 0.003000

dross rate (lb/ft^2)


mfAleff

0.001800 0.002500

0.001600 0.002000

nominal dross rate


0.001400 0.001500

0.001200 0.001000
10-11-99 18:00 11-11-99 0:00 11-11-99 6:00 11-11-99 12:00 11-11-99 18:00 12-11-99 0:00 12-11-99 6:00 12-11-99 12:00
time (dd-mm-yy, hh:mm)

sample mfAleff predicted mfeffAl dross rate effAl target = 0.20%

Figure 3: Predicted and actual effective Al for a GA Æ GI transition using 5wt% Al brightener bars, including
variations in the dross rate during the transition. Note that the effective Al target is 0.20wt% [27]. As per
Figure 3, mfeffAl indicates the mass fraction of effective Al in the bath.

A noteworthy feature of this test is that a constant feed of 0.45%Al jumbos was not able to maintain the bath Al
level at the target level, indicating that the combined Al demands of the strip and intermetallic conversion were
higher than the Al this material could supply. This shortcoming was compensated for by several undesired
brightener bar additions to raise the effective Al level to the target of 0.20wt%Al [27]. Subsequent mass
balance model calculations based on the results of this and other tests indicated that the Al demand of the bath
necessitated the feeding of 1wt%Al jumbos until the desired effective Al target was attained, followed by a
mixture of 1%Al and SHG ingots to maintain the bath effective Al until the transition was over after which the
normal GI feed practice could resume. This practice was implemented, with the results shown in Figure 4 [27]
showing that the target GI bath Al level was reached and maintained soon after the transition spike addition,
eliminating the need for additional brightener bar additions [27]. These examples serve to illustrate that bath Al

862 Galvatech '04 Conference Proceedings


control during GA Æ GI transitions must be altered to compensate for the additional Al demands of bottom
dross conversion and additional short-term dross production through modified feeding practices. It should also
be noted that the Al demands of the bath can be reduced somewhat by producing high coating weight products
and lowering the incoming strip superheat during the transition period as these lower the %Al in the coatings
[28].

Figure 4: An example of a GA Æ GI transition carried out using the modified 1wt%Al CGG ingot addition
practice, showing that the target Al content of 0.20wt% was achieved early in the process and maintained [27].

BATH CONTROL DURING GI Æ GA PRODUCT TRANSITIONS

The GI Æ GA transition essentially consists feeding only SHG ingots to the bath such that the bath Al decreases
via removal of Al from the bath by the strip and dross production from the GI production composition in the
η+L field to the desired GA bath composition in either the three phase region or the δ+L phase field. This
process is, therefore, a gradual one. Upon reaching the desired GA composition, any remaining η intermetallics
will transform into δ phase (or the appropriate mixture of δ and η phases if the bath composition is in the three
phase field) via the interfacial/diffusion transformation mechanism advanced by Gauthier et. al. [34, 35]. The
principle difficulty encountered with respect to bath chemistry control in this transition stems from the reaction
in which the η phase intermetallics decrease in size by dissolution into the bath as the bath Al decreases [37],
given by the reaction:

η-Fe2Al5ZnX Æ 2[Fe] + 5[Al] + XZn (5)

where the [] brackets indicate the element in the liquid Zn solution. This dissolution reaction is a basic
consequence of the bath soluble Fe increasing with decreasing Al content, as per Figure 1, and Le Châtelier’s
Principle. This reaction tends to increase the bath Al and can delay the attainment of the desired GA
composition. This can result in scheduling difficulties if not managed properly as timing the attainment of the
GA bath composition to match the production schedule for the start of GA production is the bath management
goal in this case. Some simple remedies for increasing the rate of Al output via the strip, such as light coating

Galvatech '04 Conference Proceedings 863


weights, use of wide strips and higher strip entry temperatures, can be used but these solutions will not always
achieve the desired timing.

The GI Æ GA transition process can be represented on the Zn-Fe-Al phase diagram as pictured in Figure 5. An
example of this process in an industrial bath is plotted on the Zn-Al-Fe diagram in Figure 6 [40]. The transition
process in this case essentially consists of the η-Fe2Al5ZnX intermetallics dissolving into the bulk bath until the
bath composition is very close to the Fe/Al solubility curve at the bath temperature, after which further
decreases in the bath Al content result in the bath composition following the solubility curve until the desired
GA start composition is reached. At this point, the feeding of Al containing ingots (usually in ratio with SHG
ingots) is re-started and the bath Fe content tends to climb slightly above the solubility curve as per normal
steady-state GA operation.

0.050

0.045

0.040
B A’
0.035

0.030 dissolution of additional


Fe2Al5Znx
wt%Fe

0.025
A
0.020
dissolution of Fe2Al5Znx
0.015

0.010

0.005

0.000
0.08 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30

wt%Al
Figure 5: Schematic representation of the GI Æ GA transition process on the Zn-Al-Fe phase diagram [2]. The
transition starts at point A or A’ and proceeds to point B.

The bath management difficulty posed by the dissolution of the η-Fe2Al5ZnX intermetallic by equation (5) is
illustrated in Figure 5 via the points A and A’, which represent different starting total Al/Fe for the transition.
In this case, points A and A’ lie on the same tieline and have the same bath effective Al. However, point A’,
due to its higher total Al/Fe content, has a larger mass of η phase intermetallics will dissolve back into the bath
in the initial stages of the transition. The greater mass of Al contained in the dross particles versus the case of
having point A as the starting composition will tend to retard the attainment of the GA composition. It should
be pointed out that basing the GI Æ GA transition computations on the bath starting effective Al content (which
is the normal quantity used to control bath chemistry, as stated above) will lead to errors as, for example, both
points A and A’ have the same effective Al but represent baths that contain differing amounts of η-Fe2Al5ZnX
intermetallics, which will give up differing masses Al to the bath by following the process illustrated in Figure
5. Fortunately, this difficulty can be overcome by some simple modifications to the steady-state GI mass
balance.

864 Galvatech '04 Conference Proceedings


GA to GI transtion process - sol. curves at 460C
0.04
η+δ+L

0.035 δ+L

0.03

0.025
wt% Fe

0.02 η+L

0.015

0.01

0.005

0
0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22 0.24
wt% Al
total Al/Fe data eff Al/Fe data

Figure 6: Total and effective Al/Fe bath compositions from example of a GI Æ GA transition carried out in an
industrial bath [40]. Effective Al computations on this data were carried out using the NEAC™ program.

This simple modification essentially consists of starting the GI mass balance model for the GI Æ GA transition
from the total Al/Fe bath composition rather than from the effective Al composition in order to take the
additional Al from the intermetallic dissolution into account. However, it must be cautioned that the starting
composition must be selected with care due to the possibility of having a given bath sample excessively
contaminated by intermetallics. It is strongly recommended, therefore, that multiple analyses be taken from
pre-transition bath samples in order to establish a consistent and reasonable starting point for the transition
computations. It is also be recommended that the end-point computations and timing be based on the strip area
throughput of the bath rather than time as it is the former that truly controls the Al depletion of the bath.
Ideally, the strip area input information can be provided by the production schedule. Should this not be
possible, the transition computations can be carried out using reasonable estimates of the average strip width
and other production parameters which can then be iterated over strip length increments to estimate the total
required strip area throughput until the transition end-point is reached.

A set of curves for predicting the GI Æ GA transition end-point using the latter methodology is shown in Figure
7. This plot shows the mass fraction of bath Al (mfbathAl) content as a function of strip area throughput for a
variety of starting bath total Al contents. It should be noted that due the chemistry of the transition process as
illustrated in Figure 5, by the end of the transition process the bath total and effective Al contents are likely very
close to each other and, therefore, they are considered to be the same for the case of this calculation. The line at
0.0014Al (0.14wt%) in Figure 7 represents the end point of the transition process in this particular case [40].
These curves were constructed using the steady-state GI mass balance model constructed for this particular
CGL [40]. This methodology has been successfully implemented at this line via a simple spreadsheet-based
calculation tool.

Galvatech '04 Conference Proceedings 865


GI-GA transition practice - effAl vs Cumulative Strip Area from 0.16 - 0.20% effective Al starting composition

0.00205

0.00195

0.00185

0.00175
mf bath Al

0.00165

0.00155

0.00145

0.00135
0.00E+00 5.00E+05 1.00E+06 1.50E+06 2.00E+06 2.50E+06 3.00E+06 3.50E+06
Cumulative Strip Surface Area Throughput (one-side ft^2)

effAl = 0.16% effAl = 0.17% effAl = 0.18% effAl = 0.19% effAl = 0.20%

Figure 7: Application of a GI mass balance model to predicting the strip area throughput required to reach the
GI Æ GA transition end-point of 0.14wt%Al from a given starting total Al. Note that “average” production
parameters were used in this computation with the mass balance being updated at strip area increments [40].

CONCLUSIONS

The current state of knowledge of some bath management tools from the open literature and industrial practice
applicable to the management of the CGL bath during GA Æ GI and GI Æ GA product transitions was
summarised. Specifically, these tools are: (i) the Zn-rich corner of the phase diagram, (ii) numerical and
physical modeling of bath flows and, in particular, the effect of the non-isothermal transient flows during CGG
ingot melting and their impact on the transient Al distribution in the bath, (iii) the current state of knowledge of
the kinetics and transformation mechanisms of CGL bath intermetallics and (iv) the use of tools such as mass
balances to be able to predict bath effective Al levels as a function of CGL production parameters.

For the GA Æ GI product transition, it was shown that the main bath management difficulty in meeting and
maintaining the desired GI bath effective Al target lies in the additional Al required for the transformation of the
GA bath δ-FeZn7 intermetallic to the GI bath η-Fe2Al5ZnX intermetallic and the additional formation of η-
Fe2Al5ZnX intermetallics in the GI bath due to the lower Fe solubility in this region of the phase diagram versus
the GA bath composition. It was shown that higher Al feeds can be used in combination with modifications to
the GI mass balance model to meet and maintain the desired bath effective Al during the transition period. It
was further recommended that the transition material used for the Al “spike” addition contain approximately
5wt%Al so that this material will sink in the CGL bath, avoiding Al losses through oxidation and excessive
dross generation. This methodology for managing the GA Æ GI transition has been successfully demonstrated
on the industrial scale.

866 Galvatech '04 Conference Proceedings


For the GIÆGA transformation, it was shown that the principle bath management challenge was to account for
the additional Al input to the bath from the dissolution of η-Fe2Al5ZnX intermetallics during the running down
of the GI bath. This dissolution process tends to delay the achievement of the desired GA bath composition and
can result in production scheduling difficulties. It was shown that this difficulty can be overcome by starting
the normal GI mass balance model from a carefully determined bath total Al composition and computing the
transition end-point on the basis of the strip surface area throughput. This methodology for managing the GI Æ
GA transition has also been successfully demonstrated on the industrial scale.

ACKNOWLEDGEMENTS

Much of the industrial work contributing to the data contained in this paper was carried out while J. McDermid
was a Senior Scientist at the Noranda Inc. – Technology Centre and the authors would like to thank Noranda
Inc. for their permission to publish this work. J. McDermid would also like to thank Stelco Inc. and the
National Science and Engineering Research Council of Canada (NSERC) for support of his present activities
through their funding of the Stelco/NSERC Industrial Research Chair in Steel Product Application.

REFERENCES

[1] N-Y Tang, “Determination of Liquid-Phase Boundaries in the Zn-Fe-Mx Systems,” J. Ph. Equil., Vol. 21,
No. 1, 2000, pp. 70 – 77.
[2] J.R. McDermid and W.T. Thompson, “Fe Solubility in the Zn-rich Corner of the Zn-Al-Fe System for Use
in Continuous Galvanizing,” 44th Mechanical Working and Steel Processing Conference Proceedings, vol.
XL, Orlando, USA, Iron and Steel Society, 2002, pp. 805 – 813.
[3] M-L Giorgi, J-B Guillot and R. Nicolle, “Assessment of the Zinc-Aluminium-Iron Phase Diagram in the
Zinc-Rich Corner,” Calphad, Vol. 25, No. 3, 2001, pp. 461 – 474.
[4] F. Ajersch, L. Trépanier and F.E. Goodwin, “Particle Size and Composition of Dross Particles from
Galvanize and Galvanneal Operations,” 44th Mechanical Working and Steel Processing Conference
Proceedings, vol. XL, Orlando, USA, Iron and Steel Society, 2002, pp. 771 – 780.
[5] P. Perrot, J-C. Tissier and J-Y. Dauphin, “Stable and Metastable Equilibria in the Fe-Al-Zn System at
450°C,” Z. Metallkd, Vol. 83, No. 11, 1992, pp. 786 – 790.
[6] N-Y Tang and X. Su, “On the Ternary Phase in the Zinc-Rich Corner of the Zn-Fe-Al System at
Temperatures below 450°C,” Met. Mat. Trans. A, Vol. 33A, 2002, pp. 1559 – 1561.
[7] N-Y Tang, M. Dubois and F.E. Goodwin, “Progress in Development of Galvanizing Bath Management
Tools,” Proc. 4th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech ’98), Chiba, Japan, Iron
and Steel Institute of Japan, 1998, pp. 76 – 83.
[8] N-Y Tang, “Characteristics of Continuous-Galvanizing Baths,” Met. Mat. Trans. B, Vol. 30B, 1999, pp.
144 - 148.
[9] Private communication with J.R. McDermid, 2002.
[10] J.R. McDermid, “Hot Dip Galvanizing Bath Chemistry and Control,” 21st Century Steelmaking Program:
Metallic Coated Sheet Production, American Iron and Steel Institute and the Iron and Steel Society,
Indianapolis, USA, June 2000.
[11] F. Ajersch et. al., “Validation Studies of the Numerical Simulation of Flow in the Bethlehem Steel, Burns
Harbour Galvanizing Bath,” Proc. 4th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech
’98), Chiba, Japan, Iron and Steel Institute of Japan, 1998, pp. 642 – 647.
[12] C. Binet and F. Ajersch, “Application of Mixing Models to Evaluate the Transport of Aluminum in a
Continuous Galvanizing Bath,” Proc. 4th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech
’98), Chiba, Japan, Iron and Steel Institute of Japan, 1998, pp. 648 – 653.

Galvatech '04 Conference Proceedings 867


[13] F. Ajersch et. al., “Numerical Analysis of the Effect of Operating Parameters on Flow in a Continuous
Galvanizing Bath,” Proc. 5th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech 2001),
Brussels, Belgium, Verlag-Stahleisen, Düsseldorf, Germany, 2001, pp. 511 – 518.
[14] F. Ajersch et. al., “Numerical Analysis of the Effect of Temperature Variation on Flow in a Continuous
Galvanizing Bath,” 44th Mechanical Working and Steel Processing Conference Proceedings, vol. XL,
Orlando, Florida, Iron and Steel Society, 2002, pp. 863 – 873.
[15] F. Ajersch, F. Ilinca and J-F. Hétu, “Simulation of Flow in a Continuous Galvanizing Bath: Part I.
Thermal Effects of Ingot Addition,” to be published in Met. Mat. Trans. B, Vol. 34B, 2003.
[16] F. Ajersch, F. Ilinca and J-F. Hétu, “Simulation of Flow in a Continuous Galvanizing Bath: Part II.
Transient Aluminum Distribution Resulting from Ingot Addition,” to be published in Met. Mat. Trans. B,
Vol. 34B, 2003.
[17] M. Gagné, A. Paré and F. Ajersch, “Water Modelling of a Continuous Galvanizing Bath,” Proc. 84th
Galvanizer’s Association Meeting, Pittsburgh, USA, Galvanizer’s Association, 1992, pp. 147 – 163.
[18] M. Gagné and F. Ajersch, “Galvanizing Bath Water Model Tests be Monitoring pH Changes from
Localized Additions,” Proc. 3rd Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech ’95),
Chicago, USA, Iron and Steel Society, 1995, pp. 687 – 694.
[19] M. Gagné and M. Gaug, ‘Numerical Modeling of Fluid Flow Patterns in a Continuous Galvanizing Bath,”
Proc. 4th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech ’98), Chiba, Japan, Iron and
Steel Institute of Japan, 1998, pp. 90 – 95.
[20] E. Baril, M. Dubois and M. Gagné, “Investigation of Fluid Flow in the Snout of a Continuous Galvanising
Bath Using Numerical Modelling,”. Proc. 5th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet
(Galvatech 2001), Brussels, Belgium, Verlag-Stahleisen, Düsseldorf, Germany, 2001, pp. 435 –442.
[21] J.R. McDermid and B. Maag, “Numerical Modelling of Ingot Charging Configurations at Pro-Tec CGL2”,
Proc. 94th Galvanizer’s Association Meeting, Dearborn, USA, Galvanizer’s Association, 2002.
[22] K. Otsuka, M. Arai and S. Kasai, “Development of Dross Control Methods in a Continuous Galvanizing
Pot by Numerical Bath Flow Analyses,” Proc. 4th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet
(Galvatech ’98), Chiba, Japan, Iron and Steel Institute of Japan, 1998, pp. 96 – 101.
[23] K.J. Evans and C.J. Treadgold, “Modelling and Measurement of Transient Conditions in the Galvanizing
Pot,” Proc. 91st Galvanizer’s Association Meeting, Jackson, USA, Galvanizer’s Association, 1999, pp.
131 – 151.
[24] R.M.M. Mallens et. al., “Prevention of Dross Contamination in the Corus Ijmuiden Hot Dip Galvanizing
Line,” Proc. 5th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech 2001), Brussels,
Belgium, Verlag-Stahleisen, Düsseldorf, Germany, 2001, pp. 255 – 261.
[25] C.R. Shastry and J.J. Galka, “Analysis of Zinc Melt for Aluminum, Iron and Dross Intermetallics,” Proc.
90th Galvanizer’s Association Meeting, Indianapolis, USA, Galvanizer’s Association, 1998, pp. 1 – 23.
[26] F.E. Goodwin, “An Integrated View of Galvanizing Bath Flow and Dross Management,” Proc. 93rd
Galvanizer’s Association Meeting, Portland, USA, Galvanizer’s Association, 2001.
[27] J.R. McDermid and C.E. Dewey, “Optimizing the GA to GI Transition at ProTec CGL#1,” Proc. of the
92nd Galvanizer’s Association Meeting, Toronto, Canada, Galvanizer’s Association, 2000.
[28] N-Y. Tang, “Demystifying CGL Bath Chemistry Management,” Proc. of the 92nd Galvanizer’s
Association Meeting, Toronto, Canada, Galvanizer’s Association, 2000.
[29] T. Le and M. Gagné, “Material Balance in the Zinc Bath at Dofasco No. 4 CGL,” Proc. 5th Int. Conf. on
Zinc and Zinc Alloy Coated Steel Sheet (Galvatech 2001), Brussels, Belgium, Verlag-Stahleisen,
Düsseldorf, Germany, 2001, pp. 533 – 539.
[30] Y.H. Liu and N-Y. Tang, “Automating and Optimizing Bath Management Using the Computer Program
PAL,” Proc. 94th Galvanizer’s Association Meeting, Dearborn, USA, Galvanizer’s Association, 2002.
[31] J. Faderl, W. Maschek and J. Strutzenberger, “Spangle Size and Aluminum Pick-up for Hot-Dip Zinc
Coatings” Proc. 3rd Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech ’95), Chicago, USA,
Iron and Steel Society, 1995, pp. 675 – 685.

868 Galvatech '04 Conference Proceedings


[32] S. Yamaguchi, “Thermochemical Stability and Precipitation Behaviour of Dross Phases in CGL Bath,”
Proc. 4th Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech ’98), Chiba, Japan, Iron and
Steel Institute of Japan, 1998, pp. 84 – 89.
[33] Y.H. Liu et. al., “Dross Formation and Control During Transitions from Galvannealing to Galvanizing,”
44th Mechanical Working and Steel Processing Conference Proceedings, vol. XL, Orlando, USA, Iron and
Steel Society, 2002, pp. 781 – 790.
[34] M. Gauthier, F. Ajersch and J.R. McDermid, “Phase Transformation Mechanisms of Intermetallic
Particles Suspended in Hot-Dip Galvanizing and Galvannealing Baths,” Proc. 5th Int. Conf. on Zinc and
Zinc Alloy Coated Steel Sheet (Galvatech 2001), Brussels, Belgium, Verlag-Stahleisen, Düsseldorf,
Germany, 2001, pp. 352 – 358.
[35] M. Gauthier and F. Ajersch, “Analytical Modeling of the Phase Transformation Kinetics f Intermetallic
Particles Suspended in Hot-Dip Galvanizing and Galvannealing Baths,” 44th Mechanical Working and
Steel Processing Conference Proceedings, Vol. XL, Orlando, USA, Iron and Steel Society, 2002, pp. 791
– 804.
[36] T. Arioka et. al., “Growth Mechanism of Dross Particles in Molten Zn Baths” Proc. 5th Int. Conf. on Zinc
and Zinc Alloy Coated Steel Sheet (Galvatech 2001), Brussels, Belgium, Verlag-Stahleisen, Düsseldorf,
Germany, 2001, pp. 393 – 400.
[37] R.J. Barnhurst, S. Bélisle and M. Gagné, “Intermetallic Particle Formation in Continuous Galvanizing
Baths,” Proc. 4th Int. Conf. on Zinc Coated Steel Sheet, Paris, France, European General Galvanizers
Association, Putney, London, U.K., paper SE5.
[38] N-Y Tang and F.E. Goodwin, “A Study of Defects in Galvanized Coatings,” Proc. 5th Int. Conf. on Zinc
and Zinc Alloy Coated Steel Sheet (Galvatech 2001), Brussels, Belgium, Verlag-Stahleisen, Düsseldorf,
Germany, 2001, pp. 49 – 55.
[39] J.R. McDermid, internal Noranda project report (2002).
[40] J.R. McDermid, “CGL Bath Management During GA Æ GI and GI Æ GA Transitions”, presentation to
the AISI Metallic Coated Steel Technical Committee Meeting, Peru, OH, USA, February 27,2003.

Galvatech '04 Conference Proceedings 869

View publication stats

You might also like