AIR KNIFE COATING IN HOT-DIP GALVANIZING

A.N. Hrymak, J.F. Forbes*, J. Knight, E. Elsadaawy, A. Balthazaar and Y. Sheng

Department of Chemical Engineering

McMaster University
Hamilton, Ontario, Canada L8S 4L7
[email protected]

*University of Alberta
Department of Chemical and Materials Engineering
536 Chemical and Materials Engineering Building
Edmonton, Alberta, Canada T6G 2G6

Key Words: hot-dip galvanizing, air knife, jet sripping

INTRODUCTION

We have undertaken a study of the air knife used in hot-dip galvanizing strip. The focus is to develop and validate a
model to predict average coating thickness. Computational studies have included: development of impingement
pressure and shear stress profiles, sheet end effects relative to knife position, and effects of gap profiling in the air
knife. Pressure profile and wall shear stress results have been incorporated into a coating thickness prediction, based
on the previous work of Ellen and Tu (1994), by using simplifying correlations that take into account plenum
pressure, knife to strip distances, gap and line speed. Results show good prediction with available data over a wide
range of product coating weights, using coil-average coating weights from industrial data. Sensitivity studies show
that certain operating regions may be preferable for certain coating weights, considering factors such as the air knife
pressure and line speed.

MODELING

The air knife jet influences the coating thickness through the wall pressure and shear stress distributions on the
strip. A precise determination of the wall pressure and shear stress distribution is key to air knife coating
analysis. An experimental and computational fluid dynamics investigation was conducted into the effect of
knife to jet distance, pressure, and air knife gap. Pressure measurements with a transducer and local velocity
measurements with a hot-wire anemometer were made. Fluent TM a finite volume code from Fluent Inc. has
been used with standard k-e-model parameters fitted to match pressure measurements, to calculate velocity and
shear stress components at the wall.

The CFD and experimental pressure data match the literature results well However the shear stress CFD
values are not in agreement with the literature, specifically Beltaos (1976). Predictions of shear stress at the

44 th MWSP Conference Proceedings, Vol. XL, 2002 853

wall are critical in obtaining realistic predictions of the coating thickness. The earlier work of Beltaos (1976)
and Ellen and Tu (1994) use a simplified wall shear stress model, which is updated here using the CFD
predictions and simplified correlations.

• " ~ trip X

Figure 1 Schematic of region near air knife

In the coating process, before solidification, it is assumed that the variations in temperature and properties such
as dynamic viscosity, density, and thermal conductivity are small enough to be neglected. The liquid zinc flow
is considered steady, laminar, isothermal, incompressible flow. Assuming negligible surface tension and slowly
varying coating thickness the Navier Stokes equation for a thin film on a plate reduces to:

,U dy 2 - jog + ~ -0 (1)

Boundary conditions are:

u=V at y = 0 (no slip)

du/dy = x at y = t (2)

where u is the velocity in the zinc film, p the pressure due to the air knife (see Figure 2), and • the shear stress
due to the impinging jet. The solution to (1) based on boundary conditions (2) is;

(3)

where:

f
T - t,/pg non-dimensional film thickness (4a)
VvV

854 44 th MWSP Conference Proceedings, Vol. XL, 2002

 S m non-dimensional shear stress (4b)
q;.vg
1@
G-I+-~~ effective gravitational acceleration (4c)
log d x

The molten zinc flow flux, q is;

q - dy- Vt 1 (5)
o 2 3

Introducing the non-dimensional withdrawal flux, Q (Ellen & Tu, 1984)

(6)
Q-V ¢zV

Gives

GT 3 ST 2
Q - - ~ + ~ + T (7)
3 2

Qmax can be determined by solving dQ/Dt = 0 and be a function of S & G at coordinate x.

dQ
= GT 2 - ST- 1- 0
dt

1
S --~ V S 2 - 4 G
V- (8)
2G

Therefore for a given line speed the maximum non-dimensional withdrawal flux Qmax, will be a function of
dp/dx and • at the value of x. Based on mass continuity, the physically available Q will be the minimum of the
values of Qmax corresponding to every x value. Once Qmin is found it can be substituted back into (7) and the
coating thickness T can be calculated for each x point beyond the Qmin. The final non-dimensional liquid
coating thickness is the steady state T value calculated at x/D > 20, where the T values becomes virtually
constant. Substitution o f T into (4a) provides the actual coating thickness t.

T final
t- ~ (9)

pV

44 th M W S P Conference Proceedings, Vol. XL, 2002 855

 1.0

0.9
• . [ , Z/D=4Exp
0.8 I I / " Z/D=12Exp
ii l ] . Z/D=20Exp
0.7 | = I ,,,..--~D = 4 C F D
I ~ D = 12 C F D
0.6 "l i _re.Z/_ D -_20 CFD_
Sg
l ii
~0.5

~" 0.4

0.3

0.2

0.1
•

0.0
-10.0 -8.0 -6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 10.0

xlD

Figure 2" Pressure Profiles" Fluent and Experimental results match well

Figure 3 illustrates the overall results for predicted versus measured values for over 3000 coil samples. As the
figure clearly indicates, the model seems to adequately represent the coating mechanism at measured coating
weights less than 60 g/m 2. The ability of the model to predict the coating weight becomes poor as the measured
coating weight increases. This may be due to a number of reasons: less accurate results for the higher coating
weight families or model inadequacy due to the neglect of inertial force in the film.

f
/
f

J
/,,/

/
J
#*

,I j

Figure 3 Comparison of model to operating data coil averages

856 44 th MWSP Conference Proceedings, Vol. XL, 2002

 S E N S I T I V I T Y STUDIES

The objectives of the sensitivity studies were to determine regions of the operating space (defined in terms of
line speed, pressure, and Z/D) that may prove beneficial for improving galvanizing line operation. Figure 4
shows coating weight contour plots for one line speed where the contours represent curves of constant coating
weight. Note that these plots show generally that coating weight increases as knife-to-strip distance increases
and as knife pressure decreases. Note also that the slope of the coating weight surface increases as knife
position is increased and knife pressure is decreased, simultaneously (i.e., as we move toward the upper left-
hand comer of the figure).

30

25

Z/D20
15

10

1.5 2 2.5 3.5 4.5

P

Figure 4: Coating Weight Contour Maps for Line Speed of 1.6 m/s,
coating weights in g/m 2, pressures in psig.

The conclusion that can be drawn from Figure 4 is that a high sensitivity coating weight region occurs at very
low knife pressures and high knife-to-strip distances. This may explain the large variation observed in very high
coating weight products. A flat surface indicates relative ease of control (i.e, it is not particularly difficult to
stay on the operating surface). A very steep surface is more difficult to maintain a position on, since small
changes would result in large deviations in coating weight. The sensitivity studies indicate that a number of
issues must be considered when choosing the operating region for production of a specific coated product.
Three of the most important issues are: transmission of variance (e.g., pass-line variation), accuracy of control
(or controllability), and sensitivity to uncertainty in the measurements (e.g., minimizing the impact of
uncertainty in the knife position measurements or knife plenum pressure).

Selecting an operating region to minimize the effects of knife position measurement uncertainty follows the
same line of argument as that for minimizing the effects of knife position variation. An operating region should
be selected that is comparatively "flat" with respect to Z/D. A good choice would be very small knife positions
(i.e., as close to the strip as possible) and medium pressures. The difficulty with this operating strategy is that if
the knife position is uncertain, proximity to the strip would be undesirable. Based on the above argument, a
favourable operating regime should be chosen such that the knife positions are as close as tolerable (without the
risk of running the knives into the strip) and as high a pressure as possible. Line speeds are often dictated by the
processing time required in the annealing furnace for the specific product,

44 th MWSP Conference Proceedings, Vol. XL, 2002 857

Figures 5 and 6 show the relationship between knife pressure and line speed at two different Z/D ratios. Note
that both show linear curves of constant coating weight. The general shift of these curves is towards the upper
left-hand comer of the plot as Z/D is decreased. Also, the slope of these curves appears to be a function of Z/D
(i.e., the slope seems to increase as Z/D decreases).

Z/D=6

'~ 1.5

1
!

°51 1.5 2 2.5 a 3.5 4 4.5

P0 (psi)

Figure 5" Line Speed versus Pressure (Z/D = 6)

Z/D : 12
2.5 ............................ ..................
...... ~ ~.

1
i

0.5
1 1.5 2 2.5 3 3.5 4 4.5
P0 (psi)

Figure 6: Line Speed versus Pressure (Z/D = 12)

858 44 th M W S P Conference Proceedings, Vol. XL, 2002

To illustrate the relationship that the model predicts between coating weight and knife position, two sets of plots
were produced (Figures 7 and 8). They all show a similar trend" nearly linear behaviour in the Z/D range of 7 to
11 and a increasing slope beyond Z/D of 12.This seems to agree with observations at a number of galvanizing
lines.

V = 1.5 m/s, P = 2psi V = 1.5m/s, P = 4 p s i

114 68! ........

112 /
66~ //
110 ' #
,..108
¢.-
1~64i

~1o6
~-1o4
8
°102 O 60

100
I 58
98
i
96
12 14 16
56L6 8 10 12 14 16
6 8 10
Z/D Z/D

Figure 7: Coating Weight versus Z/D (V=l.5 m/s)

V = 2.5 rn/s, P = 2psi V = 2.5 m/s, P = 4psi

165 100

160 I-
/
94
~55
c~ 92
.=_ e--
J
~
-.~

150 r- 90 i-
O

88
145
86

140L .......... x_ . . . . . . . . . . .

6 8 10 12 14 16 6 8 10 12 14 16
Z/D ~D

Figure 8" Coating Weight versus Z/D (V=2.5 m/s)

SUMMARY

The coating weight model of Ellen and Tu has been modified using correlations for pressure and shear stress
profiles due to the impinging jet calculated by computational fluid dynamics over a range of parameters of
industrial interest. Coating weight predictions up to 100 g/m 2 are within 15% of the experimental coil coating
weight average. Current work is focussing on pressure and shear stress correlations appropriate for knife-to-
strip versus gap ratios of 6 or less. The model has been used to provide sensitivity information which can be
used to determine appropriate operating regions for a given coating weight.

44 th MWSP Conference Proceedings, Vol. XL, 2002 859

 ACKNOWLEDGEMENTS

The authors thank Dofasco Inc. and the International Lead and Zinc Organization for support of this work. In
particular, we would like to acknowledge the contributions of Mr. Norm Forrester of Dofasco and Dr. Frank
Goodwin of ILZRO.

REFERENCES

Beltaos, S.," Oblique Impingement of Plane Turbulent Jets", Journal of the Hydraulics Division, Sept., 1976,
pg. 1177-1193

Ellen, C. H., Tu, C. V., "Letters to the Editor", AIChE Journal, Vol. 30, No. 5, 1984, pg. 879

Ellen, C. H. ,Tu, C. V., "An Analysis of Jet Stripping of Liquid Coatings", Journal of Fluids Engineering, Vol.
106, 1994, pg. 399-404

APPENDIX

Pressure and Shear Stress Profiles

P - Pmax ec~p2 ~p= x/bp, c=-0.693

- -1

max 22254/Cz /2+99747(Cz / Z

- 0.924/Po

The half width of the pressure profile, or, the value of x where P=Pmax/2

bp - - 0.0019 C z-~- +0.0551C z - + 0 . 4 0 3 5 D

D

1.782Z"max~: r
z- - -1.2 < ~ < 1.2
e(O.75G)2

1.023r
T : max
°.4
~'c > 1.2

1.023Zmax
z" - ~ < - 1.2
[ ~ r l 0"4

--x/b~

860 44 th MWSP Conference Proceedings, Vol. XL, 2002

 (
rmax -- --0.2328
~ ~)~
C z
~ ~)1
+0.1216 C~ - 0.0017/Po

The peak shear stress location is estimated using the following equation

b~ - 0.003 C z + 0.0511C~ ~ + 0.4142 D

D

44 th M W S P Conference Proceedings, Vol. XL, 2002 861