An experimental and numerical study on the modelling of ﬂ uid ﬂ ow, heat transfer and solidi ﬁ cation in a copper continuous strip casting process

. Anexperimentaland “ numericalstudy( ﬁ rstpartofthisstudystudy)wascarriedouttoinvestigatethe solidi ﬁ cationprocessinacoppercontinuousstripcastingprocess. Themodelhasbeentunedbyexperimentalresults (i.e. cooling water ﬂ ow measurements, temperature measurements and metallographic analysis). Further, the resultshavebeenusedtostudythepossibilityofimprovedproductivity. Inthisreport(secondofthestudy)the ﬂ ow patternofthemoltencopperduringastripcastingprocessasamanufacturingmethodhasbeenstudiedusingafull-scalewatermodel.Thedynamicsimilaritybetweenmodelandrealsystemhasbeenstudied.Sixdifferenttypesof inletsystemtothemouldhavebeenstudied:inletnozzlejetswithfreestream,submergednozzlejets,slot-submerged inlet system, semi slot-submerged inlet system, submerged-slot inlet nozzle jets and ﬁ nally submerged-slot inlet nozzle jets with jet killer. Moreover, the effects of nozzle angle, nozzle diameter, casting speed, tundish adjustment and misalignment of the inlet nozzle jets on the ﬂ ow pattern have been investigated. The vortex formationandbubbleentrainment,dependinguponthenozzlecon ﬁ guration,immersiondepthandthe ﬂ uidlevelin the mouldhave also been studied.It was foundthat the slot-submerged inlet nozzle jets with jet killer arrangement showed an obvious improvement of the ﬂ uid ﬂ ow characteristics, yielding better tracer distribution in the ﬂ ow pattern, lower values of back mixing ﬂ ow, lower turbulence and lower vortex and recirculation ﬂ ow.


Introduction
The velocity distribution of molten copper contained within the solidifying shell of a continuous casting machine is very influential on the distribution of inclusion particles, which is important to the internal cleanliness and quality of the copper that are mentioned by Szekely and Yadoya [1].Moreover, Robertson et al. [2] and Thomas et al. [3] are shown that the higher casting speed leads to higher meniscus turbulence and vortex formation resulting in entrainment of slag in to the liquid.
In addition, Gupta and Lahiri [4] and Chiang [5] are shown that the flow pattern has a great influence on heat transfer to the shell during the critical initial stages of solidification.
In the other study, Tsai and Green [6] show that tundish working as a buffer and distributor of liquid metal between the ladle and mould, plays a key role in affecting the performance of the caster, solidification process productivity and also quality.Also, Sheng et al. [7] are indicated that the liquid flow and the temperature distribution are the basic factors governing operation of the tundish process.
With the growth of continuous casting, a large number of studies have been reported about the fluid flow inside both the mould and tundish.Both theoretical and experimental works have been made to assess the flow pattern inside the mould and tundish.Matsushita et al. [8] studied the free surface fluctuation in the actual system to understand the relevance between the surface wave motion of molten steel near the mould wall during casting and mould oscillation.They observed the meniscus of the molten steel directly through a quartz glass window mounted on the mould wall.They concluded that the meniscus is not stationary but fluctuates as those of mould oscillation.Later, Andrzejewski et al. [9] studied a full-scale water model to find out the flow pattern, liquid-velocity profile, and gas injection inside the continuous steel casting mould.Based on their flow pattern studies, they recommended an optimum operating condition in terms of casting rate, immersion depth and gas injection rate.However, little attention has been paid to the disturbance at the meniscus.
There have already been significant efforts in the field of mathematical modeling of transport processes in the mould.He and Sahai [10] and Mahmoudi and Vynnycky [11] have considered steady-state flow conditions and forced convection system.Hence, the momentum equations and the equations for convection of thermal energy were solved separately.These mathematical modelling results provide useful information for understanding and improving the casting process.It has been shown by Mahmoudi and Sivesson [12] that numerical simulations can accurately predict the solidification pattern of a casting, provided that proper boundary conditions for inlets and the mould surfaces are included in the model.
The focus of this study is the development of a coupled flow model to quantitatively describe the strip casting process which is used in casting copper base alloys.Experiments have been made to study the heat flow and the solidification process in the strip caster.Results from these measurements have been used in numerical simulation for heat flow and solidification.The model is used to study the solidification behaviour under various operating conditions and with different casting velocities.The simulation results give important information for caster design optimisation.The results obtained can also be used for the prediction of the behavior and characteristics of the cast material.
In order to develop accurate and efficient modelling tool for this project, three different objectives have been considered: -Water-model study which will ultimately be applied to predict and understand the effect of different design variables on the fluid flow in the mould.The result will be used to develop a mathematical model of the flow pattern in the liquid pool, and thereby to determine how molten copper is distributed through the nozzles.-To develop a mathematical model of the flow pattern and temperature field within the molten copper.The resulting heat flux to the solidifying shell will also be studied.
-Eventually, this model will represent the development of a comprehensive mathematical model for fluid flow, heat transfer and solidification within a continuous strip casting mould.The model will ultimately be applied to predict and understand the effects of different variables on the solidification process.
The purpose of the present work is to study the first objective of this project.The second and third steps will be discussed in later publications.

Strip casting as manufacturing method
Most of the articles defined sustainable manufacturing as a 'creation' or 'production' of product and services.The majority of these papers used the definition proposed by U.S. Department of Commerce.Moldavska and Welo [13] reported that the U.S. Department of Commerce (DOC) defines sustainable manufacturing as "the creation of manufactured products using processes that minimize negative environmental impacts, conserve energy and natural resources, are safe for employees, communities, and consumers, and are economically sound".
As described by Haapala et al. [14], opportunities to improve the environmental quality of casting processes exist in the development of sand mould and permanent mould coatings, binders, and lost foam materials, as well as improved thermal management and process-based models to support environmental assessment.
Increased use of heat recovering technologies can decrease energy in addition to greenhouse gas footprints of casting.Improved methods of casting are investigated by Gunasegaram et al. [15] and Tharumarajah [16] which leads towards net-shape casting could make possible a reduction or even elimination of machining or finishing steps downstream in production.
Strip casting is a type of Near-Net-Shape Casting (NNSC) technique.NNSC generally refers to the casting processes in which the products of the cast possess dimensions which are close to the final products, and therefore post-casting, size-reduction, and finishing steps can be minimized.Strip casting, in particular, is defined as the casting of molten metal into metal strips (thickness in the order of several mm) in a single processing step, which may subsequently be hot-rolled, in-line (if required), into thin-strips of desired product thickness.Since the process of strip casting merges casting and rolling into a single process, Ge et al. [17] show that the steps of slab re-heating and repeated hot-rolling which are required in conventional hot-strip production can be removed.This can provide many economic, environmental as well as technical benefits.
For example, Wechsler [18] has reported that in the industrial-scale CASTRIP operation in Crawfordsville, the overall energy consumption, which contains the energy requirement for a 50% size reduction in hot-rolling of the TRC process were 81-89% lower, in comparison with conventionally produced steel strips.Greenhouse gas emissions were also reduced by 71-80%.
Clearly, the control of flow patterns, both in tundish and mould, plays a significant role in having a better product quality.Therefore, in this paper, a study has been implemented on this discussion during a sustainable manufacturing method to provide sustainability in addition to new and effective results, because sustainability affects many aspects, and providing it, even in a small amount, has tremendous effects.

Complexity of interaction phenomena
There are several methods for studying phase transformations during a solidification process.Quenching during unidirectional solidification is often very useful since it gives good information about the complete transformation process.However, a major drawback is that the experiments have to be compared with previously not quenched solidified cast materials.
-Dynamic motion of the free liquid surfaces and interfaces, including the effects of surface tension, withdrawal program and gravity.-Transport of solute, inclusion bubbles and porosity.
-Heat transport through the solidifying shell, which contains growing air gap.-Thermal shrinkage of the solidifying shell.
-Stress generation within the solidifying shell.
-Segregation, on both microscopic and macroscopic scales.
Because of this complexity, no model can include all of the phenomena at once.An essential aspect of successful model development is the selection of the key phenomena of interest to a particular modelling objective and making of reasonable assumptions.Mechanistic models are based on satisfying the laws of conservation of mass and energy in an appropriate domain with appropriate boundary conditions.The equations are discretized using finite volume methods and are solved numerically with computers, which are becoming increasingly fast and affordable.
This report will focus on the optimal casting practise in the present machine set-up.It will also include a discussion of the priority of developments in the process to reach the target of improved productivity.
The numerical model will ultimately be applied to predict and understand the effects of different variables on the casting process to gain: -Improved casting practise (Standard Operating Procedure = SOP).-Increased productivity.

Objectives and part goals
The objective can be divided in the following part goals: -Investigation of the effect of the adjustment in height of the tundish on the flow pattern.-Investigation of the effect of the position misplacement of the tundish on the flow pattern.-Investigation of the effect of liquid level in the tundish at the casting temperature on the flow pattern.-Investigation of the effect of the misalignment of the inlet nozzle jets on the flow pattern.-Investigation of the effect of the dimension of the inlet nozzle jet on the flow pattern.-Investigation of the effect of the plug in of the inlet nozzle jet(s) on the flow pattern.-Investigation of formation of the air/gas bubbles.
-Trying to measure the fluid/particle velocity in the liquid part of the water-model.-Investigation of the effect of number of the inlet nozzle jets tundish on the flow pattern.-Investigation of the effect of casting speed/inlet velocity on the flow pattern.

Experimental procedure
Experiments have been made at Outokumpu Copper R&D Västerås, Sweden, to study the heat flow during the solidification process for pure copper.Strips are cast in a vertical oscillating mould.The temperature in the tundish was measured at the beginning of each experiment.An inconel-sheathed thermocouple of type "K" with a diameter of 1.5 mm was inserted in three different positions of the liquid pool width and allowed to follow the solidifying strip downwards in the caster.A number of experiments were performed for different tundish nozzle jets.
In order to investigate the heat flow through the mould wall during the solidification process, the temperature in the graphite mould wall was measured using nine thermocouples placed 2 mm from the mould surface, and compared with three additional measurements obtained from three thermocouples placed 4 mm from the mould surface.Temperature measurements in the outlet cooling water were also performed.Measurements were made on both sides of the mould.
The thermocouples were connected to a data acquisition system that allowed the collection of temperature-time data with an interval of 10 milliseconds.The data were plotted as a cooling curve.
The amount of cooling water and average heat released through the mould was also measured during the experiments.
The water model experimental set-up was specially constructed for studying the fluid flow in the liquid zone during the solidification process for the vertical strip caster used by Outokumpu Copper.Experiments are conducted using a transparent glass water model with a full-scale of mould and tundish of the system aiming to investigate the effects of the design variable of the inlet nozzle jets and operating parameter in adjusting the tundish/nozzles on the flow pattern.A series of trial with different nozzle type, size, and configuration has been undertaken in the water model.Different adjustment configuration of the tundish has also been studied.Due to A general view of the full-scale water model consists of the mould, the tundish, the storage tank, and the tubing with pump and various fittings.The mould of 300 mm Â 27 mm size is made of glass so that the direct flow observation is possible.The water flow through the model is impressed by two volumetric flow meter acting on the mould reservoir.A compensating storage tank is placed between the pump and the tundish.The maximum flow rate of water through the model is 21 lit/min.The filling heights in the tundish and in the mould remain on the adjusted levels during all time of the experimental run.
In order to simulate the solidification front and gain a better understanding of the fluid flow in the liquid zone during the solidification process, a V-shaped transparent plastic plexiglass wall with lots of very small holes at the bottom, has been inserted inside the water model.Using a pump, water was recirculated from the reservoir to the mould through a tundish and 18 inlet nozzle jets.All the experiments were carried out in the casting speed range of 0.9-2.1 m/min.The flow pattern is visualized by mixing of color trace, KMnO 4 into the flow.The inlet flow rate was measured by a volumetric measuring device, Vision 2008, with a sensor material of Grilamid TR55 and accuracy of ±1.5% with about 1000 plus/lit.The variation of local flow rate was recorded through a data collecting equipment connected to the computer.The experiments were recorded using a TV camera from the front side of the mould.The experimental apparatus is shown in Figure 1.Important parameters and the necessary data for the water model experiments are given in Tables 1 and 2. Note that only hydrodynamic, and not thermal, aspects will be discussed in this work.

Results
Flow visualization experiments were performed on a fullscale physical "water caster" under a variety of conditions.This "water caster" is a transparent glass/plastic presentation of an actual strip caster used at Outokumpu copper, Finspång.The result shows how the jets spread outward until it impinges against the inclined "V-shape" solidification front.It then spreads in all directions, resulting in a 3-D flow pattern locally.However, the bulk of the flow splits upward to the meniscus to flow back along the free surface to the nozzles.The result also shows the intense turbulence and swirling of the jets.
The programme of systematic experiments was performed in order to trace the parameters influencing the flow field and the free surface fluctuations in the mould.These parameters of practical interest are the casting speed, nozzle configuration, the factors affecting the immersion depth of the entry nozzle, and the rate of air/gas bubbles formed during a casting process.The investigation on the water model was performed with an original 18 free stream entry nozzles (trial P1T1) with a diameter of 5.2 mm and V cast = 1.0-1.5 m/min.Descriptions of different trials are given in Table 3.

General view
Depending on the operational parameters, very different types of flow patterns can be observed in the upper part of  the mould volume.Trial No. 1 (P1T1) consists of a typical example of the meniscus profile and fluid flow pattern for a shallow inflow of the nozzle spout in to the bulk liquid.Note that the profile is not always symmetric on either side of the nozzle, but the time-average value could be considered as a symmetric pattern.Along the narrow sidewall, the average meniscus profile remains horizontal with rather slight convex elevation from the normal level.This elevation increases with the increase in casting speed.Normally, surface disturbances are dependent on density, r, viscosity, m, and surface tension, s.Since the flow is turbulent, viscous forces are not important, and inertial force and gravity dominate the flow.

Effect of inlet system 3.2.1 Effect of nozzle (size, angle and position)
In order to simulate the casting process in the water model, the fluid is assumed to leak out through the V-shaped transparent plexiglass wall at the same rate as it is replenished by means of N jets placed at the bottom of the tundish.This means that solid is extracted from the solidification interface at the real caster.The jets each have a diameter of d jets , and hence an area of A jets ¼ pd 2 jets =4, and inlet speed of P1T1-P2T3 clearly show that across the meniscus, there is a crest and a trough around each nozzle depending on the inlet velocity and free stream depth of the liquid.No prominent wave formation occurs near the nozzle; instead, oscillating broken waves are found in that region.The operational parameters are given in Table 2.The results in trial No. 2 and 3 (P1T2-3) show a strong clockwise recirculation loop in the right part of the flotation of incoming stream.Moreover, local recirculation zone can be observed in vicinity of each inlet nozzle.It is clear that the flow pattern of trial No. 2 prevents freezing at the meniscus but promotes slag entrapment.It becomes apparent from the P1T1-3 that an intensive surface stream is directed from the nozzle to the narrow side of the mould.Here, the flow changes direction downward forming a far-reaching, clockwise rotating roll structure.
The result also indicate that the surface fluctuations are rather modest across the upper part of the mould under the operational conditions when any type of submerged system has been used; while for the free stream condition, vortices and intensive free surface fluctuations can be observed.
Free surface fluctuations of molten copper in the mould region of the continuous strip caster are considered to have a significant effect on the surface quality of the cast.Stabilization of surface quality in continuous casting is of great importance for the improvement product qualities.
Also, intensive turbulence develops by the abrupt change of the flow direction for the nozzle inlet system in comparison with the slot case.This will be discussed in details in the second part of this work.
P1T2 and P2T1 shows the effect of plug in the inlet nozzle jets on the flow pattern.The flow pattern can strengthen the mixing in the bulk; a statement that can be classified is that the "dead zone" at the upper right corner of the mould obviously can be observed when the most outside nozzles are plugged in.Also, a big recirculation loop occupies almost all of the right-hand part of the water model when the most left-hand side nozzle jets are plugged in.
Comparing the results presented in P1T2-3 and P2T1-3 indicate that the flow pattern, especially near the free surface, changes substantially when the entry nozzle is immersed deeply into the liquid according to the experimental run in submerged inlet system experiments.It becomes apparent from the trials that heavy surface fluctuations are developed on the free surface of the meniscus.
Trial No. 5 shows that slight misalignment of the inlet nozzle jets causes asymmetric flow from opposite sides of the nozzle.It should also be noted, according to these results, a great deal of variability exists in the physical model, particularly for the free stream case.
In general term, the flow field for nozzle case in all types, show that the inlet stream, (and in particular, inclined flow stream for misalignment) splits in one clockwise and one counterclockwise flow stream.However, this double-roll flow structure is directed to the entry nozzles.

Effect of slot (semi and full size)
The incoming speed of molten copper to the mould throughout the slot can be taken as: The results in P2T4 and P3T4 show a strong clockwise recirculation loop in the right part of the domain due to the flotation of the incoming stream.This will affect in preventing of the freezing at the meniscus and promoting slag entrapment.However, no local recirculation can be found in this case, which in passing to real casting process will have a positive effect on the solidification process.The result also indicates that, similar to the submerged system, the surface fluctuations are rather modest across the upper part of the mould in comparison with the free stream, when vortices and intensive surface fluctuations can be observed.
As explained earlier, it can be found that the turbulent mixing is very intense near the inlet region due to high inlet velocities for the nozzle (submerged or free stream) case comparing with the slot inlet system.This will be discussed in details in the second part of this work.

Effect of casting speed
Trial No. 2 (P1T2) together with trials P1T3 and P1T5 show a comparison of three different "inlet velocity/casting speed" in the mould.A big clockwise recirculation loop can be found in all cases.Note that the size of the loop and the immersion depth under the inlet nozzle are stronger for the higher casting speed.However, the nature of the general recirculation pattern in the upper mould is influenced very little by an increase in casting speed, although higher U c results in an increase in the length of the recirculation zone.This will of course affect the depth of liquid pool and the solidification shell thickness.Note also that the meniscus profile keeps on fluctuating at any casting speed.
Furthermore, stronger turbulent flow can be observed and also calculated for the higher casting speed.This can be explained by the Reynolds number (Re), given by: where U ch and L ch are characteristic velocity and length scales.With U ch = U c and L ch = L, Re is of the order of 10 4 (using the data of Tab.1), indicating that the flow will most likely be turbulent, specially at the higher level of the water model.This is indeed consistent with the experimental observation in trials P1T2, 3, 5, and P2T2.It is also interesting to note that observation of the water model showed a significant increase in surface turbulence with increased casting speed, all the more so for opened inlet nozzle jets (P1T2).

Air/Gas bubbles
It comes apparent from the trials P1T2 and 3 that beaver surface fluctuations are stimulated in the surface of the liquid when the U c are increased.It should additionally be noted that significant air bubbles appeared, especially at the upper part of the flow.Also, it can be seen that more air bubbles are formed when the length of the free stream is increased or the misalignment nozzles are used.However, the number of air bubbles in the water model experiments can be adjusted, and often eliminated, by altering the level of the meniscus.An increase in the water level inside the mould entails a decrease in the percentage ratio (fraction) of air bubbles.
The observation of the water model indirectly shows a significant increase in bubble generation with decreasing submergence depth (P1T2, 5).The model predicts that if the jets remain entirely below the surface, less amount of formed air/gas bubble would be expected.
The bubble entrapment takes place in the trough region and it occurs in two ways.Fluid coming from the narrow sidewall drags a layer of the bubbles at the trough.This is the primary mechanism of the bubble entrainment.Second, once a deep trough is formed, the fluid flowing from the nozzle side also entraps air bubbles at the trough.
Passing to continuous casting this means that the more sensitive copper shell in the meniscus is overfloated in an irregular manner.It is likely that the choppy flow favours near-surface slag inclusions in the strand.Note also that the buoyant action of the dispersed gas/air bubbles lifts the inlet stream more to the surface of the liquid.With respect to gas bubble, it can be stated that the rising velocity of gas bubbles is not very different in water and copper liquid; the gas hold should be nearly the same provided that the bubble size spectrum is identical.Hence, the lifting effect of the gas on the mould is similar in both systems.

Effect of tundish misplacement
When the tundish is adjusted to a higher level as in the experimental run P1T4, the disintegrated flow pattern develops.P1T4 shows the displacement of the crest and trough across the meniscus due to misplacement of the tundish, resulting in an asymmetric flow pattern along the narrow sidewall, while P1T5 shows an irregular manner overfloated meniscus level due to misplacement of the tundish.In both cases, deviation of displacements and random meniscus fluctuations along the mould width can be observed.
According to these results, it should be noted (P1T.4 and 5) that a great deal of variability exists in the flow pattern of the physical water model; amongst other factors, this could be due to slight misplacement of the tundish, causing asymmetrical flow on opposite sides of the channel centreline, as well as turbulent stochastic fluctuations.

Effect of cooling water flow on the casting temperatures
The In/Out cooling water flow, In/Out temperatures of the cooling water and casting temperatures have been measured on both strands during casting.The data has been used to estimate the cooling power.
Figure 2a shows a typical (cooling water) flow and temperature as a function of time, measured directly during a casting process.The evaluated heat flux variation and in/out cooling water temperature difference is also given in Figure 2a for both strands.A typical measured casting temperature as a function of time is also given in Figure 2b.
Unfortunately due to size of the files (127 MB) we cannot reproduce them here.
It was found that the average heat flux, flow rate and temperature values for the north strand were higher than those for the south one.Very strong variations/fluctuations were noticed in all measurements in spite of assuming to have fairly constant casting speed (so far we could not measure the casting speed, we are going to measure it in a near future): -The total mean heat flux can vary between 140 and 350 kW.-The flow rates vary between 1.2 to 2.2 l/s.
-The temperature difference vary between 289 and 313 K.
-The casting temperatures fluctuates between 1253 and 1313 K.
Microstructure analysis has been performed-during the casting campaigns-to find out the correlations between casting operating parameters and quality of produced materials.Different samples with different extrusion qualities (grade A-D) have been prepared for metallographic observation.The structures have been analysed before and after extrusion operations.
Figure 8 shows microstructure of extruded rod.The microstructure consists of two different phases.The a phase, containing around 39% Zn, has been shown as light area, when b phase with almost 46% Zn as dark region.The structure often shows segregated b phase with shape of round bands (Fig. 8b).
The position of the metallurgical centre has been found by analysing the macrostructure of cross-sections of the ingot.The composition variation has been also been recorded during casting campaigns.
Investigations also were performed to find out how well a and b are distributed during a casting process.Different samples were taken for microstructural analysis using light optical microscope.
The results will be discussed in details elsewhere [20] even though some specific discussion regarding the present work will be drawn in next section.

Discussion
Of particular interest is the significance of these experimental results and observation considerations for the copper strip casting process.

Similarity considerations
The design and operation of flow models need due consideration of similarity criteria.The liquid velocity pattern measured in the water model corresponds to that in liquid copper, if the criterion of the Reynolds number and that of the Froude number (Re = r l U ch L ch /m, F ¼ U 2 inlet =r l L ch Þ, are fulfilled simultaneously.This is the case, when the geometric ratio is in agreement with the equation: The indices m and r denoting the model and the real subject respectively; n being kinematic viscosity of the two liquids under consideration.It should be noted here that although the density and dynamic viscosity, m of water are different to those of copper (see Tab. 2 and note v = m/r), the dimensionless parameter of the Reynolds number (Re = r l U ch L ch /m) which governs the fluid flow, is roughly the same for a given value of U c and geometry.Note that the flow is characterised by the ratio of r s /m, which is roughly 10% higher for copper than for water; nonetheless, one should expect the velocity fields to be essentially similar.

The criterion of convection patterns
The convection pattern of the mould and tundish can be determined by the forces exerted on the molten copper.The buoyancy force (F b ) and inertia force (F i ) will give rise to natural convection and forced convection respectively.Therefore, the ratio of F b /F i can be manifest the convection pattern of fluid flow in the mould.This means: or From the definition of Grashof number (Gr = gl 3 bDT/v 2 ) where b is the coefficient of thermal expansion, DT is appropriate temperature difference, u is local velocity and g is acceleration due to gravity, the following is reached: Therefore, the dimensionless number, Gr/Re 2 , can be the criterion to determine the convection pattern in the mould system, as follows: -Gr/Re 2 <<1: internal force dominates fluid flow, forced convection pattern.-Gr/Re 2 @1: both internal and buoyance forces dominate fluid flow, mixed convection pattern.-Gr/Re 2 >>1: buoyance force dominates fluid flow, natural convection pattern.
For copper casting, it is appropriate to take DT as the difference between the temperature of the incoming liquid copper and the melting temperature (e.g.40 K), and L ch as the channel half-width, 0.0135 m.In this case Gr/Re 2 varies between 0.0015 at the top and 16.1 at the bottom of the liquid pool depth.This implies that the buoyancy forces will be much weaker than inertia forces at the top and can therefore be neglected; while the mixed convection can be predicted further down.
The effects of different casting operating variables on the casting problems are given in Tables 4 and 5.This can ultimately be applied to predict and understand the effects of different variables on the solidification process.
The numeric order in the first column of Table 4 describes ranking of the effect of different casting operating parameters, which means, for instance, nozzle (No. 1) has more important effect than tundish (No. 3).Moreover, the alphabetic order in that column determines the degree of importance of different variable for that special parameter.This, again, means that the variable of how to plug in the nozzle probably has more effect than nozzle dimension (to be over sized).The second and third columns in Table 4 describe the problems caused by those parameters (described in the first column) and possible solution for those problems, respectively.
Note that the casting velocity/inlet speed has no number.That's because the casting speed, basically, can't be considered as a parameter that varies under a casting process.The intention is to have V cast as constant as possible.Also, it is worth to note that for a constant casting velocity and consequently inlet velocity, the liquid velocity through the nozzles can be changed.That effect should be studied by looking to some other parameters like plug and dimension.
From the results presented in Tables 4 and 5, the following instruction can be drawn to define how the nozzles should be plugged in and out: -In order to avoid Asymmetric flow, it is desirable to have nozzle plugs as symmetric/ordered as possible (e.g.ON and OFF).One should keep in mind that having higher turbulence will cause more random mixing and consequently more random heat releasing/freezing.-In order to have the flow as even as possible, it is recommended to start to plug in/out from the center part of the nozzles, keeping in mind instruction No. 1.This will help to have more homogenous flow at the long side (it is named x-z plane) of the mould and having smaller up-ward recirculation in the narrow side (x-y plane) of the mould.-Avoiding to plug in the two-three (of 14-18 nozzles) outermost nozzles.
* Plugging the outermost nozzles gives more room for upward recirculation flow near the end of the wall.* Plugging the outermost nozzles (with same inlet velocity) gives also more room for formation of the dead zone at the upper corner of the liquid (near the end of the wall).
The results are valuable for analysing the effects of different casting parameters, such as cooling power, superheat of the molten material and the casting speed.Those parameters control the solidification behaviour and consequently quality of the cast material.On the other hand, casting quality directly contribute to the extrusion efficiency.
The microstructure study in this work (Figs.8a-d) indicates having segregated zinc often with shape of round bands of b phases.One also could find colonies of longitudesegregated-rich Pb bands almost everywhere.Having segregated areas zincs resulting in b and even g phases at the surface of the ingots, emphasizing the effect of cooling process after the mould on the quality of the produced billet.Therefore, using secondary cooling process after the mould can be recommended in order to control cooling process of the solidified billet passing through mould region.Figure 3a shows a rather considerable miss-match between the position of the metallurgical centre and geographic centre of the billet.This indicates that solidification is not ending at the centre of the billet, meaning applying of non-uniform heat transfer coefficient distribution in the mould region It can be found that the turbulent mixing is very intensive near the inlet region due to high velocities.The following suggestions can be considered:

Using of slot-submerged inlet nozzle with jet killer
From the results presented and discussed earlier, the following suggestion can be drawn: -Damping the turbulence.This can be done by using the slot inlet system.-Less gas/air bubble formation.This can be done by using submerged nozzle system.-Decreasing the free surface (meniscus) disturbances, in order to avoid the random heat releasing from the liquid to the mould.That can be done by: * Using submerged inlet system.

Problem Effecting parameter Solution
* Plugging out the first cooling channel to postpone beginning of the solidification process.-Avoiding of having local (and big) up-ward recirculation zone.This can be done by using jet-killer as explained earlier.
-Avoiding of having Asymmetric flow.This can be done by using slot inlet system together with using the jet-killer.
Considering the presented and discussed results, the following inlet system of slot-submerged inlet nozzle with jet killer can be proposed.More experimental and especially theoretical (simulation obtained from mathematical modelling) is needed for a final design.

Mathematical modelling
This study has revealed a great deal about the flow pattern, which has been confirmed with experiments.However, it is difficult to use the physical models to study the accompanying heat flow, which is more easily can be done using a mathematical model.That model will ultimately be applied to predict and understand the effects of different design variables on the fluid flow in the mould.
In order to develop accurate and efficient modelling tools to have a better control on the casting process, two more different objectives in two different steps, as an option, can be considered: -PART 1: Mathematical model of the flow pattern in the liquid pool.
To develop a mathematical model of the flow pattern in the liquid pool, and thereby to determine how molten copper is distributed through the nozzles.To verify acceptable accuracy of the model, its predictions will be compared with experiments performed earlier in part 1.
-PART 2: Mathematical model of the entire strip casting process.
To model a comprehensive mathematical model for fluid flow, heat transfer and solidification within a continuous strip casting in Finspång.This model can be used to investigate the effects of important casting operation and design variables on the solidification behavior.Experiments will be performed in order to evaluate the accuracy of the model.

Direct observation of molten copper meniscus in the mould
Free surface fluctuations of molten copper in the mould region of the continuous strip caster are considered to have a significant effect on the surface quality of the cast.To find out the relationship between the surface wave motion of molten copper near the mould wall during casting and mould oscillation, the meniscus of the molten copper in the mould can be directly observed through a quartz glass window mounted on the mould wall.The schematic drawing of the apparatus used for direct observation of the meniscus can be found in the manuscript of Mahmoudi and Vynnycky [19].
The results are valuable for analysing the effects of different casting parameters, such as the mould wall temperature, the superheat of the molten material, the heat of fusion, the specific heat capacity and the casting speed, on a continuous casting process.
It was observed in this study that the casting speed has a significant effect on the solidification process.It controls the solidification front location and shape, which directly contribute to the microstructural quality of the cast material and determining of the mould design.It can be seen that with the increase of the casting speed, the solidification front moves downstream and the slope of the front become steeper.On the other hand, lower values of the casting speed decrease the productivity.Therefore, a cost-effective combination between these parameters is very much desired.Thus, one should strictly control the casting conditions in order to improve the strip quality.An understanding of transport phenomena in the continuous casting system can help in optimizing a casting process.
As discussed before, the shape of the solidification front is very important for controlling the structural uniformity of the cast material.A steeper solidification front results in higher structural non-uniformity, while a flat interface is desirable to get better uniformity in the cast product.The shape and location of the solidification front may be controlled by varying the cooling condition in the mould or the casting speed.The results presented here are of interest and importance in obtaining a better control of existing continous casting processes leading to improved product quality and increased productivity.

Remarks and suggestion
The operation of strip casting machine is difficult problems arising from the heavy inlet stream into the mould, which can become even twofold as compared with conventional continuous casting.Choppy flow, wave and vortex formation in the surface of the liquid metal, favor the entrapment of mould powder in the strand and also reduces the cleanliness of the solidified strip shell.On the other hand, sufficient heat has to be transferred to the narrow side of the wide mould in order to avoid re-melting in this sensitive zone.It can be stated, therefore, that properly designed flow in the mould is decisive for the successful operation of wide mould casters.
However, control of flow pattern both in tundish and mould, plays an important role in having a better product quality.Liquid copper flow and temperature distribution are the basic factors governing operation of the solidification process.

Conclusion
The following main conclusions can be drawn: -A melt recirculation zone develops at the first few centimetre of the mould, where the solidification begins.That can be a cause for solidification defect.
-With the increase of superheat, the solidification front moves further toward the outlet of the mould.Moreover, it can be noted that the centre and surface temperatures are higher for higher super heat.-It will be very important to have homogenous and well distributed a and b phases, also to avoid of having g phase.-Higher cooling effect resulted in shorter liquid pools.This means for increasing productivity one need to use higher cooling effect.-The temperature increases at the billet surface after passing through the mould.That could cause re-heating of the billet, which in turn could affect structure changes.Plant observation also confirms that the billet will be reheated when it passes the mould.-It was found, however, that by increasing the heat flux/ heat transfer condition at the billet surface, it was possible to change the temperature distribution at the billet surface and gain continuous temperature drop along the casting length.
-With an increase of withdrawal speed, the outlet surface temperature of the cast material remains at a higher temperature.
-With a decrease of the casting speed, the solidification front moves toward the entrance of the mould.-Very high variations in operating parameters have been observed during the last 2 months.That can be riskyfrom the quality point of view-for casting and consequently extrusion operations.-It's highly recommended to use the operating data, extracted from this study and presented here for next experimental campaigns.Moreover, very careful quality studies on the produced billet structure should be performed in relation to plant measurements to optimize casting practice.

Fig. 2 .
Fig. 2. (a) Effect of number of the inlet nozzle jets; (b) Effect of number of the inlet nozzle jets; (c) Effect of number of the inlet nozzle jets.

Table 3 . 3 4 5 1 2 2 Particle measurements under submerged inlet nozzle jets 3 3 4 1
Trials of the inlet nozzle jets on the flow pattern Effect of the plug in of the inlet nozzle jets on the flow pattern Effect of inlet velocity on the flow pattern 1 Effect of casting speed/inlet velocity on the flow pattern Formation of the air/gas bubbles 1 Effect of the misplacement of the tundish on the flow pattern Effect of the adjustment in height of the tundish on the flow pattern 1 Effect of the misalignment of the inlet nozzle jets on the flow pattern Effect of the dimension of the inlet nozzle jet on the flow pattern 2 Effect of number of the submerged inlet nozzle jets on the flow pattern Effect of the plug in of the submerged inlet nozzle jets on the flow pattern 2 Formation of the air/gas bubbles under submerged inlet nozzle jets 2 3 Effect of T-profile submerged inlet nozzle jets on the flow pattern 2 4 Effect of Slot nozzle on the flow pattern 2 5 Effect of Semi-Slot nozzle on the flow pattern 2 6 Effect of Submerged-Semi-Slot nozzle on the flow pattern 3 1 Particle measurements under free stream jets 3 Particle measurements under Semi-Slot inlet nozzle jets 3 Particle measurements under Submerged-Semi-Slot inlet nozzle jets 4 New inlet system: Submerged-Semi-Slot inlet nozzle jets with jet killer

Fig. 7 .
Fig. 7. (a) Typical result of variation of flow, temperature and heat flux during casting campaign.(b) Typical casting temperature.

Fig. 8 .
Fig. 8. (a, b) Microstructure of the sample (extruded bar).(c) Microstructure of the cross section of the cast bar.(d) g phase at the surface.

Table 1 .
Important parameter for water model.

Table 2 .
Physical data for water and copper.