Sie sind auf Seite 1von 36

Book: “Scale-up in Metallurgy”

Editor: M. Lackner
ISBN: 978-3-902655-10-3

Book Chapter: "Scale-up fundamentals"


number [MA]

Authors: Kamil Wichterle* and Marek C. Ruzicka**

*Department of Chemistry
Faculty of Metallurgy and Material Engineering
VSB-Technical University of Ostrava
17. listopadu 15/2172
708 33 Ostrava-Poruba
Czech Republic
E-mail: kamil.wichterle@vsb.cz
Tel: +420 596 994 304
Fax: +420 596 918 647
www.vsb.cz

**Department of Multiphase Reactors


Institute of Chemical Process Fundamentals
Academy of Sciences of the Czech Republic
Rozvojova 135
16502 Prague
Czech Republic
E-mail: ruzicka@icpf.cas.cz
Tel: +420 220 390 299
Fax: +420 220 920 661
www.icpf.cas.cz

Abstract
This chapter covers the basic material about the use of the dimensionless numbers in the broad
field of metallurgy. This area contains a wide number of different processes of very diverse nature.
The goal is to identify these particular processes, determine the fundamental physical phenomena
underlying these processes and recast the complex picture of the metallurgical science and
technology into a simpler framework in terms of subsystems that concern (i) the transport
phenomena together with (ii) the reaction phenomena. Within this framework, the experience from
the solid background of the chemical engineering can be transferred on to the metallurgical
problems and its 'unit operation' - not essentially different from the chemical ones (grinding,
leaching, extraction, flotation, separation, heating, melting, oxidation/reduction, gas-liquid reaction,
cooling, etc.). These particular processes are assigned the relevant dimensional numbers that
reflect the dominant quantities affecting the process. From the chemical-engineering point of view,
all the metallurgical operations involve the mass, momentum and energy transport, and are
essentially multiphase in nature (the solid, liquid, and gas phases are involved in various
combinations and importance). In addition, the reaction phenomena also occur and even dominate
in many metallurgical operations. The difference between various approaches is reflected by
different sets of dimensionless numbers (e.g. pyrometallurgy vs. hydrometallurgy). The
particularities of the individual metallurgical processes (e.g. steel making technology, liquid metals)
as compared with the 'usual' chemical technologies (water and common liquids) are reflected by
the different accent on the dimensional numbers employed and by their relative magnitude. The
characteristic length and time scales encountered in the realm of the metallurgy are to be
accounted for by the values of the respective dimensionless numbers. Altogether, it is shown that
metallurgy essentially features the following: multiphase nature, multiscale structure, substantial
heat transport phenomena, liquid metal hydrodynamics, strong coupling between the transport and
reaction, usually under extreme ambient conditions. The general absence of good laboratory
1
models in the metallurgy has its rationale: a 10-cm model cannot grasp all the phenomena
occurring in the real furnace, and even if, their 'proportions' would highly be misleading.

Contents

Introduction

Part I. Basic: Equations, Scaling, Dimensional numbers


1. What is scale-up ?
2. How to obtain dimensionless numbers ?
2.1 Dimensional analysis
2.2 Scaling of equations
2.3 Comparison of DA and SE
4. Governing equations in metallurgy and their scaling
4.1 Balance equations
4.2 Closure relations
4.3 Thermodynamic equations
5. Balance of momentum
5.1 Fluid mass conservation
5.2 Fluid momentum conservation
6. Balance of heat
7. Balance of mass
8. Other processes, effects, and DN related to metallurgy

Part II. Application: Processes and Operations in Metallurgy


9. Metallurgy: art and technology
10. Front end operations
(size reduction, classification, mechanical separation, granulation etc.)
11. Pyrometallurgy
Basic procedures
Dimensionless numbers
12. Hydrometallurgy
(typical processes)
13. Electrometallurgy
(electrolysis, heating, plasma cutting)
14. Liquid metals
(unit operations)
15. Solid metals
16. Conclusions

Symbols
References

Introduction

This chapter presents a review over the scale-up fundamentals and its applications in the
metallurgy. It has two parts. In the first Part I., the basic facts are presented about the governing
equations, their scaling and the dimensionless numbers thus obtained, which are relevant and
recommended for further usage in making empirical correlations and for scale-up purposes. The
equations present the fundamental conservation laws of the macroscopic physics, verified by long-
term experience. They comprise the three main quantities, the momentum, the heat (energy), the
mass. The result of their scaling is twofold: the dimensionless equations, and, the list of
dimensionless numbers that are relevant and have clear physical meaning. The Part I. starts with a
brief introduction to the scale-up, followed by a concise presentation of the dimensional analysis. In
the second Part II., the broad and diverse area of metallurgy is mapped in terms of the individual
2
technologies, the basic processes, and the key operations. In these 'unit operations', the
importance of the momentum, heat and mass transport phenomena are assessed, together with
the possible reaction phenomena, and their corresponding particularities are highlighted. The
adequate dimensionless numbers suitable for the given system description are recommended and
the dominant effects likely prevailing under given conditions are estimated.

Part I. Basic: Equations, Scaling, Dimensional numbers

1. What is scale-up ?

The scale-up means to find a description of a given system that involves the system dimensions.
Usually, a small model is studied first, on a small length scale L1, and its description is established.
Then, this knowledge is transferred onto a larger system, a prototype with a length scale L2 > L1,
under the assumption of the similarity of these two. This means that the description must explicitly
involve the dependence on the system size L.
Advantageously, the description is formulated in dimensionless variables, dimensionless
numbers (DN), whose physical dimension is unity, denoted as [-]. These DN contain L but their
values are independent of the absolute system size. The change of L must be compensated by a
suitable variation of the other quantities gathered in DN, to secure no change in its magnitude. It is
also called the modeling.

Consider a small laboratory model (index M) that is described with a dimensionless quantity ΠM by
a relation
ΠM = ΦM(ΠiM) , i = 1,2,... (model) (1.1)
where Πi are the important dimensionless parameters affecting the master quantity Π and the
symbol Φ stands for the functional relationship, often called a correlation.
We want to describe a bigger system, the prototype (index P), for which it holds
ΠP = ΦP(ΠiP) . (prototype) (1.2)
When these two systems are not similar, the relations (1.1) and (1.2) are generally different. When
these two systems are similar, the relations (1.1) and (1.2) are identical:
ΦM = ΦP (scale-up relation)
ΠM = ΠP (similarity law)
Πi M = Πi P ∀i (similarity criteria) . (1.3)

For instance, the dimensional period of a simple pendulum is P ~ (l/g)1/2 [s]. Applying the similarity
law for the dimensionless periods, PM/(lM/g)1/2 = PP/(lP/g)1/2 [-], gives the relation between the
periods of big and small pendula: PP/PM = (lP/lM)1/2, which is called the scaling rule. Increasing the
pendulum length 25-times gives only 5-times longer period, in virtue of (lP/lM)1/2; the factor of 5 is
called the scaling coefficient. Here, we have no parameters affecting the period (no Πi ). In the
following example, we have two of them.

In a long-pipe flow, choose the dimensional drag force F as the master quantity. May it depend on
the following properly selected parameters
F = f(D, d, ρ, µ, V) , (1.4)
where D is pipe diameter, d wall roughness, ρ and µ are the fluid density and viscosity, V the
typical fluid speed. After performing the dimensional analysis, the dimensionless formulation reads:
F /ρV2D2 = Φ(µ/ρDV, d/D) , (1.5)
where the term (ρV2D2), which is (dynamic pressure)x(cross section area), is the relevant scaling of
the drag force, F ~ ρV2D2, used for the nondimensionalization of F. The equation can be written
as:
Π = Φ(Π1, Π2) , (1.6)
where the two Pi-terms are the Reynolds number Π1 = (ρDV/µ) and the dimensionless wall
roughness Π2 = d/D. The r.h.s. of (1.6) is usually written as C = C(Re, d/D) and expressed by
3
empirical correlations (Colebrook formula) or diagrams (Moody chart) for the pipe-flow friction
factor C [-]. The above relations are valid for both small and big pipes, within the limits where the
selected quantities in (1.4) remain relevant for F and the similarity paradigm holds (e.g. not in
micro/nano-systems, mega-systems with open surface and gravity effects, flows in external force
fields, etc.).

Where there is no exceptional variable to be called the 'master quantity', we have an implicit
formula for the system description,
Φ(Πi) = 0 , i = 1,2,... , (1.7)
where any of the Pi-terms can be explicated, in case of need.

2. How to obtain dimensionless numbers (DN) ?

We have seen that the dimensionless numbers (DN) are the most suitable quantities to be used for
the scale-up. The natural question is how to find them for a given practical problem. There are two
basic ways. One is the dimensional analysis and the other is the scaling of equations.

2.1 Dimensional analysis (DA)

Here, we have neither the governing equations describing our problem nor the list of the relevant
variables needed for the description. The method of DA is simple but the result is not guaranteed.

First, choose a dimensional master quantity a and make a list of variables (parameters), ai and bi,
that in your opinion affect the value of a. The list should be complete (no important parameter
omitted) and the variables should be relevant (real influence on a) and mutually independent (e.g.
not to take both dynamic and kinematic viscosity). There are no strict rules how to create such a
list, one should be guided by the deep insight into the problem, profound experience, intuition and
good luck. Nevertheless, there are many suggestions and useful tips in the literature.
Recommendations are available regarding some aspects of the problem the parameters should
reflect (e.g. system geometry, material properties, kinematic and dynamic aspects, external
conditions, etc.).

Second, write formally the unknown relation f, how your master quantity depends on the chosen
variables,
a = f(a1, a2, ... ak ; b1, b2, ... bm) , (2.1.1)
where the k-variables ai have mutually independent dimensions (e.g. length, time) while the
dimensions of m-variables bi together with a can be composed from the former by combination
(e.g. speed = length/time).

Third, combine the independent dimensions of ai to reproduce the dimensions of a and bi, on
assuming the form of the product of simple powers:
[a] = [a1]p01. [a2]p02. [a3]p03 ... [ak]p0k ... ≡ A ,
[bi] = [a1]pi1. [a2]pi2. [a3]pi3 ... [ak]pik ... ≡ Bi , i = 1,2,...m , (2.1.2)
where the unknown exponents pij are found by solving the linear algebraic system obtained by
equating the dimensions of the both sides of (2.1.2). Also, denote the r.h.s. of (2.1.2) by A and Bi
respectively.

Fourth, make the dimensional variables a and bi dimensionless, by dividing them by A and Bi.
a/A = Φ(b1/B1, b2/B2, b3/B3, ... bm/Bm) , (2.1.3)
where the new symbol Φ stands for the former function f, whose form has changed, since it
became a function of less variables (all ai disappeared). Rewrite (2.1.3) into the usual notation:
Π = Φ(Π1, Π2, ... Πm) , (2.1.4)
where the dimensionless terms (Pi-terms, DN) are Π ≡ a/A, Πi ≡ bi/Bi.
DA relies on the Buckingham Pi-theorem that says that it is possible to go from (2.1.1) to (2.1.4).

4
The advantage of DA is the reduction of variables, from (k+m) to (m), and, making the system
description dimensionless. Moreover, we get the basic scaling for the master quantity,
a ~ A = a1p01. a2p02. a3p03 ... akp0k , (2.1.5)
which is very useful for making the scale-estimates. The unknown dimensionless function Φ is the
'correction' to the rough but vital scale estimate,
a = A.Φ(Π1, Π2, ... Πm) , (2.1.6)
and cannot be determined by DA itself, but some other methods are needed, e.g. experiments,
simulations, etc.

2.2 Scaling of equations (SE)

Here, we have the governing equations describing our problem that contain the full list of the
relevant variables needed for the description. The method of SE is simple and the result is
guaranteed.

Consider a simple mechanical example, the free-fall under gravity g of a body of mass m along the
x-axis in a fluid medium with resistance factor k. The dimensional governing equation is the
Newton force law,
m(d2x/dt2) = mg - k(dx/dt)2 , (dimensional) (2.2.1)
with two kinds of quantities, the variables (t,x) and the parameters (m,g,k). Note that all these
quantities are relevant for the problem, since they stand in the governing equation. The initial
condition be: t, x,(dx/dt) = 0, with no new variable. There is no boundary condition.
Choose the typical yet unspecified quantities for the variables, to scale them. Take T for the
characteristic time scale and L for the characteristic length scale. The dimensionless variables are:
t* = t/T, x* = x/L, dx*/dt* = (dx/dt)/(L/T), d2x*/dt*2 = (d2x/dt2)/(L/T2).The dimensional govering
equation takes a new form:
mL/T2(d2x*/dt*2) = mg - kL2/T2(dx*/dt*)2 , (dimensional) (2.2.2)
where each term is a product of a dimensionless part (*) and dimensional part (no *).
In the scaling analysis, it is important to estimate the relative magnitude of the terms in the
equation. If we choose the scales T, L in such a way, that the starred variables will not much
exceed unity, will vary roughly between 0 and 1, which is denoted o(1), the relative magnitude is
given by the unstarred terms, (mL/T2):(mg):(kL2/T2). This is very convenient, because we can
adjust the proportion by changing values of the control parameters (m,g,k). This way, we can make
some terms negligible and the other one dominant. Doing this, we can select the wanted regime of
the behavior of our system. It will be dominated by the chosen effect or the specific process
(viscous/inviscid, gravity/inertia, conductive/convective, unsteady/steady, etc.). Also, reducing the
number of terms in the equation, by neglecting the 'small' terms with respect to the 'large' terms,
presents a great simplification and the reduced equation can be amenable to solution.
Knowing all this, we can skip the manipulation with the starred and unstarred variables and
treat the original equation in a bit simpler way. Substitute T for t, L for x, directly into (2.2.1).
Choose also the characteristic speed V and put it for L/T. Obtain the dimensional scale equation:
mV/T = mg - kV2 , (dimensional) (2.2.3)
which represents the scale estimates of the three forces involved:
(inertia) = (gravity) - (resistance) . (2.2.4)
Diving by one of the three terms we obtain the dimensionless scale equation and the relevant DN.
Which one to divide with? Choose the gravity term (mg) since it is the driving force for the system,
i.e. it is an important term that should be retained. The best way how to conserve it is to put it into
the denominator of the DN thus produced:
V/gT = 1 - kV2/mg . (dimensionless) (2.2.5)
Here, the two products are the wanted DN obtained by SE. They have the following meaning:
Π1 ≡ V/gT = (inertia)/(gravity) ,
Π2 ≡ kV2/mg = (resistance)/(gravity) . (2.2.6)
If we put Π1 = 1, it means that the inertia and gravity forces are comparable, also denoted
O(1), i.e. the same order of magnitude. Consequently, the speed scale can be estimated as V ~
gT, where T is the current time t over which the acceleration proceeds. We have the unsteady
speed behavior at the startup of the motion, the inertial response of the mass to switching-on the
driving force. If we put Π2 = 1, it means that the resistance and gravity forces are comparable. This
5
yields the scale estimate for the steady speed V ~ (mg/k)1/2, resulting from the balance of the two
opposite forces. Note that Π1 decays with time as ~ 1/T, while no time scale is involved in Π2,
which represents the timeless force equilibrium. Thus the dimensionless equation
Π1 = 1 - Π2 , (2.2.7)
on the short time scales, simplifies to:
Π1 ≈ 1 , (2.2.8)
where Π2 ~ V2 can be neglected because of the smallness of V. In contrast, on the long time
scales, it simplifies to:
Π2 ≈ 1 , (2.2.9)
where Π1 ~ 1/T can be neglected because of the largeness of T.

2.3 Comparison of DA and SE

DA is a simple semi-intuitive method that, without almost any previous knowledge and without
having the governing equations, generates a great deal of the solution to a possibly very complex
practical problem. We get the scale-estimate of the wanted quantity and the DN that should be
considered when looking for the correction function (Φ). Also, the number of parameters is
reduced, whence the number of degrees of freedom when planning experiments, performing
measurements and correlating our data. Moreover, the parameters are in the form of DN, which
are suitable for further modeling (scale-up, scale-down).
The price is that we cannot be sure whether the results obtained by DA are reliable. Our list
of variables may be incomplete, these may be inter-dependent as well as irrelevant. The choice of
good variables may be a very tricky task, indeed. DA usually fails at regime transitions, since
different effects are important on both sides of the transition, whence different lists of the
parameters should be chosen for description. DA is not a universal method, it has certain limits of
application: the problem under study must be solvable by DA, which is not always clear
beforehand.

SE is a simple rigorous method for manipulating the fundamental governing equations to obtain all
the variables relevant for the problem (those involved in the equation) and to generate all the
relevant DN related to it. Moreover, the DN usually represents the ratio of two different effects, e.g.
forces, time scales, etc., which enables their physical interpretation and gives a deeper insight into
the particular problem. However, like DA, it does not give us the relation between the DN. It must
be found by other means (solving the equations at least in a simplified form, experiments, etc.).
Why to use SE when we have the full equations? Because these may be difficult to solve,
the solution is not a closed formula but a single realization in space and time (a pile of data), we
cannot see the effect of parameters, we do not know the rough patterns of the forces or effects
dominating the system behavior under various conditions (different regimes, modes of behavior,
limits of behavior, etc.) Very important is the effect of the initial and boundary conditions. They
must be involved in the SE procedure too, and can produce important DN. Note that only some
material properties are present in the equations (e.g. density, viscosity), while the others enters via
the boundary condition (e.g. surface tension). Also, at the boundaries, the fluxes are defined, that
makes the system open, far from equilibrium. These are highly important control parameters.

4. Governing equations in metallurgy and their scaling

The metallurgy comprises many complex technologies, ranging over a broad spectrum of individual
processes, of highly different nature. Behind this extreme and apparent complexity, all the
metallurgical operations can be, in principle, described by the balance equations for the three
following quantities: the mass, the momentum and the energy. The transport and reaction
phenomena are traditionally studied and systemized in the area of the chemical engineering.
During a century, huge progress has been reached in forming the equations, manipulating them,
scaling them, solving them, and interpreting the results obtained, in a large number of particular
situations of various degrees of complexity, under both steady and unsteady conditions, for highly
nonlinear problems, in strongly-coupled multiphase systems, spanning a broad range of time and
6
space scales. Thus, it seems to be appropriate to apply the chemical-engineering approach also to
the complicated problems in the field of the metallurgy, treating them in terms of the fundamental
balance equations.

4.1 Balance equations

There are several classes of equations we use for describing the system. The very basic ones are
the balance equations for the conserved quantities that represent the ultimate laws of nature as
they operate within the limits of our everyday experience, in the macroscopic world of our day life
(no extremely small and large length and time scales). They are the conservations of mass,
momentum and energy. These laws are very deep and fundamental, and reflect some basic
properties of the invariance of these physical quantities with respect to certain geometrical
transformations of the space and time. They have the mathematical form of the evolution partial
differential equations, with the proper initial and boundary conditions, and are typically nonlinear
(impossible to solve analytically, difficult to solve numerically). They have the general form of:
accumulation = input - output ± internal sources/sinks .
They are the equations of change and the balance base is the volume unit. Their physical
dimension then is [balanced quantity/m3.s]. The terms on the r.h.s. represent different contributions
to the total change standing on the l.h.s. These contributions corresponds to different processes
whose relative magnitude determines the character of the resulting solution. Their proportion is
crucial to determine the prevailing regime or mode of operation, as well as to neglect the weak
effects with respect to the strong ones, simplifying thus the description enormously and making the
problem near-solvable. The balance equations are those to be scaled to find the main list of the DN
relevant for the given problem. They are the main source of our knowledge about the system under
study.

4.2 Closure relations

These formulas represent the 'stuffing' put into the balance equations. They are the particular
expressions we substitute for the general terms called 'input', 'output', and 'sources' in the balance.
We have closures for transport and kinetic phenomena.
There are closures for the fluxes, whose spatial difference results in a volumetric
contribution to the balance of a 'small control element'. Consider that a conductive flux j of a
quantity q depends on this quantity,
j = f(q) , (4.2.1)
which sounds sensible. Generally, we do not know the function f. Therefore, we expand it near
some suitable point q0,
j = aq0 + b(q-q0) + c(q-q0)2 + ... , (4.2.2)
and neglect the higher-order terms to get:
j = aq0 + b(q-q0) . (4.2.3)
Be the suitable point q0 the reference point, where there is no flux, the constant term is thus zero,
j = b(q-q0) , (4.2.4)
which is the usual form of the relation between the flux and its driving force, from the linear non-
equilibrium thermodynamics. In the differential form, we have:
j = - B∇ q , (4.2.5)
where there is a smooth spatial profile of q(x). The transport coefficients b, B are usually found in
the tables. The closures for the conductive fluxes are typically linear (by our approximation), but
can be multidirectional or mutually coupled. Then the proper vector notation is needed, allowing
the transport coefficients be the tensor quantities. The convective fluxes are products of q and the
speed v, which are nonlinear when q involves v (e.g. momentum - the quadratic nonlinearity of the
flow equation).
Besides the fluxes, there are closures for the reaction rates, mostly power-law in simple
chemical kinetics,
r = kcn , (4.2.6)
or rational-function form for biochemical and biological kinetics,
r = k1c/(k2+c) , (4.2.7)

7
in the simplest case. The kinetic closures are typically nonlinear, making the problem difficult.
The closure equations are usually embodied into the balance equations and are therefore
not scaled solitarily, but within the main equations. However, when they enter boundary conditions,
they are scaled there.

4.3 Thermodynamic equations

Paradoxically, the common (i.e. equilibrium) thermodynamics should be named the thermostatics,
since the are no forces, there is no macroscopic motion, there is no time and all the changes are
only virtual, i.e. 'infinitely slow' and 'infinitely small'.

Here we have important formulas for the equilibrium conditions, where there are no fluxes and no
reactions, which are useful to set the base line (reference state) for the quantities like q0 above.
The equilibrium means the equality of potentials, within the space, between phases, across
boundaries, etc. Various quantities are used, e.g. thermodynamic potential, chemical potential,
fugacity, concentration, etc., depending on the particular equilibrium situation and on the
approximation adopted. Typically, we use the phase distribution coefficient, the Henry constant,
equilibrium constant of reaction, adsorption isotherm, etc.

Another formulas express the equations of state, which specify the values of some important
properties of matter under given circumstance. For instance, the state of the ideal gas, the state of
the real gas, the state of condensed matter, values of the material and transport and rate
coefficients at given temperature and pressure (density, viscosity, surface tension, specific heat,
rate constant, etc.). This information can be found in numerous tables on thermodynamic data.

When these equations enter the balances, they are scale together with them. If they are used as
external auxiliary formulas, they may or may not be scaled, depending on our choice.

It follows that in the metallurgy, all the above classes of equations are relevant, of which only the
balance equations together with their proper initial and boundary conditions are the main source of
DN. Consequently, only these equations will be presented in some detail further, will be scaled, the
DN thus generated will be discussed, and various situations analyzed, where some of them
dominate over the other.

5. Balance of momentum

The conservation of the momentum is the basic physical law and the very ground of the
macroscopic mechanics. One main discovery was that force F changes the state of motion of a
body. The other was that 'state of motion' can be expressed by a quantity called 'momentum',
which is a product (mass)x(speed). Formally, in symbols, we have the Newton force law, d/dt(m.v)
= F, which is an ordinary differential equation. With constant mass, it becomes the basic-school
formula: F = m.a. Note that the term (momentum/time) is often called the inertial force, to have the
equation terminologically homogeneous. This model describes the translation of a body. To
capture also its rotation, besides the above liner momentum, we have to consider also its angular
momentum. From single bodies, we can move to systems of many bodies or particles, with mutual
interactions of different character and strength. Depending on the latter, i.e. on the ability of a body
to deform, we have three traditional states of matter: solid (strong bonds, low deformation), liquids
(medium bonds, condensed state, easily deformable, low compressibility, surface tension), gases
(weak or no bonds, easily deformable and compressible). Other peculiar states of matter are often
added to these three: plasma, granular matter (powders, soil), rheologically complex materials
(hybrids between solid and liquid), etc. The very large ensembles of particles can be considered as
continua. In a continuum, the macroscopic quantities are smooth in space and time. Consequently,
they can be described by partial differential equations.
In this section, we will consider the balance of the linear (not angular) momentum for a fluid
continuum. For simplicity, the fluid will be of simple rheology (Newtonian) and incompressible. The
8
governing equation is the Navier-Stokes equation, which merely is the Newton force-law written for
a fluid element fixed in space (Euler's view). It is a nonlinear partial differential equation whose
solution describes the flow field: the velocity and pressure distribution within the flow domain and
its time evolution. It is mathematically extremely difficult to solve it in general. It is impossible to do
it analytically (theoretically - with pen and paper), it is possible to do it numerically within severe
limits on the resolution and the value of the Reynolds number. However, after simplification, it is
even possible to solve it exactly. These simplifications come from neglecting some terms in the
basic equation. The neglection is made upon proper scaling analysis and estimation of the relative
magnitude of the individual terms. Here, the DN play a crucial role.
Therefore, the basic balance equations for the single-phase fluid flow will be presented, will
be scales, the DN obtained will be discussed, the possibility of neglecting some terms will be
highlighted, with a special attention to the typical situations occurring in metallurgy.

5.1 Fluid mass conservation

Together with the fluid momentum, the fluid mass balance is considered, since these two are
coupled and are solved together. The mass balance sounds for ρ and imports v from the
momentum balance. On the contrary, the momentum balance sounds for v and imports ρ from the
mass balance. Thus, these two are two-way coupled.
The fluid is the basic host environment for the other transport and reaction processes.
When the environment is stagnant (motionless fluid, solid), we have the conduction (molecular
diffusion). When the environment is not stagnant but flows (flowing fluid, melt), we have the
convection (macroscopic motion). The heat transport in considered in Sec. 6, both in stagnant and
flowing environment. The mass transport in considered in Sec. 7, where the 'mass' means some
other (chemical) component present in the carrying fluid, not the fluid mass itself. If the component
is presented not in the form of a true solution but it is a mechanical individual (particle), we have a
dispersion, a multiphase flow.

The mass balance (continuity equation) for a compressible fluid reads:


(∂ρ/∂t) + ∇ (ρv) = 0 , (5.1.1)
with the following physical meaning:
(unsteadiness) + (convection) = 0
Choosing the variables (ρ, v, x, t), their scales (ρ0, V, L, T), and the parameters (-), we get by
scaling the dimensional scale equation:
ρ0/T + ρ0V/L = 0 . (5.1.2)
The density ρ0 stands at all terms, so we can divide to obtain:
1/T + V/L = 0 . (5.1.3)
To make it dimensionless, we must choose one term to divide with. Choose the convection term
(V/L) to get the dimensionless scale equation:
(L/VT) + 1 = 0 , (5.1.4)
with one DN only, (L/VT) ~ (unsteadiness)/(convection). We can obtain the unknown scales by
putting some DN equal to unity, i.e. to demand that the two effects, whose the DN is the ratio, are
comparable. Assume we know T and L and look for V. If we put (L/VT) = 1, we get V = L/T saying
that (unsteadiness) ≈ (convection). Vice versa, by our own choice, if we take V = L/T because we
wish so, as a consequence (L/VT) must be 1, which is only possible when (unsteadiness) ≈
(convection).
With incompressible fluid, ρ is constant and the continuity reads:
∇ .v = 0 , (5.1.5)
which contains the speed only and is highly suitable to be plugged in the momentum balance to
simplify it. Here we have no DN.

5.2 Fluid momentum conservation

The momentum balance is also called the force balance. It is the Newton force law m.a = F written
for the fluid continuum. To see the similarity more closely, write it first in a general form, using the
convective derivative, when the control box flows with the fluid:

9
D/Dt (ρv) = Fsurface + Fvolumetric . (5.2.1)
The momentum change is caused by two kinds of force, the surface (pressure, viscosity) and the
volumetric (gravity) ones. Substituting the closure relations for the forces for a simple Newtonian
single-phase incompressible fluid, we get the famous Navier-Stokes equation:
ρ(∂v/∂t) + ρ(v. ∇ )v = - ∇ p + µ ∇ 2v + ρg . (5.2.2)
The physical meaning of the five individual terms is the following:
(1 unsteadiness) + (2 convection) = (3 pressure) + (4 viscosity) + (5 gravity) .
It is useful to say few words about each of these five terms, because their ratios are the important
hydrodynamic DN. To interpret these DN, we should understand their components.
The l.h.s. represents the change of the fluid momentum and has two terms of the
acceleration. They are also called the inertial forces, namely the latter one. The first one ρ(∂v/∂t) is
the Eulerian acceleration (unsteadiness), as it would be if the control box would travel with the
fluid. It is what the engineers would call the 'accumulation term'. This term is important in unsteady
flows, e.g. start-up flows, decaying flows, periodic flow, etc. The second one ρ(v. ∇ )v is the
Lagrangian (convective) accelerations, the price for fixing the control box in space and letting the
fluid flow through. It is the main source of the problems since it is the non-linear term. For instance,
it the very cause of the turbulence (vortex stretching) and prevents the mathematicians to prove
the existence and unicity theorems for the flow equation. Scaling it off the equation in a particular
problem presents a great relieve. If not to cancel it completely, at least some clever approximation
is a great help.
The r.h.s. represents the individual forces acting on the fluid element currently in the control
box and has three terms. The first one is the pressure force, the pressure difference across the
control box. It is the normal surface force, an elastic conservative force, long-range in nature.
There is nothing much to be said here about this force. Perhaps, that with an incompressible fluid,
the pressure is undetermined and is calculated using some tricks through the iterative coupling
with the continuity equation. With a compressible fluid, the problem is well-posed and the pressure
is obtained with help of an additional state equation p = p(ρ), we have 3 unknowns (ρ,v,p) and 3
equations (mass, momentum, state). The hydrodynamic pressure coincides with the
thermodynamic pressure, provided the flow situation is not too brutal (no extreme speed, small
time and length scales, rarefication, etc.). In complex fluids, this 'simple' pressure cannot be used
and elaborated concepts for the normal stresses must be developed. The suitable scales for
pressure may be difficult to obtain. The pressure can be a pretty bad guy.
The second one is the viscous force, the shear-stress difference across the control box. It is
the tangential surface force, an inelastic dissipative force. It slows down any motion. It causes the
transport of momentum (velocity) from fast regions to slow regions. It takes a tax for this transfer:
some momentum is 'lost', more precisely, the energy of fluid motion is dissipated or degradated
into heat. It is responsible for the formation of the boundary layers near walls, where the flow must
meet the no-slip boundary condition, i.e. to drop from some large bulk speed to speed zero. There
are severe gradients at the walls, whence severe dissipation. In turbulence, the viscosity eats the
small vortices.
The third one is the gravity force that acts on the whole body of fluid. It is the body
(volumetric, mass) force, a conservative force with a potential, an external field force. If the density
is constant, it does not cause the flow. If the density varies, it can cause buoyancy-driven flows. In
both cases, it attracts the whole fluid to the ground. If this tendency is not counter-acted by firm
solid bottom and side walls, the fluid just 'falls down'. It causes that the fluid flows 'down the hill',
which is omnipresent in nature (rivers) and employed also in technologies (drinking water plants
and distribution, flow through wastewater plants, etc.). Besides the gravity, other similar forces can
operate. The external force field can be generated by the rotation (centrifuge, cyclone) or by
electromagnetic fields.
Note that the balance (5.2.2) is based on a unit volume of fluid, a small control box that is
completely inside the fluid. The dimension of the force-like r.h.s. is [N/m3]. The l.h.s. has the
dimension [momentum/m3s]. Therefore, there are no surface effects included. The balance
contains only two material parameters who are the bulk properties, density ρ and viscosity µ. The
third important one, the surface tension σ, comes through the boundary condition, together with the
corresponding DN.

10
Now we perform the scaling analysis of the balance (5.2.2). Choosing the variables (v, p, x, t), their
scales (V, P, L, T), and the parameters (ρ, µ, g), we get by scaling the dimensional scale equation:
ρV/T + ρV2/L = P/L + µV/L2 + ρg . (5.2.3)
Dividing the five terms by the convective inertia term (ρV2/L), because of tradition and also to keep
it, we have the dimensionless scale equation:
L/TV + 1 = P/ρV2 + µ/ρLV + gL/V2 , (5.2.4)
with only four DN (the convective term is unity). The four DN have proper names:
Sr + 1 = Eu + 1/Re + 1/Fr . (5.2.5)
The four DN are relevant for fluid motion and have a clear physical meaning in terms of forces:

Strouhal number, Sr = L/TV, (unsteadiness)/(convection)


Euler number, Eu = P/ρV2, (pressure)/(convection)
Reynolds number, Re = ρLV/µ, (convection)/(viscosity)
Froude number, Fr = V2/gL, (convection)/(gravity) . (5.2.6)

These numbers can be interpreted in many different ways, in terms of many different quantities.
For instance, Re = (LV/ν) contains three things: length L, speed V, kinematic viscosity ν, which is
(length2/time). Thus, Re can be the ratio of the forces (µV/L2)/(ρV2/L), ratio of the length scales
(L)/(ν/V), ratio of (viscosity)/(inertia) time scales (V/L)/(ν/L2), etc. Note that we use only one
common length scale L for all terms and for all spatial directions. This is possible only here, where
the main purpose is to demonstrate the general methodology, unrelated to any specific problem. In
particular applications one should respect the multiscale character of the nature and be ready to
assign different processes different scales, they deserve.

The relevance and use of these DN follow from their definitions and depends on the particular flow
problem. The basic features of the five terms of the governing equation are important to interpret
the meaning and usage of the four DN. Briefly, the Strouhal number is used in unsteady flow, in
technology when we start the process or finish it. It is often used in mixing, where the time scale it
the period of the impeller motion and enters the correlations for the power input. The Euler number
usually is not too often used in practical application, mainly because the pressure scale P is
estimated by the dynamic fluid pressure ~ ρV2 (see Bernoulli equation), which particular choice
makes Eu = 1. The Reynolds number is of the paramount importance, since it determines the
character of the fluid flow. There are two limits, where the flow equation becomes linear and can be
solved (!): viscous limit (Re = 0) and inviscid limit (Re = ∞) with potential flow. These two case are
qualitatively very different. From these limits, using various approximations, we enter the broad
region of the 'finite' Re flows (0 < Re < ∞): the further from the limit, the more difficult the progress.
The interplay of the viscous and inertial effects at finite Re creates very complex flow phenomena.
Re enters most of the engineering correlations for processes where the fluid flow is involved. Re in
notorious for demarcation the border between the laminar and turbulent flow, the two canonical
flow regimes, with disparate difference in their impact on the other transport phenomena and on
the retention time distribution in technological equipment, which is crucial for the reaction
phenomena. The Froude number (sometimes defined as (V2/gL)1/2) is important where the gravity
(buoyancy) and inertia interplay, e.g. open channel flow, free surface problems, hydraulic
engineering, floating vessels, surface waves at long length scales, buoyancy driven flows, etc.

Since we divided (5.2.3) by the convective term, all four DN are related to the inertia force. If we
want to compare the importance of other effects, we must create new DN by combining the original
ones. We can obtain the following new numbers:

(pressure)/(viscosity) = Eu.Re ,
(pressure)/(gravity) = Eu.Fr ,
(viscosity)/(gravity) = Fr/Re, (5.2.7)

The ratio (gravity)/(viscosity) is often introduced as the Galileo number,

Ga = Re2/Fr = gL3/ν2, (5.2.8)

11
where the square of Re is used for a practical reason: the velocity cancels. This is helpful, when V
is difficult to estimate or is a part of the solution. For instance, when we look for speed of bodies
and particles moving in fluids (bubbles, drops, solids), we have to solve the iteration process:
speed depends on drag, drag depends on Re, Re depends on speed, i.e. a cycle. Applications are
e.g. in sedimentation, fluidization, bubble columns, flotation, etc. When buoyancy effects are
relevant, as it usually is in these applications, Ga is corrected by (∆ρ/ρ) and the product is called
the Archimedes number,

Ar = (∆ρ/ρ)Ga = (∆ρ/ρ)gd3/ν2. (5.2.9)

This number is also obtained by scaling the Newton force law for a body falling/rising in a fluid
under gravity.

We assumed that the scales (V, P, L, T) are chosen properly, so that the hydrodynamic
dimensionless variables (v*, p*, x*, t* + their derivatives) are o(1), i.e. between zero and unity.
This can be made and verified for a particular problem, but not here when working on a general
level. Only then we can see the relative importance of the individual terms from the values of DN.
When some DN is too small or too large, one of the effect dominates over the other and the latter
can be neglected.
Sometimes it is difficult to find suitable scales. We can proceed by the 'trial and error'
method, intuitively choose the scales and then verify whether the starred variables really are o(1).
Also, we can select one particular DN that contains two effects that we think should be in balance
and set it unity.
Usually, we can find suitable scales for length L and speed V. The problem often is how to
scale time and pressure. In case of time, there are few general choices: velocity scaling (T = L/V),
periodic scaling (T = 1/frequency), relaxation scaling (T = relaxation time), energy scaling (T =
L(M/E)0.5), diffusion scaling T = L2/(diffusivity). Others particular choices come from the
hydrodynamic equation. We can compare the unsteady term L/TV, which contains the time scale
T, with the other terms in (5.2.4), to express their similar order of magnitude. Consequently, we
obtain the following scalings for the time:

convection scaling: T = L/V ,


pressure scaling: T = ρVL/P ,
viscosity scaling: T = L2/ν ,
gravity scaling: T = V/g . (5.2.10)

The basic scaling T = L/V makes the Strouhal number equal to unity, Sr = 1.
Similar holds also for the pressure scale P (P also means ∆P). We can compare the
pressure term P/L with the other terms in (5.2.4) to express their similar order of magnitude. We
obtain the following scalings for the pressure:

unsteady scaling: P = ρLV/T,


convective scaling: P = ρV2,
viscous scaling: P = µV/L,
gravitational scaling: P = ρgL . (5.2.11)

To the governing equation (5.2.2), inevitably belong the initial and boundary conditions. The initial
conditions are given functions of the spatial distribution of the variables ρ0(x), v0(x), p0(x) in the
initial time. There is nothing to be scaled, they merely describe the flow field at the beginning of our
observation. On the other hand, the boundary conditions can be more complicated. They are
equations that the flow variables must obey at the boundary of the flow domain during the flow
observation. Because they are equations, they can be scaled. In single phase flow, the fluid is
bounded by a rigid wall where the no-slip condition applies, vwall = 0. It is a simple equation,
actually nothing to scale. In multiphase flow, there is a deformable fluid interface that separates the
two immiscible phases (e.g. air/water, water/oil, metal-1/metal-2, etc.). The interface possesses

12
certain surface energy that gives rise to the surface (interface) tension. The boundary condition
here is the free-slip condition, which is a complicated force balance at the interface. The interfacial
force is decomposed into the normal and tangent components.
By scaling the normal component we get a number that compares the fluid pressure and
the Laplace pressure produced by the interface tension, N = P/(σ/L), sometimes called the Laplace
number. Using the different scalings for P, the number N takes the following forms and names:

unsteady P: Un = ρL2V/σT , (unsteadiness/capillarity)


convective P: We = (∆)ρLV2/σ , (inertia/capillarity)
viscous P: Ca = µV/σ , (viscosity/capillarity)
gravitation P: Bo ≡ Eo = (∆)ρgL2/σ . (gravity or buoyancy/capillarity) (5.2.12)

Weber number compares the inertia and capillary forces. It applies to deformation of
bubbles and drops at free rise or on collisions with an obstacle, where dynamic effects are
important. We contains speed V and therefore applies to situations where the fluid particles moves,
not where are stationary (bubble columns, extractors, flotation, etc.).
Capillary number compares the viscous and capillary forces. It is relevant when viscous
forces dominate, e.g. in slow flows at small scales. The typical situation is the drainage of a thin
liquid film between two interfaces, at least one of them is fluid (to have σ in play), for instance,
interactions of fluid particles with themselves, with solids, and with walls (coalescence, bouncing,
adhesion), frequently encountered in bubble columns, flotation columns, extractors, etc.
Bond number (Eotvos number) is the ratio of gravity (ρ) or buoyancy (∆ρ) forces to the
capillary forces. The typical situation is deformation of a stagnant drop sitting on a horizontal plane,
where Bo compares the hydrostatic pressure across the drop (ρgL) with the capillary pressure
(σ/L) inside the drop. Naturally, the length scale L must be the drop size (not the bubble/flotation
column size!). From the equilibrium deformation of bubbles and drops, the static value of σ can be
obtained. Bo does not contain speed and therefore applies to situations where the fluid particles
are stationary, not where they move. Bo = 1 determines the capillary length Lcap = (σ/(∆)ρg)1/2,
where the liquid surface starts to elevate, near the wall.
There is also Morton number, Mo = We3/Re4.Fr, composed in such a way to contain the
material properties of fluids.
By scaling the tangential component we get a number that compares the surface gradient
of the interface tension ∇ Sσ and the fluid shear stress τ, N = ∇ Sσ/τ. The number N is called the
Marangoni number Ma:

Ma = (∆σ/L)/(µV/L) = (∆σ/µV) . (5.2.13)

It may be difficult to estimate the velocity scale V for complex motions near interfaces. Therefore,
often the diffusional scaling is used, V ~ (diffusivity)/L. Surface tension gradients are very important
for the capillary phenomena. The gradient is usually produced by uneven distribution of some
agent that affects the surface tension (the gradient is along the interface not from the interface to
the bulk). In practice, the agent usually is a surfactant or 'contaminant', or the temperature. The
surfactants are of two basic kinds. The organic compounds (detergents, tensides) are attracted to
the interface and reduce the tension. They are less polar than water and tend to nonpolar air. The
inorganic compounds (electrolytes, salts) are repelled from the interface and increase the tension.
They are highly polar and escape from the nonpolar air.
The uneven distribution of the agent along the interface cause a difference in the surface tension
along the interface. The difference results in interfacial stress, which is a force that drives the
motion of the near-surface fluid elements from the region of small tension to the region of large
tension. This interfacial flow is often directed against the bulk flow and slows it down. It exerts
strong effects in applications: reduction in terminal speed of bubbles and drops (hence large gas
holdup in bubble columns), suppression of the coalescence by slowing down the drainage process
(bubble columns, flotation, foams, extraction, etc.). These effects are difficult to study, even in lab
under well-define conditions. The primary information about the surfactant distribution along the
interface is practically inaccessible. Some aspects of the complex adsorption/desorption processes
are reflected by the dynamic surface tension. The tricky thing is that even a small amount of a
13
contaminant can produce strong surface gradients while the value of the static tension is
unchanged. Using the (static) surface tension σ in correlations and scale-up relations is usually
helpless or even misleading. Note that, under comparable conditions (same Re etc.), a bubble
behaves identically in both pure water and pure organic solvents. In contrast, it behaves very
differently in pure water (liquid) and slightly contaminated water (liquid).

For the practical considerations and successful scaling, one should know what are the sources of
the fluid motion. These terms are of a high importance and should not be scaled-off the equation.
They can often be used for dividing the other terms when making the equation dimensionless and
producing the relevant DN.
In the hydrodynamic equation (5.2.2), there is only one driving term, the gravity term (ρg)
with its rather limited sphere of action: down-hill flows at constant density (gravity flows, open
channel flows, etc.) and buoyancy driven flows at variable density (thermal convection, solutal
convection, halinoconvection, etc.).
In the initial condition, there can be the nonzero velocity field as the 'startup investment' of
motion. It conserves energy at the absence of viscosity (idealized situation, perpetum mobile). It
dissipate energy at the presence of viscosity (realistic situations). Accordingly, this term cannot
support the motion for long enough times. The characteristic time of the decay can be estimated by
comparing the unsteady term with the viscous term in (5.2.4), T ~ ρL2/µ. In modified formulation,
these two terms can form a velocity scale-relaxation equation: ρ(dV/dt) = -(µ/L2)V, yielding the
same relaxation time.
In the boundary conditions, there are several driving forces, with practical impact. First, the
pressure difference is imposed across the flow domain: it is the most common way how to make
fluids flow in technology (e.g. pipe flow). We need an external source of pressure (pumps).
Second, the boundary or its part is moving, which is commonly employed in applications. In
mechanical mixing, the propeller is the moving boundary. In pneumatic mixing, the rising bubbles,
as the internal moving boundaries, generate the liquid motion. We need an external source of
speed (moving element(s)). While the pressure driving is more common in internal flows (inside
confining walls) the velocity driving can be expected in external flows (flows past bodies).

For applications, it is helpful to know the possible importance of the five terms in the hydrodynamic
equation (5.2.2) in particular flow situations, which we can encounter in our technology. This helps
to choose the appropriated DN to describe the system and to build the scale-up correlations.
One division is along the configuration of the flow domain and the flow boundary. The low
Re flow in pipes is a typical representative of internal flows, where we usually consider a balance
(i.e. equilibrium) between the pressure driving force and the viscous resistance force, i.e. the terms
3 and 4 should be comparable. The flow over an obstacle placed in a big (ideally infinite) domain
represents the typical external flow. Here, the (ideally uniform) convected stream interacts with the
body, and its main speed must accommodate to the no-slip condition at the body surface. It
happens in the boundary layer with help of severe speed gradients, whence viscosity effects. Thus
the terms 2 and 4 interplay, the former dominates far from the body, the latter near the body.
The turbulence is often present in high Re flows, in many applications. Three terms are
typical. The nonlinear inertia term 2 give birth to the energy transfer between the various spatial
scales (energy cascade), which is the very essence of the turbulence and the very root of its
complexity. Since this term endlessly spreads the energy on and on over all possible scales (which
is beautifully seen in the spectral dynamics in the Fourier space), the flow field is essentially
unsteady and the term 1 must be included. With the terms 1 and 2 we have the ideal Eulerian
turbulence, important in the theoretical studies but rather unrealistic (Re = ∞). Therefore, we must
add also the dissipative viscous term 4, to have a real picture and finite Re. Similarly like with the
boundary layer, the inertia term 2 dominates in one range (large scales or 'eddies') while the
viscous term 4 in the other (small scales or 'eddies'), separated by the Kolmogorov scale. The
three terms 1,2,4 are basically enough to produce a valid turbulence, but the other two may be
included (3 in internal pressure-driven flows, 5 in gravity driven flows). Note that the notion 'steady
turbulence' is only an engineering approximation, assuming the fully developed turbulence and
approaching it with the statistical methods, as it would be a random process (Reynolds' averaged
equations). This is a great help in practice, but cannot replace the physical insight needed for

14
choosing the proper DN. Note that turbulence is a typical 3D phenomenon (2D turbulence is
perverse - no vortex stretching for spatial reasons).
In unsteady flows, the unsteady term 1 is relevant. Depending on the specific situation, it is
combined with other terms. In starting up/restarting a production line, this term determines the
time scale estimate of the transient period, which is a valuable practical quantity. Also, at the
change of the control parameters, it can say how long will the system settle down in the new
regime. On switching off, the relaxation time for reaching the state of rest can be obtained from the
this term too.
The gravity term 5 can be important for gravity driven flows (downhill flows) and buoyancy
driven flows (nonuniform density). Likewise the unsteady term, also this term must be combined
with other terms. For instance, in open channel flows one can assume the balance between the
driving gravity 5 and the resisting viscosity 4. By adjusting the slope properly, a steady flow may
result. Generally, the convective term 2 comes into play, as well as the unsteady term 1
(nonuniform depth, obstacles, flow past a step, hydraulic jump, etc.). Another class comprises the
flows either fully driven or only influenced (co-driven) by the density difference. In applications,
joining two streams of water, one is clear and one is a dense suspension (e.g. ore particles) may
cause vigorous mixing. The thermally driven convection starts not at an arbitrary small temperature
difference (or gradient), but at some critical one, given by the stability analysis (e.g. Rayleigh-
Benard problem). Strong thermal gradients are encountered in metallurgy. Needles to say that the
configuration must be unstable, i.e. top-heavy layering.
Besides the five terms of the governing equation (5.2.2) and their DN, other effects can
come from the boundary conditions, and may be important under certain circumstances. We
introduced few DN related to the deformable boundary between two immiscible fluids (Ca, Bo, We,
Ma). Their physical meaning was briefly mentioned in that section. In practical situations, we can
use them in many multiphase flows with a variety of dispersed particles and free surfaces. For
instance, when mixing a batch pneumatically with the stream of bubbles (aeration tanks in
wastewater plants, mixing liquid metals, ore flotation, etc.), the bubble behavior will likely depend
on the character of the flow past it (inertial/viscous, Re), on the bubble shape and its dynamic
deformation (fluid pressure/capillary pressure, We), and on the surface tension gradients (Ma),
since under real operating conditions all liquids contain 'contaminants'. The flotation efficiency may
depend on similar DN.

This Section 5 on the momentum transfer is of larger extent than those dealing with the heat (Sec.
6) and mass (Sec. 7) transfer. One reason is that the momentum transport (i.e. fluid flow) is the
most complex since the momentum is a vectorial quantity, while heat and mass are mere scalars.
Second, the hydrodynamics is the most important of these three, because it is explicitly involved in
the other two: see the convection term in heat and mass balances. On the contrary, the effect of
heat and mass phenomena on the hydrodynamics is not explicitly involved in the hydrodynamic
equation (5.2.2). These two affect the flow only implicitly, e.g. by variations in fluid density and
viscosity. In gas flows, the fluid pressure is affected too, through the state equation.
The mutual coupling between the flow, heat, mass and reaction phenomena is depicted in
Fig. 1. The Momentum exports the velocity (v) into the both heat and mass balances, where it
produces the convection terms. Besides, it contributes to the total energy balance
(thermodynamical + mechanical) by the heat generated by the viscous dissipation in the fluid
(ediss). On the other hand, the Momentum is affected by the density variations (∆ρ) caused by the
temperature variations (∆T) in non-isothermic systems, and by the concentration variations (∆c),
provided they are strong enough to drive the fluid motion.
The Heat imports from the fluid the important convection term (v) and the dissipated energy
(ediss). From the Reaction, it obtains the source/sink term of the reaction heat (enthalpy), depending
on its thermal character (exothermic/endothermic). As a feedback, the Heat affects the Reaction
with the overall temperature variations in the environment.
The Mass imports from the fluid the important convection term (v). From the Reaction, it
obtains the source/sink term of the mass, depending on the particular components
(product/reactants). As a feedback, the Mass affects the Reaction with the overall concentration
variations in the environment.
If all the transport and reaction phenomena are presented and strongly intertwined, the
solution to our problem is a formidable task. We have to solve the three balances simultaneously
15
numerically, which is very difficult even nowadays when strong computer power is available,
especially when we want to have a fine resolution in space and time, and the situation should be
close to reality. Consequently, any possible simplification made by the scaling analysis is of great
help. Neglecting the less important phenomena against the few dominant ones make it possible to
describe the system with few DN only, to a good approximation.

6. Balance of heat

In a simple form, often used in engineering, the heat balance involves only the two basic
mechanisms of the heat transport, the conduction and the convection. The first one, also called
heat diffusion, occurs in any material environment, both stagnant and moving (i.e. flowing), where
there are spatial variations in the temperature. In a stagnant environment (v = 0), the diffusion is
the only transport mechanism. The second one, occurs only in the moving environment, where the
speed v is obtained from the hydrodynamic equation, see the Momentum → Heat coupling in Fig.
1. The balance reads:
(∂θ/∂t) + (v. ∇ )θ = κ ∇ 2θ . (6.1)
The physical meaning of the three individual terms is the following:
(unsteadiness) + (convection) = (diffusion) ,
and their role is easily seen. We perform the scaling of the heat balance. Choosing the variables
(θ, v, x, t), their scales (Θ, V, L, T), and the parameters (κ), we get by scaling the dimensional
scale equation:
Θ/T + VΘ/L = κΘ/L2 . (6.2)
Dividing by Θ, the temperature scale disappears, and we have:
1/T + V/L = κ/L2 . (6.3)
Dividing the three terms by the diffusion term (κ/L2), because of tradition and also to keep it, we
have the dimensionless scale equation:
L2/κT + LV/κ = 1 , (6.4)
with only two DN (the diffusion term is unity). The two DN have proper names:
Fo + Pe = 1 . (6.5)
The two DN are relevant for the simplified heat transport and have a clear physical meaning in
terms of the process rates:

Fourier number, Fo = L2/κT, (unsteadiness)/(heat diffusion)


Peclet number, Pe = LV/κ , (hydrodynamic convection)/(heat diffusion) . (6.6)

These numbers can be interpreted in many different ways, in terms of many different quantities,
e.g. time scales, length scales, etc. Note that we use only one common length scale L for all terms
and for all spatial directions, which is acceptable for the demonstration purposes. In treating a
particular problem, this may not be true. For instance, the convection and diffusion length scales
may be different. Then, Peclet would contain (Ldiff2/Lconv).
When using the basic time scaling, T = L/V, the Fourier number becomes the Peclet
number, and the unsteady and convective effects are then comparable, as a direct consequence of
our choice of scaling. This may not correspond to reality. The Peclet number facilitates the
coupling between the hydrodynamics and heat transfer, see the Momentum → Heat coupling in
Fig. 1. It compares the transport by hydrodynamic convection (flow of medium, V) and the
molecular diffusion of heat (heat diffusivity κ).
Since we customarily divided (6.2) by the diffusion term, both DN are related to the diffusion
rate. If we want to compare the importance of other effects, we must create new DN by combining
the original ones. For instance:

(unsteadiness)/(convection) = Fo/Pe = Sr = L/TV. (6.7)

Few frequent combinations are shown here, involving the heat and hydrodynamic effects. The
Prandtl number,

16
Pr = Pe/Re = ν/κ , (momentum diffusion)/(heat diffusion) , (6.8)

compares two material properties, the diffusivity of momentum and that of heat. It is often used in
thermal processes to show how much the flow is affected by heat diffusion (thermal convection,
boiling, etc.). At large Pr, the flow field evolves faster than it is affected by the heat penetration. At
small Pr, the thermal effects are important and may affect the hydrodynamics.
The heat Grashof number stems from the Archimedes number, where the general density
difference ∆ρ is replaced with its specific cause, the temperature difference ∆Θ,

Gr = α∆ΘgL3/ν2 , (buoyancy force)/(viscous force) (6.9)

which is typical for heat-driven buoyancy effects in fluids. It is basically a hydrodynamic number,
where the heat enters as the 'buoyant agent' and causes density gradients, see the Heat →
Momentum coupling in Fig. 1.
The thermal Rayleigh is closely related to Gr,

Ra = Gr.Pr = α∆ΘgL3/νκ , (buoyancy force)/(viscous force) (6.10)

and is crucial for the onset of the thermal convection and its evolution via a series of bifurcations. It
compares the driving thermal disturbance ∆Θ acting on a parcel of fluid with the rates of the
transport processes who tend to smear it out (diffusion of momentum ν and heat κ).

The thermal boundary conditions usually are of two kinds. Either the temperature θw is given at the
boundary or the heat flux jw through it. The former case is trivial, θw/Θ. The latter one, in case of
heat conduction, is given by:
jw = - λ( ∇ θ)w , (6.11)
where λ is the heat conductivity. To calculate the flux, the temperature field must be known to
evaluate the wall gradient, which is difficult. Therefore, another expression for the flux is
introduced, with help of the empirical heat transfer coefficient kh:
jw = kh (∆θ)w , (6.12)
where (∆θ)w is the bulk-boundary temperature difference. Equating (6.11) and (6.12) yields the
formula for the coefficient:
kh = - λ( ∇ θ)w /(∆θ)w . (6.13)
A simple scaling of (6.13) by (λ/L) leads to the Nusselt number (also Biot number),

Nu = kh /(λ/L) , (6.14)

which is the dimensionless heat transfer coefficient (the temperature scales Θ cancel). The length
scale L should relate to the thickness of the thermal boundary layer. Nu it is the most desired
quantity in the heat transfer, since it contains the valuable Pischwejtz coefficient kh. Therefore,
numerous correlations for Nu are available in the engineering literature.
Another situation occurs when we have the heat radiation at the boundary. The heat flux is
given by the Stefan-Boltzmann equation,
jw = εσθw4 , (6.15)
where ε is the relative radiation (real body vs. ideal black body), σ is the Stefan-Boltzmann
constant, and θw is the wall temperature. At scaling, a proper temperature scale Θ should be
chosen.

So far, a very simple form of the heat equation (6.1) was considered. One extension is to add the
heat generated/consumed by a reaction. The reaction term q is the volumetric source/sink of the
energy [J/m3.s]. We can choose a scale Q for q, to obtain an extended version of (6.2):
Θ/T + VΘ/L = κΘ/L2 ± Q , (6.16)
which is not linear in Θ so that the temperature scale cannot be removed by dividing. A new DN
appears, denoted here as:

17
heat Damkohler number, Da = Q/(κΘ/L2) , (heat by reaction )/(heat by diffusion) (6.17)

which facilitates the Reaction → Heat coupling in Fig. 1. The term q depends on the reaction
energy ∆E taken from the thermodynamic data and on the reaction rate r. The reaction rate r is
given by a kinetic closure equation and depends on the concentration of the reacting component c
taken from the mass balance (together with its scales) and on the kinetic constant k. The kinetic
constant k is given by a closure equation and depends on the temperature θ and possibly also on
the pressure p, for a reaction in the gas phase (a new direct Reaction ↔ Momentum pressure-
coupling not shown in Fig. 1; e.g. combustion, explosion, etc.). The relation between k and θ is
typically exponential, k ~ Exp(-a/θ) (Arrhenius equation). If the reaction heat scale Q is of a similar
form as q, we have Da ~ Exp(-a/Θ)/Θ. It is a problem, since the proper DN should be a ratio of
power monomials, i.e. ratio of products of simple powers of individual quantities. Such an improper
DN cannot be obtained by DA, because the physical dimensions are the power monomials. It can
however be obtained by SE, as we can see here.

So far, we did not consider the energy balance of the carrying fluid. In its simplest form with an
ideal fluid, it is the Bernoulli equation:
(1/2)ρv2 + p + ρgh = const. (6.18)
The physical meaning of the three individual terms is the following:
(convection) + (pressure) + (gravity) = const.
Choosing the variables (v, p, h), their scales (V, P, L), and the parameters (ρ, g), we get by scaling
the dimensional scale equation:
ρV2 + P + ρgL = const. (6.19)
Here, we can easily see the convective and gravitational scaling for the pressure, P ~ ρV2 and P ~
ρgL. Dividing by the convective term (ρV2), because of tradition, we get:
1 + P/ρV2 + gL/V2 = const. (6.20)
Assigning the DN their proper names we have:
1 + Eu + 1/Fr = const. , (6.21)
which is the counterpart of the steady and inviscid equation (5.2.4), so only Eu and Fr are present.
In engineering applications, with real fluids, the r.h.s. of (6.18) is not a constant but the energy
dissipated within the flow, Ediss. Accordingly, a new DN would appear, Ediss/(ρV2). This rather
coarse approximation can be very useful in solving practical problems. However, a more detailed
mechanical energy balance can be obtained from the full fluid momentum equation.

Another extension is to consider the full energy balance, where we have the simple heat transport,
the reaction energy effects, and the contribution of the mechanical energy from the moving
continuum, one part of which - the dissipative heat, facilitates the Momentum → Heat coupling in
Fig. 1. The equation is rather complicated and is not presented here. The interested reader can
find it easily in the literature and perform its scaling, in case of need.

7. Balance of mass

The mass conservation equation is written for a one component, one chemical compound, that is
dissolved in the carrying fluid in the form of a true chemical solution, i.e. is microscopically mixed
with the molecules constituting the fluid. Like with heat, there are two basic mechanisms of the
mass transport, the conduction and the convection. The first one, mass diffusion, occurs in any
material environment, both stagnant and moving (i.e. flowing), where there are spatial variations in
the concentration. In a stagnant environment (v = 0), the diffusion is the only transport mechanism.
The second one, occurs only in the moving environment, where the speed v is obtained from the
hydrodynamic equation, see the Momentum → Mass coupling in Fig. 1. Like with heat, there is a
reaction term r [mole or kg/m3.s] due to the possible reaction, the volumetric source/sink of the
mass of the component, which can be produced/consumed. The balance reads:
(∂c/∂t) + (v. ∇ )c = D ∇ 2c + r . (7.1)
The physical meaning of the four individual terms is the following:

18
(unsteadiness) + (convection) = (diffusion) + (reaction) ,
and their role is easily seen. We perform the scaling of the mass balance. Choosing the variables
(c, v, r, x, t), their scales (C, V, R, L, T), and the parameters (D), we get by scaling the dimensional
scale equation:
C/T + VC/L = DC/L2 + R . (7.2)
Dividing the four terms by the diffusion term (DC/L2), because of tradition and also to keep it, we
have the dimensionless scale equation:
L2/DT + LV/D = 1 + RL2/DC , (7.3)
with only three DN (the diffusion term is unity). The three DN have the proper names:
Fo + Pe = 1 + Da2 . (7.4)
The three DN are relevant for the mass transport and have a clear physical meaning in terms of
the process rates:

Fourier number, Fo = L2/DT, (unsteadiness)/(mass diffusion)


Peclet number, Pe = LV/D, (hydrodynamic convection)/(mass diffusion)
Damkohler number, Da = (R/(DC/L2))1/2 , (mass by reaction)/(mass by diffusion) . (7.5)

These numbers can be interpreted in many different ways, in terms of many different quantities,
e.g. time scales, length scales, etc. Note that we use only one common length scale L for all terms
and for all spatial directions, which is acceptable for the demonstration purposes. In treating a
particular problem, this may not be true. For instance, the convection and diffusion length scales
may be different. Then, Peclet would contain (Ldiff2/Lconv).
When using the basic time scaling, T = L/V, the Fourier number becomes the Peclet
number, Fo = Pe, and the unsteady and convective effects are then comparable, as a direct
consequence of our choice of scaling. This may not correspond to reality. The Peclet number
facilitates the coupling between the hydrodynamics and the mass transfer, see the Momentum →
Mass coupling in Fig. 1. It compares the transport by the hydrodynamic convection (flow of the
medium, V) and the molecular diffusion of mass (mass diffusivity D). The Damkohler number
compares the rate of reaction with the rate of diffusion, and facilitates the Reaction → Heat
coupling in Fig. 1.
Since we customarily divided (7.2) by the diffusion term, both DN are related to the diffusion
rate. If we want to compare the importance of other effects, we must create new DN by combining
the original ones. For instance we can form:

(unsteadiness)/(convection) = Fo/Pe = Sr = L/TV,


(unsteadiness)/(reaction) = Fo/Da ,
(convection)/(reaction) = Pe/Da . (7.6)

Few frequent combinations are shown here, involving the concentration and the hydrodynamic
effects. The Schmidt number,

Sc = Pe/Re = ν/D , (momentum diffusion)/(mass diffusion) (7.7)

compares two material properties, the diffusivity of momentum and mass (also called mass or
diffusion Prandtl number). It is often used in mass processes to show how much the flow is
affected by mass diffusion. At large Sc, the flow field evolves faster than it is affected by the
diffusion of the dissolved component. At small Pr, the mass diffusion effects are important and may
affect the hydrodynamics.
The heat Grashof number stems from the Archimedes number, where the general density
difference ∆ρ is replaced with its specific cause, the concentration difference ∆C,

Grashof number, Gr = β∆CgL3/ν2 , (buoyancy force)/(viscous force) (7.8)

which is typical for mass-driven buoyancy effects in fluids. It is basically a hydrodynamic number,
where the mass enters as the 'buoyant agent' and causes the density gradients, see the Mass →
Momentum coupling in Fig. 1.
The concentration (salinity, mass) Rayleigh number is closely related to Gr,
19
Ra = Gr.Pr = β∆CgL3/νD , (buoyancy force)/(viscous force) (7.9)

and is crucial for the onset of the concentration-driven convection. It compares the driving
concentration disturbance ∆C acting on a parcel of fluid with the rates of the transport processes
who tend to smear it out (diffusion of momentum ν and mass D).

In heterogeneous catalysis, where the diffusion and reaction interplay, specific DN are employed.
Solving the corresponding equation in case of a model situation (a cylindrical pore in a catalyst, a
spherical pellet), we obtain the concentration profile c(x), whose mean value cm normalized by the
bulk concentration c0 is the effectiveness factor φ = cm/c0, which also is ~ (mean reaction
rate)/(maximum rate). The model solution for a pore gives φ ~ Tanh(Th)/Th, where the Thiele
number (modulus) is Th = L(k/D)1/2. Here, L is the pore length and k the rate constant. Note that for
r = kc, the Damkohler number coincides with the Thiele number, Da = Th. Often, a 'generalized' Th
is introduced, to retain the last equality also for reactions of higher orders. Reaction and mass
transfer is combined in the Hatta number. There also are numbers typical for the reaction kinetics
itself. For instance, the Arrhenius number compares the activation energy and kinetic energy of
molecules.

The boundary conditions in the mass transfer usually are of two kinds. Either the concentration cw
is given at the boundary or the mass flux jw through it. The former case is trivial, cw/C. The latter
one, in case of the mass conduction, is given by:
jw = - D ( ∇ c)w , (7.10)
where D is the mass diffusivity. To calculate the flux, the concentration field must be known to
evaluate the wall gradient, which is difficult. Therefore, another expression for the flux is
introduced, with help of the empirical mass transfer coefficient km:
jw = km (∆c)w , (7.11)
where (∆c)w is the bulk-boundary concentration difference. Equating (7.10) and (7.11) yields the
formula for the coefficient:
km = - D( ∇ c)w /(∆c)w . (7.12)
A simple scaling of (7.12) by (D/L) leads to the Sherwood number Sh (also Sherman number),

Sh = km /(D/L) , (7.13)

which is the dimensionless mass transfer coefficient (the concentration scales C cancel). The
length scale L should relate to the thickness of the mass diffusion boundary layer. Sh it is the most
desired quantity in the heat transfer, since it contains the valuable Pischwejtz coefficient km.
Therefore, numerous correlations for Sh are available in the engineering literature.
Another situation occurs when we have the phase change at the boundary, and the
corresponding mass flux is to be found, which may be difficult.

8. Other processes, effects, and DN related to metallurgy

When injecting gas into molted metals at high speed and pressure, for instance for mixing and
reacting purposes, the compressibility effects may be important. They are reflected e.g. by the
Mach number, which appears in various correlations.
When working with systems of complex rheology, non-Newtonian materials, new
phenomena come to play (elasticity, plasticity, memory, etc.). There are specific DN used in
rheology to accounts for these complicated effects (e.g. Bingham number, Deborah number,
Weissenberg number, etc.).
Molten metals are conductive fluids and interact with external electric and magnetic fields.
Additional DN are used to describe these effects (e.g. Alfven number).
Flow of rarified gasses and diffusion in porous media require a special treatment, since the
continuum hypothesis may not be fully applicable. The Knudsen number then compares the mean

20
free path of the fluid molecules with the flow domain size (e.g. pore). The fluid tends to be
'continuous' when the former is much smaller than the latter.
When the fluid contains particles that are not molecules of a chemical compound but
individual bodies much larger than the fluid molecules, we have a dispersion. For very fine
particles, e.g. ~ submicron or micron size, where the particle inertia is negligible and the gravity
forces are weak, we have a colloidal dispersion. Here, the stochastic Brownian motion is the
important physical mechanism. For bigger particles, we have the multiphase hydrodynamics that
deals with dispersions of solids, drops and bubbles in fluids. In nature, almost all flows are
multiphase. In technology, most of them are multiphase. This difficult field presents an active
research area with its own nomenclature, jargon, methods and approaches. There are many
typical multiphase operations in the chemical engineering, which can be relevant also for
metallurgy (separation by sedimentation, flotation, filtration, extraction; pneumatic mixing, mass
transport, reaction and mixing in gas liquid systems - bubble columns, airlifts, etc.). There are
many specific DN and many correlations for the target quantities in the engineering literature, to
describe the system, to correlated experimental data, and to perform the scale-up.
An important issue is the retention time distribution in the technological apparatuses and
equipments, which is well elaborated in the chemical engineering. There are two ideal types of the
flow through the operating units (ideal mixing, piston flow). These are combined as the main
building blocs into schemes of various complexity, to reflect the real flows and to fit the
experimental tracer data (axial dispersion, Peclet number). Newly, the fluid age can be considered
as the field variable and a simple transport equation for it can be written. The solution gives us the
space-time distribution of the fluid age inside the equipment.
In metallurgy, there are important operations with the solid phase, e.g. grinding, mixing,
separation, transport, manipulations, etc. It seems that these very complex processes have not
been systematically described and treated fundamentally, from point of the scaling analysis.
Presently, the science of the granular media becomes fashionable, with new insight from the basic
physics. However, there is a long tradition in this area, thanks to the research in the soil
mechanics, building industries, and engineers working with powders and grains.
Under given conditions, the system or process settles in a certain regime of behavior, which
can be described by a set of DN, correlation relations, etc. The important question is the regime
stability. If the regime occurs near the critical point, the threshold of its instability, the ever-present
fluctuations in the state variables and in the values of the control parameters may cause its
unwanted transition into another regime of operation. To prevent this, one should know the
regimes' map in the space of the operational parameters, and to keep the process far enough from
the regime boundaries. Here, the dimensional analysis is of a little help. The methods of trial and
error is rather unsuitable for a full-scale operation unit, to find the safe values of the control
parameters. One way is to do experiments on a small-scale lab model, assuming the similarity.
The other way is to perform the stability analysis on the governing equations. However, it can be a
difficult task and the theoretical results obtained under simplified assumptions may be of a little
relevance for the real process. Nevertheless, when successful, the both ways lead to the critical
values of DN that separate the neighboring regimes (e.g. critical Re for laminar-turbulent flow
regime transition).

Part II. Application: Processes and Operations in Metallurgy

9. Metallurgy - art and technology

Metallurgy is a very old art [Genesis 4:22]. When science of chemical metallurgy and technology
emerged in the middle of 19th century, large industrial metallurgy plants which had been developed
quite empirically were successfully operating. During the following 150 years, the metallurgy
science was more explanatory than predictive. It contributed significantly to better understanding of
the structure of alloys depending on composition and thermal treatment, and to development of
new advanced materials. However, it has not contributed revolutionarily to the technology
development. For example, the blast furnace equipment has not changed essentially and in
21
steelmaking there appeared only three principal improvements – Bessemer converter, Siemens-
Martin furnace and the oxygen process. In this period, a satisfactory way of scaling up was based
on increasing size of these equipments in small steps not exceeding few percent. This contrasts
with the chemical industry, where the innovation period is 5-10 years and the scale up procedure
from test tube to the industrial size comes either through one pilot plant step or directly.
Nevertheless, we would like to point out that the dimensional analysis and scaling of the
equations can still contribute to better understanding of the metallurgy processes and can be
helpful for proper choice of laboratory experiments (scale-down) providing through physical
modeling reliable information on actual technology and on its optimum performance.

There are several classes of the material treatment in metallurgy:

 Front end operations:


• Size reduction - crushing and grinding
• Classification, sieves
• Mechanical separation (sedimentation, froth flotation, magnetic separation)
• Agglomeration, granulation, briquetting

 Pyrometallurgy
• Drying
• Calcination
• Pyrolysis
• Roasting
• Smelting
• Converters
• Substitution of metals

 Hydrometallurgy
• Suspension processes
• Leaching by water, acids, bases
• Filtration, sedimentation
• Crystallization, precipitation
• Liquid extraction, ion exchange, membrane processes

 Electrometallurgy
• Electrolysis
• Electric heating, plasma

 Liquid metals
• Storage and transport
• Stirring
• Sedimentation, inclusions, bubbles, vacuum process
• Casting

 Solid metals
• Material engineering
• Forging

10. Front end operations

Front end operations are mechanical operations with raw materials of metallurgy, processing of
ore.

22
Size reduction - crushing and grinding with aim to reduce mineral material to manageable
size and to liberate particles of important metal compounds from the waste rock (gangue).
Essential quantitative concept is the energy W required for size reduction of given volume V. It is
theoretically proportional to the energy ES for formation of the unit area of a new surface and
inversely proportional to the diameter L of particles. Therefore efficiency of the comminuition is
1/Kr , where the Rittinger number is
Kr = (W/V) L/ES . (10.1)
The main problem is that the surface energy ES is related to a number of more or less objective
quantities as yield strength, brittleness, propagation of cracks etc. and it depends significantly on
the way of crushing or grinding and on the equipment applied.

Classification (sieving) – scale-up problem is trivial.

Mechanical separation from the waste by:


- Sedimentation - using the difference in densities.
- Laws of sedimentation are given by relation of Fr and Re. Specific combinations of the
variables are sometimes useful, to eliminate the velocity (e.g. Archimedes number Ar ~
Re2/Fr) or to eliminate the particle size (Lyaschenko number Ly ~ Re.Fr).
- Flotation – using different surface properties depending on the chemical composition of the
surfaces froth: concentrated minerals are usually water-repellent, the tailing is low of metal
minerals and is retained as a sediment.
- Magnetic separation - these processes are controlled by the forces having analogous
effects on both the laboratory and industrial scales. The scale-up of such processes by
dimensionless numbers is less efficient.

Granulation, briquetting and mechanical agglomeration are, again, the mechanical


processes and their scale up from the laboratory to the industrial equipments is obvious.

11. Pyrometallurgy

11.1 Basic procedures

The pyrometallurgy comprises several classical techniques of the metal preparation by the effect of
heating, and a large number of the modern procedures is carried out at high temperatures. From
the viewpoint of the metallurgy, the low temperatures are about 200-500°C, the most of processes
dealing with the common non-iron metals (copper, lead, zinc, tin, etc.) work at about 500-1000°C,
the contemporary iron metallurgy operates at about 1500-2000°C, and some special metals (W,
Os, Pt, etc.) do require even higher temperatures.
In any case, the thermal treatment brings on the chemical and/or the phase transitions of
the materials. A high temperature is usually required because the corresponding chemical
processes are very slow at the room temperature and, moreover, only the high temperature makes
some endothermic equilibrium reactions possible, with reliable yield of the products. This is
quantitatively described by the thermodynamics, and it is well known which range of temperature is
suitable for the particular metal production and purification.

Drying is an introductory step eliminating the free water from the subsequent reactions. It
employs lower temperature heating, which spares energy. The rotary kiln is the most common
equipment for the continuous drying of particulate matter. Another options are the fixed bed dryer
for coarse particles and the fluid bed dryer for smaller particles of well-defined size. Unlike the
drying of the food products, here is no problem with the temperature sensitivity, so that the flue
gases can directly be used as the drying medium.
Water is present mostly in the form of the free moisture, kept on the particle surface by
surface forces (Bond number). More complicated is the presence of the moisture in pores, where
the characteristic size of the pores or cracks should be introduced when evaluating Bo. The heat

23
transfer is controlled mostly by the heat conduction in particles, characterized by the Fourier
number.

Calcination takes place at higher temperatures. This term was originally used for the
thermal decomposition of limestone at 900°C in the lime manufacture,
CaCO3 → CO2 + CaO .
Now, this term is more generally applied to any process of thermal decomposition accompanied
with the evolution of a gas. In metallurgy, it covers mostly removing of CO2 from carbonate rocks or
the bonded water from various minerals. The bonded water in hydroxides or the water in hydrated
crystals appear also in various downstream products of the hydrometallurgy and should be
eliminated by the calcination prior the following pyrometallurgy treatment. Similarly are treated
some final products e.g. pigments like TiO2, Fe2O3 or microcrystalline materials like alumina Al2O3,

2Al(OH)3 → 3H2O + Al2O3 .


The shaft furnace or the rotary kiln are the common equipments for the continuous calcination of
minerals. Special low tonnage materials are calcinated periodically in the muffle furnaces. The
thermal conduction and the kinetics of the decomposition control the process, and the
characteristic times typically are tenth of hours. When nanostructured products are demandable,
we cannot speed up this process by a considerably higher temperature, because of the danger of
the material sintering.

Pyrolysis is an auxiliary process in metallurgy. One aim of the indirect heating of the
carbonaceous materials (historically - wood, in contemporary technology - hard coal) is to prepare
a carbon material (coke) of a high caloric capacity, suitable as a heating and oxygen reducing
medium for various metallurgy processes. The retort furnace with the indirect heating is the usual
equipment. The pyrolysis of coal dust also takes place in the agglomeration of fine particles of
minerals by heating with the coal powder to obtain pellets. In any case, this is a conduction heating
of a solid bed, slowly transported through the heating zone, usually accompanied by the evolution
of volatiles.

Roasting (heating, including reaction with gas environment). The mechanical mixing of the
treated granular material is feasible in rotary kilns or multiple shaft kilns. For the materials in form
of small uniform particles, the fluid bed roasting can be applied too. The heat transfer from the
flowing gas to the solid particles controls the process. An important issue is the reaction heat of the
reaction. For the oxidative roasting with O2, a significant amount of heat is evolved; less heat
comes from the oxidation by CO2. The reductive heating by CO or H2 is endothermic. In the last
years, the direct reduction of the iron ore is considered as an alternative to the classic blast
furnace. The chloride metallurgy use also the compounds like Cl2, F2, HCl, HF as a gaseous
medium at various temperatures. The gaseous products like the halogenides of Ti, Sn, U and
others are obtained at a high purity. Subsequently, these metals can be liberated from the
halogenides by less electronegative elements.

Smelting (reaction with gas followed by phase transition) may include the formation of
liquid metal, matte, slag, in some cases also volatile metals or volatile salts. Therefore, a fixed bed
of the material is the usual starting point of the process. Traditional smelting equipment developed
over centuries quite empirically is the blast furnace, where the bed of feed moves slowly down due
to the burnout of coke and drainage of iron and slag. The melted products are collected in the blast
furnace hearth, separated, and periodically discharged.
Typical arrangement for smelting in non-iron metallurgy is an open hearth heated by the flue gas
from upside. The batch may contain different immiscible components of different density in several
layers. The most common case is a molten metal covered by a layer of oxides either in a form of
solid particles of as a molten slag. In the smelting processes starting with sulfide minerals the layer
of matte (molten sulfide like Cu2S) appears. Under certain circumstances, there may be more
immiscible layers, e.g. slag, matte, and one or more immiscible layers of liquid metals.
The simple smelting is a downstream operation for preparation of Fe from iron bloom or Cu
from blister copper. Small scale smelting also occurs at the torch welding. Other operations with
melted metals occur in the tin or zinc plating.
24
Converters are special smelting equipments, where a fast exothermic oxidation takes
place. Historically, it occurred during the puddling (reaction of carbon from steel with air
environment),
2C + O2  2CO .
A similar process was later popular in the Siemens-Martin open hearth reverberatory
furnaces where the oxidation medium was CO2 from the hot flue gas,
C + CO2 2CO ,
and the ore was added as a powder,
C + FeO  Fe + CO .
The important inventions were the Bessemer converter in 1868, with the pressure air blown
through the bottom of the ladle with the molten pig iron, and the application of the pure oxygen
since 1962. Removing of the other elements like Si, P, S from the steel is enabled by the reactions
with the basic refractory material or with the lime powder introduced to the ladle with the oxygen,
4P + 5O2 + 6CaO  2Ca3(PO4)2 , etc.
Other converters are used in the non-iron metallurgy for the conversion of matte to metal,
e.g. in the copper production, by the summary reaction
Cu2S + O2  Cu + SO2 .
A highly exothermic reaction in converters is also employed to the melting of recycled metal
scrap where the heat of the fusion is transported to the scrap pieces by the convection of the liquid
metal. When the liquid steel is not at hand, the melting of the scrap should be initiated by the
electric arc.
Both the mass and the heat transfer between the flowing gas and the stationary level take
place in the open hearth furnaces. In the converters, the superficial gas velocity is extremely high
and the classical scale up procedures for bubbled beds are inapplicable.

Pyrometallurgy substitution of metals is a reaction of some specific elements at a high


enough temperature with a compound of a metal with the goal to liberate the elementary metal in
the liquid or vaporized states. These reactions are usually exothermic; nevertheless they have to
be initiated by a local overheating.
• Carbon is the most popular driving element, liberating namely the iron from the iron oxides
in the blast furnaces. It is also used for the preparation of Si, Mn, P, H2, Nb, Ta, Mg from
their oxides.
• Hydrogen is used at the reduction of W, Mo, Ge oxides.
• Silicone or ferrosilicium can liberate Mg or V.
• Calcium can substitute V, Zr, U, Th, Nb, Ta in their compounds.
• Sodium can liberate also K and Ca
• Aluminium is applied to produce Mn, Nb, Ta, V.
The corresponding reactions are quite fast and the stoichiometry of the reactions is trivial.

11.2 Dimensionless numbers

From the viewpoint of the scaling, all these processes are essentially the chemical reactions
controlled by the heat transfer. There are several options how to introduce the heat into the
process:
• Heat transmission from the wall to stationary or moving bed of granular material;
- Nusselt number for heat transmission through the wall.
- Fourier number for heat conduction through the bed consisted of continuous gas and
dispersed solid particles.
- Grashof number for possible natural convection in the gas.
• Heat introduced to the bed of granular material by hot gas (usually by the flue gas);
- Nusselt number for convective flow around a particle.
- Fourier number for heat conduction inside the particle.
• Heat introduced through the level of melt;
- Nusselt number for convective flow around a plate.
- Fourier number for heat conduction in the liquid layer or multilayer.
25
• In high temperature processes, the heat radiation becomes significant or predominant. The
heat introduced by the flowing gas is transmitted to the material either directly or through the
reflection of refractory (reverberatory ovens). However, the effect of the radiation is significantly
dependent on the absolute temperature of gases and surfaces and it can hardly be modeled at
different temperatures. Any dimensionless number does not appear to be essentially useful in
the radiation scaling-up.
• Significant (positive or negative) contribution of chemical reactions to the heat balance;
- Damkohler number.
• Refractory wear and dissolution;
- Sherwood number as a function of Schmidt and Reynolds numbers.
• Chemical reactions in pyrometallurgy are non-linear from view of temperature; the temperature
can affect non-linearly direction of the reactions and the reaction rate. Any similarity of different
kinds of reactions is quite exceptional, and introduction of dimensionless numbers is not very
practical here.
• Batch reactors are usual here, because the reactors should be as simple as possible. The
lining of the reactor walls is purposely or unintentionally consumed by side reactions and it has
to be restored very often. The refractory wear is a mass transfer process with a reaction
between the solid wall and the stirred liquid bath.

12. Hydrometallurgy

The hydrometallurgy covers different processes of the metal liberation and purification where
an essential step is the dissolution of certain metal compounds in liquid solvents at normal or
slightly elevated temperature. The typical processes comprise:
- dissolution of metal elements (e.g. gold amalgam in mercury),
- oxidation of metal elements and dissolution by acids (e.g. to form CuSO4, ZnSO4…),
- dissolution of soluble metal salts in water (as a downstream process of pyrometallurgy),
- dissolution of oxides in acids to form salts or complex salts (leaching of titanium as [TiO]2+,
uranium as [UO2(SO4)3]4-, etc.),
- dissolution of elements to form soluble complex compounds (cyanide leaching of gold or
silver to obtain [Au(CN)2]- or [Ag(CN)2]-),
- dissolution of oxide in alkali solution (bauxite in sodium hydroxide to obtain [Al(OH)4]−),
- bacterial leaching can be used to catalyze oxidation of sulfides at low temperatures.
The solution obtained, the “pregnant liquor”, is separated from the waste rock by common
techniques; i.e. by filtration, sedimentation, possibly aided with centrifugation. The pregnant liquor
usually contains more metallic compounds and the classical chemical separation techniques are
applied to the formation of the new solid phase by:
- sequential crystallization (e.g. radium from barium by M. Curie-Sklodowska),
- precipitation using various reagents (usually by ionic compounds),
- oxidation or reduction,
- complex formation,
- hydrolysis,
- cementation by less noble metal.
In any case, either the filtrate or the precipitate or both are valuable products and an efficient
separation is required, typically employing the following:
- filtration,
- sedimentation.
Another option is no need for formation of any solids and the separation is based e.g. on:
- liquid extraction by organic solvents,
- membrane separation,
- ion exchange,
- adsorption,
- electrolysis.

26
Specific problems of hydrometallurgy are connected with the treatment of heavy
suspensions. For a suspension of solid particles, an efficient agitation is necessary. Small scale
apparatuses are usually equipped with impellers (a nice example from the 16th century is
presented by G. Agricola). It is worth to note that the impellers are essentially characterized by the
dimensionless power number,
Po = W/(ρ N3 d5) , (12.1)
where W is mechanical energy transmitted to liquid, N is frequency of rotation, and d is the impeller
diameter.
A better option how to prevent the sedimentation of the particles and to re-suspend the sediments
is the air (pneumatic) mixing, typically used in the Pachuca tanks, with the airlift riser in centre of a
high slender tank. The large size of the industrial equipment means that there is a significant
hydrostatic pressure difference along the tank height, and the following dimensionless number may
be useful: ∆ρ VP/(ρ V0), where VP is the settling velocity and V0 the superficial gas velocity.

Generally, all hydrometallurgy processes are similar to the corresponding ones commonly
encountered in the chemical industry and any instruments for scale-up, like dimensionless
numbers, can be easily be taken from the chemical engineering practice.

13. Electrometallurgy

The electrolysis is mainly a mass transfer process, where the mass flow is proportional to the
electric current and the transfer coefficient can be expressed in terms of electric quantities as:
kS = I/(γ F L2) , (13.1)
where F [A.s/mol] is the Faraday constant. The ratio of the electric current to the area of the
electrode, I/L2 [A/m2], is another important quantity. The voltage U mostly lies in the range 1-10 V.
The electric power is W = U.I and is somehow related to the metal production.
The electrolysis of solutions to win the elementary metals is typical for refining of the copper
from the water solution of the copper sulfate. The purification of the copper uses the
electrochemical dissolution of the crude copper at the anode and the deposition of the pure Cu at
the cathode. The electrolysis of the solutions is suitable also for the formation of the deposits on
metal surfaces, e.g. plating them by Zn, Cd, Cu, Ag, Cr, Ni. The electrolysis of the melted salts
have to be performed at higher temperatures.
• Liquid elementary sodium is a product of the electrolysis of NaCl mixed with CaCl2 to decrease
the melting point. The chlorine is a by-product. Sometimes, magnesium is also manufactured
by the salt electrolysis.
• Aluminum is electrolyzed from alumina Al2O3 in the melted kryolite Na3(AlF6).

Electric heating is suitable whenever a reaction can take place at a very high temperature only.
Sometimes, the presence of oxygen in the process should be avoided and, therefore, the
conventional combustion heating cannot be applied even at moderate temperatures. The electric
arc furnaces are necessary also in the case when a high temperature is required for the metallurgy
process. Practically, all the processes at temperatures above 2000°C are driven by the electric
heating. The electric energy is also consumed by the electric arc welding. The electric arc furnace
is also the starting equipment for smelting solid materials like metal scrap or iron bloom obtained
by the direct reduction. The electric conductivity of the batch in the metallurgy processes is
satisfactory in most of the cases. Here, hundreds of Volts is a typical voltage, and the electric
power W = U.I is converted primarily to the heat. When the electrolysis takes place simultaneously
with the heating (like in Al manufacture), the direct current should be applied. For the heating
alone, the alternate voltage and sometimes high frequencies are preferred.

Plasma cutting occurs when a high temperature and high speed plasma is targeted to the metal
surface where an intensive heat transfer is followed by melting and simultaneous discharging of
metal particles. The transfer of heat to the metal is facilitated by the plasma, formed form an inert
gas blown from a nozzle through a strong electric arc.

27
14. Liquid metals

Gravity flow is practically the only way of liquid metal transporting. Both the slag and pig
iron is discharged from a blast furnace through open channels. Plugs made of clay are used to
stop the metal flow. The other ways of the transport are either in the torpedo cars or simply in
ladles moved by a crane. The hydrodynamic principles of the modeling should take into
consideration in these problems. Note that the kinematics viscosity of liquid metals is considerably
lower than that of water and does not depend much on temperature. The viscosity of slag is higher
and it increases with decreasing temperature, like the viscosity of molten glass.
Torpedo ladles and torpedo cars are used for temporary storage and transport of the liquid
pig iron, namely when transported from a blast furnace to the steelmaking plant. They are
constructed as large horizontal lined cylindrical vessels that may slowly rock to homogenize the
batch up to 200 tons. During this time, the desulfuration reaction takes also place with calcium,
sodium or magnesium compounds injected by the immersion lance. Gaseous or vapor products stir
also the batch efficiently.
Basic oxygen furnace (BOF) has been the key equipment of steel mills in the last
decades. The oxygen is blown into the liquid iron either by a lance immersed to the metal from
upside, or through a set off tuyeres in the bottom of a converter. As the reaction at about 2000°C is
very fast, typical superficial gas velocity is 1.5 m/s. Such extreme conditions cannot be found in
any other industrial process and are beyond the limits of laboratory models.
Vacuum treatment is a way for removal of dissolved gases (CO, H2, etc.) from liquid
metals. The snorkel system is essential for one class of steel degassing equipments. It should be
noted that the steel rises in vacuum to the height no more than 1.5 m.
Mixing of metals at high temperatures by mechanical agitators manufactured from
ceramics is quite exceptional. Mixing is usually aided by pneumatically, by gas stirring. The most
suitable driving gas is the inert argon. In contrast, both the oxygen and nitrogen contained in air
may react with metals. Argon is usually introduced through a porous brick at the ladle bottom. The
bubbling rate of the argon is moderate; it cannot be compared with introduction of oxygen into
BOF. Therefore, all scale-up rules known in the chemical engineering for the bubble formation,
motion and breakup in the bubble columns and air-lift reactors are applicable here. A gentle
introduction of the argon can also help to support the flotation of slag and other non-metal
inclusions. At the end, it forms an inert atmosphere above the surface. The single bubble behavior
as well as the behavior of a plume of bubbles is usually described by the Bond (Eötvös) and
Reynolds numbers, and possibly also the Morton number. Note that with fluid particles (bubbles,
drops), Bo relates to the static problems while We to the dynamic problems. What is surprising is
that the dimensional analysis indicate, that the system air-water is usually more adequate as a
laboratory model for the liquid steel than systems with mercury, melted tin or Wood metal, that are
used by certain investigators.
Continuous casting is a modern alternative to the casting of massive ingots. By cooling of
the stream of liquid iron, slabs, blooms, beams, billets or round ingots are prepared. In comparison
with the extrusion of glass, polymers or pastry, the iron lacks the highly viscous or the plastic
rheological behavior above its melting point. Due to this sharp phase transition from the low-
viscosity liquid to the completely solid body, the heat transfer during the continuous casting should
be delicately controlled. The heat convection with a significant axial heat conduction takes place
here. The cooling is mediated by the water-cooled walls of the vibrating copper mold. The
entraining billet with solidified perimeter is intensively cooled directly by the water jets. To obtain a
high quality material, the discharged metal should be free of bubbles and non-metal inclusions.
Therefore, the tundish has several weirs to assure the plug flow, as close as possible, and to keep
the level of the pure steel in the last compartment with a discharge nozzle. Then, the problems of
continuous casting are described by the momentum and heat balances and the usual tools for the
scale-up. The modeling, both the physical and mathematical, is plausible here.
Foundry casting is a process where the liquid metal is filled into a mold and solidified by
the controlled cooling to obtain the shaped metal components. The hydrodynamics of the metal
flow, heat transfer and the volume contraction due to the crystallization and cooling are the
28
important features. The prevention of the occluded bubbles and the perfect filling of the mold by
the metal can be aided by application of the vacuum. The centrifugal casting employs the force
induced by the mold rotation, to overcome the increasing viscosity at the verge of the solidification.
In the particular case of the lost wax casting, the permeation of the wax to a porous wall is another
flow problem. The classical methods of hydrodynamics can be applied to describe these
processes.

15. Solid metals

The material engineering investigates the properties of solid materials. There are some well
defined physical constants, like elastic modulus, cohesive strength or plasticity, while the other
quantities are more or less qualitative, like brittleness, hardness, occurrence of cracks, etc. They
characterize the final material and depend not only on composition and the actual temperature but
also on temperature and deformation history.

Forging is processing solid materials, employing the plastic deformation either at normal
temperature or more usually at the temperature closer to the melting point. The creep, as a very
slow flow, occurs at high stresses and is followed by the hardening and recovery - softening of the
metallic body. The plastic flow model can be applied to the scaling of the process. The rheology
measurements of plastometers give essential data on the time dependent shear, strain, and strain
rate, which can be employed in modeling the process with metals. Different experiments with 3D
strain, at least for the material stretched, pulled back, or twisted, should be designed. The physical
modeling using softer plastic materials like e.g. lead or extremely soft plasticine were used for the
scale up of forging. Recently, the mathematical models employing the basic plastometry data were
found to be efficient too.

16. Conclusions

The problems of the metallurgy that can be described by the well defined quantities are generally
identical with the problems we traditionally meet in the classical chemical engineering. They
comprise the hydrodynamics (often multiphase), the transport phenomena, namely the heat and
mass transfer, and chemical reaction phenomena. The metallurgical problems can be attacked by
the methods developed in the chemical engineering, respecting the specific features of the
metallurgical processes, often extending the parameter ranges we are used to have in chemical
technology. Consequently, it is likely that no essentially new dimensionless numbers will appear in
metal processes. Even in the processes driven by the electric force the ampere, volt or ohm
quantities are linked to metallurgy simply through the energy, either transformed to heat or to
chemical changes. The only new dimensionless quantities are constructed as simplexes
comparing similar quantities having synergetic effect to the processes under interest. They are
constructed ad hoc. The complex processes with chemical reactions and with different kinds of
heat transmission current in metallurgy do not leave much space for the complete scale up from
the laboratory data or the scale down to small scale modeling. Nevertheless, all instruments of the
approach employing the dimensionless numbers in hydrodynamics, in heat and mass transfer
considerations, can give a better insight to the metallurgy processes and can help to generalize our
particular observations.

Symbols

There are many symbols used in the text and only the main ones are presented here. Those of
local validity, auxiliary notation and symbols easily understandable from the context are not listed
here.

29
a variable in DA; acceleration [m/s2]
b variable in DA
c concentration [kg/m3]; cp heat capacity [J/kg.K]
C coefficient (drag, added mass) [-] ; concentration scale [kg/m3]
d length, size, [m]
D diameter [m]; mass diffusivity [m2/s];
f function symbol [f]
F force [N]; function symbol;
g gravity [m/s2]; reduced gravity g' = (∆ρ/ρ)g [m/s2]
j flux, of heat [J/m2s], of mass [kg/m2s]
k transfer coefficient [J/m2sK]; rate constant [e.g. 1/s]
l length [m]
L length scale [m]
m mass [kg]
M mass scale [kg]
p pressure [Pa]
P pressure scale [Pa]; period of oscillation [s]
q heat source [J/m3.s]
Q scale of heat source [J/m3.s]
r radius, position [m]; reaction rate [kg/m3s]
R reaction rate scale [kg/m3s]
t time [s]
T time scale [s]
v velocity (continuous phase) [m/s]
V velocity scale, mean speed [m/s]
x coordinate, distance, position [m]

Greek letters

α thermal expansivity [1/K]


β concentrational expansivity [1/kg]
∆ difference, variation [-]
θ temperature [K]
Θ temperature scale [K]
Φ function symbol [Φ]
κ heat diffusivity, λ/ρcp [m2/s]
λ heat conductivity [J/msK]
µ dynamic viscosity (fluid) [Pa.s]
ν kinematic viscosity (fluid) [m2/s]
Πi Pi-term, dimensionless number [-]
ρ density (fluid) [kg/m3]
σ interfacial tension [N/m, J/m2]
∇ Nabla operator [1/m]

Others

* dimensionless,
O(1) order of unity
o(1) from 0 to 1
[] physical dimension
DN dimensionless number(s)
DA dimensional analysis
SE scaling of equations

References
30
The following list of references is a selection of the few from the many sources of information on
the subject reviewed in this Chapter. There are no references given in the text to avoid disturbing
the reader and are lumped in this section, possibly with a brief commentary at the topical entry.

Transport and reaction phenomena

Bird R.B., Stewart W.E., Lightfoot E.N. 1965 Transport phenomena. J. Wiley 1965, NY.

Carslaw H.S., Jaeger J.C. 1947 Conduction of heat in solids. J. Wiley, N.Y.

Crank J. 1956 Mathematics of diffusion. Clarendon Press, Oxford.

Cussler E.L. 1997 Diffusion mass transfer in fluid systems. Cambridge University Press, UK.

Deen W.M. 1998 Analysis of transport phenomena. Oxford University Press, NY.

Davidson P.A. 2004 Turbulence. An introduction for scientists and engineers. Oxford University
Press.

Fogler H.S. 2005 Elements of Chemical Reaction Engineering. Prentice Hall.

Froment G.F., Bischoff K.B. 1990 Chemical Reactor Analysis and Design. Wiley.

Incropera F.P., DeWitt D.P. 1996 Fundamentals of heat and mass transfer.

Johnson R.W. (Ed.) 1998 The handbook of fluid dynamics. CRC Press, Boca Raton, USA.

Levenspiel O. Chemical reactor engineering. J. Wiley, NY.

McCabe W.L., Smith J., Harriott P. 2005 Unit Operations of Chemical Engineering. McGraw-Hill.

Munson B.R., Young D.F., Okiishi T.H. 1990 Fundamentals of fluid mechanics. J. Wiley, NY.

Rohsenov W.M, Choi H.Y. 1961 Heat, mass and momentum transfer. Prentice-Hall, London.

Slattery J.C. 1972 Momentum, energy, and mass transfer in continua. McGraw-Hill, NY.

Thomson W.J. 2000 Introduction to transport phenomena. Prentice Hall, NJ.

Tritton, D. 1988 Physical fluid dynamics. Clarendon Press, Oxford.

Welty J.R., Wicks C.E., Wilson R.E. 1969 Fundamentals of momentum, heat, and mass transfer. J.
Wiley, NY.

White F.M. 1974 Vicous fluid flow. McGraw-Hill, NY.

Handbooks and data books

IidaT., Guthrie R.I.L. 1988 The physical properties of liquid metals. Oxford University Press,
Oxford.

Perry R.H., Chilton C.H. 1973 Chemical Engineers´ Handbook. McGraw-Hill Book Company, NY

31
Reid R.C., Prausnitz J.M., Poling B.E. 1987 The properties of gases and liquids. McGraw-Hill,
Boston.

Weast R.C., Astle M.J. (Eds.) 1981 CRC Handbook of chemistry and physics. CRC Press, Boca
Raton, Florida.

Dimensional analysis

There are tens of books fully devoted to the dimensional analysis and the related topics (similitude,
modeling, scale-up, etc.), the powerful tools often used in engineering and physics. This simple
method has a long tradition, having been used semi-intuitively for almost centuries. Its
mathematical footing is dated back to ~ 1900 and is connected with the name of E. Buckingham.
The book by G.I. Barenblatt (2003) touches the cutting edge of the scaling approach and shows
the limits of the dimensional analysis, as well as introduces its generalization (intermediate
asymptotics). The Part I. of this chapter bears the fingerprint of the recent review by M.C. Ruzicka
(2008).

Barenblatt G.I. 2003 Scaling. Cambridge University Press.

Becker H.A. 1976 Dimensionless parameters. Applied Science Publishers, London.

Astarita G. 1997 Dimensional analysis, scaling, and orders of magnitude. Chem. Eng. Sci. 52,
4681-4698.

Boucher D.F., Alves G.E. 1959 Dimensionless numbers. Chem. Eng. Prog. 55 (9), 55-64.

Boucher D.F., Alves G.E. 1963 Dimensionless numbers-2. Chem. Eng. Prog. 59 (8), 75-83.

Bridgman P.W. 1922 Dimensional analysis. Yale University Press, New Haven.

Isaacson E. De St Q., Isaacson M. De St Q. 1975 Dimensional Methods in engineering and


physics. Edward Arnold, London.

Jerrard H.G., McNeill D.B. 1992 Dictionary of scientific units. Including dimensionless numbers and
scales. Chapman & Hall, London.

Langhaar H.L. 1951 Dimensional analysis and theory of models. John Wiley, NY.

Massey B.S. 1971 Units, dimensional analysis and physical similarity. Van Nostrand, London.

Ruzicka M.C. 2008 On dimensionless numbers. Chem. Eng. Res. Des. 86, 835-868.

Schuring D.J. 1977 Scale models in engineering: Fundamentals and applications. Pergamon
Press, Oxford.

Szirtes T. 1998 Applied dimensional analysis and modeling. McGraw-Hill, NY.

Taylor E.S. 1974 Dimensional analysis for engineers. Clarendon Press, Oxford.

Zierep J. 1971 Similarity laws and modeling. Marcel Dekker, NY.

Reactors, scaling and scale-up

Naturally, the scale-up topic is partly treated in many books on the dimensional analysis listed
above. Here are sources where this topic is the main one, or applied to reactor engineering. In
32
particular, the excellent book by W.B. Krantz (2007) is highly recommended to anybody who dare
to perform the scaling analysis in detail to solve specific problems.

Branan C.R. 2005 Rules of Thumb for Chemical Engineers Gulf Professional Publishing.

Euzen J.P. et al. 1993 Scale-up methodology for chemical processes. Gulf Publ. Co. Houston,
Texas.

Grassmann P. 1971 Physical principles of chemical engineering. Pergamon Press, Oxford.

Johnstone E.R., Thring M.W. 1957 Pilot plants, models, and scale-up methods in chemical
engineering. McGraw-Hill, NY.

Krantz W.B. 2007 Scaling analysis in modeling transport and reaction processes. AICHE and J.
Wiley, N.J.

Nauman E.B. 2008 Chemical Reactor Design, Optimization, and Scaleup.Wiley-Interscience.

Stichlmair J.G. 2002 Scale-up engineering. Begell House, Inc., NY.

Zlokarnik M. 2006 Scale-up in chemical engineering. Wiley, Weinheim.

Related topics

The theoretical stability analysis means a sophisticated manipulation with the governing equations.
It is demanding and is usually not a part of the engineering education.

Chandrasekhar, S. 1961 Hydrodynamic and hydromagnetic stability. Oxford University Press.

Drazin, P.G., Reid, W.H. 1981 Hydrodynamic stability. Cambridge University Press.

The complex rheology of the flowing materials (suspensions, dispersions, melts, etc.) occurring in
metallurgy was not considered in this chapter, so the reader is advised to consult one of these
books for the basic information in this field.

Astarita G, Marrucci G. 1974 Principles of non-Newtonian fluid mechanics. McGraw-Hill, NY.

Barnes H.A., Hutton J.F., Walters K. 1989 An introduction to rheology. Elsevier, Amsterdam.

Macosko C.W. 1994 Rheology. Principles, measurements, and applications. Wiley-VCH, NY.

Morrison F.A. 2001 Understanding rheology. Oxford university Press. NY.

Stickel J.J., Powel R.L. 2005 Fluid mechanics and rheology of dense suspensions. Ann. Rev. Fluid
Mech. 37, 129-149.

As indicated already, in most of the technological applications, the multiphase systems typically
occur. Some references bellow can be useful for obtaining both general and specific knowledge
about this complicated subject.

Brady J.F., Bossis G. 1988 Stokesian dynamics. Ann. Rev. Fluid Mech. 20, 111-157.

Brennen, C.E. 2005 Fundamentals of multiphase flow. Cambridge University Press.

Crowe, C., Sommerfeld, M., Tsuji, Y. 1998 Multiphase flows with droplets and particles. CRC
Press, Boca Raton, USA.
33
Kleinstreurer C. 2003 Two-phase flow: theory and applications. Taylor and Francis, NY.

Nguyen A.V., Schulze H.J. 2004 Colloidal science of flotation. Marcel Dekker, NY.

Sadhal S.S., Ayyaswamy P.S., Chung J.N. 1997 Transport phenomena with drops and bubbles.
Springer, NY.

Kolev N.I. 2002 Multiphase flow dynamics. Vol. 1, 2. Springer, Heidelberg.

Soo S.L. 1990 Multiphase fluid dynamics. Science Press, Beijing.

One particular area of the multiphase flow is the behavior of the granular matter, which is highly
relevant in some areas of metallurgy. The granular state of matter is so specific that it surely
deserves its own identity. Several references are given to familiarize with this exciting subject.

Brown R.L., Richards J.C. 1970 Principles of powder mechanics. Pergamon Press, Oxford.

Duran, J. 2000 Sands, powders, and grains. Springer, NY.

Forterre Y., Pouliquen O. 2008 Flows of dense granular media. Ann. Rev. Fluid Mech. 40, 1-24.

Hinrichsen H., Wolf D.E. 2004 The physics of granular media. Wiley, Weinheim.

Nedderman R.M. 1992 Statics and kinematics of granular materials. Cambridge University Press.

Taylor D.W. 1948 Fundamentals of soil mechanics. J. Wiley NY/Chapman & Hall London.

It has long been known that both the natural and man-made processes do possess a broad
spectrum of the length and time scales. Only recently this important fact was given the adequate
impact within the newly formed research field called the 'multiscale science'. This new paradigm is
equally valid for both the chemical engineering and the metallurgy engineering.

Glimm J., Sharp D.H. 1997 Multiscale science: a challenge for the 21-st century. SIAM News
30(8), 4.

Li J., Kwauk M. (Eds.) 2004 Complex systems and multi-scale methodology. Chem. Eng. Sci. 59
(8-9), 1611-1904. (special issue)

Marin G.B. (Ed.) 2005 Multiscale analysis. Adv. Chem. Eng. vol. 30.

Perspectives. The future of the chemical engineering research: complex systems. 2005 AICHE J.
51(7), 1839-1884. (special section)

Metallurgy

The books on the subject are well-known to the researches in this area, so only few specimen are
mentioned, of which Guthrie (2002), Seetharaman (2005) and Szekely (1979) contains much of the
fundamentals.

Agricola G. 1556 De re metalluca libri XII, Basilae.


(online: http://knihovna.vsb.cz/knihovna/agricola/234.html)

Bodsworth C. 1994 The extraction and refining of metals. CRC Press, Boca Raton.

34
Borchers W. 2007 Metallurgy. A brief outline of the modern processes for extracting the more
important metals. Negley Press.

Brandt D.A., Warner J.C. 1999 Metallurgy Fundamentals. Goodheart-Willcox, Illinois, USA.

Chandler H. 2006 Metallurgy for the Non-Metallurgist. ASM Int., OH, USA.

Davenport W.G., King M., Schlesinger M., Biswas A.K. 2002 Extractive Metallurgy of Copper.
Elsevier Science.

Engh T.A., Simensen C.J., Wijk O. 1992 Principles of metal refining. Oxford University Press, NY.

Geiger G.H., Poirier D.R. 1980 Transport phenomena in metallurgy. Addison-Wesley, Reading
M.A.

Gore G. 2003 Electrolytic Separation, Recovery And Refining Of Metals. Wexford College Press,
Wexwood.

Guthrie R.I.L. 2002 Engineering in process metallurgy. Oxford University Press, N.Y.

Gupta C.K. 2003 Chemical Metallurgy. Wiley-VCH Verlag GmbH & Co.

Han K.N. 2002 Fundamentals of aqueous metallurgy. SMME Inc., Colorado, USA.

Kubachewski O., Alcock C.B. 1983 Metallurgical thermochemistry. Pergamon Press, Oxford.

Lackner M. (Ed.) 2009 Scale-up in Combustion. ProcessEng Engineering GmbH.

Newton J., Wilson C.L. 1942 Metallurgy of Copper. ISBN-13: 978-1443725804

Sahai Y., Guthrie R.I.L. 1986 Recent advances in the hydrodynamics of metallurgical processing.
In: Mujumdar A.S. & Mashelkar R.A. (Eds.), Advances in transport processes, Vol. IV, Wiley
Eastern Ltd, New Delhi, 1986, pp. 1 - 48.

SahaiY., Pierre G.R. (Eds.) 1992 Transport Processes in Engineering: Advances in Transport
Processes in Metallurgical Systems, Vol 4. Elsevier, Amsterdam.

Seetharaman S. (Ed.) 2005 Fundamentals of metallurgy. Woodhead Publishing Ltd., Cambridge.

Szekely J. 1979 Fluid Flow Phenomena in Metals Processing, Academic Press, NY.

Szekely J., Evans J.W., Brimacombe J.K. 1988 The mathematical and physical modelling of
primary metals processing operations. Wiley, N.Y.

Verhoeven J.D. 2007 Steel Metallurgy for the Non-Metallurgist. ASM Int., OH, USA.

Waseda Y., Isshiki M. 2001 Purification Process and Characterization of Ultra High Purity Metals:
Springer Verlag, Berlin

Wichterle K., Obalova L. 1999 Suspended particle circulation in gas-lift tanks. Chem.Papers 53,
384-389.

35
Momentum
∆c → ∆ρ
∆Θ → ∆ρ

v
v
ediss

Heat Mass

qreact rrate

∆c
∆Θ
Reaction

Figure 1

The scheme of the inter-relation between the four main processes in the metallurgy: transport of the
three conserved quantities (Momentum, Heat, Mass) and the reaction phenomena (Reaction). The
mutual coupling between the processes is shown by the arrows. The physical quantities that
facilitate the coupling are also indicated. See text for details.

36

Das könnte Ihnen auch gefallen