Process Development 1

Process Development
Herbert Vogel, BASF Aktiengesellschaft, Ludwigshafen, Federal Republic of Germany

1. Introduction . . . . . . . . . . . . . . 1 3.4.1.4. The Limitations of the Miniplant

1.1. The Objective of Industrial Re- Technique . . . . . . . . . . . . . . . . 22
search and Development . . . . . . 1 3.4.2. Pilot Plant . . . . . . . . . . . . . . . . 22
1.2. The Production Structure of the 4. Process Evaluation . . . . . . . . . . 23
Chemical Industry . . . . . . . . . . 2 4.1. Preparation of Study Reports . . 23
1.3. The Task of Process Development 4 4.1.1. The Summary . . . . . . . . . . . . . . 23
2. Process Data . . . . . . . . . . . . . . 4 4.1.2. Basic Flow Diagram . . . . . . . . . 23
2.1. Chemical Mechanism . . . . . . . . 4 4.1.3. Process Description and Flow Dia-
2.2. Physicochemical Data . . . . . . . . 5 gram . . . . . . . . . . . . . . . . . . . 23
2.3. Processing . . . . . . . . . . . . . . . 6 4.1.4. Waste-Disposal Flow Diagram . . . 24
2.4. Patenting and Licensing Situation 6 4.1.5. Estimation of Capital Expenditure . 24
2.5. Development Costs . . . . . . . . . . 7 4.1.5.1. Introduction . . . . . . . . . . . . . . . 24
2.6. Location . . . . . . . . . . . . . . . . . 7 4.1.5.2. ISBL Investment Costs . . . . . . . . 24
2.7. Market Situation . . . . . . . . . . . 7 4.1.5.3. OSBL Investment Costs . . . . . . . 26
2.8. Raw Materials . . . . . . . . . . . . . 8 4.1.5.4. Infrastructure Costs . . . . . . . . . . 26
2.9. Plant Capacity . . . . . . . . . . . . 8 4.1.6. Calculation of Production Costs . . 26
2.10. Waste-Disposal Situation . . . . . . 10 4.1.6.1. Feedstock Costs . . . . . . . . . . . . 27
2.11. End Product . . . . . . . . . . . . . . 10 4.1.6.2. Utilities . . . . . . . . . . . . . . . . . 27
3. The Standard Course of Process
4.1.6.3. Waste-Disposal Costs . . . . . . . . . 30
Development . . . . . . . . . . . . . . 11
4.1.6.4. Staff Costs . . . . . . . . . . . . . . . . 30
3.1. The Iterative Nature of Process
4.1.6.5. Maintenance Costs . . . . . . . . . . 31
Development . . . . . . . . . . . . . . 11
4.1.6.6. Overheads . . . . . . . . . . . . . . . . 31
3.2. Drawing up an Initial Version of
4.1.6.7. Capital-Dependent Costs (Deprecia-
the Process . . . . . . . . . . . . . . . 13
tion) . . . . . . . . . . . . . . . . . . . . 31
3.2.1. Tools used in Drawing up the Initial
Version of the Process . . . . . . . . 15 4.1.7. Technology Evaluation . . . . . . . . 31
3.2.1.1. Databases . . . . . . . . . . . . . . . . 15 4.1.8. Assessment of the Experimental
3.2.1.2. Simulation Programs . . . . . . . . . 15 Work . . . . . . . . . . . . . . . . . . . 35
3.2.1.3. Expert Systems . . . . . . . . . . . . . 16 4.2. Return on Investment . . . . . . . . 35
3.3. Checking the Individual Steps . . 17 4.2.1. Static Return on Investment . . . . . 36
3.4. Testing the Entire Process on a 4.2.2. Dynamic Return on Investment . . . 36
Small Scale . . . . . . . . . . . . . . . 18 4.3. Economic Risk . . . . . . . . . . . . 37
3.4.1. The Miniplant Technique . . . . . . 18 4.3.1. Sensitivity Analysis . . . . . . . . . . 37
3.4.1.1. Introduction . . . . . . . . . . . . . . . 18 4.3.2. Amortization Time . . . . . . . . . . 37
3.4.1.2. Construction . . . . . . . . . . . . . . 19 5. Trends in Process Development . 37
3.4.1.3. The Limits of Miniaturization . . . . 21 6. References . . . . . . . . . . . . . . . 38

1. Introduction 302], an amount which is often of the same order

of magnitude as the profit of the undertaking.
1.1. The Objective of Industrial It is the job of research management to
Research and Development use these resources to achieve competitive ad-
vantages. After all, for the chemical industry
In the chemical industry, ca. 6 % of turnover is the halcyon days of the sellers’ market (de-
spent on research and development [1, pp. 282, mand > supply) have receded into the distant

c 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

10.1002/14356007.b04 437
2 Process Development

past; it is now the customer who calls the tune

(buyers’ market).
Unlike consumer goods such as cars and
clothes, most commercial chemical products are
faceless (e.g., hydrochloric acid), and as a rule
the customer is therefore only interested in sales
incentives such as price, quality, and availability.
All the research activities of an industrial en-
terprise must therefore ultimately boil down to
two basic competitive advantages, namely those
of being “cheaper” and “better” than the com-
petitor.
At the same time, “better” now refers not only
to availability and product quality, but also envi-
ronmental compatibility of the process, and the
quality assurance approach of the supplier, etc.

1.2. The Production Structure of the

Chemical Industry
If the production structure of the chemical in-
dustry is examined, it is seen that there are only
a few hundred major basic products and inter-
mediates that are produced on a scale of at least
a few thousand to several million tonnes per an-
num worldwide.
This relatively small group of key products,
which are in turn produced from only about ten
raw materials, are the stable foundation on which
the many branches of refining chemistry (dyes,
pharmaceuticals, etc.), with their many thou-
Figure 1. Product family tree of the chemical industry:
sands of often only short-lived end products, are starting from raw materials and progressing through the ba-
based [2, p. 57 ff]. This has resulted in the well- sic products and intermediates, to the refined chemicals and
known family tree (see Fig. 1), which can also final consumer products [3], [4]
be regarded as being synonymous with an in-
tegrated production system, with synergies that
are often of critical importance for success.
A special characteristic of the major basic
products and intermediates is their longevity.
They are statistically so well protected by their
large number of secondary products and their
wide range of possible uses that they are hardly
affected by the continuous changes in the range
of products on sale. Unlike many end products,
which are replaced by better ones in the course
of time, they do not themselves have a so-called
life cycle. However, the processes for producing
them are subject to change. This is initiated by
new technical possibilities and advances opened
up by research but is also dictated by the current Figure 2. How the raw material base of the chemical in-
raw material situation (see Fig. 2). dustry has changed with time
 Process Development 3

Figure 3. Life cycles of the acrylic acid processes

a) Cyanohydrin,acrylonitrile,and propiolactone processes; b) Reppe process; c) Propylene oxidation; d) New process?[4]

Here it is not the individual chemical product, technology to a new trend-setting technology,
but the production process or technology which the productivity in the research and development
has a life cycle. Figure 3 shows the life cycles of sector increases appreciably, thus enabling sub-
the acrylic acid processes [4]. stantial competitive advantages to be achieved.
Enterprises which already have competitive It is precisely on this innovative activity that
advantages must take account of this technol- the prosperity of highly developed countries
ogy S curve [5] in their research and develop- with limited raw material sources such as Ger-
ment strategy (see Fig. 4). The curve shows that many and Japan is based.
as the research and development expenditure on
a particular technology increases, the produc-
tivity of such expenditure decreases with time
[6]. If enterprises are approaching the limit of
a particular technology, they must accept dis-
proportionately high research and development
expenditure, with the result that the contribution
made by these efforts to the research objectives
of “cheaper” and/or “better” becomes increas-
ingly small, thereby always giving the competi-
tor the opportunity of catching up on the techni-
cal advantage.
Once an enterprise has reached the upper re-
gion of the product or technology S curve, the
question arises whether it is necessary to switch
from the standard technology to a new trend-
setting technology in order to gain a new and suf- Figure 4. The technology S curve [8]: The productivity of
ficient competitive advantage [7]. Figure 4 de- the research and development expenditure increases con-
picts this switch to a new technology schemat- siderably by switching from the basic technology to a new
trend-setting technology
ically and shows that on switching from a basic
4 Process Development

To assess whether a “better” and/or “cheaper” Thermodynamic Equilibrium. Many reac-

research and development strategy is still accept- tions, such as esterifications and dimerizations,
able in the long term for a given product or pro- proceed to chemical equilibrium. The thermody-
duction process, the R & D management must namic equilibrium provides information about
develop an “early warning system” that deter- the maximum possible conversion, and to assess
mines the optimum time for switching to a new the potential conversion it is essential to know
product or a new technology [9]. how far the intended reactions are from chemical
equilibrium.
Example 2.1. The formation of trioxane from
1.3. The Task of Process Development formaldehyde

The task of process development is to extrap-

olate a chemical reaction discovered and re-
searched in the laboratory to an industrial scale,
taking into consideration the economic, safety, in the gas phase is a typical equilibrium reaction.
and ecological boundary conditions. The equilibrium constant, which is given by
The starting point is the laboratory equipment

pT 4 p0F −p
and the outcome of development is the produc- Kp = =
3
(2.2)
p3F 3p−p0
F
tion plant; in between, process development is
required. The following account shows how this where
task is generally handled. Although the sequence p = total pressure at equilibrium
of steps in the development process described p0F = initial partial pressure of formaldehyde
are typical, they are by no means obligatory, and
pT , pF = equilibrium partial pressure of triox-
it is only possible to outline the basic framework.
ane or formaldehyde, respectively,
can be expressed according to [10] as:
  7350
2. Process Data log Kp /bar −2 = −19.8 (2.3)
T /K
When the laboratory phase has been completed It follows that if the process conditions as-
and before the actual process development is sumed are T = 353 K and p0 F = 0.2 bar, a maxi-
started, further information must be obtained mum formaldehyde conversion of:
since the latter stage is normally associated with 3·pT
high costs. U= ·100 = 35% (2.4)
p0F
can be achieved in one throughput. The conver-
sion actually achieved will be < 35 % and de-
2.1. Chemical Mechanism pends on the catalyst and reactor systems.
The thermodynamic data (standard enthalpy
The core of a chemical plant is the reactor. Its
and entropy of formation, specific heat of the
input and output decide the structure of the en-
reactants) required to calculate the equilibrium
tire plant built up around it. Therefore, detailed
constant can be found in published tables or
knowledge of the chemical reaction must be
can be estimated by approximation methods (see
available at the earliest possible stage. In partic-
→ Estimation of Physical Properties ) [11].
ular, the following questions must be clarified:
1) Thermodynamic equilibria Kinetics. In order to specify the type of reac-
2) Kinetics of the main, secondary, and side re- tor to be used later, information must be available
actions on the potential reaction routes to the main and
3) Dependence of selectivity and conversion on secondary products and the byproducts. The rate
the process parameters of formation and its dependence on process pa-
4) Heats of reaction rameters such as temperature, pressure, and cat-
alyst concentration should, if possible, be known
quantitatively.
 Process Development 5

Example 2.2 [12, pp. 173 – 175]. The starting

k1 ·c0A
material A reacts not only to form product P, but cP =
k1 −k2
also (1) to form an undesirable byproduct Y in
· {exp (−k2 ·t) −exp (−k1 ·t)} (2.14)
a parallel reaction or (2) to form an undesirable
secondary product X in a sequential reaction:


k1
cX =c0A · 1+ ·exp (−k2 ·t) −
k2 −k1

k2
1) First-order parallel reaction ·exp (−k1 ·t) (2.15)
k2 −k1
For the starting material A:

dcA
= − (k1 +k3 ) ·cA (2.6)
dt 2.2. Physicochemical Data
with cA = c0A at time t = 0, integration results
in: A knowledge of exact physicochemical data has
become more important for a number of reasons:
cA =c0A ·exp {− (k1 +k3 ) ·t} (2.7) 1) The increasing use of simulation programs
has made the requirements for exact physi-
For the useful product P: cochemical data increasingly important. The
result of a simulation calculation can only be
dcP
=k1 ·cA (2.8) as good as the quality of the physicochemical
dt data.
with cP = c0P for t = 0, integration gives 2) To approve chemical plants, the authorities
demand information on the toxicity, degrad-
k1 ·c0p ability, and safety of the materials involved
cp = c0p + [13], [14].
(k1 +k3 )
· {1 − exp [− (k1 +k3 ) ·t]} (2.9) 3) The public are demanding more information
on the effect of the materials being handled
The same applies to cY . on the environment.
2) First-order sequential reaction
This is covered by three simultaneous differ- At the start of process development, mate-
ential equations: rial data files for pure substances as well as bi-
nary and ternary mixtures are started. As devel-
dcA opment advances, these grow and must be con-
= −k1 ·cA (2.10) tinually updated. They subsequently form a doc-
dt
ument which is passed to the planning and plant
construction departments.
dcP The layout shown in Table 1 has proved
=k1 ·cA −k2 ·cP (2.11) useful as a model for the collection of pure-
dt
substance data.
Only data which have been reliably evaluated
dcX
are kept in the files and these are the only values
=k2 ·cP (2.12) fed into the simulation programs.
dt
At the start of process development, the first
If c 0P = c0X = 0, integration of these equations step is to collect all available literature data on
yields: a material (see→ Estimation of Physical Prop-
erties).
cA =c0A ·exp (−k1 ·t) (2.13)
6 Process Development

Many companies now maintain their own ma- Table 1. Important pure-substance data frequently required in pro-
cess development using acrylic acid as an example [15]
terial databases. The data have been evaluated
and should have been found satisfactory when Product name acrylic acid
CAS no. [79-10-7]
used in practice. The times when a single com- Formula C 3 H4 O2
pany had 10 different sets of Antoine parameters Molar mass, kg/kmol 72.06
for water between 0 and 100 ◦ C ought now to be
Melting point, ◦ C 13.5
over. Boiling point, ◦ C 141.0
If no material values can be found in the lit- Antoine parameters ∗
erature, which is often the case for binary and A 9.135
B −3 245
ternary data, they should initially be estimated C 216.4
by using empirical formulae or realistic val- Vapor pressure at 20◦ C, mbar 10
ues that have proved themselves for many sub- Heat of evaporation at bp, kJ/kg 633
stances (e.g., heat capacity≈2 kJ kg−1 K−1 for Heat capacity, kJ kg−1 K−1 1.93 (liquid)
Density, kg/m3 1 040 at 30◦ C
most organic liquids). However, experimental Viscosity, mPa · s 1.149 at 25◦ C
determination or confirmation of the most im- Heat of formation, kJ/mol
portant values on which the plant design is based Gross calorific value, kJ/kg 19 095
Heat of transformation, kJ/kg 1 075 (polymerization)
is unavoidable.
Heat of fusion, kJ/kg 154 at 13◦ C
It is sheer negligence to rely on physicochem- Solubility in H2 O miscible
ical data originating from a single literature ref- Solubility of H2 O in acrylic acid miscible
erence. MAK
Toxicity (LD50 , rat, oral) 340 mg/kg
Water hazard class 1 (“low water hazard”)
2.3. Processing Odor threshold pungent
Flash point, ◦ C 54
Processing is intimately linked to the chemical Ignition temperature, ◦ C 390
reaction. While the chemical aspects are still Lower explosion limit, vol % 2.4 (47.5◦ C)
Upper explosion limit, vol % 16 (88.5◦ C)
subject to extensive modification (for instance,
in the choice of catalyst, solvents, etc.), there ∗ ln (p/bar) = A + B/(C + T /◦ C).
is little point in paying a great deal of attention
to processing. It is only when the reaction mix-
ture is being produced in a representative man-
ner that an initial processing procedure can be 2.4. Patenting and Licensing Situation
devised (see Section 3.2). There is no generally (See also → Patents)
accepted way of going about this. It is more an
art than craft, and reliance must be placed on in- If a new idea is to be used in developing a pro-
spiration and experience. It has, however, always cess, a careful check should be made to deter-
been helpful to gather as many experienced spe- mine whether any third-party proprietary rights
cialists as possible and to discuss the problems would be infringed.
regularly. Although tentative efforts are now be- The best situation is one in which the ideas are
ing made to compile collective know-how in ex- completely new and the company can file its own
pert systems for processing strategies, these are patents to ensure it can use the ideas without re-
still far from being universally applicable. striction. However, patent applications not only
The above deliberations culminate in a sep- provide protection but also represent a source of
aration concept which can be broken down into information for competitors [16]. Careful con-
individual unit operations. Initially, these oper- sideration must therefore be given to the ques-
ations can be examined in the laboratory either tion of whether it is desirable to keep an idea
on a batch basis, or in a continuous process to secret or whether it should be patented.
determine whether they are feasible in princi- The risk involved in deciding against patent-
ple (e.g., are azeotropes formed?, how difficult ing in-house knowledge can be reduced by elimi-
is separation?, what is the dissolution rate?, can nating the patentable novelty of that information
the phases be separated?, etc.). This gives the ini- on a worldwide basis through publication in a
tial process concept, which can be used as a basis suitable form, e.g., in an in-house journal or a
for starting the actual process development. technical bulletin, etc.
 Process Development 7

If the idea is restricted by third-party propri- The development costs determined in this
etary rights, the following questions should be way do not contain any contributions arising
answered: from personnel not directly involved in devel-
opment, such as management or patent special-
1) When do those rights expire?
ists. Nor is account taken of unsuccessful de-
2) How can they be rendered void?
velopments. It nevertheless seems reasonable to
3) How can they be circumvented by modifica-
apportion the direct development costs to the
tions?
product and to allow for the other research and
If the idea is completely covered by third-party development costs in some other way, for exam-
proprietary rights, the only way out is to enter ple by means of a general cost provision (see
into licensing negotiations. Section 4.1.6.6). Although this method only ac-
Normally, a process should only be developed counts for the costs of successful development
after the legal situation has been clarified. projects, the financial burdens on the product
will be relatively high if it must bear all the de-
2.5. Development Costs velopment costs of the first plant.

Development costs often constitute a consider-

able proportion of the total cost of a project. If 2.6. Location
the process is completely new, these costs are
often around 50 % of the investment costs for The main competitive advantages such as price
the industrial plant, and perhaps more if a pilot and availability are nowadays often dependent
plant must be built. on the location of the production plant. Thus,
Development costs can be quoted with suf- there is now an increasing tendency to move
ficient accuracy if the following items are as- the production of major basic chemicals such as
sessed: methanol to the oil-producing countries. How-
ever, products whose production requires con-
Man years: This involves determining how siderable know-how are less location-dependent
many chemists and engineers will be employed and will continue to be produced in the indus-
on the project and for how long. Since the costs trialized countries, even in the future. Extreme
associated with each of the specialists working cases are those in which legal requirements de-
on the individual investigations are known, the cide the location, as is the case in the field of
man-year costs can then be calculated. gene technology.
In deciding on location, consideration should
Setting up costs: A rough estimate of the be given to the following aspects:
costs of setting up a small-scale plant is suffi-
cient since this item is normally the smallest in 1) Proximity to the sales market
the development costs. Once some notion of the 2) Proximity to the sources of raw materials or
process is available (a rough process flow dia- precursors
gram), these costs can be determined with rea- 3) Transport facilities (ship, rail, road)
sonable accuracy. 4) Quality of the available labor
5) Availability of cooling water
Operating costs: The operating costs can be 6) Waste-disposal management (sewage treat-
estimated from the assumed development time ment plant, incineration plant)
and the supervisory staff (shift personnel) re- 7) Political considerations (tax, investment aid,
quired. A rule of thumb is that the annual op- approval situation)
erating costs of an experimental plant will be at
least equal to those required to set it up.
More difficult to estimate are the amounts re-
quired annually for modifications and repairs to 2.7. Market Situation
the experimental plant. These depend very much
on whether new or already proven technologies The competitive advantages generated by pro-
are being used. cess development should ultimately benefit the
8 Process Development

market and its customers. It is therefore impossi- Since the exact purchase prices (transfer
ble to target process development correctly with- prices) are an important source of information
out a precise knowledge of that market. The fol- in assessing manufacturing costs, they must al-
lowing information is required if the basic data ways be available and kept up to date in a chemi-
needed to plan production (plant size, maximum cal company. Figure 6 shows the variation in the
production costs) are to be obtained price of some basic chemicals.
It is not only the price of potential raw ma-
1) Market price (how it will develop and vary
terials which must be determined but also their
with time)
availability and quality (purity, state as deliv-
2) Demand, broken down into in-house and
ered).
third-party demand
It is never possible to reach a generally appli-
3) Market growth
cable decision on a particular raw material since
4) Utilization of capacity in existing plants,
much depends on its availability at a particular
with a breakdown in terms of location
location and variation with time.
(United States, Western Europe, Japan) and
production process
5) Competitive situation (who is the largest
competitor?) 2.9. Plant Capacity
6) Customer situation (are there many small
customers or a few large ones?) Optimum plant capacity, i.e., the size of plant
which yields the maximum return (see Sec-
The question of whether a process development tion 4.2), will be achieved if a degree of capacity
or improvement will be worthwhile can be ana- utilization of 100 % is reached 3 – 5 years after
lyzed by using a portfolio representation (Fig. 5). start-up. However, since the plant capacity must
Portfolio analysis is a strategic planning tool, be decided about 4 – 5 years before start-up for
used to concentrate research investment on prod- planning and legal approval reasons, and plants
ucts for which market prospects appear favor- which are not fully utilized over a period of time
able and competitive advantages can be ex- increase the specific production costs, it is essen-
ploited [17]. tial to be able to look 7 – 10 years into the future
in determining the optimum capacity. Therefore,
precise knowledge of the market potential and of
2.8. Raw Materials market growth trends is required.
The relationship between production price
Raw materials, their availability, and their price and degree of capacity utilization is as follows
structure have always been crucial factors re- [19]
sponsible for shaping the technological base
of the chemical industry and, consequently, its P 1+F/V
= (2.16)
growth and expansion (see Fig. 2). P0 1+F/V ·Keff /K0
The chemical industry produces a wide vari-
ety of products from only a few inorganic and where
organic raw materials. P = production price in monetary units per
Thus, raw materials such as natural gas, pe- kilogram,
troleum, air, and water are used to produce ba- P0 = production price at rated capacity,
sic chemicals such as synthesis gas (CO – H2 F = fixed costs,
mixture), acetylene, ethylene, propene, benzene, V = variable costs,
ammonia, etc. These are used in turn to manu- K eff = effective plant capacity, in tonnes per
facture intermediates such as methanol, styrene, annum,
urea, ethylene oxide, acetic acid, acrylic acid,
cyclohexane, etc. (see Fig. 1). K0
Any change in the price and availability of = rated capacity, in tonnes per annum.
raw materials will therefore have a substantial Thus, as the degree of capacity utilization
effect on the production processes for secondary drops, the production price increases hyperbol-
products. ically and this dependence is the greater, the
 Process Development 9

Figure 5. Portfolio matrix [17]

A) Sleeping dog products These are problem products whose competitive situation is weak because of their low market share.
In such cases process development would not be worthwhile.
B) Cash cow products These are products which, although their market growth is low, have a high market share. To exploit the
cost advantages to the full, there is little point in heavily investing in process improvements. Instead, these products should be
“milked” and the resulting cash flow used to support the star products.
C) Star products All the efforts should be concentrated on these products since they will safeguard the survival of the company
in the long term.
D) Question mark products In these cases the company must decide whether to increase the market share (turn them into star
products) or to abandon them if the prospects are slim.

Figure 6. Variation in the price of some basic chemicals with time [18]
10 Process Development

higher the contribution made by the fixed costs be undecided and the process should therefore be
(see Fig. 7). designed to produce only waste products which
can be readily incinerated. Processes which pro-
duce salts or large amounts of dilute aqueous
solutions containing materials that are difficult
to degrade biologically are inherently location-
dependent.

2.11. End Product

It is essential to determine the demand for the
end product as well as its specification and
achievable price. A strongly competitive situa-
tion will mean particularly severe changes with
time, and in this case it is necessary to adopt a
particularly critical approach to forecasting. The
product specification is of particular significance
for process development. In the simplest case, it
Figure 7. Dependence of the relative production price
consists of a minimum purity, for example, that
on degree of capacity utilization for different fixed the end product should contain at least 99 % of a
cost/variable cost ratios F/V particular chemical compound. The permissible
impurity content may vary considerably from
The greater the uncertainty in the market
one product to another. Thus while it may be a
studies and the newer the technology used, the
few percent in the case of dyes, the limit is often
greater will be the preference for choosing a
only a few parts per million for monomers.
lower capacity. Many Japanese chemical com-
However, in most cases the end product can-
panies adopt this approach and prefer to set up
not be specified so simply since its subsequent
several plants with smaller capacities. However,
use may be affected in different ways by individ-
lower capacities result in higher specific plant
ual impurities. In such cases, it is not sufficient
costs and therefore in higher production costs.
to specify an upper limit for the sum of impuri-
This is represented by the following empirical
ties; instead, limits must be specified for impor-
relationship:
tant individual components. Often, the chemical
I2 =I1 · (K2 /K1 )χ (2.17) analysis is not sufficient or is not sufficient on
its own for specifying the end product, a case
where in point being, for example, polymers. In such
I = capital expenditure (investment), cases chemical analysis is supplemented by pro-
K = capacity, cess engineering properties which, in the sim-
χ = degression exponent. plest case, can be reduced to simple physical
properties.
For individual items of equipment, χ = 0.6, The process development engineer endeav-
and for plants containing the same number of ours to obtain as detailed a specification as pos-
items of equipment χ = 2/3 [20]. sible for the end product before he starts work,
but usually the specification will only be final-
ized during process development.
2.10. Waste-Disposal Situation (see Statements such as “as pure as possible” are
Section 4.1.6.3) inadequate since the removal of the final traces
of impurities is time consuming and expensive.
Waste-disposal facilities (incineration, sewage Individual specifications which are of no signif-
treatment plant, waste-disposal site) depend on icance in characterizing the end product should
the location. However, at the beginning of pro- be avoided. If possible, a composition which
cess development the location will generally still can be determined analytically is preferable to a
 Process Development 11

purely physical characterization since it is more the maximum capacity of the next larger plant
closely related to the process sequence. whose operating performance can be calculated.
It is here that the process development engineer
has an opportunity to save time and money. If
3. The Standard Course of Process reliable documents can be drawn up which en-
able the maximum scale-up of four powers of
Development ten (100 g/h×104 = 1 t/h) to be spanned in de-
signing a production plant, omitting the interme-
3.1. The Iterative Nature of Process diate stage also eliminates the costs of the pilot
Development plant and 3 – 4 years’ development time [23],
providing an important advantage in terms of
The development of chemical processes is a cost effectiveness and marketing. Consequently,
complex procedure. efforts are now made to extrapolate directly from
The first hurdle in establishing a new process the miniplant to the production scale.
is overcome when a promising synthetic route, The following list summarizes some of the
usually with associated catalysts, is discovered, important process steps and typical maximum
but many issues must be clarified and many prob- scale-up values above which reliable scale-up is
lems solved before industrial implementation is no longer possible [24]:
feasible and all the documents necessary to de-
sign and operate a chemical plant have been as- Reactors
sembled [21], [22]. How are these planning doc- Multitubular reactor,
uments arrived at? homogeneous tube,
homogeneous stirred tank >10 000
The conventional procedure is carried out in Bubble column < 1000
three stages (see Fig. 8): Fluidized-bed reactor
An initial process concept is developed on 50 – 100
the basis of an optimized laboratory synthesis.
This is not usually continuous (see Section 3.2),
and the individual process steps are examined Separation processes
Distillation and rectification 1000 – 50 000
independently of each other in the laboratory Absorption 1000 – 50 000
(see Section 3.3). A continuous laboratory plant Extraction 500 – 1 000
(the so-called miniplant) is then designed, set up, Drying 20 – 50
and operated (see Section 3.4.1). This is a small Crystallization 20 – 30

but complete plant handling production quanti-

ties of ca. 100 g/h and consisting of a synthesis
section, working-up, and all recycling streams. Larger scale-up factors are possible for gas-
Once the process concept has been confirmed phase reactors such as multitubular reactors.
in the miniplant, the next step is to design and An overview of the main types of reactor and
set up a trial plant with a much higher capacity, a discussion of the criteria for the choice of a par-
the scale of this pilot plant being between that ticular reactor and reactor scale-up during pro-
of the miniplant and that of the industrial plant. cess development can be found in Parts A and
The quantities produced are a few kilograms per B of this volume. The use of dimensional analy-
hour or tonnes per annum and enable, for ex- sis in scale-up is discussed in→ Scale-Up in the
ample, application tests to be carried out on the Chemical Industry.
product or large-scale deliveries to be made to The unit operations of distillation, rectifica-
customers. Operation of the pilot plant makes it tion, and absorption can be scaled up without
possible to complete and verify data and docu- an intermediate stage. This explains, for exam-
mentary information obtained at an earlier stage ple, why much effort is expended on obtaining
of process development (see Section 3.4.2). operating conditions that allow the reaction and
The scale-up factor from one stage to the processing of gases and liquids. The behavior
next is always limited by the “minimum princi- of gases and gas – liquid equilibria can be ex-
ple”, i.e., the process stage or piece of equipment plained well in physical terms and calculation is
with the lowest scale-up feasibility determines therefore straightforward.
12 Process Development

Figure 8. The stages involved in developing a process

The time intervals specified for design, installation, and operation are only tentative and, in specific cases, the real values may
differ greatly from those given

Example 3.1. Vapor – liquid equilibria can The vapor pressure of the pure component i,
readily be measured experimentally and can be P0i (T ), is often represented by a three-parameter
described well in mathematical form. The gen- Antoine expression:
eral relationship for the equilibrium between a Bi
liquid phase and an ideal gas phase takes the Pi0 (T ) =Ai + (3.2)
Ci +T
form:
Several models are available for describing the
γi (xi ) ·Pi0 (T ) activity coefficient γ i . Examples are the Wilson
yi (xi ,T ) = ·xi (3.1) model, which can be applied to completely mis-
P
cible mixtures, and the NRTL model, which is
where suitable for systems with a miscibility gap [25],
yi = mole fraction of component i in the gas [26].
phase, For solids, scale-up factors are several or-
xi = mole fraction of component i in the liq- ders of magnitude lower because of complica-
uid phase, tions such as deposits, encrustations, and abra-
P0i (T ) = saturation pressure of pure component sion. This is true not only for synthesis (e.g., in
i at the system temperature T , fluidized-bed reactors), but also for processing
P = vapor pressure of the mixture, (e.g., drying and crystallization).
γi = activity coefficient of component i. Process development does not, however, take
place as a one-way street. Assumptions are made
for the individual development stages which are
 Process Development 13

only confirmed or refuted when the next stage is A start should be made on choosing between
being worked on. It may be necessary therefore various processes or process variants as early as
to go through the individual stages several times possible so that consideration of a larger num-
with modified assumptions, resulting in a cyclic ber of possibilities is restricted to the labora-
pattern (see Fig. 9). tory. Prolonged investigation of two variants on
a trial-plant scale should be avoided.
Most mistakes are made at the beginning of
the activity, but it is still relatively easy and cheap
to eliminate them at the miniplant stage. How-
ever, the further process development advances,
the more expensive it becomes to eliminate mis-
takes (see Fig. 10). In the final production plant,
corrections can only be made with an enormous
expenditure of time and money.
As shown in Figure 9, each development
stage is followed by an evaluation to de-
cide whether development should be continued,
stopped, or started again at an earlier develop-
ment level (see Section 4).

3.2. Drawing up an Initial Version of the

Process
Once all the information has been assembled, an
initial version of the process is drawn up [27].
In general, it has been found useful to adopt the
following procedure:
After the initial versions of the process, which
should be the most reasonable ones possible at
that stage, have been drawn up, they should be
discussed by the project team. This discussion
should take account of all the information gath-
ered hitherto and, after all the advantages and
disadvantages have been reviewed, only one ver-
sion of the process should remain.
There is at present no method which will re-
sult in the “best” version of the process being
invariably chosen at the outset and there will
Figure 9. The cyclic pattern of process development presumably not be one in the future either [28],
[29]. Finding the optimum version of a process
The most important task is to find the weak requires not only knowledge and experience but
points and subject them to particularly close also a considerable creativity, since the number
scrutiny. The entire process will then be exam- of possible ways of carrying out a given task is
ined again with the improved data obtained in almost infinite and many of the rules can still not
this way, and so on. The fact that many deci- be quantified [30].
sions must be taken with incomplete knowledge Example 3.2 : The esterification of an im-
is a fundamental but inevitable difficulty. To de- pure organic acid with methanol yields a reac-
lay development until all uncertainties have been tion mixture containing at least 5 components
eliminated would be just as wrong as starting in- (N = 5): unreacted acid, unreacted alcohol, es-
dustrial development on the basis of laboratory ter, volatiles such as H2 O, lower esters, di-
discoveries alone. methyl ether, and high-boiling components such
as higher esters.
14 Process Development

Figure 10. The costs of eliminating mistakes and the investment costs increase by a factor of 10 from one development stage
to the next

According to Stephanopolous [31] the pos- solution is unknown. To calculate all possible
sible number of arrangements for the separation variants, to compare them, and then select one is
operations can be calculated from the relation- not feasible because of the large number. Many
ship: of the variants can of course be eliminated at the
[2· (N −1)] ! (N −1) start for trivial reasons (industrial, economic, le-
Z= ·S (3.3) gal reasons, etc.). However, even the number of
N !· (N −1) !
where potential variants left is still so large that it is im-
possible to calculate investment and production
Z = number of alternative arrangements, costs for all of them. Furthermore, at this point
N = number of components, the information available for doing this is often
S = number of separation processes. inadequate and there is consequently a risk of
eliminating promising variants.
If all components can be separated by distil- Perhaps this example reveals why inventing
lation, (N − 1) = 4 separation columns are nec- processes is still an art rather than a craft. How-
essary, which can be arranged in 14 differ- ever, the tools available for performing this art
ent ways. If a second separation process – for are being continuously improved [30], [32], [33]
example, an extraction, if the ester, alcohol, (see Section 3.2.1).
and water form azeotropes – must be added, On the basis of this initial version, an indus-
this already results in 224 possible arrange- trial plant is designed. The individual unit op-
ments. If the number of possible reaction pro- erations (reactor, absorber, distillation, etc.) are
cedures (e.g., batch/continuous; gas/liquid; ion designed on the basis of the existing informa-
exchanger/mineral acid) is also taken into con- tion (approximate size and diameter of columns,
sideration, there will be an almost infinite num- etc.).
ber of ways of carrying out the process (a so- By scaling down this hypothetical large-scale
called combinatorial explosion). plant a trial plant is designed, which is nowadays
To find the best process among all these vari- generally a miniplant (see Section 3.4.1). At the
ants is quite impossible simply because the ideal
 Process Development 15

same time, the industrial plant is simulated by Figure 11 provides an overview of the structure
using a computer program and an initial com- of an on-line data bank.
plete mass and energy balance sheet is drawn Databases differ as to whether they contain
up. the information itself (primary information or
factual data) or whether they refer to other infor-
mation sources (secondary information or liter-
3.2.1. Tools used in Drawing up the Initial ature data).
Version of the Process Some types of data which are important for
process development are [36]:
Requirements which must be met by modern
EDP tools are as follows: 1) Material properties [37]
2) Physicochemical data
1) Fast access to the information 3) Ecological and toxicological data
2) Information which can be conveniently han- 4) Costs of raw materials, intermediates, and
dled at the workplace end products
3) The information must be continuously up- 5) Energy and equipment costs
dated 6) On-line literature searches

Table 2. A selection of simulation programs for steady-state pro-

cesses

Program Supplier

ASPEN PLUS Aspen Technology Inc.,

251 Vassar Street
Cambridge, MA 02139
United States
CHEMASIM BASF AG,
Ludwigshafen,
Germany
CHEMCAD Chemstations Engineering
Software COADE
10375 Richmond Ave.,
Suite 1225
Houston, TX 77042
United States
DESIGN II Chemshare
FLOWPACK ICI
HYSIM Hydrotech. Ltd.
400, 119 14th Street N.W.
Calgary, Alberta T2N1Z6
Canada
PROCESS Simulation Science Inc.,
PRO/II 1 051 W Bastanchury Road,
Fullerton, CA 92633
United States
Figure 11. On-line service components [35]
VTPLAN Bayer AG,
(CONTI) Leverkusen,
Germany

3.2.1.1. Databases [34]

3.2.1.2. Simulation Programs [38]
To carry out process development as efficiently
as possible, it is important to avoid duplicate Simulation using computer programs furnishes
activities. This includes taking account of ex- the process development engineer with a very
isting knowledge or exploiting it. For this rea- effective tool. The development of very fast,
son, chemical companies not only build up inter- large computers and effective physical models
nal databases but also utilize external databases. has now made it possible to simulate individual
16 Process Development

units (e.g., distillation), networks of units of the where complex problems are solved by draw-
same type (e.g., a stirred tank cascade, a heat ing conclusions and using experimental knowl-
exchanger network) and networks of dissimilar edge. In this case, however, the knowledge does
units (e.g., chemical plants). not only consist of facts which can be stored in
In practice, the simulation program used de- databases but is acquired by carrying out process
pends on the objective; there is, after all, little development, experiencing failures, being suc-
point in trying to solve detailed problems with cessful, repeating the same process, and learning
a simulation program which, although it is all- at the same time; in short, by acquiring a feel for
embracing, is slow if there are very much more a problem. Knowledge is gained as to when it is
efficient programs available which are designed necessary to stick to the rules and when they can
for the specific requirement. Programs for solv- be broken; a stock of practically proven, heuris-
ing specific problems are usually developed by tic knowledge is built up [30].
the user himself and are not generally available Some examples of heuristic rules in process
or are available only in rare cases. development are:
Numerous programs for dealing with flow
sheet problems are available from universities, 1) Carry out each reaction or separation only
industry, or the commercial market (see Tables 2 once (e.g., volatiles should only be removed
and 3) [33]. once)
2) Remove substances which can be readily iso-
lated immediately after they are produced
Table 3. A selection of simulation programs for dynamic systems and do not spread them over the entire
[39–42] working-up process
Program Supplier 3) Design recycling paths so that, if possible,
they only result in small feedbacks
CHEMADYN BASF AG,
Ludwigshafen,
4) If possible, use substances which are in any
Germany case present in the reaction output as aux-
DIVA Prof. Dr. Gilles, iliary solvents since this may reduce the
University of Stuttgart, working-up costs considerably
Germany
SATU Höchst AG,
Frankfurt,
Combinatorial problems such as drawing up
Germany an initial version of a process or developing a
SIMUSOLV The Dow Company complex catalyst system result in a combinato-
Midland, Mi 48674, rial explosion (see Section 3.2). Time does not
United States
SPEEDUP Prosys Technology Ltd.,
allow all the possible solutions to be tested. The
Cambridge, human expert copes with such a combinatori-
England, al explosion by eliminating those possibilities
(since 15.10.1991 Aspen which seem to him to be unfruitful, concentrat-
Technology Inc.)
ing on those which seem to be feasible and using
heuristic knowledge which steers him towards
the optimum solution but does not guarantee it.
3.2.1.3. Expert Systems The main components of an expert system are
as follows (see also Fig. 12):
The difference between simulation programs
and expert systems is that in the former the mod- 1) Knowledge base: heuristic knowledge is re-
eling is rigidly embodied by an algorithm, while presented in this case by a system of symbols
in the latter, knowledge of the model is stored in 2) Knowledge editor: a module which assists
the knowledge base independent of the deduc- in changing, adding, or removing rules and
tive mechanism [43–45]. formalizes their evaluation
An expert system is a computer program 3) The deductive mechanism, which combines
equipped with knowledge and ability which can knowledge and problem data by deriving fur-
carry out complex tasks by imitating human in- ther data from the rules stored in the knowl-
telligence. Such systems are therefore particu- edge base
larly suitable for use in process development 4) input – output systems
 Process Development 17

gramming languages such as LISP (List Proces-

sor) were introduced. These are able to process
symbols and symbol structures using a computer
[48]. In addition a wide range of computer tools
became available. Although these have made
the development and handling of expert systems
easier, they have contributed almost nothing to
an understanding of the phenomena [49].
It is now beyond dispute that expert systems
Figure 12. Structure of an expert system [46] have justified themselves, although the eupho-
ria of the 1970s has given way to a more sober
This technique was first applied in 1969 for assessment. The belief that a lack of knowledge
describing and designing composite heat ex- and understanding can be overcome by an ex-
changer systems [47]. It was rule-oriented, i.e., pert system has been found to be completely er-
the expert system involved a series of rules roneous. On the contrary, an expert system can
whose combination led to different results. The only be used efficiently by experts.
system was controlled by weighting factors,
which were in turn dictated by success. The sys-
tem “learned” through this method of changing 3.3. Checking the Individual Steps
weights and was therefore capable of yielding
increasingly better results. Like the reaction step or steps, the cho-
sen preparatory and separation steps are first
checked individually and independently of one
another on a laboratory scale. Even at this stage,
new requirements may be imposed on the reac-
tion stage. For example, difficulty in removing
a byproduct may make it necessary to alter the
reaction conditions or modify the catalyst.
Along with the experimental work, e.g., test-
ing an absorber (see Fig. 13), thermodynamic
simulation calculations are carried out for the
unit concerned.
Laboratory experiments are generally carried
out with pure, well-defined materials. Thus, the
solvent selected for the absorption is used as re-
ceived from the chemical stores. This differs,
of course, from the subsequent situation, where
the solvent must be recycled for cost reasons
and inevitably becomes enriched with byprod-
ucts, some of which are unknown, however, at
this point in time.
The feedstock used is a synthetic mixture of
gases prepared from pure materials which also
does not contain byproducts. The experiment
shows whether it is physically possible to absorb
Figure 13. Checking the individual steps in the laboratory the useful product from the gas stream with rea-
using pure feedstock with the absorber stage as an example sonable amounts of solvent. If the absorber can
(see Fig. 14)
be simulated with the available material data in
a calculation run in parallel, work on this step
The first expert systems were laboriously will be complete. The unit is then another piece
written in FORTRAN, but the technique ac- which can be fitted into the jigsaw puzzle of the
quired a new impetus in the 1980s when pro-
18 Process Development

entire process, which is the next item which must tion, the solvent circuit may become enriched in
be considered in an integrated trial plant. byproducts and this may lower product purity.
In principle, these problems can only be
solved by mathematical simulation, but since
many of the quantities that are required for a
mathematical description are unknown, an ex-
perimental approach must be adopted.
Example 3.3. When a plant which includes
recycling streams (see Fig. 15) is operated, the
solvent circuit becomes enriched in a high-
boiling substance whose existence was not pre-
viously known because it was below the analyt-
ical limit of detection in the reaction product.
This high-boiling substance also has the trou-
blesome property of depositing when it reaches
a certain concentration in the solvent and forms
a coating in the column.
Since the high-boiling substance was not
known before, it was not possible to include it
in the mathematical simulation. Now that it has
been discovered as a result of operating an inte-
grated plant, it can be characterized and incor-
porated in the simulation.

Figure 14. An initial version of a process for oxidizing a

starting material with air and then isolating the product by
3.4.1. The Miniplant Technique
absorption and distillation
3.4.1.1. Introduction
Once all the individual steps have been suc- Improvement in computer programs for mod-
cessfully tested, it is possible to draw up a reli- eling processes has resulted in the integrated
able flow sheet of the entire process (see Fig. 15). miniplant technology acquiring ever increasing
However, if any subsidiary step is found not to be importance because the synergism between the
feasible at this stage, the process concept must miniplant technique and mathematical simula-
be changed. tion means that the margin of safety in scaling
This preliminary flow sheet, which has not up is as good as that obtained by setting up a
as yet been fully tested as a whole, can be used pilot plant [50–52], [53, pp. 179 –182].
to draw up an initial rough cost estimate (see The miniplant technique has the following
Section 4). characteristics:
1) The trial plant includes all the recycling paths
3.4. Testing the Entire Process on a (it is a so-called integrated miniplant) and it
Small Scale can consequently be extrapolated with a high
degree of reliability.
As Figure 15 shows, when the individual steps 2) The components used (columns, pumps,
are put together, recycling streams are created condensers, pipelines) are often the same
(in this case one gas and one solvent recycling as those used in the laboratory. In addition,
path). These recycling streams or feedbacks are they are largely reusable standardized com-
an economic necessity, but they raise new pro- ponents which have been tried and tested in
cess engineering problems. Thus, material re- continuous operation. For these reasons, in-
cycled to the reactor may drastically affect the vestment costs are low and flexibility is high.
activity and service life of the catalyst. In addi-
 Process Development 19

Figure 15. Example of a simplified process flow chart produced after testing the individual steps

3) The plant is operated round the clock on a 3.4.1.2. Construction

shift basis for weeks and it therefore must
be as fully automated as possible to keep op- The following design documents are required as
erating costs low, but not, however, at the a minimum for constructing a miniplant:
expense of flexibility.
The measuring and control instruments used 1) A process flow sheet showing all the quan-
are standard components which are nowa- tity flows in the miniplant and the associated
days generally connected to a small process temperature and pressure conditions
control system [54, pp. 183 – 186], thus mak- 2) Engineering flow diagram
ing it possible to carry out modifications in Required information:
the measurement and control rapidly. a) All equipment and machines
4) The entire plant is normally set up in an b) Internal diameters and pressure ratings of
extracted chamber so that explosion-proof pipelines and construction material
operation is possible and safety-at-work re-
quirements can be met more easily.
20 Process Development

c) The objectives of the instrumentation efficiently (because of excessive temperature

d) Information on the insulation of the equip- differences, foaming, etc.).
ment and pipelines Example 3.4. A miniplant produces a high-
3) List of measuring points (measuring point boiling residue at a rate of 1 g/h. This residue
number, specification, parts list, point of in- is removed from the bottom of a distillation
stallation) column with a distillation boiler capacity of
4) Safety concept, i.e., how the sensors (e.g., for 200 mL, so that the mean residence time is 200 h.
temperature, pressure, flow) are linked to the The time required to reach the steady state is
actuators (e.g., valves, pumps) by the safety consequently about 25 d (about 3 times the mean
logic residence time). Only then will it be possible to
assess this bottom residue and draw conclusions
about its fouling behavior, composition, proper-
ties, etc.

Specification of Material Balance. In Sec-

tion 3.2 it was shown that a miniplant should be
designed by scaling down a hypothetical large-
scale plant. The scale-down factor is determined
by finding the points in the process where the
smallest and largest quantity flows occur. The
levels at which these flows can be handled de-
termine the scale-down factor.
Normally only flows of
1 g/h and 10 kg/h
can be handled in a miniplant. At the same time,
the minimum flow is primarily determined by
the time required to reach a steady state (see Ex-
ample 3.4), while the upper limit of the flow is
set by the maximum hydrodynamic load of the
laboratory column used.
Example 3.5. In a hypothetical large-scale
plant producing 5 t/h, the minimum flow is
50 kg/h of a high-boiling substance which must
be removed and incinerated, while the maximum
flow is 100 t/h of the reaction throughput, which
contains the useful product as a 5 % aqueous so-
lution.
Scale-down factor:
Figure 16. Example of a miniplant
for the minimum flow: 50 kg/h ÷ 1 g/h =
50 000,
for the maximum flow: 100 t/h ÷ 10 kg/h =
Specification of Equipment Size. In a 10 000.
miniplant, the hold-up for a given throughput is Thus the miniplant should be designed to pro-
almost always greater than in an industrial-scale duce not less than 100 g or not more than 500 g
plant. The size of the vessels and column bot- of useful product per hour.
toms should therefore be minimized, otherwise
it will take too long for the miniplant to reach Specification of Type of Equipment. Un-
a steady state because of long residence times. like normal laboratory experiments in which the
In the column bottoms, which are particularly equipment operates only during the daytime,
critical, a hold up of 100 – 200 mL is normally miniplants must run continuously over a period
the lower limit. Below this level, a laboratory of weeks to be able to fulfil their task. This is
evaporator can usually no longer be operated due to the fact that the time required to reach
 Process Development 21

a steady state is often long and because it is monitor liquid levels, deposits, foam formation,
only possible to draw reliable conclusions about etc., than in the large-scale plant. Schulz-Walz
long-term effects such as corrosion, fouling, has given examples of measuring and metering
catalyst deactivation after a long time (1 week techniques for miniplants [55]. In the last two
to 6 months). decades, important advances have been made
Unlike laboratory plants, the requirement im- in sensor technology and, along with electron-
posed on the reliability of miniplant compo- ics and processor technology, this has consid-
nents are therefore very high and almost the erably simplified the miniaturization of sensors
same as for a large-scale plant. Occasionally this for miniplants (see Table 5).
sets very tight restraints on choosing equipment
and in some cases necessitates in-house devel-
opment if nothing suitable is available on the Table 5. Established miniaturization limits for instrumentation com-
market. ponents [51]

Measured quantity Principle of Established

measurement minimum
3.4.1.3. The Limits of Miniaturization measurement range
Gas flow thermal 0.02 – 0.6 L/h
Table 4 lists the minimum dimensions for which volumetric 2 – 200 L/h
it is possible to scale-up a process. Liquid flow thermal 2 – 30 g/h
Coriolis force 0.07 – 1.5 kg/h
magnetic inductance 0.6 – 6 L/h
volumetric 0.1 – 1 L/h for piston
Table 4. Examples of tried and tested miniaturization limits [51] meters
2 – 40 L/h for oval
Example Tried and tested limit disk meters
1 mL/h upwards for
Columns 30 mm diameter packed columns metering pumps
35 mm diameter columns with structured gravimetric 2 – 50 g/h (balance
packing, dual-flow plate columns range 0 – 5000 g)
50 mm diameter bubble-cap columns Temperature resistance change T min range 30◦ C
Pumps 1 mL/h injection pumps (maximum range
10 mL/h piston pumps −200 to 600◦ C)
100 L/h rotary pumps thermoelectric T min range 50◦ C
Belt filter 50 mm belt width voltage (maximum range
0 – 1000◦ C)
Centrifuges delivery 5 L/h
Level conductive >50 mm
Pipelines 1.5 mm diameter, metal or Teflon
capacitive >50 mm
hydrostatic >4 mbar
Differential pressure strain gauge 0 – 1.25 mbar
piezoresistive
Columns can still be operated effectively capacitive
down to 35 mm diameter when filled with an
ordered packing. For plate columns having one
bubble cap per plate, the limit is ca. 50 mm. At
still smaller dimensions, wall effects are difficult There are four stages in the automation of a
to suppress. miniplant:
Experimental equipment for handling solids Stage 1. At this stage, the electrical instru-
or dealing with extreme operating conditions is mentation of the miniplant consists of compact,
often not available. This often necessitates in- self-contained individual components, e.g., bal-
house developments and cooperation with spe- ances, metering controls, metering pumps, ther-
cialist firms, university institutes, and research mostats, temperature controllers, pH sensors,
companies. individual controllers, control valves, pressure
Expenditure on instrumentation is often very sensors, and plotters. Monitoring and operation
much lower for a miniplant than for the corre- is the responsibility of the shift personnel.
sponding large-scale plant. For this to be the Stage 2. At this stage the structure of the sen-
case, however, most of the miniplant compo- sors, actuators, and individual components is the
nents must be made of glass and therefore oper- same as in Stage 1, but the individual compo-
ate at low-pressure, thus making it much easier to nents may be connected to a data processing
22 Process Development

computer via a data interface. Monitoring and 3.4.2. Pilot Plant

operation is again the responsibility of the shift
personnel. It may be necessary to construct a pilot plant if
Stage 3. Here the system for logging the mea- one of the following conditions applies:
sured data is the same as in Stage 2, but the 1) The scaling-up risk is too large to proceed
lower-ranking self-contained individual compo- directly from the miniplant to the industrial-
nents are controlled and provided with set points scale plant. The reasons for this may be:
by a higher-ranking automation software, so that a) The process involves several critical
shift personnel are able to concentrate on moni- stages (e.g., handling of solids) which can-
toring tasks. not be described by physical models
Stage 4. At this stage the miniplant is b) a difficult or completely novel technology
equipped with a compact process control sys- is being used
tem. While the actuator and sensor system is the
2) It is necessary to provide representative
same as at Stage 1, all the other functions such as
product quantities, e.g., for the market
controlling, measuring, regulating, calculating,
launch, and these cannot be produced by the
processing data, optimizing, keeping records,
miniplant in a reasonable time
alerting, logging, operating, and observing are
performed by the process control system. The operation of the pilot plant should clar-
The choice of automation stage is decided ify all the issues which have not been fully dealt
during the miniplant design phase on the basis with in the miniplant.
of the specific plant requirements and boundary These may include, for example [56], [57]:
conditions as well as on the results of trials. The 1) Checking design calculations
higher automation Stages 3 and 4 are only justi- 2) Solving scale-up problems
fied if a miniplant is to be operated over a long 3) Checking experimental results obtained with
period of time. the miniplant
4) Measuring the true temperature profiles in
the reactors and columns under adiabatic
3.4.1.4. The Limitations of the Miniplant conditions
Technique 5) Gaining process know-how (dynamic behav-
ior of the plant, start-up and shut-down pro-
A disadvantage of the miniplant approach cedures)
emerges if individual steps in the process are 6) Producing representative sample quantities
subject to an unduly large risk in extrapolat- in fairly large amounts
ing from the miniplant to the industrial-scale 7) Precisely assessing fairly small flows (e.g.,
plant. For example, scale-up factors of ca. 104 residues)
are still not feasible for extraction and crystal- 8) Training the personnel who are to run the
lization steps even today. This disadvantage can plant
often be circumvented by isolating critical sec- 9) Improving the estimate of the expected ser-
tions of the process and working on them in an vice life
intermediate stage (pilot stage). If enough of the 10) Carrying out material tests under realistic
unmodified product can then be collected from conditions
an integrated miniplant at an acceptable cost for
it to be possible to operate, e.g., an extraction A pilot plant should be designed as a scaled-
column or a crystallizing plant on a pilot scale down version of the industrial-scale plant and
with a representative feed for a sufficiently long not as a larger copy of the existing miniplant
time, the entire plant can again be extrapolated [58]. Nowadays it costs almost as much to de-
upwards with a calculable risk. This may obvi- sign, construct, and operate a pilot plant as an
ate the need for constructing a cost-intensive and industrial-scale plant.
time-consuming pilot plant for the entire process The decision to build a pilot plant results in
[56]. considerable costs. As a rule of thumb, the ini-
tial investment is at least 10 % of the invest-
ment costs of the subsequent industrial-scale
 Process Development 23

plant. Moreover, if they use particularly toxic 1) A summary

substances, pilot plants now often require ap- 2) A basic flow diagram
proval, and this may considerably lengthen the 3) A process flow diagram and a description of
time required to put them into operation. the process
Once process development on a pilot scale 4) A waste-disposal flow diagram
has been successfully concluded, the pilot plant 5) An estimate of the capital expenditure
must be kept on stand-by until the industrial- 6) A calculation of the production costs
scale plant is running satisfactorily. Normally, 7) A technology evaluation
when the larger plant is started up, the pilot plant 8) The level to which the experimental details
is operated simultaneously so that any problems have been worked out
which occur in the former can be dealt with
rapidly. 4.1.1. The Summary

The summary must start with a precise defini-

4. Process Evaluation tion of the objective. This is followed by a short
description of the process and of the most im-
After each development stage (see Section 3.1) portant process stages.
the existing knowledge should be documented The chemistry of the process should be re-
and the status of the process evaluated. This in- presented by overall reaction equations, with the
volves preparing study reports and is now stan- assumed conversion and selectivity being spec-
dard in many enterprises. ified.
At this stage it is necessary to answer the Details of the result of the study, the produc-
question posed in Section 1 as to whether the tion costs, the capital expenditure for a specified
process is “better” and “cheaper” than that of annual production output with an indication of
the competitors. the accuracy of the estimate and the planned lo-
cation of the plant, if known, are other important
points.
4.1. Preparation of Study Reports The summary concludes with information on
the state of knowledge and on the major risks.
The study report documents knowledge about
the status of the development of a process af- 4.1.2. Basic Flow Diagram
ter specified time intervals or specified develop-
ment stages to provide the basis for making a The basic flow diagram provides a quick
decision. The objective of the study report is to overview of the total process. It shows the in-
answer three important questions: put, output, and recycling streams and identifies
the individual process stages (see Fig. 14).
1) Can the production process be implemented
in this way in principle? 4.1.3. Process Description and Flow
2) What is the return on investment? Diagram
3) How big is the risk in economic and techno-
logical terms? All the individual steps of the process should be
At least three project study reports must be described as accurately as possible on the basis
completed in the course of the development of a of all the information available at the time, ref-
process: erence being made to the process flow diagram
(see Fig. 15).
1) One at the beginning of process development The latter should clearly show the process in
2) One in the course of working out the details detail and the composition of the flows.
of the process The following information should be in-
3) One at the end of process development cluded in the process flow diagram:

A study report should include the following 1) All equipment and machines (columns, ves-
items: sels, heat exchangers, reactors, etc.)
24 Process Development
Table 6. Example of a material balance table

2) All product streams (with numbering) 2) Incineration in a power station or an inciner-

ation plant
Beneath the process flow diagram the material 3) Central hazardous waste incineration
balance should be presented in a clearly laid out 4) Waste-disposal site
table (see Table 6).
When development is at an advanced stage
or is complete, the process flow diagram should
4.1.5. Estimation of Capital Expenditure
show the following additional information:
1) Identification of the energy carriers (type of 4.1.5.1. Introduction
steam, cooling water, compressed air, elec- An important reason for preparing study reports
tric power, nitrogen, deionized water, heat- is to determine the capital expenditure (invest-
ing gas, refrigerants, etc.) ment) of a project [59], [60]. This is made up
2) Details on the construction materials, size of the independent location ISBL (inside bat-
and power of the equipment and machines tery limits) process plant costs such as invest-
3) Typical operating conditions (e.g., pressure ment in the production plant and the associated
and temperature in columns and pipelines) control rooms, laboratories, employee facilities,
4) Important fittings tank farms, and loading and unloading stations,
5) Purpose of instrumentation as well as the location-dependent OSBL (outside
battery limits) costs of all the service facilities,
such as investment in storage buildings, cooling
4.1.4. Waste-Disposal Flow Diagram towers, and waste-disposal facilities. Infrastruc-
ture costs are also to be included (see Fig. 17).
The waste-disposal flow diagram illustrates the The accuracy of the investment cost estimate
waste streams for which there is no further eco- depends critically on the level of maturity the
nomic use in the plant (see Section 1.3) and in- process has reached [62].
dicate their flow numbers from the process flow
diagram. Precise information should be given on
4.1.5.2. ISBL Investment Costs
the toxicity, degradability, water danger, flash
point, ignition temperature, MAK, odor thresh- Of the many methods for determining the cap-
old, etc. of the individual components, their mass ital required to set up a plant, a method of es-
flows, and the state of aggregation of the flow timation which derives the costs of a new plant
(viscosity, sediment content, etc.). This infor- from the known costs of existing plants with an
mation makes it possible to specify the type of accuracy of ± 30 % is now firmly established
waste disposal required, for example: (the additional-cost calculation method). Such
a method is only possible because the structure
1) Sewage treatment plant
of chemical plants is always very similar, i.e.,
 Process Development 25

Figure 17. Subdivision of a project into ISBL, OSBL, and infrastructure activities [61]

they are made up of interlinked pieces of equip- mined from the average value of the machines
ment of similar construction whose number does and equipment (see Fig. 18).
not vary widely. For this reason, the overall in- This results in the following simple relation-
vestment cost of the entire plant when installed ship:
exceeds the average value of the machines and
equipment (= total machine and equipment costs directplantcosts = overallfactor×
divided by the number of machines and pieces Σ (valuesof machinesanditemsof equipment) . (4.1)
of equipment) by a factor which depends on the
capacity. Since direct plant costs are always estimated
The total machine and equipment costs are for some date in the future, it is necessary to de-
determined from the parts list, while the costs of termine the effect of increases in price. Use is
individual units can be obtained from in-house therefore made of price indexes which are pub-
databases or directly from the manufacturers. lished for chemical plants in the individual coun-
Example 4.1 [59]. The overall factor for the tries (see Table 7). The price index i determined
chemical plant construction projects handled by by extrapolation is divided by the index i0 at
BASF in 1984 was on average 3.9; i.e., the direct the date of the cost determination. The ratio i/i0
plant costs of an average chemical plant is 3.9 is then a measure of the increase in investment
times the machine and equipment costs. costs to be expected.
The overall factor is essentially determined The direct plant costs make no allowance for
by the following items: the cost of engineers (usually 10 – 20 %) or for
contingencies (ca. 5 – 10 %).
1) Operating conditions (pressure, tempera- This method of estimation has the disadvan-
ture) tage that while the machine and equipment costs
2) Type of construction materials used (steel, can be determined fairly precisely, the average
stainless steel, special materials) value of these items is multiplied by a compara-
3) Size of machines and pieces of equipment tively inexact overall factor. The accuracy can be
considerably increased by breaking the overall
According to Miller [63], the effect these factor down into individual factors.
items have on the overall factor can be deter-
26 Process Development

Figure 18. Overall factor for the direct plant costs as a function of the average value of the machines and pieces of equipment
in 1984 [59]:
a) Best-fit curve; b) Confidence limits for a statistical confidence interval of 95 %

Table 7. Price indexes i for chemical plants in Germany , [64, p. 280] while for larger plants it is likely to be close to
Year i
the lower limit.
If the location is known, the OSBL costs can
1976 100
be estimated very accurately.
1980 125.3
1981 132.9
1982 140.5
1983 144.4 4.1.5.4. Infrastructure Costs
1984 147.2
1985 150.4
1986 154.0 The infrastructure costs depend on the location
1987 159.6 of the plant. They are not, however, related ex-
1988 164.7 clusively to the project since the infrastructure,
1989 170.9
such as workshops, the central store, and the
canal system, is also used by other production
facilities. Obviously, this type of cost can only
4.1.5.3. OSBL Investment Costs be specified if the precise position of the plant
at a location is known.
The OSBL investment depends on the loca-
tion of the plant. It includes, for example, the
cost of storage buildings, cooling towers, waste-
4.1.6. Calculation of Production Costs
disposal facilities, linking the OSBL facilities to
the location in question (e.g., water and electric- The cost effectiveness of a new process depends
ity supply), etc. Since the question of location is on the production costs of a product (unit of
generally still unresolved at an early stage in de- currency divided by unit of quantity) (see Sec-
velopment, it is not possible to specify the OSBL tion 4.2). The following information is required
costs explicitly but they can be tentatively taken to determine these costs:
as 20 – 30 % of the ISBL costs. For small plants
the percentage tends to lie near the upper limit, 1) Material balance
 Process Development 27

2) Waste-disposal flow diagram this does not overcome the problem of the im-
3) Utilities pact of the proposed plant on the market – it only
4) Investment shifts it. The significance of raw material costs in
cost estimation varies widely. Thus, in processes
The production costs can be calculated by involving a high level of material refinement (as
adding up the following items: is the case for some pharmaceuticals) or in those
1) Feedstock costs (raw materials and auxil- with high specific energy costs (e.g., chlorine
iaries) production), precise determination is not as im-
2) Manufacturing costs, which can be subdi- portant as in the case of cheap, mass-produced
vided into: products, such as petrochemicals, manufactured
in large plants.
a) Utility costs
Credits for byproducts can be set against the
b) Waste-disposal costs
raw material costs from the outset.
c) Staff costs (wages and salaries)
d) Maintainance
e) Miscellaneous costs (concern overheads) 4.1.6.2. Utilities
3) Capital-dependent costs (depreciation, inter-
est, insurance taxes) For large-scale processes, the cost effectiveness
depends appreciably on the cost of the utilities.
The combination of utilities required in the plant
therefore acquires considerable importance and
consideration must be given to the ideal thermal
coupling even at an early stage in the process de-
velopment since it can have a considerable im-
pact on the configuration of the process [66]. For
example, whether the energy derived from con-
densing vapors can be utilized depends on the
behavior of the materials. Clean and stable con-
densates usually present no problem in using the
heat of condensation to raise the temperature of
cold streams, but in the case of products which
tend to form deposits (by decomposition, poly-
merization, etc.) optimum thermal utilization of
Figure 19. Temperature – enthalpy diagram this type of energy is not possible since the va-
pors must be quenched to prevent deposits and
this results in loss of energy. This type of ques-
4.1.6.1. Feedstock Costs tion should be resolved in the miniplant phase.
A suitable aid for systematizing such con-
The raw material requirements are given by the siderations is Linhoff analysis [67], [68] (see
material balance for the chosen capacity, but to → Pinch Technology). The Linhoff analysis de-
determine the feedstock costs the prices of the termines the process configuration for which the
raw materials are required. The change in raw energy supplied from or lost to the surround-
material prices while the process is being devel- ings is a minimum. The problem which must be
oped can be estimated, but it is more difficult to solved is how to couple the heat sinks and heat
allow for the fact that the erection of the plant sources within the plant such that minimum en-
will have an impact on the raw material mar- ergy consumption is achieved. The initial step in
ket and consequently alter the price structure. the calculation is to determine the so-called com-
Therefore, in some cases it is necessary to enter posite curves and pinch temperature, which is
into negotiations with the raw material suppliers the temperature above which heat can be lost and
at an early stage. If the raw materials originate below which heat can be absorbed (see Fig. 19).
from another plant in the same concern, market For a given interconnection system, the Lin-
prices are replaced by transfer prices [65], but hoff analysis shows how far the heat utilization
28 Process Development

is from being ideal. However, there are often rea- An initial estimate of the steam requirement
sons for not choosing an ideal interconnection: of a column is given by:
1) Start-up operations often require start-up
t/hsteam = 1/5 (reflux + headproduction) t/h (4.2)
heat exchangers, and this means higher in-
vestment costs This is based on the assumption that:
2) Energy utilization only becomes possible as
a result of increasing the column pressure, 1) The thermal combination is optimum (e.g.,
and this means higher investment costs and the hot discharge from the bottom is used to
may result in material problems (decompo- heat the feed)
sition, side reactions, etc.) 2) Normal organic liquids are involved (heats
3) Problems associated with the formation of of evaporation ca. 110 kWh/t)
deposits during heat transfer, which necessi-
tates direct heat removal (quenching) Electrical energy is more than twice as ex-
Generally, as the plant capacity increases, re- pensive as steam, and because of the explosion
ducing the operating utilities (i.e., optimizing the hazard it is normally not suitable for heating
thermal combination) rather than increasing the columns. Its most important applications are in
investment will have a more beneficial effect on feeding material (pumps, impellers, compres-
production costs. sors) and in carrying out operations such as size
The energy analysis is used to find the reduction, mixing, and separation.
amounts of energy required or released and to The pump delivery N required can be esti-
calculate the energy per unit quantity. The deter- mated using the following formula [70]:
mination of energy costs requires information on V̇ ·H ··g
the prices of the different types of energy, which N /kW = (4.3)
3600·103 ·ηtot
depend on the quantities required and the loca-
tion. where

Steam is the energy carrier most widely used V̇ = the useful feed flow in m3 /h,
for heat columns because it offers the following H = delivery head in m,
advantages:  = density in kg/m3 ,
g = acceleration due to gravity (10 m/s2 ),
1) It is inexpensive since it can often be gener- η tot = overall efficiency.
ated directly in the synthesis section of the
plant The overall efficiency is made up of the
2) There are no explosion hazards provided the mechanical and the electrical efficiencies:
hot pipe surfaces are properly insulated
3) The temperature range covered is large (see ηtot =ηmech ·ηelectr. (4.4)
Table 8) For rotary pumps, the overall efficiency is
ca. 0.7 – 0.8 while for piston pumps it is
Table 8. Enthalpy and temperature of some frequently used steam
pressures [69] 0.8 – 0.9. For small pumps (< 5 m3 /h), the over-
all efficiency may, however, be much lower.
Gauge pressure, p, Temperature, ◦ C Heat of
bar vaporization ∗, Another application for electrical energy in
kWh/t chemical plants is in auxiliary electrical heating
0 100 627 systems (e.g., frost protection).
1.5 127 605
4 153 585
16 204 534
Fuels are only used in chemical plants if
25 226 508 streams must be heated to such high tempera-
40 252 477 tures that steam is no longer economical (the
100 312 361 limit is ca. 100 bar of steam, i.e., 310◦ C). How-
∗ With indirect heating, only the difference between the ever, the use of fuels in chemical plants always
incoming stream and the discharged condensate can be utilized. entails a risk from the ignition source.
 Process Development 29

Only the net calorific value of the fuel can be The cooling water required to remove a given
used since the steam formed can generally not quantity of heat depends on the temperature of
be condensed. the water and its quality. If more detailed infor-
A few typical values of the standard enthalpy mation is not available, it is sufficient as a first
of combustion are given in Table 9. approximation to determine the amount of cool-
ing water required by assuming its temperature
Table 9. Standard enthalpy of combustion at 25◦ C [71]
rises by 20◦ C.
Substance kWh/t For high product temperatures, an alterna-
tive to water cooling is to use air coolers. Al-
Acetaldehyde 7 360
Acetic acid 4 040 though these require a higher investment than
Carbon 9 110 water-cooled condensers, they are often cheaper
Carbon monoxide 2 810 to operate and are less prone to fouling.
Ethane 14 440
Ethanol 8 250
However, unless there is a shortage of cooling
Formaldehyde 5 190 water at a known location, it is generally suffi-
Formic acid 1 590 cient to assume water cooling in an initial cost
Hydrogen 39 740 effectiveness calculation.
Methane 15 460
Methanol 6 310
Refrigeration Energy. If temperatures be-
low approximately 25 – 30◦ C must be achieved
in a process, energy must be expended on refrig-
Cooling water can be drawn from a body eration. Refrigeration is generally achieved by
of surface water or from the groundwater [72]. compressing, cooling, and adiabatically expand-
Compared with surface water, brackish water or ing coolants in the plant itself; i.e., the refriger-
seawater has the advantage of a more constant ation requirement can be expressed as a require-
temperature and a lower infestation with algae, ment for electrical energy and cooling water. To
but requires special measures to be taken against remove a given quantity of heat, about 20 – 50 %
corrosion. A recirculating procedure (Fig. 20) is of that quantity, depending on the temperature
preferred to a once-through system, with only level required, must additionally be expended in
sufficient cold river water being used for top- the form of electrical energy, thereby increasing
ping up as is required to maintain the speci- the amount of heat which must be removed to
fied temperature (the advantages of this are that about 120 – 150 %.
economies in river water are possible in the cold
season, the minimum temperature can be kept Compressed Air. Small quantities of com-
constant, and high flow rates in the circuit pre- pressed air are drawn from pipelines. An ex-
vent corrosion and fouling). ample is the control air required by closed-loop
control instruments (the rule of thumb is 1 m3 /h
(STP) of control air per instrument).
Larger amounts are produced by an air com-
pressor in the plant and the requirement can then
be expressed in electrical energy [73]. The power
N expended on compressing gases is given by:
 n−1

n P2 n
Ntheo =P1 ·V̇1 · · −1 (4.5)
n−1 P1

where
n = cp /cv ,
P1 or P2 = initial or final pressure, respectively,
cp or cv = heat capacity of the gas at constant
Figure 20. Example of a recirculation system for cooling
a chemical plant with river water pressure or constant volume,
V̇ 1 = initial volumetric flow.
30 Process Development

For air compression (n = 1.4, overall effi- should have a sufficiently high biological degra-
ciency η ≈ 0.7): dation rate, regardless of its total content in the


stream. This requirement prevents individual un-
N/kW = 0.14·V̇1 / m3 /h · degradable substances being lost in an otherwise
  readily degradable mixture. To assess whether
(P2 /bar)0.286 −1 (4.6) disposal of a stream via a sewage treatment plant
is possible, the following information must be
collected for each substances it contains:
1) BOD5 and COD value [79]
4.1.6.3. Waste-Disposal Costs 2) Zahn – Wellens test [80]
3) Bacteria and fish toxicity [81–84]
The waste-disposal situation has become much 4) Water hazard class [85]
more difficult for chemical plants, a trend which
is continuing. The delivery of effluent and wastes Incineration Plants [86, pp. 213 – 217].
to central sewage plants, waste incineration Waste streams which consist only of compounds
plants, or waste-disposal sites is restricted by of- containing carbon, hydrogen, and oxygen as
ficial regulations, and as a result waste-disposal aqueous solutions with a concentration >10 %
costs have risen dramatically in Germany in re- are ideally suitable for disposal in an incinera-
cent years (see Fig. 21). tion plant or a power station (if the content is
>10 %, the heat of incineration is roughly equal
to the heat of evaporation of the water).
Disposing of waste by incineration is more
difficult if other elements such as nitrogen, chlo-
rine, sulfur, and metals are present. Such wastes
can only be disposed of economically in a cen-
tral waste incineration plant equipped with suit-
able absorption systems for NOx , HCl, SO2 , etc.
[87], [88].

Waste-Disposal Site. The restrictions im-

posed on wastes which can be dumped are con-
tinuously increasing, and new dumping space is
no longer available. Consequently, this waste-
disposal option is virtually no longer available
for new processes.
Figure 21. Expenditure on environmental protection in-
vestment in the chemical industry in Germany [74, p. 303]

This poses a challenge to process develop- 4.1.6.4. Staff Costs

ment. The slogan “built-in rather than add-on
environmental protection” expresses the current Operating staff costs in chemical plants can vary
demand for environmental pollution to be re- widely. For large automated plants, they may
duced in the manufacturing process itself and for be less than 3 % but for small batch processes
efforts to be made to put even the byproducts of they may amount to 25 % of the manufacturing
a chemical process to good use if possible [75– costs. Often they are 5 – 10 % of the manufac-
78]. turing costs.
In new plants, whether batch-operated or of
Sewage Treatment Plant. Effluents dis- the single-train type, the trend is towards the
charged into a sewage treatment plant must fullest possible automation. The introduction of
be readily capable of biological degradation and process control engineering has made it possible
must not be toxic to the bacteria in the plant. In to operate and monitor virtually every aspect of a
addition, each substance in an effluent stream plant using pictorial displays [89, pp. 158 –162].
 Process Development 31

Almost regardless of the size of the plant, a well- 4.1.6.6. Overheads

planned and automated single-train installation
requires the following minimum staff: The overheads of the company consist of the
total costs of the central facilities (research, per-
1) One shift foreman or his deputy sonnel, energy-supply department), works man-
2) Two shift employees for the control room agement, staff facilities, internal plant traffic,
3) One shift employee for outside operations etc. These costs are very dependent on the struc-
4) One shift employee as a reserve ture of the company. As a rule they are ca. 5 –
20 % of the combined energy, staff, and depre-
This minimum staff of five should be multi-
ciation costs.
plied by a factor, specific to the company, which
reflects the type of shift model, the sickness
statistics, and vacations to give the total num-
4.1.6.7. Capital-Dependent Costs
ber of employees required. Special operating re-
(Depreciation)
quirements such as frequent changes in produc-
tion schedule, periodic start-up and shut-down
The plant capital can be depreciated by various
of the plant, loading and unloading of tankers
methods [92]. For a cost efficiency calculation
may make additional staff necessary.
it is generally sufficient to assume linear depre-
ciation over a ten-year utilization period. The
annual depreciation is then 10 % of the plant
4.1.6.5. Maintenance Costs
capital:
Annual repair and maintenance costs generally I
Depreciation = (4.7)
amount to 3 – 6 % of the invested capital [90]. 10·N
About half of this amount is material costs and
where
the rest wages. Their long-term increase can
therefore be assumed to be between that of the I = total investment in currency units
investment costs (see Table 7) and that of the N = nominal capacity in kg/a
wage rates (see Table 10). Experience shows that
high repair costs are incurred shortly after a plant
is put into operation, in eliminating start-up diffi-
4.1.7. Technology Evaluation
culties, and after a prolonged operation because
of the need to carry out major repairs. Depending Technology evaluation should provide informa-
on how new a process is, annual costs of 5 – 10 % tion on the technical risk associated with a pro-
of the estimated total investment should be al- cess. This risk is equivalent to the economic
lowed for in an initial estimate. damage which would occur in the extreme case
if the process did not work at all or could only be
rendered operational by carrying out subsequent
Table 10. Gross hourly earnings of an average worker in the chem- improvements.
ical industry in Germany , p. 286[91] Technology evaluation should also reveal
Year Earnings, ¤/h whether the individual unit operations or the
equipment and machines used are of techni-
1980 7.22
1981 7.68
cally established types and whether particular
1982 8.10 risks are incurred by breaking new ground (size,
1983 8.47 construction, material, scale-up factor, etc.). It
1984 8.72 is also necessary to include a statement as to
1985 8.99
1986 9.30
whether the process is based on a technol-
1987 9.69 ogy with which the company is familiar (e.g.,
1988 10.06 high pressure, gas-phase oxidation, phosgena-
1989 10.42
tion technology, etc.) or whether fundamental
1990 11.11
innovations are involved. If limits are exceeded
(e.g., if the dimensions of the largest extraction
32 Process Development

column previously operated by the company are train plant increases as the number of process
considerably exceeded in the new project), the steps and the probability of failure of individ-
limits which have been tested should be speci- ual steps increase: a single-train plant can only
fied. function if each step is operational.
This information should be elaborated in de- If partial streams are fed back to earlier
tail as early as possible in order to recognize, stages, the coupling between them is particu-
consider, and evaluate potential risks and causes larly strong. This is also true of the energy cross-
of failure. During operation of the miniplant, links within a process (see Section 4.1.6.2). The
close attention should be paid to deposits, fre- increase in risk as the number of process steps
quently occurring malfunctions, etc., since such increases is independent of whether the process
problems often subsequently lead to insupera- is continuous or batch.
ble difficulties in the industrial-scale plant (e.g., In practice, various options are now adopted
foaming, emulsion formation, aerosol forma- in resolving this dilemma (see Figs. 22 and 23):
tion, etc.).
The technology evaluation gives a list of 1) Decoupling by inserting tanks for intermedi-
weak points arranged in order of technical risk ates (the greater the buffer tank, the greater
(e.g., catalyst service life, corrosion, etc.). This the independence). For larger quantities of
list can then be used to take steps which reduce product or gaseous intermediates, this ap-
such risks. In principle, two steps are possible: proach is virtually impracticable.
2) Decoupling by installing parallel arrange-
1) Expenditure on R & D can be increased at the ments of equipment which is particularly
weak point (e.g., the unit concerned may be susceptible. For example, filters and fittings
checked on a pilot scale or a well-established that are susceptible to failure should be pro-
solution may be sought). This option should vided with a by-pass if safety considerations
always be chosen if, in an extreme case, fail- permit (e.g., heat exchangers in the main flow
ure of the unit concerned would be accom- in which severe fouling is to be expected).
panied by a total investment loss.
2) Failure scenarios can be developed, i.e., what
can be done later if problems occur or if a unit
fails completely?
Normally, these two options have to be
weighed against each other, i.e., it is necessary to
clarify whether it is more economic to minimize
the risk by increasing R & D expenditure (e.g.,
by building a pilot plant), or to reduce or elimi-
nate it at a later date by technical measures (e.g.,
additional instrumentation or backup units).
Some general aspects, which may be useful
in assessing the technical reliability, are summa-
rized below.
From the estimate of the risk and the mea-
sures it suggests, the economic effect can be
quantified very exactly (surcharges on the de-
velopment costs or on the investment and repair
costs). This transforms technical reliability into
cost effectiveness considerations and results in
the possibility of expressing the risk of a devel-
opment project in monetary units.

Improving Technical Reliability Figure 22. Example of decoupling a single-train plant:

Decoupling Process Steps. The susceptibil- a) Intermediate tank to smooth out variations in feed flow;
b) Recycling to smooth out variations in feed flow
ity to failure of a process carried out in a single-
 Process Development 33

If solids are processed, the process must be

assessed with particular care. The scaling-up of
equipment for separating solids should be sub-
jected to particularly careful scrutiny. Even the
bunkering and transportation of solids present
difficulties which are often specific to the mate-
rials.
If new technology is employed, the opera-
tional safety can be expected to be lower than
that of well-established processes. It can be in-
creased if similar types of equipment are used in
the experimental plant as in the industrial-scale
plant. This allows scaling-up to be possible but
in the case of reaction equipment, it is not always
the case.
Materials to be Processed. If toxic, inflam-
mable, caustic, or otherwise hazardous sub-
stances are to be handled, a higher degree of op-
erational safety is required from the outset than
in the case of harmless substances.
Corrosion is an ever-present risk and neces-
sitates the use of special materials. In making an
initial selection, use may be made of published
tables (see → Corrosion), but in most cases it
Figure 23. Decoupling of a single-train plant by including is necessary to carry out corrosion tests. Trace
redundant equipment impurities can lead to considerable difficulties if
A) Filter station with single by-pass; B) Pumping station they accumulate at particular points in the equip-
with A and B pump; C) Preheater with flushing facilitya)
Auxiliary preheater; b) Double block and bleed; c) Preheater ment. The most reliable information in this con-
nection is provided by operating an integrated
Each measure which is aimed at increasing miniplant.
the uptime of a single-train plant results in an If the flow rate of mixed phases is high, ero-
increase in investment costs. In practice it is of- sion of the wall material can be expected. The
ten the case that such steps are only taken after extent of erosion can often be lowered by the
the plant is put into operation for the first time. choice of material and by effective design. In-
Choice of Process Steps which can be Safely creases in wall thickness are generally of no help
Scaled-Up. An example of the dependence of because the erosion usually takes the form of
technical reliability on the operational method pits. Miniplant experiments are of no assistance
employed is mass transfer. While simple stan- in this connection since they are usually not re-
dard equipment can be used for rectification, a presentative in hydrodynamic terms.
wide variety of equipment in which phases are In the case of moving solids, not only the
mixed in several stages and then separated has walls of the equipment but also the solids them-
been developed for extractions. Such equipment selves are subject to abrasion. If a process is in-
generally requires more expenditure for safe op- tended to give a particular size, the abrasion loss
eration than distillation columns. may have a considerable effect and, moreover,
The complexity of an operation increases as an increase in this abrasion is always expected
the number of substances and phases it involves on scaling up a process.
increases and as the operating conditions be- Solids often deposit inside equipment and re-
come more extreme. Thus, the presence of two sult in fouling. Valuable information about this is
liquid phases requires special precautions to be provided by miniplant results. In heat exchang-
taken. The same applies if particularly high or ers, allowance is made for fouling by increasing
low temperatures or pressures are involved. the heat transfer resistance by empirically deter-
mined amounts (fouling factors) (see Fig. 24).
34 Process Development

der to regenerate the contaminated catalyst, but

fluidized-bed catalysts can usually be regener-
ated in the reactor. Often it is also possible to
remove, regenerate, and replace the catalyst in
a sidestream while the plant is operating. Foul-
ing that arises both from the material being pro-
cessed and from the material of which the equip-
ment is made is more troublesome. For example,
in many polymerization reactors, wall deposits
can be suppressed by suitable choice of the wall
material and processing method. In such cases,
steps should be taken to ensure that the same
wall material is used in the trial plant and in the
Figure 24. Effect of the fouling factor Fi on the heat trans-
industrial-scale plant.
fer coefficient k If wall erosion or deposit growth occurs
F 1 = 7×10−4 m2 K W−1 (typical value for river water); at a constant rate, the difficulties can be ex-
k 0 = 1000 W m−2 K; pected to decrease in bulk-storage equipment
Q̇ = k A ∆T ;
1/k = 1/k 0 + Fi ;
and pipelines with increasing scale-up but to re-
Fi = Si /λs main unchanged in columns and heat exchangers
where Si = thickness of fouling layer; λs = thermal conduc- (equipment with large surface areas).
tivity of fouling layer; Q̇ = heat flux; A = surface area of heat Machines. Since machines have moving
exchanger
parts, they are always a potential source of mal-
If the expected operating time is inadequate, function and the reliability of a process de-
a back-up heat exchanger must be installed in creases as the proportion of machines it em-
parallel. In larger plants, heat exchangers are ploys increases. In general, machines having re-
usually subdivided into several units for design ciprocating parts are regarded as less reliable
reasons. Thus, if suitable fittings (double block than those employing only rotating parts. For
and bleed) are provided to ensure that each heat this reason it is particularly important to pro-
exchanger can be cleaned while the plant is op- vide a reasonable level of back-up. Because of
erating, the effect on the availability of the en- their relatively low cost, duplicate full-capacity
tire installation is small (see Fig. 23). However, pumps are usually installed, but occasionally
this is only true provided the cleaning frequency two pumps having different functions can be
is technically feasible. The design of heat ex- provided with a common backup. In the case
changers and the cleaning operations should be of reciprocating machines, in addition to a com-
carefully matched to the type of fouling, and ex- plete backup (2×100 % capacity), 2×66 % ca-
tensive automation is always worthwhile in the pacity, or in exceptional cases 3×50 % capacity,
case of large pieces of equipment. may also be considered.
If fouling is expected in columns, larger Turbo machines are usually installed at
columns should be designed for whole-body ac- 1×100 % capacity, but provision is made for
cess to each plate, while in small columns a rapid replaceability of parts which are subject
manual-access hole should be provided for each to wear. Alternatively, two machines with 50 %
plate to allow cleaning. Furthermore, designs capacity are often installed.
should be used whose operation is relatively in- Instrumentation. The narrower the limits in
sensitive to fouling and which can readily be which quantity flows and operating conditions
cleaned. have to be kept for reliable operation, the higher
The choice of reaction equipment can be will be the requirements imposed on the instru-
considerably influenced by fouling. As foul- mentation.
ing increases, the pressure drop in fixed-bed To increase plant availability, provision is
reactors increases, whereas fluidized-bed reac- made for n measurements instead of one mea-
tors are virtually unaffected. Moreover, fixed- surement in the case of critical circuits and con-
bed reactors must be opened and emptied in or- trol loops. The complete unit will operate as
specified provided m of the n measurements
 Process Development 35

(m<n) are in the “good” range (m of n circuits). grated miniplant) should be specified and the du-
The greater m is, the higher will be the safety ration and conditions of the experiments given.
margin and the lower the plant availability. In The aim is to arrive at an as objective assess-
the case of chemical plants, “two out of three” ment as possible of the quality of the data ob-
circuits have been found to be satisfactory, thus tained since this is of major importance for the
providing a compromise between safety margin accuracy of the project cost estimate.
and availability.
Multiple-Train Design. According to Equa-
tion (2.17), the increase in investment costs 4.2. Return on Investment
on scaling up the plant is approximately
proportional to the capacity to the power After the individual cost factors have been esti-
2/3. Consequently, the specific investment costs mated, a rough cost effectiveness calculation is
for a single-train plant are approximately 20 % possible. This should provide information on the
lower than for a twin-train plant of the same to- return on investment (= ratio of profit to capital
tal capacity. In addition, the personnel costs are employed) for the planned project. Investment
lower for a single-train plant than for several in- in the process under consideration will be prof-
dividual plants of the same total capacity. How- itable if the sum of the revenues exceeds the total
ever, certain plant components can often not be outlay and the profit (revenue – outlay) makes it
scaled up to the required total capacity with an possible to amortize and pay reasonable interest
adequate safety margin. Thus, the maximum ca- on the capital invested.
pacity of polymerization vessels may be limited The return on investment provided by a pro-
because a breakdown of the cooling system must cess can be increased by minimizing the pro-
be insufficient to bring about dangerous operat- duction costs. Exposing the main cost factors
ing conditions, or it may be impossible to in- (raw material, energy, waste disposal, personnel
crease the size of an oxidation reactor for cost costs, depreciation) will therefore indicate the
reasons. In such cases, the total capacity is pro- direction the development should take in order
vided by using the requisite number of individual to improve the process (see Table 11).
pieces of equipment.
A significant advantage of multiple-train de- Table 11. Examples of measures which may result in the reduction
of the main cost factors
sign is that it is easier to cope with a temporary
reduction in load, which would normally mean Main cost factor Measures

shutting down and then starting up again. If the Raw materials a) Increase selectivity of catalyst
plant is underloaded, the specific consumption b) Minimize processing losses
c) Use raw materials of lower quality
figures of multiple-train plants are lower. Fur- Energy a) Optimize the thermal integration
thermore, the capacity of such plants can more b) Use multistage evaporation
easily be increased at a later date, particularly if c) Extraction instead of distillation
this has been planned from the outset. The ad- d) Air condenser instead of water
condenser
vantages of multiple-train design can, of course, Waste disposal a) Integrated waste utilization
only be exploited if it is possible to link any line b) Choice of location
in a process stage with any of the lines of the sub- Depreciation a) Simplify the reaction control and
work-up
sequent stage. For this purpose manifolds will be b) Reexamine choice of materials
needed. c) Reexamine dimensioning of pipeline
system
d) Exploit synergies of the location

4.1.8. Assessment of the Experimental Work

Minimizing the production costs does not,
This part of the project study should contain in- however, maximize the return on investment if
formation on whether the existing experimental the full capacity of the plant is not reached.
results have been obtained in-house or are de- In such cases, a plant with lower capacity and
rived from third parties. If in-house knowledge higher production costs might provide a larger
is available, the test units in which the exper- return on investment despite being less econom-
iments were carried out (e.g., laboratory, inte- ical.
36 Process Development

Some of the indicators frequently used in case, the sales are, of course, unchanged but the
practice to assess return on investment are de- production and investment costs are:
scribed below.
P C2 =P C1 −∆P C
I2 =I1 −∆I (4.10)
4.2.1. Static Return on Investment

In its simplest form, the return on investment cal- where

culation widely adopted in practice relates the PC 1 , I 1 are the production and investment costs before the
change,
expected annual profit to the capital invested: PC2 , I 2 are the production and investment costs after the
change.
(E −P C)
R= ·100 (4.8)
I
The return offered by a particular approach is
then given by the ratio of the changes:
where
∆P C
R is the percentage return in %/a R12 = (4.11)
E is the annual sales proceeds (= amount sold ∆I
multiplied by production or transfer price) in
monetary units per year
The change in the overall return resulting from
PC is the annual production costs in monetary units per adopting this particular approach is then:
year
I is the capital expenditure in monetary units R1 ·I1 +∆P C
R2 = (4.12)
E − PC is the annual profit in monetary units per year I1 +∆I
The weakness of static return-on-investment in-
The method can be made more informative by dicators is that they take a short-term view and,
taking the sales into consideration separately: since they do not allow for future changes, they
favor short-lived investments. Their advantage
(E −P C) E
R= · ·100 (4.9) is that they can easily be determined.
E I
The first factor indicates the sales success and
the second the capital turnover. 4.2.2. Dynamic Return on Investment
Instead of R, use may also be made of the
reciprocal value, i.e., the payback time 100/R. The dynamic return on investment method is su-
This indicates the time required to recover the perior to the static method because allowance
invested capital so that the plant begins to make is made for the timing of the cash movements.
a profit. Since only contemporaneous monies are com-
The return on capital from the process being parable, the investment outlay and the earnings
studied should be compared with that of known from the profit must be discounted to a fixed
processes. Above a threshold value, which varies point in time, usually the start of production. If
from case to case, the process will provide a re- a certain rate of interest p is assumed, all the cap-
turn on investment. A minimum return R of 20 % ital movements can be referred to that point to
is often employed, but because of the consider- obtain the so-called net value of a capital move-
able vulnerability to technical risk and market ment:
certainty, this is only a guideline. While the value
(E −A)t=0 = (E −A)t · (1+p)−t (4.13)
may be lower if the risk is low, it is prudent to
impose a minimum value up to a factor of two where
higher in the case of new high-risk products.
Generally the process development engineer E = expected earnings, e.g., from product
is not faced with the question of how profitable sales in monetary units per year
a project is but of whether a particular process A = expected outlay, e.g., investment costs,
engineering approach is more profitable than an production costs, in monetary units per
other, alternative solution within the framework year
of the process being studied or not [61]. In this p = interest rate
 Process Development 37

t = number of years (negative before start- 4.3.1. Sensitivity Analysis

up and positive after start-up)
For example, when started up, a plant has Sensitivity analysis yields information about the
to bear double the cost of an outlay incurred sensitivity of the process to variations in individ-
five years before start-up if the rate of interest ual influencing variables and therefore about the
is 15 %. Conversely, a revenue five years after reliability of the result of the return on capital
start-up will only carry half the weight and one analysis. It also provides pointers to the direc-
ten years after start-up will be virtually insignifi- tion in which the process should be developed
cant. The capital value K of a project is therefore further. The development is particularly worth-
the sum of the net values: while if it alters influencing variables to which
the return on capital is highly sensitive. The pro-
t=+n

K= (E − A)t · (1+p)−t (4.14)
cedure is to vary the different influencing vari-
t=−m ables in Equation (4.15) one after the other in the
where unfavorable direction (while keeping all the oth-
ers constant) by a certain fraction (for example,
m = duration of process development in 10 %) and then determine how much the internal
years rate of return p0 differs from the original value
n = useful life of project in years (usually (see Table 12).
10 years)
Table 12. Relative change in the internal rate of return on varying
To compare different projects, a useful life of the influencing variable x ∗
ten years is usually assumed for chemical plants. Parameter x ∆x/x
An investment is only beneficial if its capital
Development costs +10 %
value is equal to zero or is positive. Investment costs +10 %
The internal rate of return method ( dis- Product price −10 %
counted cash flow, DCF) tries to find the rate Capacity utilization −10 %
Net raw material costs +10 %
of interest which will result in a capital value Operating resources costs +10 %
of zero, i.e., the rate at which the net values of Staffing costs +10 %
the incoming and outgoing payments are equally
∗ By successively changing the parameters x in the unfavorable
large (the internal rate of return). The internal direction, the change in the internal rate of return can be
rate of return p0 is determined by setting the calculated from Equation (4.15).
above capital value function equal to zero:
4.3.2. Amortization Time
t=+n

−t
(E − A)t · (1+p0 ) =0 (4.15)
t=−m The risk as a function of time can be quantified
Projects are compared by comparing their cal- by means of the payback time t w . This indicates
culated internal rates of return, the project with the time which will be necessary for the cost
the highest internal rate of return being the most of the capital employed to be recovered. As an
profitable. initial estimate, t w is the ratio of the investment
costs I to the annual surplus income (E − A):
I
4.3. Economic Risk tw ≈ (4.16)
(E −A)
The return on capital calculation indicates
whether or not a project is economically worth-
while. To calculate the return, assumptions must
5. Trends in Process Development
be made about the most probable product prices,
The pattern of process development should be
the production outlay, the plant loading (actual
such that it produces competitive advantages,
production/nominal capacity), investment costs,
i.e., a process should be developed rapidly and
etc. Time will see whether these assumptions are
cheaply. Some trends which promote this objec-
correct. The impact of these uncertainties on the
tive are highlighted below [93]:
cost effectiveness of the project can be quanti-
fied by using the following variables.
38 Process Development

1) Nowadays only a team made up of members have a noticeable effect on the cost effective-
from all the scientific disciplines is capable ness of a new process.
of coping with the complex task of develop- 4) Another trend is the development and use of
ing a process. For this reason, a project team, expert systems. These are already available
usually consisting of a chemical process en- for simple, limited systems such as the op-
gineer, a chemist specializing in preparation, timization of individual unit operations. For
an engineer, and possibly the future works much more complex systems such as cata-
manager should be brought together at the lyst development or the optimization of en-
beginning [94]. tire plants, expert systems are at present still
If necessary, this group may recruit or con- a long way from general practical applica-
sult other specialists. Team work and team tion.
spirit are essential to the success of the ac-
Future tasks in process development will be
tivity, the synergies resulting from this type
in the following fields [95]:
of activity being exploited to the full.
When process development is complete, it 1) Improvement of plants and processes with a
is advantageous for at least one member of view to an integrated environmental protec-
the project team to continue working on the tion [96–98]
project during planning, construction, com- a)
missioning, and start-up, since otherwise a Residue reduction: Reduction in the
great deal of detailed knowledge would be amounts of byproducts by improving the
lost. catalysts and the reaction control
2) In the case of short-lived consumer-related b)
products (dyes, active substances, disper- Integrated residue utilization: Integrated
sions, etc.), which are preferably produced work-up of unavoidable waste streams
in batch processes, the trend is towards within the main plant, i.e., exploitation
equipment which is as flexible as possible of on-site synergies (thermal and material
(multipurpose production plants, multipur- coupling) and effective reuse of worked-
pose small-scale plants). Innovations in such up wastes
products are usually due to changes in for- 2) Development of new processes for the basic
mulation, and much less a result of modifi- products and intermediates on the basis of:
cations to the equipment. The ideal is to pro-
a) natural gas (C1 chemistry)
duce a wide variety of products in a batch
b) renewable raw materials
plant over a long period of time.
c) waste streams as raw material
3) In the case of basic products and interme-
diates produced in continuous plants, pro- 3) Development of processes that result in
cess development can make a contribution smaller, more economical, and cheaper
to competitiveness by reducing the time and plants.
the cost required to develop improved man-
ufacturing processes. This objective can be
achieved by increasing the scale-up factors 6. References
(see Section 3.1). In order to limit the re-
1. Jahrbuch der Chemiewirtschaft, VCH
sulting increase in technical risk, use is now
Verlagsgesellschaft, Weinheim 1991.
made of the synergy between an integrated 2. H.-B. Amecke: Chemiewirtschaft im
miniplant and a mathematical simulation. Überblick, VCH Verlagsgesellschaft,
This makes it possible to dispense with the Weinheim 1987.
pilot plant stage, i.e., the synergism bet- 3. H.-J. Quadbeck-Seeger: “Chemie für die
ween an integrated miniplant technique and Zukunft,” Angew. Chem. 102 (1990)
a mathematical simulation results in scale- 1213 – 1224.
up safety margins of similar quality to those 4. W. Jentzsch: “Was erwartet die Chemische
obtained by using a pilot plant. As a result Industrie von der Physikalischen und
it is possible to shorten the time required for Technischen Chemie?” Angew. Chem. 102
process development by ca. 30 %, which can (1990) 1267 – 1273.
 Process Development 39

5. C. Marchetti: “Die magische 26. J. M. Praussnitz: Molecular Thermodynamic of

Entwicklungskurve,” Bild Wiss. 10 (1982) Fluid-Phase-Equilibria, Prentice Hall,
115 – 128. Englewood Cliffs, N.Y. 1969.
6. E. G. Krubasik: “Angreifer im Vorteil,” 27. I. Hacker, K. Hartmann: “Heuristische Regeln
Wirtschaftswoche (1984) no. 23, 48 – 56. zum Entwurf verfahrenstechnischer Systeme,”
7. M. Perlitz, H. Löbler: “Brauchen Unternehmen in Probleme der modernen chemischen
zum Innovieren Krisen? ”, Zeitschrift für Technologie, Akademie-Verlag, Berlin 1980,
Betriebswirtschaft 55 (1985) 424 – 450. pp. 194 – 244.
8. G. Specht, K. Michel: “Integrierte 28. E. Blaß: “Methodische Entwicklung
Technologie- und Marktplanung im verfahrenstechnischer Prozesse,”
Innovationsportfolio,” ZfB 58 (1988) Chem.-Ing.-Tech. 57 (1985) 201 –210.
502 – 520. 29. G. Kaibel, E. Blaß, J. Köhler: “Gestaltung
9. M. E. Porter: “Technology and Competitive destillativer Trennungen unter Einbeziehung
Advantage,” Journal of Business Strategy thermodynamischer Gesichtspunkte,”
(1985) 60 – 78. Chem.-Ing.-Tech. 61 (1987) 16 – 25.
10. W. Busfield, D. Merigold, J. Chem. Soc. (A) 30. H. H. Erdmann, J. Kussi, K. H. Simmrock:
1969 2975 – 2977. “Möglichkeiten und Probleme der
11. H. Knapp: “Methoden zur Beschaffung von Prozeß-Synthese,” Chem.-Ing.-Tech. 56
Information über Stoffeigenschaften,” (1984) 32 – 41.
Chem.-Ing.-Tech. 59 (1987) 367 – 376. 31. G. Stephanopoulos, A. W. Westerberg:
12. E. Fitzer, W. Fritz: “Technische Chemie,” “Studies in Process Synthesis-II,” Chem. Eng.
Springer-Verlag, Berlin 1975. Sci. 31 (1976) 195 –204.
13. B. Streit: Lexikon Ökotoxikologie, VCH, 32. E. Blaß: Entwicklung verfahrenstechnischer
Weinheim 1991. Prozesse – Methode, Zielsuche, Lösungssuche,
14. L. Roth, G. Rupp: Sicherheitsdatenblätter, Lösungsauswahl, 1st ed., Verlag Salle und
VCH, Weinheim 1991. Sauerländer, Frankfurt 1989.
15. BASF, Acrylsäure rein, technical information, 33. K. Hartmann, L. Dietzsch: “Modellierung und
TI/ED 1330 d, Ludwigshafen 1987. Simulation verfahrenstechnischer Systeme,” in
16. F. Dolder: “Geheimhalten oder patentieren?” Probleme der modernen Technologie,
Management Zeitschrift 60 (1991) 64 – 68. Akademie-Verlag, Berlin 1980, pp. 43 – 77.
17. E. D. Dichtl, O. Issing: Vahlens Großes 34. E. Blaß: “Aufgabengerechte
Wirtschaftslexikon, Verlag C. H. Beck, Verlag Informationssammlung für die
Franz Vahlen, München 1987. Verfahrensentwicklung,” Chem.-Ing.-Tech. 56
18. Chemical Week, International ICIS/Chemical (1984) 272 – 278.
Week Price Report, Europe spot prices, $/m.t. 35. M. Buder, W. Rehfeld, T. Seeger: Grundlagen
19. Ullmann, 4th ed., 4, 24. der praktischen Information und
20. K. W. Schneider: “Großanlagen in der Dokumentation, vol. 1, K. G. Saur, München
Mineralölindustrie und Petrochemie,” 1990, pp. 366 ff.
36. “Dechema-Informationssysteme für die
Chem.-Ing.-Tech. 37 (1965) 875 – 877.
Chemische Technik und Biotechnologie,”
21. G. Lück: “Verfahrensentwicklung,” Fortschr.
Achema-Jahrbuch 1991, vol. 1, Frankfurt
Verfahrenstech. 21 (1983) 473 – 486.
1990, p. A60.
22. H. Harnich, R. Steiner, K. Winnacker:
37. R. Berenz: “Das Rad nicht zweimal erfinden,”
Chemische Technologie, Carl Hauser Verlag,
Jahrbuch der Chemiewirtschaft, VCH
München 1984.
Verlagsgesellschaft, Weinheim 1991,
23. W. Hofen, M. Körfer, K. Zetzmann:
pp. 172 – 174.
“Scale-up-Probleme bei der experimentellen
38. M. Zeitz: “Simulationstechnik,”
Verfahrensentwicklung,” Chem.-Ing.-Tech. 62
Chem.-Ing.-Tech. 59 (1987) 464 – 469.
(1990) 805 – 812.
39. G. Wozny: “Software macht den Fortschritt,”
24. J. Krekel, G. Siekmann: “Die Rolle des
Jahrbuch der Chemiewirtschaft, VCH
Experiments in der Verfahrensentwicklung,”
Verlagsgesellschaft, Weinheim 1991,
Chem.-Ing.-Tech. 57 (1985) 511 – 519.
pp. 153 – 158.
25. J. Gmehling, U. Onken, W. Aret: 40. D. Gilles, P. Holl, W. Marquardt: “Dynamische
Vapor – Liquid Equilibrium Data Collection, Simulation komplexer chemischer Prozesse,”
DECHEMA, Frankfurt 1977. Chem.-Ing.-Tech. 58 (1986) 268 – 278.
40 Process Development

41. W. Scheiding, H. Hartmann, P. Tönnishoff: 58. R. P. Palluzi: “Understand Your Pilot-Plant

“SATU 86 – Simulationsrahmen zur Options,” Chem. Eng. Prog. 87 (1991) no. 1,
dynamischen Prozeßsimulation für den 21 – 26.
industriellen Einsatz,” Chem.-Ing.-Tech. 61 59. P. Prinzing, R. Rödl, D. Aichert:
(1989) 292 – 299. “Investitionskosten-Schätzung für
42. W. Marquardt: “Rechnergestützte Erstellung Chemieanlagen,” Chem.-Ing.-Tech. 57 (1985)
verfahrenstechnischer Prozeßmodelle,” 8 – 14.
Chem.-Ing.-Tech. 64 (1992) 25 – 40. 60. E. Muthmann: “Ermittlung der
43. W. Klar: “Expertensysteme,” Elektronik 8 Investitionskosten von Chemieanlagen in
(1991) 108 – 123. verschiedenen Projektphasen,”
44. J. J. Ferrada, J. M. Holmes: “Developing Chem.-Ing.-Tech. 56 (1984) 940 – 941.
Expert Systems,” Chem. Eng. Prog. 86 (1990) 61. W. Frey, F. Heimann, S. Maier:
34 – 41. “Wirtschaftliche und technologische
45. A. Lieberam: “Expertensysteme für die Bewertung von Verfahren,” Chem.-Ing.-Tech.
Verfahrenstechnik,” Chem.-Ing.-Tech. 58 62 (1990) 1 – 8.
(1986) 9 – 14. 62. J. Jung: “Aspekte der Vorausberechnung von
46. P. Trum, N. R. Judica: Anlagenkosten bei der Projektierung
“Expertensysteme – Aufbau und verfahrenstechnischer Anlagen,” Chem. Tech.
Anwendungsgebiete,” Chem.-Ing.-Tech. 58 (Heidelberg) 12 (1983) 9 – 18.
(1986) 6 – 9. 63. C. A. Miller: “New Cost Factors Give Quick,
47. A. H. Masso, D. F. Rudd: “The Synthesis of Accurate Estimates,” Chem. Eng. 13 (1965)
System Designs,” AICHE 15 (1969) no. 1, 226.
10 – 17. 64. Jahrbuch der Chemiewirtschaft, VCH
48. G. L. Steele: Common LISP, Digital Press, Verlagsgesellschaft, Weinheim 1991, p. 64.
1984. 65. K. E. Lunde: “Transfer Pricing,” Chem. Eng. 4
49. L. Puigjaner, A. Espuña: Computer-Oriented (1985) 85 – 87.
Process Engineering, Elsevier Science 66. H. Körner: “Optimaler Energieeinsatz in der
Publishers, Amsterdam 1991. Chemischen Industrie,” Chem.-Ing.-Tech. 60
50. D. Greß, H. Hartmann, G. Kaibel, B. Seid: (1988) 511 –518.
“Einsatz von mathematischer Simulation und 67. B. Linnhoff, J. A. Turner: “Heat Recovery
Miniplant-Technik in der Networks: New Insights Yield Big Savings,”
Verfahrensentwicklung,” Chem.-Ing.-Tech. 51 Chem. Eng. 88 (1981) no. 22, 56 – 70.
(1979) 601 – 611. 68. B. Linnhoff: “New Concepts in
51. S. Maier, G. Kaibel: “Verkleinerung Thermodynamics for Better Chemical Process
verfahrenstechnischer Versuchsanlagen – was Design,” Chem. Eng. Res. Des. 61 (1983)
ist erreichbar?” Chem.-Ing.-Tech. 62 (1990) 207 – 223.
169 – 174. 69. D’Ans Lax: “Taschenbuch für Chemiker und
52. L. A. Robbins: “The Miniplant Concept,” Physiker,” 3rd ed., vol. 1, Springer Verlag,
Chem. Eng. Prog. 75 (1979) no. 9, 45 – 48. Berlin 1967.
53. W. Brust, F. Heimann, G. Kaibel, S. Maier:
70. F. A. Henglein: Grundriss der chemischen
“Klein, aber fein,” Jahrbuch der
Technik, Verlag Chemie GmbH, Weinheim
Chemiewirtschaft 1991, VCH, Weinheim
1963.
1991, pp. 179 – 182.
71. Chemiker Kalender, 3rd ed., Springer Verlag,
54. U. Wörsdörfer, U. Müller-Nehler: “Ein
Berlin 1984.
Gewinn für die Forschung,” Jahrbuch der
72. W. Hellmund, R. Auelmann, J. Wasel-Nielen,
Chemiewirtschaft 91, VCH, Weinheim 1991,
A. Rothert: Chem. Ind. (Düsseldorf) 114
pp. 182 – 186.
(1991) no. 9, 12 – 13; 114 (1991)
55. A. Schulz-Walz: “Meß- und Dosiertechnik
no. 10,10 – 12; 114 (1991) no. 11, 10 – 12.
beim Betrieb von Miniplants,”
73. J. M. Coulson, J. F. Richardson: Chemical
Chem.-Ing.-Tech. 62 (1990) 453 – 457.
Engineering, 4th ed., Pergamon Press, Oxford
56. J. M. Berty: “The Changing Role of the Pilot
1990.
Plant,” Chem. Eng. Prog. 75 (1979) no. 9,
74. Jahrbuch der Chemiewirtschaft 1991, VCH
48 – 51.
57. J. G. Lowenstein: “The pilot plant,” Chem. Verlagsgesellschaft, Weinheim 1991, p. 303.
Eng. 9 (1985) no. 23, 69 – 76.
 Process Development 41

75. W. Swodenk: “Umweltfreundlichere 86. P. Reher: “Produktion mit umweltgerechter

Produktionsverfahren in der chemischen Entsorgung,” Jahrbuch der Chemiewirtschaft,
Industrie,” Chem.-Ing.-Tech. 56 (1984) 1 – 8. VCH Verlagsgesellschaft, Weinheim 1991,
76. G. Lipphardt: “Produktionsintegrierter pp. 213 – 217.
Umweltschutz, Verpflichtung der Chemischen 87. W. Hüning, H. Nentwig, C. Gockel: “Die
Industrie,” Chem.-Ing.-Tech. 61 (1989) verbrennungstechnische Entsorgung eines
855 – 860. Chemiebetriebes,” Chem.-Ing.-Tech. 56
77. T. Bakay, K. Domnick: (1984) 521 – 526.
“Produktionsintegrierter 88. W. Hüning, P. Reher: “Verbrennungstechnik
Umweltschutz – Verminderung von für den Umweltschutz,” Chem.-Ing.-Tech. 61
Reststoffen, gezeigt anhand ausgeführter (1989) 26 –36.
Beispiele,” Chem.-Ing.-Tech. 61 (1989) 89. G. Färber: “Standardtechnik löst Probleme,”
867 – 870. Jahrbuch der Chemiewirtschaft 1991, VCH
78. H. Lenz, M. Molzahn, D. W. Schmitt: Verlagsgesellschaft, Weinheim 1991,
“Produktionsintegrierter pp. 158 – 162.
Umweltschutz – Verwertung von Reststoffen,” 90. D. Wagener: “Kostenermittlung am
Chem.-Ing.-Tech. 61 (1989) 860 – 866. Anlagenbau für die chemische Industrie,”
79. DIN 38 409,Bestimmung des Biochemischen Tech. Mitt. 65 (1972) 542 –547.
Sauerstoffbedarfs (BSB), DIN 38 409 part 51; 91. “Bruttostundenverdienst,” Jahrbuch der
DIN 38 409, Bestimmung des Chemischen Chemiewirtschaft 1991, VCH
Sauerstoffbedarfs (CSB); DIN 38 409 part 41. Verlagsgesellschaft, Weinheim 1991, p. 286.
80. “Modifizierter Zahn-Wellens-Test,” OECD
92. H. Kölbel, J. Schulze: “Neuentwicklungen zur
Guidelines for Testing of Chemicals, Paris
Berechnung von Preisindices für chem.
1981, p. 302 B.
Anlagen,” Chem. Ind. (Düsseldorf) 19 (1967)
81. I. Juhnke, D. Lüdemann: “Ergebnisse der
340 – 343, 701 – 706.
Untersuchung von 200 chemischen
93. G. Bogenstätter, H. Gerrens: “Die chemische
Verbindungen auf akute Fischtoxizität mit dem
Fabrik der Zukunft,” Chem. Ind. 37 (1985)
Goldorfentest,” Z. Wasser Abwasser Forsch.
275 – 277.
11 (1978) 161 – 164.
82. G. Bringmann, R. Kühn: “Grenzwerte der 94. J. Platz, H. J. Schmelzer: Projektmanagement
Schadwirkung wassergefährdender Stoffe in der industriellen Forschung und
gegen Bakterien (Pseudomonas putida) und Entwicklung, Springer, Berlin 1986.
Grünalgen (Scenedesmus quadricauda) im 95. R. Shinnar: “The Future of the Chemical
Zellvermehrungshemmtest,” Z. Wasser Industries,” CHEMTECH 1991, 58 – 64.
Abwasser Forsch. 10 (1977) 87 – 98. 96. M. Zlokarnik: “Umweltschutz – eine ständige
83. DIN 38 412,Testverfahren mit Herausforderung,” Chem. Ing. Tech. 61 (1989)
Wasserorganismen, 1991, part 8. 378 –385.
84. DFG Deutsche Forschungsgemeinschaft: 97. G. Scharfe, B. Sevekow: “Das Beispiel Bayer,”
Toxikologische Untersuchungen an Fischen, Chem. Ind. (1991) 17 – 20.
Verlag Chemie, Weinheim 1983. 98. J. Wiesner: Produktionsintegrierter
85. Kühn Birett Merkblätter Gefährliche Umweltschutz in der chemischen Industrie,
Arbeitsstoffe, 25th suppl., Nov. 1984, “VI-4 DECHEMA, Frankfurt, 1990.
Wassergefährdende Stoffe.”