Computers and Chemical Engineering 33 (2009) 947–953

Contents lists available at ScienceDirect

Computers and Chemical Engineering

journal homepage: www.elsevier.com/locate/compchemeng

Chemical Product Engineering—The third paradigm

Michael Hill ∗
Department of Chemical Engineering, Columbia University, New York, NY 10027, United States

a r t i c l e i n f o a b s t r a c t

Article history: New chemical products have historically been created by combining a broad knowledge of existing
Received 5 February 2008 chemical products with scientific experimentation. Since a combinatorial explosion of product options
Received in revised form 31 October 2008 will inevitably limit all experimental techniques, it should be preferable to minimize experimentation
Accepted 9 November 2008
through a systematic consideration of product formulations prior to experimentation. This is the essence
Available online 27 November 2008
of product design and engineering.
While the design of a chemical product and its manufacturing process are analogous, some critical
Keywords:
differences are so fundamental that a new paradigm and new approaches are needed to successfully solve
Product design
Product engineering
product design problems. In addition, chemical product design requires a methodology or algorithm to
Chemical engineering paradigm apply chemical engineering fundamentals. Product design techniques should draw largely on heuristics
when data are limited, followed by more detailed calculations when data become available. Significant
work is still needed to establish a comprehensive generic methodology for engineering chemical products
in the absence of complete data.
© 2008 Elsevier Ltd. All rights reserved.

1. Introduction Engineers of various disciplines have longed worked in the

Chemical Product Engineering and the related area of chemi-
cal product design have recently received much attention within
the chemical engineering community, with an exponential increase
in published papers over the past decade (Costa, Moggridge, &
Saraiva, 2006). This interest in Chemical Product Engineering
is a direct consequence in the shift in the chemical indus-
try towards the manufacture and sale of high value-added
materials marketed on performance rather than compositional
specifications (Cussler & Wei, 2003; Seider, Seader, Lewin, &
Widagdo, 2009; Villadson, 1997). These high value-added mate-
rials, often termed chemical products (Cussler & Moggridge, 2001),
include performance chemicals, semi-conductors, paints, cosmet-
ics, inks, pharmaceuticals, personal care products, household
products, and foods (Favre, Marchal-Heusler, & Kind, 2002; Hill,
2004).
A chemical product may consist of an individual chemical, but
more frequently it will be a mixture of chemicals with a set formu-
lation and often a set microstructure. These chemical products may
be described as formulated products (Favre et al., 2002), in contrast
with products consisting of assembled solid-state components (like
automobiles, personal computers, and microwave ovens), termed
assembled products.
∗ Tel.: +1 212 854 6166; fax: +1 212 854 3054.
E-mail address: mih16@columbia.edu.
field of product design, specifying the physical form, function and
aesthetics of assembled products. Yet it has been chemists and
not chemical engineers who have typically specified formulated
chemical products (Hill, 2004). Chemical engineers have generally
restricted themselves to processing, and in the minds of many,
chemical engineering has been synonymous with process engi-
neering. Be that as it may, opportunities are now emerging where
chemical engineers are specifying the products themselves (Cussler
& Wei, 2003).
Chemical engineering education already provides many of the
essentials needed for product design. This should not be surpris-
ing, as whenever chemical products are used by consumers, they
undergo some transformation due to applied stresses, tempera-
ture gradients, physicochemical hydrodynamics, mass transfer, etc.,
making product use a “process” in the chemical engineering sense
(Edwards, 2006; Hill, 2004). Hence an analysis of product behav-
ior ultimately requires the same fundamentals as those needed
to analyze process behavior, such as thermodynamics, transport
phenomena, and reaction kinetics.
Nevertheless, there are some fundamental differences between
the design of a chemical product and its manufacturing process,
suggesting the need for a new paradigm to successfully solve prod-
uct design problems. Before addressing this, however, the context
and vocabulary of product design must be clearly established, espe-
cially since many manufacturers of chemical products do not use
product design in the sense described in this paper, but rather per-
form “product development”.

0098-1354/$ – see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compchemeng.2008.11.013
948 M. Hill / Computers and Chemical Engineering 33 (2009) 947–953
2. Design, development, and engineering
The words design and engineering both convey a sense of arrang-
ing and planning (Simpson & Weiner, 1989). Thus, whether an
electrical engineer is identifying the circuitry required for a new
electronic product, a mechanical engineer is creating a new type of
aircraft, or a civil engineer is devising a new bridge, these engineers
follow a methodical approach and refer to their task as design.
On the other hand, new formulated chemical products are typi-
cally created by combining a broad knowledge of existing product
formulations with scientific experimentation, an approach histor-
ically called product development. While chemical engineers are
trained in scientific experimentation, our strength is more typically
in systems analysis, whereas chemists tend to have their primary
strength in experimentation. Thus the highly experimental nature
of product development has tended to keep formulated products
the domain of chemists.
Chemical product development may be guided by scientific
hypothesis, intuition, or simple trial-and-error. Product develop-
ment may also at times be accelerated through high-throughput
experimentation, where large numbers of products are simultane-
ously made in small quantities, or through experimental design,
where statistical techniques reduce the number of experiments
required. Nevertheless, these techniques have their limitations. For
example, it is impractical to use high-throughput experimentation
to make large numbers of structured products (like emulsions or
composite powders) in small quantities. Similarly, experimental
design can help determine optimal levels of a specified component
but a combinatorial explosion will frequently prevent selection
from a list of all potential components. Thus, chemical product
development is all too often random trial-and-error experimenta-
tion.
However, the systematic identification of problem solutions
should be superior to a random identification of solutions, either
because better solutions can be identified or because acceptable
solutions can be identified sooner or with less resource. So while
it is unrealistic to eliminate all experimentation, it is desirable to
minimize it through the introduction of a systematic considera-
tion of product formulations prior to experimentation, an activity
properly termed product design. From this perspective, the object of
product design is to specify a small set of formulations likely to meet
the product requirements, and which can be confirmed or refined
through experimentation. Thus, chemical product design should be
viewed as a phase of chemical product development that should
precede a more focused experimental program.1 This is analogous
to chemical process design, which is generally recognized as the
first phase of chemical process development (Ng & Wibowo, 2003).
Since a process to manufacture a chemical product must also be
specified, the overall domain of (1) chemical product design and
development, and (2) product-oriented process design and devel-
opment, is best termed Chemical Product Engineering (Costa et al.,
2006).
However, an even broader view of product design is possi-
ble. Product development projects typically progress through a
series of stages. While the definitions of these stages vary, they
typically include: (1) concept development, where product ideas
are created and undergo initial screening, (2) system-level design,
where the product is conceptually decomposed into subsystems, (3)
detailed design, where an initial product composition and process
1
This terminology is similar though not equivalent to that used in the domain
of assembled products, where product development includes marketing, design,
and manufacturing (Ulrich & Eppinger, 2004). The domain of chemical products has
displayed a narrower usage of the term product development, restricting it to the
applied scientific activity required to specify a new product.
are proposed,2 (4) testing and refinement, accomplished through
prototyping and experimentation, (5) production ramp-up, where
manufacturing capability is established, and (6) product launch and
on-going manufacture (Ulrich & Eppinger, 2004). As apparent from
the names of the stages, design is generally associated with the
detailed design stage.
Nevertheless, elements of design may occur in other stages of
product development as well. For example, the creation and ini-
tial screening of product ideas during concept development should
be based not just on an analysis of what the market would like
to buy, but also on an analysis of what is possible to manufac-
ture. Back-of-the-envelope engineering calculations during this
phase can ensure that all new product ideas are grounded in real-
ity. Similarly, product launch and manufacture generally require
some form of product quality assurance, and since the develop-
ment of quality assessment techniques requires an understanding
of the product’s intended behavior, engineering calculations should
be a part of the development of quality assessment techniques.
All these engineering calculations may be thought of as product
design, regardless of the product development stage in which they
occur.
Hence, while we may refer to the activity in the detailed design
stage of product development as product design, it is also possi-
ble to collectively consider all design activities across all product
development stages as product design. Since organizations tend to
be structured along product stages, opportunities for synergy by
similar activity across the various stages should not be overlooked.
3. A new paradigm
Having considered some of the elements of product engineer-
ing, we can now address the matter of paradigms. While various
comments on chemical engineering paradigms have appeared over
the years (Costa et al., 2006; Mashelkar, 1995; Stephanopolous &
Han, 1996; Villermaux, 1993), an overuse of the word paradigm by
society in general may have led to some confusion over the mean-
ing of the term. One is reminded of the Dilbert comic strip where
every engineer says his project is a paradigm but no one seems to
know what that means!
The term paradigm was popularized by Kuhn (1996) in his book,
The structure of scientific revolutions, first published in 1962. Bor-
rowing the word from linguistics, Kuhn used the term to indicate
a specific way of viewing scientific reality, the mindset of a sci-
entific community. Some of Kuhn’s examples include Copernican
astronomy, Newtonian dynamics, and quantum mechanics. Each of
these paradigms affected the choice of problems that were consid-
ered worthy of solution, as well as acceptable approaches to solving
those problems.
As pointed out by Kuhn, even when paradigms are known to be
inadequate, their inadequacies are frequently minimized or even
ignored by a scientific community. But if and when a paradigm
reaches a crisis where its technical inadequacies are brought into
focus, perhaps driven by social requirements, a new paradigm will
arise to explain what the prior paradigm could not. Thus the inad-
equacies of Newtonian mechanics in explaining some observations
that had been viewed as anomalies eventually led to Einsteinian
dynamics, and the success of this new paradigm in explaining those
observations opened up an entirely new set of problems as worthy
of consideration (Kuhn, 1996). Of course Newtonian mechanics is
still useful and may even be considered as a special case of Ein-
steinian dynamics.
2
Unlike assembled products which typically undergo detailed design prior to test-
ing and refinement, most chemical products have historically been developed by
experimentation without prior detailed design, as noted above.
 M. Hill / Computers and Chemical Engineering 33 (2009) 947–953
From this perspective, it should be appreciated that originally
chemical engineering had no paradigm. Chemical processes were
studied within the context of various industries, and so engineers
studied processes to make soap, dyestuffs, sugar, etc. Without the
mindset of a unifying principle, engineers did not look for and
hence failed to see commonality between these processes (Wei,
1996).
Chemical engineering received its first paradigm in 1915 with
the introduction of the unit operations concept (Committee on
Chemical Engineering Frontiers, 1988; Favre et al., 2002; Wei,
1996). This mindset allowed engineers to recognize commonalities
between elements of chemical processes despite their use in dif-
ferent industries. Under this paradigm, chemical engineering was
no longer the study of how to manufacture a specific commodity,
but rather the study of unit operations. As a consequence, chemical
process design became a matter of deciding which sequence of unit
operations was most appropriate to manufacture a desired product.
While still useful to the present day, the unit operations
paradigm proved inadequate for solving some important classes of
problems. This awareness led to the emergence of chemical engi-
neering science as a second paradigm in the late 1950s, as best
exemplified by the textbook Transport Phenomena (Bird, Stewart,
& Lightfoot, 2002; Committee on Chemical Engineering Frontiers,
1988; Favre et al., 2002; Wei, 1996). This approach taught engi-
neers to analyze problems by thinking in terms of their underlying
fundamental chemical and physical sciences, writing mathemati-
cal equations to describe the phenomena, and then solving those
equations. The chemical engineering science paradigm may also be
described as the “first principles” approach.
The chemical engineering science paradigm is widely used
today. In fact, its application has been broadened by the incorpora-
tion of biological science and new information technology tools.
But as important as these latter elements have been, they have
been incorporated into the existing chemical engineering science
paradigm rather than lead to a new mindset. Similarly, specific tech-
niques for solving various classes of chemical engineering problems
are not new paradigms, for they fall within the current chemical
engineering way of thinking.
On the other hand, until recently the chemical engineering com-
munity largely ignored all product issues other than purity as
irrelevant, focusing exclusively on processing while leaving prod-
uct development to chemists. The purpose of chemical engineering
has been primarily to obtain the lowest cost process. Even process
related issues like reliability, controllability, and pollution control
have ultimately translated into costs that must be minimized. Thus,
process design has easily lent itself to a mathematical treatment.
In contrast, chemical product design seeks to obtain the most
added value for a product through enhanced product properties.
This is far more complex than a mathematical treatment to maxi-
mize profit, as profit will depend in some unidentified way upon a
complex set of product properties that may not even be identified
at the outset. As a consequence, product design and engineering
problems are not amenable to traditional chemical engineering
approaches. Their solution requires not just additional chemi-
cal engineering approaches, but even more fundamentally, a new
mindset, and this is why product design and engineering should be
recognized as a third chemical engineering paradigm, as first hinted
in 1988 (Committee on Chemical Engineering Frontiers, 1988).
Of course, product engineering as a paradigm does not preclude
other paradigms from emerging, nor does it replace unit operations
or chemical engineering science as paradigms. Process engineering
may even be considered as a special case of product engineering. But
a product-engineering mindset is essential if chemical engineers
are going to be able to solve problems where both the product and
its manufacturing process must be identified, an entirely new and
important class of problems.
949
4. Product design methodologies
Just as designing a chemical process requires a methodology
or algorithm to apply the chemical engineering fundamentals, so
does designing a chemical product. Cussler and Moggridge (2001)
have proposed a generic framework for chemical product design,
suggesting a 4-step algorithm: (1) identify customer needs, (2) gen-
erate ideas to meet those needs, (3) select among the ideas, and (4)
manufacture the product. They also admit that this framework is
a simplification that tries to come down on the side of universal
applicability rather than effectiveness in specific cases. While this
framework is an excellent starting point, it may be useful to expand
on it.
People who specialize in understanding consumers and mar-
ket trends typically identify customer needs, often before chemical
engineers are assigned to a product design project. These analyses
are generally based on the results of opinion surveys, which can
give misleading direction if not correctly fashioned and interpreted
(Spirer, Spirer, & Jaffee, 1998) Hence statisticians and consumer
psychologists must continue to play a central role in the creation,
administration and analysis of opinion surveys. Nevertheless, sur-
vey results must be translated into technical targets for a product
design, and this requires a technical understanding of how an
intended product can perform during use. Chemical engineers are
well equipped to perform this role, and should participate in the
development and subsequent interpretation of customer surveys.
Thus chemical engineers should have a supporting role in the iden-
tification of customer needs.
Issues surrounding the design of a manufacturing process for
complex chemical products have been discussed elsewhere (Hill,
2004; Meeuse, 2006; Meeuse, Grievink, Verheijen, & Van der
Stappen, 2000), so I will focus on the remaining two steps, namely
how ideas can be generated to meet customer needs, and how to
best select from among those ideas. These steps must be at the heart
of a chemical product design methodology.
In practice, significant guidance is needed as to how to gener-
ate options and how to best select from among them. This is not
simply a matter of brainstorming ideas and selecting the best from
among them. While brainstorming and other creativity techniques
are often useful for generating novel approaches to problems, a
generic methodology is needed to systematically transform each
novel approach into a specific set of product alternatives, and to
quantitatively analyze those alternatives so as to select from among
them.
5. Design of homogeneous products
5.1. Overview
Having taught chemical product design to undergraduates at
Columbia University, my colleagues and I have developed a useful
methodology for designing homogeneous chemical products when
limited data are available. This methodology has eight steps, and
can guide a student team through a product design problem of
moderate difficulty, each step requiring one week for a team of 3–4
students to complete, except steps 3, 4, 5, and 8 which likely need
two weeks each. Thus the course neatly fits into the time constraints
of a university semester.
The methodology assumes that the target behavior of the new
chemical product has already been identified, eliminating the need
for customer opinion surveys. Also, since the required products
are assumed homogeneous, their properties will result solely from
their components and not a product microstructure generated dur-
ing processing. This allows us to design the product and process
sequentially rather than simultaneously, greatly simplifying the
methodology and making it well within the grasp of undergrad-
950 M. Hill / Computers and Chemical Engineering 33 (2009) 947–953
uates. Thus the procedure loosely follows the 4-step procedure
of Cussler and Moggridge (2001), but adds additional important
details.
For example, this procedure includes an analysis of market
realities. A chemical product cannot be designed without a consid-
eration of whether the proposed product will be profitable. Hence
specific steps to assess the marketplace and determine profitability
are included. In addition, recognizing that product design is only
the first stage of product development and must be followed by a
focused experimental program, the procedure includes an analysis
of all uncertainties that should be followed up by experimentation.
In addition, the methodology recognizes a key difference
between the economics of commodities and specialty products that
is easily overlooked. There is little risk that a commodity chemical
manufactured by a new manufacturer or plant will go unsold if it is
priced comparable to the competition. This is because manufactur-
ers are generally unable to differentiate their commodity products
from those of their competitors if they are priced the same, and
so by the laws of supply and demand, any small increase in supply
will lead to a small decrease in price for all manufacturers of the
commodity as the market absorbs all the commodity produced by
the new manufacturer. As all of the commodity manufactured at a
new plant will sell at this new market price, the primary business
decision is whether the investment in a new plant is justified by its
return.
On the other hand, since chemical products are differentiated
by their performance specification, a new product will be governed
by its own supply and demand equilibrium, and there will be no
guarantee that a new chemical product can be sold at any price.
There is no point in trying to calculate the return on investments
(a cash flow transient) if the business proposition is not profitable
in the steady state, i.e. with investments ignored. Hence before a
prospective manufacturer considers whether the investment is jus-
tified by its return, ongoing profitability must be assessed first. This
is typically a calculation of the market size that must be achieved for
revenue to cover fixed costs, a situation referred to as “break-even”3
(Lawler, 2002).
The methodology follows below.
5.1.1. Product definition
The designer should begin by broadly defining the product
that will be designed. This starts with an analysis of customer
needs, and leads to a specification of technical product tar-
gets. The analysis must also investigate current products, if any,
in the marketplace—price, composition, the specific function of
any components, strengths and weaknesses (from both a cus-
tomer/consumer and a supplier perspective), any hidden costs, and
total market size.
Even if there is no product just like an intended product currently
in the market, there may be other kinds of products indirectly ful-
filling the same end function. For example, instead of using a device
to purify drinking water, people may be drinking impure water and
going more often to the doctor to treat water-borne illness. This will
all be important information for setting an appropriate price for the
new product, which in turn will be critical for determining whether
the new product will be profitable.
Given the time constraints of a university semester, we gener-
ally provide the students with information about customer needs,
thereby saving the considerable time that would be required for
opinion surveys.
3
Break-even, the market size needed for revenues to cover ongoing fixed costs,
is not the same as the payback period, the time required for cash flow to cover an
investment. Break-even deals with steady state issues and is measured in either units
of money or product volume; payback period deals with the cash flow transient and
is measured in units of time.
5.1.2. Technical product requirements
The mechanism(s) by which the new product may be able to
work should be analyzed next. The implications each mechanism
will this have on the physical properties of the product should be
considered. The underlying chemical engineering phenomena (e.g.
thermodynamics, reaction kinetics, transport phenomena, etc.)
that will be relevant to understanding the behavior of the product
should also be identified.
Where there are multiple properties that must be met simul-
taneously, it may be assumed that the complete required property
set can be decomposed into subsets of properties which can be
achieved separately through their own components. This will allow
the complete property set to be achieved by combining all compo-
nents.
It is useful at this point to consider if there are classes of com-
pounds that can provide some of the required properties if they
were present as components, as well as any classes of compounds
that would be inappropriate in the new product. This fundamen-
tal understanding will be later used to model the properties of the
product.
For example, consider the problem of formulating a biodegrad-
able deicer for airplane wings (Joback, 2006). One would likely
decide that freezing point depression is a more appropriate deicing
mechanism than raising surface temperature by heat generation, as
the latter effect would be temporary. This suggests that the product
should contain a freezing point depressant. However, the product
must also adequately wet and spread over the aircraft surface, not
cause corrosion to wet metal surfaces, and biodegrade at acceptable
rates. As it is unlikely that one compound will meet all these criteria,
it can be assumed that the product will consist of (1) compounds
that adequately depress the freezing point yet also biodegrade at
acceptable rates, (2) compounds to ensure wetting, i.e. surfactants,
and (3) compounds to prevent corrosion, i.e. anti-corrosion agents.
5.1.3. Product performance relationships
For each property subset, the understanding of the underlying
chemical engineering phenomena can be used to derive a set of
equations that can predict the relevant behavior as a function of
composition. While simplifying assumptions may be made, one
must be careful not to oversimplify, verifying qualitatively that the
models will be useful for predicting the relevant behavior.
Next, all physical parameters that will be needed to apply the
model with any candidate compound should be listed. In the
absence of direct experimental data, one must decide how the
needed physical parameters will be obtained (e.g. tabulated data,
appropriate correlations, group contribution methods, etc.).
For example, designing a biodegradable aircraft deicer would
require a model of freezing point depression so that one could pre-
dict the mass of ice melted per mass of deicing compound at a given
temperature. However, assuming ideal solution behavior leads to
the unlikely conclusion that the only property governing freezing
point depression is molecular weight, so solution ideality is clearly
an oversimplification. In addition, the design would require a model
of drainage rate off aircraft surfaces so that one could predict time
to product failure, as well as a model of biodegradation. These mod-
els in turn lead to a need for various physical parameters, including
activity coefficients, heats of fusion, and viscosity (Joback, 2006).
5.1.4. Product candidate generation
For each property subset, as large a list of potential candidates
as is possible should be generated, based on an understanding of
the underlying chemistry. This may be done by computer gener-
ation of alternatives or by searching through tabulated databases.
Using the various product property models and any other relevant
factors, each list may be culled by eliminating candidates that are
inappropriate.
 M. Hill / Computers and Chemical Engineering 33 (2009) 947–953
5.1.5. Product candidate selection
For each property subset, overall performance may be defined
by assigning a weighting factor to each property in the set. In
the example of the aircraft deicer, while there may minimum tar-
gets for freezing point depression and biodegradation that must
be simultaneously achieved for a formulation to be given further
consideration, assigning appropriate weighting factors to these
properties will allow the product designer to consider performance
tradeoffs in identifying the formulation with the best performance.
Next raw material costs for compounds that simultaneously
meet all the important criteria within that property subset should
be obtained, and using the property models and weighting factors,
all remaining candidates should be ranked for their raw material
costs on an equal overall performance basis. Any compounds that
are less expensive than those used in current products on an equal
overall performance basis, including hidden costs, may then be
identified.
Assuming that the complete required property set can be
achieved by combining the components for each property subset,
an overall composition to recommend for experimental study may
now be identified.
5.1.6. Process design
For the preferred composition, a base case plant capacity may
be chosen and a preliminary process designed. This preliminary
process design should include a process flow sheet, a material bal-
ance for each stream, and sizing of all major equipment. The capital
that would be required to build this plant as a function of capacity
should be determined.
5.1.7. Risk analysis
The key technical risks associated with this project should next
be identified, along with recommendation to manage those risks.
This should include a listing of all key assumptions that were made
in designing the product and its process, and an experimental
plan for corroborating the conclusions drawn from these assump-
tions.
5.1.8. Financial (business case) analysis
Based on cost/performance of the preferred composition and
current products, a recommended selling price for the new product
should be identified. Considering all available factors, the mar-
ket share that may be expected at the recommended selling price
should also be identified. Making reasonable estimates, one can
next identify the expected variable costs and fixed costs associated
with the new product.4 This information can then be used to iden-
tify the market share required for break-even, which can in turn be
compared to the expected market share to determine if the new
product is likely to be profitable on an ongoing basis. The net profit
expected on an ongoing basis may also be calculated.
Next, making reasonable estimates, the investment expenses
that will be required may be calculated. Finally, given the expected
market share and reasonable assumptions for the ramping up of
sales, the time required to recoup the initial investment while meet-
ing the internally required discount rate may be calculated. Based
on this analysis, a decision as to whether the investment should be
recommended or not may be made.
4
Variable costs are costs proportional to sales volume, and include items like
raw materials and delivery charges. Fixed costs are costs treated as independent of
sales volume, although more correctly they are step functions of sales volume. These
include items like depreciation and administrative overheads (Lawler, 2002). Some
costs, like labor and utilities, fall somewhere in between these two idealizations, and
may be treated as either. Note that a fixed capital investment, unlike depreciation,
is not an ongoing cost, and hence is not a fixed cost.
951
5.2. Discussion
This product design methodology will identify a product that
meets the preliminary performance specification, and although it
assesses both ongoing profitability and return on investments, it
guarantees an acceptable level of neither. However, as with all
design, product design should be approached iteratively. Once a
product designer completes the last step of this method, he will
know the various factors that influence product performance and
economics. In addition, there may have been multiple product pos-
sibilities identified by the methodology, some of which may have
been eliminated prematurely. Hence the product designer will be in
a position to take a fresh look at all previous decisions and explore
the impact of these decisions on ongoing profitability and return
on investment.
It is also possible that the product designed by this procedure
can be the starting point for mathematical optimization. Since the
product that offers maximum performance regardless of costs is
unlikely to be the product that offers maximum profitability, there
is value in simultaneously simulating and optimizing the effects
of product performance, consumer response, and microeconomics
(Bagajewicz, 2007). As the focus of the product design must be suf-
ficiently narrowed to permit this approach, the procedure outlined
above is a good starting point.
6. Design of structured products
The methodology proposed above assumes that a homoge-
neous product can achieve all the required product properties.
This ignores the class of chemical products known as structured
products, which achieve their properties through a microstructure
that is determined by the interaction of its components and the
manufacturing process (Edwards, 1998). Structured products are
ubiquitous in modern society, and include processed foods (e.g. ice
cream, margarine, peanut butter), household products (e.g. laun-
dry detergents), and beauty or personal care products (e.g. soaps,
shampoos, skin creams, toothpaste), to name just a few. Product
engineering for structured products will be particularly difficult,
as the product and process must be designed simultaneously (Hill,
2004).
Here again, a generic methodology is needed to systematically
transform each novel approach into a specific set of product alter-
natives, and to quantitatively analyze those alternatives so as to
select from among them. While I do not currently know how such
a generic methodology can be assembled, some general observa-
tions about such a methodology can be made. For example, as with
process design, this product design methodology would likely be
hierarchical and iterative. Two primary approaches are possible: (1)
generation and systematic reduction of the number of alternatives
through heuristics, and (2) optimization of the set of all poten-
tial alternatives through mathematical programming. By analogy to
what has been concluded about process design, it can be expected
that product design techniques will draw largely on heuristics when
data are limited, followed by more detailed calculations later on
(Douglas & Stephanopoulos, 1995).
Recent work has established how to mathematically repre-
sent the generic product-engineering problem (Gani, 2004), and
mathematical programming has been successfully applied to these
problems (Sunol, 1995). Of course, these techniques can only be
applied where sufficient data are available to enable a complete
mathematical representation of the product-engineering problem.
Conversely, in the early stages of design when such data are
generally lacking, heuristics are needed to systematically generate
and analyze alternatives. Others have begun to identify product-
engineering heuristics (Fung & Ng, 2003; Wibowo & Ng, 2001,
2002) but these heuristics have invariably been within a narrow
952 M. Hill / Computers and Chemical Engineering 33 (2009) 947–953
product context. A comprehensive methodology for incorporating
heuristics across many product domains has yet to be estab-
lished.
Given a broad product definition as a problem statement, heuris-
tics should help the engineer generate a list of potentially important
physical properties as well as the complete set of combinations of
product composition and microstructure that potentially lead to
those properties. This set of product composition and microstruc-
ture combinations will form a region of interest.
At the same time, laws of physics and chemistry will
likely constrain some combinations of product composition and
microstructure so that not all combinations will be physically
attainable. Available process equipment may further constrain
attainable combinations of product composition and microstruc-
ture. Thus the set of attainable product compositions and
microstructures can be compared to the set of attainable chem-
ical compositions in the design of reactor networks (Feinberg,
2002; Feinberg & Hildebrandt, 1997; Glasser, Hildebrandt, & Crowe,
1987; Hildebrandt & Glasser, 1990; Horn, 1964), and will form an
“attainable region” of all possible products (Garcia, 2004; Khumalo,
Glasser, Hildebrandt, Hausberger, & Kauchali, 2006). In principle,
this attainable region should be identifiable though the use of
heuristics and/or simplified models.
Once the attainable region is identified, it can then be com-
pared to the region of interest. If there is no overlap, no solution to
the product design problem exists, although the attainable region
will indicate other products that may be of future interest. On the
other hand, the overlapping region will define those combinations
of product composition and microstructure that deserve further
study.
Additional heuristics will be needed to solve the inverse prob-
lem, translating this overlapping region into a corresponding set of
formulation and processing options. This is analogous to the rules
that translate each point in the attainable region for chemical reac-
tions to a specific combination of continuous stirred tank and plug
flow reactors (Feinberg & Hildebrandt, 1997).
Finally, the set of formulation and processing options may be
reduced based on tradeoffs of sales revenues and costs. Quite likely,
the vast majority of formulation and processing options can be
quickly ruled out due to a lack of profitability. This will be anal-
ogous to Douglas’ approach to conceptual process design, where
processing options are eliminated as soon as they are shown to be
unprofitable (Douglas, 1988).
As with the design of homogeneous products, any design proce-
dure should include considerations like risk, ongoing profitability,
and return on investments. It would also likely be hierarchical
and iterative, and allow for incorporation of microeconomic issues
(Bagajewicz, 2007).
Only after the option set is reduced to a manageable size would
experimentation begin, either confirming the general engineer-
ing predictions or permitting further refinement and narrowing of
the recommended formulation and processing options. Mathemat-
ical programming would also become possible once sufficient data
became available.
7. Conclusions
The methodology for design of homogeneous products outlined
above highlights some of the real issues that chemical engineers
face in product design. These problems may be quite rich despite
the constraint that the product has no microstructure.
On the other hand, a comprehensive generic methodology
for structured products would be significantly more complex
and would require significant work to develop, especially for
design problems where complete data are unavailable. However,
such a methodology would accelerate new product devel-
opment well beyond the capabilities of purely experimental
techniques.
References
Bagajewicz, M. J. (2007). On the role of microeconomics, planning, and finances in
product design. AIChE Journal, 53, 3155.
Bird, R. B., Stewart, W. E., & Lightfoot, E. N. (2002). Transport phenomena. New York:
Wiley.
Committee on Chemical Engineering Frontiers. (1988). Frontiers in chemical engi-
neering: Research needs and opportunities. Washington: National Academy
Press.
Costa, R., Moggridge, G. D., & Saraiva, P. M. (2006). Chemical product engineering:
An emerging paradigm within chemical engineering. AIChE Journal, 52, 1976.
Cussler, E. L., & Moggridge, G. D. (2001). Chemical product design. New York: Cam-
bridge University Press.
Cussler, E. L., & Wei, J. (2003). Chemical product engineering. AIChE Journal, 49,
1072.
Douglas, J. M. (1988). Conceptual design of chemical processes. New York: McGraw-Hill.
Douglas, J. M., & Stephanopoulos, G. (1995). Hierarchical approaches in conceptual
process design: Framework and computer aided implementation. In L. T. Biegler
& M. F. Doherty (Eds.), Fourth international conference on foundations of computer
aided process design. AIChE Symposium Series No. 304 (pp. 183–197).
Edwards, M. F. (1998). The importance of chemical engineering in delivering products
with controlled microstructure to customers. IChemE North Western Branch Papers
No. 9.
Edwards, M. F. (2006). Product engineering: Some challenges for chemical engineers.
Chemical Engineering Research and Design, 84(A5), 1.
Favre, E., Marchal-Heusler, L., & Kind, M. (2002). Chemical product engineering:
Research and educational challenges. Transactions of IChemE, 80A, 65.
Feinberg, M. (2002). Toward a theory of process synthesis. Industrial and Engineering
Chemistry Research, 41, 3751.
Feinberg, M., & Hildebrandt, D. (1997). Optimal reactor design from a geometric
viewpoint—l. Universal properties of the attainable region. Chemical Engineering
Science, 52, 1637.
Fung, K. Y., & Ng, K. M. (2003). Product-centered processing: Pharmaceutical tablets
and capsules. AIChE Journal, 49, 1193.
Gani, R. (2004). Chemical product design: Challenges and opportunities. Computers
and Chemical Engineering, 28, 2441.
Garcia, J. A. (2004). Robust design for structured products, Thesis (M.S.Ch.E.), University
of Massachusetts Amherst.
Glasser, D., Hildebrandt, D., & Crowe, C. (1987). A geometric approach to steady
flow reactors: The attainable region and optimization in concentration space.
Industrial and Engineering Chemistry Research, 26, 1803.
Hildebrandt, D., & Glasser, D. (1990). The attainable region and optimal reactor struc-
tures. Chemical Engineering Science, 45, 2161.
Hill, M. (2004). Product and process design for structured products. AIChE Journal,
50, 1656.
Horn, F. (1964). Attainable and non-attainable regions in chemical reaction tech-
nique. In E. Bartholome (Ed.), Third European symposium on chemical reaction
engineering (pp. 293–302). New York: Pergamon Press.
Joback, K. (2006). Designing chemical products. In AIChE annual meeting San Fran-
cisco, November 13.
Khumalo, N., Glasser, D., Hildebrandt, D., Hausberger, B., & Kauchali, S. (2006). The
application of the attainable region analysis to comminution. Chemical Engineer-
ing Science, 61, 5969.
Kuhn, T. S. (1996). The structure of scientific revolutions. Chicago: University of Chicago
Press.
Lawler, W. C. (2002). Cost-volume-profit analysis. In J. L. Livingstone & T. Gross-
man (Eds.), The portable MBA in finance and accounting (pp. 102–124). New York:
Wiley.
Mashelkar, R. A. (1995). Seamless chemical engineering science: The emerging
paradigm. Chemical Engineering Science, 50, 1.
Meeuse, F. M. (2006). Process synthesis for structured food products. In K. M. Ng, R.
Gani, & K. Dam-Johansen (Eds.), Chemical product design: Towards a perspective
through case studies (pp. 167–179). Amsterdam: Elsevier.
Meeuse, F. M., Grievink, J., Verheijen, P. J. T., & Van der Stappen, M. L. M. (2000).
Conceptual design of processes for structured products. In M. F. Malone, J. A.
Trainham, & B. Carnahan (Eds.), Fifth international conference on foundations of
computer aided process design. AIChE Symposium Series No. 323 (pp. 324–328).
Ng, K. M., & Wibowo, C. (2003). Beyond process design: The importance of a process
development focus. Korean Journal of Chemical Engineering, 20, 791.
Seider, W. D., Seader, J. D., Lewin, D. R., & Widagdo, S. (2009). Product and process
design principles: Synthesis, analysis, and evaluation (3rd ed.). Hoboken: Wiley.
Spirer, H. F., Spirer, L., & Jaffee, A. J. (1998). Misused statistics (2nd ed.). New York:
Marcel Dekker.
Simpson, J. A., & Weiner, E. S. C. (1989). The Oxford English dictionary. Oxford: Claren-
don Press.
Stephanopolous, G., & Han, C. (1996). Intelligent systems in process engineering: A
review. Computers and Chemical Engineering, 20, 143.
Sunol, A. K. (1995). A mixed integer (non) linear programming approach to simul-
taneous design of product and process. In L. T. Biegler & M. F. Doherty (Eds.),
Fourth international conference on foundations of computer aided process design.
AIChE Symposium Series No. 304 (pp. 276–279).
 M. Hill / Computers and Chemical Engineering 33 (2009) 947–953 953

Ulrich, K. T., & Eppinger, S. D. (2004). Product design and development. New York:
McGraw-Hill.
Villadson, J. (1997). Putting structure into chemical engineering. Chemical Engineer-
ing Science, 52, 2857.
Villermaux, J. (1993). Future challenges for basic research in chemical engineering.
Chemical Engineering Science, 48, 2525.
Wei, J. (1996). A century of changing paradigms in chemical engineering. ChemTech,
26(5), 16.
Wibowo, C., & Ng, K. M. (2001). Product-oriented process synthesis and develop-
ment: Creams and pastes. AIChE Journal, 47, 2746.
Wibowo, C., & Ng, K. M. (2002). Product-centered processing: Manufacture of
chemical-based consumer products. AIChE Journal, 48, 1212.