Professional Documents
Culture Documents
Chemical Product Formulation Design and Optimization Methods Techniques and Case Studies Ali Elkamel Full Chapter PDF
Chemical Product Formulation Design and Optimization Methods Techniques and Case Studies Ali Elkamel Full Chapter PDF
https://ebookmass.com/product/hydrogen-sulfide-chemical-biology-
basics-detection-methods-therapeutic-applications-and-case-
studies-michael-d-pluth/
https://ebookmass.com/product/program-evaluation-methods-and-
case-studies-9th-edition-ebook-pdf/
https://ebookmass.com/product/psychoanalytic-case-
formulation-1st-edition-ebook-pdf/
https://ebookmass.com/product/tools-for-chemical-product-design-
from-consumer-products-to-biomedicine-1st-edition-mariano-martin/
https://ebookmass.com/product/chipless-rfid-authentication-
design-realization-and-characterization-zeshan-ali/
https://ebookmass.com/product/tidal-energy-systems-design-
optimization-and-control-1st-edition/
https://ebookmass.com/product/numerical-methods-and-optimization-
in-finance-2nd-ed-manfred-gilli/
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Chemical Product Formulation Design and Optimization
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Chemical Product Formulation Design and
Navid Omidbakhsh
Hesham Alhumade
Optimization
Keyvan Nowruzi
Thomas Duever
Ali Elkamel
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Authors All books published by WILEY-VCH are carefully
produced. Nevertheless, authors, editors, and
Prof. Ali Elkamel publisher do not warrant the information
University of Waterloo contained in these books, including this book,
Department of Chemical Engineering to be free of errors. Readers are advised to keep
200 University Avenue West in mind that statements, data, illustrations,
N2L 3G1 NK procedural details or other items may
Canada inadvertently be inaccurate.
Contents
Preface ix
About the Authors xi
1 Introduction 1
1.1 Chemical Product Engineering 1
1.2 Chemical Product Design 2
1.3 Product Design and Computer-Aided Product Design 4
References 6
Index 207
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ix
Preface
Chemical product design is a very important topic in the chemical industry. While
commodity chemicals have been the main area for chemical engineering focus in the
past several decades, specialty chemicals have been gaining more and more attention
in recent years. Therefore, accelerating the development process and optimizing the
formulation of chemical products would be of great benefit. With this change already
happening in the industry, chemical engineering education and training have not
changed enough to train engineers to fill positions in the product design field.
This book aims at providing the reader with a detailed understanding of the
product design, related statistical techniques, and optimization, and gives real-life
case studies for disinfectant formulations, optimization of defined medium, the
formulation of biocomposites, etc. This book can be used as a supplemental text-
book for chemical engineering students in a chemical product design course or to
R&D product formulation engineers so that they become familiar with the efficient
techniques used in developing new formulations. The book contains 11 chapters as
follows:
● Chapters 1 and 2: Introduction to the current product design process
● Chapter 3: Background to the related mathematical and statistical techniques
● Chapters 4–12: Cases studies
Chapters 1 and 2 introduce the reader to the current methodologies used for
designing new products in chemical industries and outlines the disadvantages of
the current processes and the need for improvement.
Chapter 3 gives a background about the theories of the methodologies used
to accelerate new product development. These methodologies include factorial
designs, mixture designs, optimal designs, linear and nonlinear regression analysis,
machine learning techniques (i.e. artificial neural networks), and multi-way
principal component analysis.
Chapters 4–11 present seven case studies to illustrate the process of product
design and its practical implications. The first case study covers optimization of
a disinfectant formulation, the second one presents optimization of a defined
medium, the third case deals with product improvement in a chemical toner
manufacturing process using multivariate modeling, the fourth case presents over
two chapters the design of wheat straw polypropylene composites, the fifth case
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
x Preface
employs simulation to formulate gasoline blends, the sixth case presents the design
of a corrosion protection coating using polyetherimide/graphene composites,
and finally the seventh case study deals with the optimization of the mechanical
properties of polypropylene-organically modified montmorillonite (PP-OMMT)
nanocomposites. The book ends with Chapter 12 that illustrates how to proceed in
selecting products to invest for business sustainability.
All chapters are equipped with clear illustrations, figures, and tables to help the
reader understand the included topics.
Many people contributed directly or indirectly to this book. We wish to pay our
gratitude and our respects to the late Professor Park Reily with whom we have
collaborated on research articles related to the topics in this book and have learned
a great deal from him. Also, this book would not have been possible without the
interactions we had with past graduate students. Although we give credit and refer-
ences in the appropriate chapters, we would like to vouch our words of appreciation
to Rois Fatoni, Hossein Ordouei, Youssef Al Herz, and Hassan Khorami. Special
thanks go also to the Wiley publishing team (Elke Maase, Katherine Wong, and
Lesley Jebaraj) for their professional work and for being patient with us. Last but
not least, we extend great appreciation to our friends and families.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
xi
hydrocarbon conversion and oil upgrading purposes. His research interests include
modeling and simulation of fluid dynamics in porous media and synthesis of
nanocomposites materials for various electrochemical applications including fuel
cells, supercapacitors, batteries, and corrosion mitigation. Dr. Alhumade received
the SABIC Distinguished Award in 2006.
Navid Omidbakhsh is Director of Early R&D and Advanced Research for
Advanced Sterilization Products (ASP), where he leads the innovation and techni-
cal feasibility of new concepts for future products. Prior to joining ASP, Navid was
Vice President of Open Innovation and Intellectual Property for Virox Technologies
and held a key role in the development of Virox’s globally registered products
and company’s exponential growth. Before Virox, Navid was an R&D engineer
for Henkel in surface technology field. Navid has earned his PhD in chemical
engineering from the University of Waterloo, Waterloo, Ontario, Canada, where his
main research area was on the development of a systematic method to optimize
chemical products/formulations. Navid holds several patents and peer-reviewed
publications in the area of product design, disinfectants, and sterilization formu-
lations and systems. Navid is also an alumnus of Harvard Business School, where
he completed programs on business, management, and innovation. He is also a
licensed professional engineer of Ontario, Canada.
Keyvan Nowruzi is a principal scientist at ASP. He has a BS in chemical engi-
neering from Sahand University of Technology, Tabriz, Iran; an MSc in chemical
engineering from Tehran Polytechnic University; and a PhD in biochemical engi-
neering from the University of Waterloo, Canada. Prior to joining ASP, he has served
as a post-doctoral fellow at the University of Guelph, Canada for four years and a
staff scientist for Akkim Kimya San. Ve Tic. A. Ş. for one year. He has been with ASP
for six years. Dr. Nowruzi has contributed in few inventions patented worldwide and
has several publications in peer-reviewed journals and international conferences.
Thomas Duever is Dean of the Faculty of Engineering and Architectural
Science and a professor of chemical engineering at Toronto Metropolitan University
(TMU). Prior to his role at TMU, Dr. Duever served as chair in the Department
of Chemical Engineering at the University of Waterloo for nine years, navigating
the department toward unprecedented growth. He has also taught industrial short
course in experimental design and polymer reaction engineering.
Dr. Duever is an accomplished researcher with interests including applied statis-
tics, experimental design, polymer reaction engineering, and product development.
He has written more than 100 articles in journals and conference proceedings to his
credit and has supervised the research projects of over 35 graduate students.
Dr. Duever is a registered professional engineer in the Province of Ontario, a
fellow of the Chemical Institute of Canada, and a fellow of the Canadian Academy of
Engineering. He holds PhD, masters, and bachelor degrees in chemical engineering
from the University of Waterloo.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1
Introduction
a lean and agile product development process, and therefore it is very important
that companies spend their R&D dollars very wisely and try to avoid less efficient
development methodologies.
Product design can have various interpretations, among them is the definition
as the entire procedures required to deliver a product with defined properties that
serve a specific need in society or industry based on inputs from various segments.
For instance, inputs from the industry of how the product may serve and what spec-
ifications should be considered during the manufacturing process. Items that can be
considered include environmental and regional regulations. An example of environ-
mentally friendly product design is the manufacturing of a greenhouse ventilation
system, where the house is designed to attenuate energy consumption and maintain
required rate of fresh air exchange. In such a process of product design of a household
air exchanger, various elements need to be considered including heat and humidity.
In addition, material selection is a significant factor in the manufacturing of such a
device to take into consideration environmental impacts such as energy conserva-
tion, corrosion, and exhaust gases, if any. The topic of product design has become
even more important with the growing changes in industry and regulation to pro-
tect the environment. For example, the manufacturing process of synthetic textile
fiber has been continuously developing since 1950. Starting wth a global produc-
tion of less than 10 million mt in the 1950 and undergoing a 10-fold increase by
2017, the effective utilization of fibers in various applications was achieved through
product design studies that were caried out on the development of various proto-
types utilizing statistical software packages. In general, the process of product design
encompasses the following steps: market needs, ideas, material selection, and finally
manufacturing and process control and optimization.
Many of the products we touch and feel today have come out of a chemical plant
one way or another. These products cannot be missed even in any quick visit to a
grocery store. Consumer products (e.g. detergents), cosmetics, health care products
(e.g. disinfectants, sanitizers), adhesives, pharmaceuticals, etc., are all examples of
chemical products. Therefore, chemical product design (CPD) is a very important
market segment and deserves enough attention in improving product development
methodologies. Chemical product engineering is the science and art of creating
chemical products, a much larger concept encompassing CPD. In other words,
chemical product engineering can be seen as the general background of knowledge
and practice supporting the concrete task of designing chemical products and their
manufacturing processes.
and governments, and media’s increased attention to this important topic, has led
governments to impose more stringent environmental regulations which puts even
more pressure on companies to try to reduce waste and carbon footprint. It would
be obvious for companies to try to optimize processes and product formulations to
deliver the same performance using “less” chemicals in a faster time and using less
resources. The million-dollar question to ask is how to achieve this, or simply how
to do more with less? In this book, we are trying to answer this question partially
and our focus will be on chemical and biological product mixtures.
In summary, the dynamic nature of the chemical and biochemical industries,
intense competition for market share, and emergence of more strict environmental
regulations require deployment of innovative product development methods to
address increasing demands for faster, leaner, and optimized products.
CAMD
Generate alternatives
Predesign
Process design
Needs and goals
Product manufacturing and testing
CAMbD
Generate alternatives
Property prediction:
Given: Obtain:
Information on compound
Properties of the compound
structure
Given: Obtain:
Information on desired Compound structures having
properties and type of the desired properties and
compound their “recipe”
Figure 1.2 Chemical product design (CAMD, CAMb D) are “reverse” of property prediction
problems.
problems, the currently available methods and tools can only solve a relatively
small percentage of these problems. This is because the property models that are
currently available are unable to predict the needed properties within an acceptable
limit of uncertainty.
The framework, however, can give a great contribution to creating property mod-
els and database development in a systematic way. This will reduce time and effort
in the early stages of the product design process and subsequently bring the product
to the market cheaper and faster.
The remainder of this book is organized as follows: Chapter 2 surveys a variety
of applications associated with CPD, while Chapter 3 covers tools commonly used
to accelerate product development. Chapters 4–12 provide illustrative case studies
related to CPD and formulation.
References
1 Lee, N.-J. and Jang, J. (1997). Performance optimisation of glass fibre mat
reinforced polypropylene composites using statistical experimental design. Polym.
Test. 16: 497–506.
2 Cussler, E.L. and Moggridge, G.D. (2011). Chemical Product Design. Cambridge
University Press. ISBN: 9781139035132.
3 Halvarsson, S., Edlund, H., and Norgren, M. (2008). Properties of medium-density
fibreboard (MDF) based on wheat straw and melamine modified urea
formaldehyde (UMF) resin. Ind. Crops Prod. 28: 37–46.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
7
Table 2.1 Automotive manufacturers, model, and components using natural fibers [1].
Audi A2, A3, A4, A4Avant, A6, A8, Roadstar, Coupe: Seat back, side and
back door panel, boot lining, hat rack, spare tire lining
BMW 3, 5, and 7 series and others: Door panels, headliner panel, boot
lining, seat back
Daimler-Chrysler A, C, E, S class: Door panels, windshield/dashboard, business table,
piller cover panel;
A class, Travego bus: exterior under body protection trim; M class:
Instrumental panel
(Now in S class: 27 parts manufactured from biofibers, weight 43 kg)
Fiat Punto, Brava, Marea, Alfa Romeo 146, 156
Ford Mondeo CD 162, Focus: Door panels, B-piller, boot liner
Opel Astra, Vectra, Zafira: Headliner panel, door panels, pillar cover
panel, instrumental panel
Peugeot New model 406
Renault Clio
Rover Rover 2000 and others: Insulation, rear storage shelf/panel
Saab Door panels
SEAT Door panels, seat back
Volkswagen Golf A4, Passat Variant, Bora: Door panel, seat back, boot lid finish
panel, boot liner
Volvo C70, V70
Mitsubishi Space star: Door panels; Colt: Instrumental panels
application. However, flax, hemp, and kenaf fibers are favored, because they have
excellent combinations of economic and functional properties [1].
The basic rule of reinforcement is that stresses to the material must be transmit-
ted from the polymer matrix to the fiber. To get the optimum reinforcement to the
polymer matrix, a fiber must have certain attributes. The length of the fibers and
the aspect ratio (length/diameter) of the fibers should be controlled to each specific
type of resin and application. Much of the research in the area of fiber-reinforced
plastics has been done using glass fibers. Glass fibers have uniform diameter and
can be made to any required length. Therefore, the length and aspect ratio are easily
controlled in the case of glass fibers. The fiber alignment is also a significant factor
for composite strength. Fibers randomly oriented will lose their reinforcement effect
by up to 80%. In cases where usage of continuous fibers is prohibited due to process
constraints, discontinuous fibers are used. In this case, stress cannot be transmitted
from the matrix polymer to the fibers across the fiber ends. Fibers with the size longer
than a critical minimum length lc are required for these discontinuous fibers [2].
Since stresses must be transmitted across the boundaries between polymer matrix
and the fiber, the properties of fiber–polymer composites are influenced by the
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2.1 Natural Fiber Plastic Composites 9
strength of the bond between the phases (interface). Providing strong interfacial
bonds can be very challenging because it is not easy to wet hydrophilic natural fiber
surfaces with generally hydrophobic viscous (molten) polymers. Coupling agents
play an important role to bind the matrix and the fibers together at the interface.
Coupling agents for more inert polymers like polyolefins are often acid-modified
versions of the matrix polymer, with maleic anhydride-grafted polypropylene
(MA-g-PP) as a prime example. MA-g-PP is widely used as a coupling agent in
composites reinforced with cellulose fibers. The treatment of cellulose fiber with
hot MA-g-PP copolymers provides the covalent bonds across the interface.
The mechanism of the reactions, which is basically divided into two steps: acti-
vation of the copolymer by heating and esterification of cellulose, is illustrated in
Figure 2.1. After this treatment, the surface energy of the fibers almost reaches the
surface energy of the matrix. Thus, a better wettability and a higher interfacial adhe-
sion are obtained. The PP chain permits segmental crystallization and cohesive cou-
pling between the modified fiber and the PP matrix.
Some recent research publications show that one of the focuses of the ongoing
research activities is to understand the effect of coupling agents on polymer blends
and composites. With the aid of scanning electron microscopy (SEM), the coupling
or dispersion mechanism and the fracture behavior can be evaluated by observing
O O
PP chain
PP chain
HO C CH2 C CH2
O
+ H2O
HO C CH C C CH C
O O
(a)
O
Cellulous fiber
O C CH2
O
O C CH C
C CH2
OH
Cellulous fiber
O
+ O
OH O
C CH C
Cellulous fiber
O C CH2
O
O O C CH C
H H
O
(b)
Figure 2.1 Illustration of coupling mechanism of cellulose fiber and maleic acid-grafted
polypropylene: (a) copolymer activation and (b) cellulose esterification.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 2 Some Typical Applications of Chemical Product Design and Intellectual Property
Composition selection
Manufacturing Property
Matrix system Filler system Additives process investigations
Polypropylene (properties) Wheat straw (properties) Product structure purpose: Processing techniques Mechanical
● Various grades (melt flow ● Coupling agent: ● Extrusion ● Tensile strength
Chemical
index, MFI) ● Injection molding ● Flexural strength
properties – composition: Product manufacturing
● Blends (mixtures of grades) ● Compression molding ● Flexural modulus
● Cellulose purpose:
● Combined with recycled PP ● Antioxidant ● Thermoforming ● Impact strength
● Hemicelluloses
pulping
● Thermomechanical:
steam explosion
● Chemo-mechanical
● Biological: fungi,
enzymes
Components’ proportion: volume percentage, weight percentage Processing variables Various methods and
standards
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12 2 Some Typical Applications of Chemical Product Design and Intellectual Property
Traditionally, modeling a composite system is carried out by dividing the system into
three main parts: composite composition selection, processing condition, and final
composite properties. Following this framework, three different kinds of modeling
strategies were conducted as illustrated in Figure 2.2.
First, the final composite properties are modeled as a function of composition
selection. These obtained models are commonly called structural property models.
Second, the final composite properties are modeled as a function of the processing
conditions. These models are called processing property models. Finally, based on
Figure 2.2 Schematic presentation of composite system and composite modeling strategy.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2.3 Modeling Natural Fiber Polymer Composites 13
Table 2.3 Discrepancies between theory and experimental work in natural fiber
composite [9].
the level of correlation between properties such as the trade-off relationship between
strength and toughness of composites, the final composite properties are investigated
to develop models that address such correlations between properties. Once the var-
ious models were constructed, they can be manipulated and then be combined to
simulate the composite system [8].
Despite the well-established mechanistic modeling in composite science,
researchers seem to face many difficulties in using the mechanistic approach
to model natural fiber polymer composites. There are several issues where the
composite theories are difficult to be applied in natural fiber plastic composite, as
summarized in Table 2.3.
A few modifications of available composite models have been proposed to
overcome these problems. Shibata and coworkers proposed a method to calculate
the fiber orientation coefficient in modeling flexural properties of PP with kenaf and
bagasse fiber. The Kelly–Tyson Modified Rule of Mixture and the Bowyer–Bader
model were used to introduce a model that can take into consideration the supercrit-
ical and subcritical length distributions of fibers in a hemp fiber PP composite [10].
A geometrical approach of cuboidal shape has been used to model the epidermal
surface and parenchyma surface of wheat straw compounded in a PP matrix. Those
modifications, however, only covered the individual issue of the complex natural
fiber plastic composite system.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14 2 Some Typical Applications of Chemical Product Design and Intellectual Property
are attributed to the weak forces that stack the layers of graphene together. Here,
the polymer adsorbs onto the delaminated graphene thin sheets, and once the
solvent is evaporated, the sheets reassemble in a form of sandwiching the resin
to form the composites. Another advantage of this method is the possibility of
synthesis of intercalated composites using polymers with low or even no polarity.
However, a considerable amount of solvent might be required to achieve an
optimum dispersion of the fillers, and this is considered a disadvantage of the
approach since solvent removal can be a critical issue. Finally, graphene composites
can also be synthesized using the melt intercalation method without the need to
a solvent as the main advantage. In this technique, graphene is mixed with the
hosting polymer matrix in molten state. Typically, a thermoplastic polymer is mixed
mechanically with graphene at certain temperatures using conventional methods,
such as extrusion and injection molding, and as a result graphene composites
are formed.
investment made in developing the product, and therefore it is an imperative for the
company to count on this significant price margin to finally have a positive return
on investment, which usually takes a few years to happen. If competitors could
copy them on day 1, the return on investment for most of these products would be
negative and no for-profit company would be encouraged to develop products such
as pharmaceuticals. In this case, the only source for new pharmaceutical drugs
would be limited to government-sponsored labs and universities. Although capable,
they are not nearly as equipped, efficient, and agile as pharmaceutical companies
to have such a robust cadence of new products in their pipeline on a consistent
manner.
Companies may decide to only protect the inventions used for to protect their
products, to avoid competitors from copying them. This is the main objective of
attempting to patent an invention and is defined as defensive strategy. Besides this,
some companies may try to protect inventive subject matter in adjacent fields to try to
hinder their competitors’ product development activities in those fields by creating
these “land mines” in the relevant patent landscape. Although effective in slowing
down competitors in these adjacent fields, this strategy does not generate tangible
return on investment, and therefore this approach is usually avoided in difficult
financial circumstances. Once patents are granted, it is the patent owner company’s
responsibility that their patents will not be infringed, by constantly monitoring the
market for any products that may fall under claims of the patents.
Although patent filing and prosecution process creates more work for their
inventors, and that they may not be directly compensated for their value added by
their inventions, they provide recognition for the inventors by their peers. Patents
are typically considered as one of the main matrices in assessing a scientist or
engineer’s creativity. The more patents a scientist or engineer has, the better their
resume will look like, and the prouder they will be in front of their peers. Some
companies also try to further encourage their prolific inventors by monetary awards
such as paying when their patent application is submitted and when they are
granted.
Before any new product development, it is wise to perform a landscape search to
understand the patent field, and the technologies and product features and meth-
ods that have already been patent protected. This activity can be extremely useful
since it avoids wasting millions of dollars on new product development only to then
realize that the developed technology is already patent-protected. It is advisable that
the landscape search be updated in regular intervals during product development to
avoid any missing relevant prior art and also to identify new publications that may
become relevant. Once the product is fully developed, a freedom-to-operate (FTO)
search is also performed to ensure that the product does not infringe any patents.
Companies often also watch their field on a regular basis for the emerging technolo-
gies and the new published patents, and therefore their subject matter experts are
usually up to date with work related to their expertise. It is also possible to gues-
timate where companies are spending their R&D dollars by reviewing their patent
portfolio, with an 18-month lag of course since patent applications are published
18 months after they are filed. This provides valuable insight for companies when
they try to monitor their competitors activities and can somewhat forecast what their
next products will be.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
References 19
Similar to scientific literatures that can help scientists and engineers with
their research, patents can also be useful to learn about the prior art; however,
patents should be considered with grain of salt, since these documents are not
peer-reviewed and more subject to error. Regardless, they provide an invaluable
source to learn from what has been done in the past in any given field. Fortunately,
patents are publicly available (except for the patent applications that have not been
published yet as well as those designated under national security programs) and
can be freely accessed. Among the most popular patent databases are the United
States Patent Office [29], the European Patent Office [30], and Google Patents [31].
These databases can be used to search by different subjects, such as keywords about
the subject of interest using Boolean expressions, assignee, inventor, title, etc. There
are also subscription-based patent databases which provide more user-friendly
access for a fee. Some of these services include Derwent Innovation, LexisNexis
TotalPatent One, and PatBase.
Many patents in the chemical, biological, and pharmaceutical fields involve a
mixture of chemicals rather than a unique new molecule. Given that each chemical
in a formulation possesses a specific function, it is often obvious to use a mixture
of chemicals to deliver a final product characteristic. For example, imagine a disin-
fectant formulation, which contains more than one antimicrobial active ingredient,
and that the inventor files a patent application for their disinfectant formulation.
Since it is generally obvious that the addition of two different antimicrobial agents
will increase the disinfection capability of the solution, the inventor has to prove
that there is an “unexpected” synergy by mixing the two chemicals. Synergy means
that the combined effect of these two chemicals is more than the mere additional
effect of one to another, and this synergy is hard to quantify and measure using
conventional one-factor-at-a-time methodology. However, using designed experi-
ments taught in this book, one can easily identify “interactions,” i.e. synergy (if the
interaction is positive). Therefore, providing such synergy proofs based on statistical
analysis provide an objective determination of synergistic activities between the two
antimicrobial active ingredients here and can easily rebut an obviousness rejection
by the examiner. Otherwise, trying to argue synergy would be very subjective and
does not provide a strong rationale. From another perspective, using these designed
experiments help identify these synergies/interactions, which otherwise can be
missed, and therefore equips inventors with reliable tools to capture and quantify
any “unexpected” results in any mixture, if exist.
References
1 Anderson, M.J. and Whitcomb, P.J. (1998). Find the most favorable formulations.
Chem. Eng. Prog. 94: 63–67.
2 Araujo, J.R., Waldman, W.R., and De Paoli, M.A. (2008). Thermal properties of
high density polyethylene composites with natural fibres: coupling agent effect.
Polym. Degrad. Stab. 93: 1770–1775.
3 Le Digabel, F., Boquillon, N., Dole, P. et al. (2004). Properties of thermoplastic
composites based on wheat-straw lignocellulosic fillers. J. Appl. Polym. Sci. 93:
428–436.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20 2 Some Typical Applications of Chemical Product Design and Intellectual Property
20 Crosby, A.J. and Lee, J.-Y. (2007). Polymer nanocomposites: the “nano” effect on
mechanical properties. Polym. Rev. 47: 217–229.
21 Noor, A.K. and Venneri, S.L. (1994). Flight-Vehicle Materials, Structures, and
Dynamics: Advanced Metallics, Metal-Matrix and Polymer-Matrix Composites.
American Society of Mechanical Engineers.
22 Yu, Z., Di, H., Ma, Y. et al. (2015). Preparation of graphene oxide modified by
titanium dioxide to enhance the anti-corrosion performance of epoxy coatings.
Surf. Coat. Technol. 276: 471–478.
23 Yang, Z., Sun, W., Wang, L. et al. (2016). Liquid-phase exfoliated fluorographene
as a two dimensional coating filler for enhanced corrosion protection
performance. Corros. Sci. 103: 312–318.
24 Ehsani, M., Khonakdar, H.A., and Ghadami, A. (2013). Progress in Organic
Coatings Assessment of morphological, thermal, and viscoelastic properties of
epoxy vinyl ester coating composites: role of glass flake and mixing method.
Prog. Org. Coatings. 76: 238–243. https://doi.org/10.1016/j.porgcoat.2012.09.010.
25 Nematollahi, M., Heidarian, M., Peikari, M. et al. (2010). Comparison between
the effect of nanoglass flake and montmorillonite organoclay on corrosion
performance of epoxy coating. Corros. Sci. 52: 1809–1817. https://doi.org/10
.1016/j.corsci.2010.01.024.
26 Sathiyanarayanan, S., Azim, S.S., and Venkatachari, G. (2008). Corrosion
protection coating containing polyaniline glass flake composite for steel.
Electrochim. Acta 53: 2087–2094. https://doi.org/10.1016/j.electacta.2007.09.015.
27 Wang, G. and Yang, J. (2011). Progress in Organic Coatings Influences of glass
flakes on fire protection and water resistance of waterborne intumescent fire
resistive coating for steel structure. Prog. Org. Coatings. 70: 150–156. https://doi
.org/10.1016/j.porgcoat.2010.10.007.
28 Santana, J.J., González, S., and Souto, R.M. (2010). Resistance of metallic
substrates protected by an organic coating containing glass flakes. Prog. Org.
Coat. 68: 240–243. https://doi.org/10.1016/j.porgcoat.2010.01.005.
29 United States Patent and Trademark Office. www.uspto.gov.
30 European Patent Office. https://worldwide.espacenet.com/.
31 Google Patents. https://patents.google.com/.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
23
Chemical Product Formulation Design and Optimization: Methods, Techniques, and Case Studies,
First Edition. Ali Elkamel, Hesham Alhumade, Navid Omidbakhsh, Keyvan Nowruzi, and Thomas Duever.
© 2023 WILEY-VCH GmbH. Published 2023 by WILEY-VCH GmbH.
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24 3 Mathematical Principles for Chemical Product Design
Since each factor is considered at high (+) and low (−) levels, a complete repli-
cate of such a design requires 2k observations and is called a 2k factorial design.
This design enables an experimenter to investigate “k” factors with a relatively small
number of runs. Consequently, it is particularly useful in the early stages of an exper-
imental investigation when there are usually many factors to be studied.
To design the experiments, the factor levels need to be established, and the
sequence of experimental trials should be randomized [1].
Since there are only two levels for each factor, the response is assumed linear over
the range of the chosen factor levels. This is often adequate, particularly in the early
stages of a study.
As the number of factors in a 2k factorial design increases, the number of runs
required for a complete replicate of the design rapidly outgrows the resources of
most experiments. The full factorial experiment provides an experimenter with
enough information to evaluate the whole set of main effects as well as all inter-
action effects. The main effects and lower order interactions are usually the most
significant terms [6]. In fact, one is usually capable of determining the main effects
and the lower order interactions by performing a fraction of the complete factorial
design with little loss of information. Such a design is called a “fractional factorial”
design (FFD). A 2k−p fractional factorial design containing 2k–p runs is called a 21p
fraction of the 2k complete design, or more simply a 2k−p fractional factorial design.
A potential concern regarding the fractional or full factorial design is the assump-
tion of linearity in the factor effects. When it is clear that the relationship is not
necessarily perfectly linear, some amount of nonlinearity can be accommodated by
adding interaction terms to a main effect or first-order model, resulting in:
∑
k
∑
k
∑
y = 𝛽0 + 𝛽j xj + 𝛽ij xi xj + 𝜀 (3.1)
j=1 i<j
where 𝜀 denotes the noise or error observed in the response y and xj is a coded vari-
able. The relationship between the coded variable xj and its natural variable X j is
defined by the following equation:
Xj − (Xj,high + XJ,low )∕2
xj = (3.2)
(Xj,high − XJ,low )∕2
The model is now capable of handling some curvature in the response function. This
curvature, of course, results from the twisting of the plane induced by the interaction
terms 𝛽 ij xi xj .
In some situations, the curvature in the response function cannot be adequately
modeled by Eq. (3.1). In such cases, a logical model to consider is
∑
k
∑∑ ∑
k
y = 𝛽0 + 𝛽j xj + 𝛽ij xi xj + 𝛽jj xj2 + 𝜀 (3.3)
j=1 i<j j=1
where the 𝛽 jj denote pure second order or quadratic effects. Two-level fractional
or full factorials do not provide sufficient data to estimate models given by 3.3.
However, adding center points to the 2k design will allow for a test to determine if
second-order effects are required. Also, they allow an independent estimate of the
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.2 Response Surface Methods and Designs 25
Often, there is some curvature in the relationship between a response and the inde-
pendent variables that fractional factorial design cannot handle. In such cases, a
method needs to be applied to detect and estimate the curvature. The 3k design is
a possible choice by an experimenter who is concerned about the curvature in the
response function [1].
However, it should be noted that this design is not necessarily the most efficient
one, and the number of experiments increases exponentially as the number of inde-
pendent factors increases. One excellent way of addressing this issue is to augment
the already performed 2k or 2k−p design with axial runs and extra center points. Such
a design is called a central composite design or CCD. A CCD can be built up sequen-
tially. Often, when we are at a point on the response surface that is remote from the
optimum, there is little curvature in the system, and the first-order model will be
appropriate. The objective here is to lead the experimenter rapidly and efficiently
along a path of improvement toward the general vicinity of the optimum. Once the
region of the optimum has been found, the linear 2k or 2k−p design can be augmented
to form a CCD and a more elaborate model, such as the second-order model, may be
employed.
This sequential strategy of experimentation is usually far more efficient than run-
ning a 3k factorial design.
85 Contours of
90 85 constant
80 response
Region of the Region of
optimum operability for
75 the process
Path of
improvement 70
65
Current
operating
conditions 60
0
1
0
−1 −1
−1 0 1
In many practical cases, for example, in our case study of disinfection formulation
optimization in Chapter 4, some prior data are available. If conventional fractional
factorial designs or D-optimal techniques are applied, one might not take advan-
tage of the information hidden in the prior data to “optimally” design further tests.
To address this issue, a Bayesian design can be contemplated, which can fully take
advantage of the prior knowledge and lead to the optimal design.
It is shown [8] that the design matrix, M, can be calculated as
{ }
1
M = Det I + 2 XUX′ (3.6)
𝜎
Defining G = M × Det{U −1 } leads to the following more practical equation:
{ }
1
G = Det U −1 + 2 X ′ X (3.7)
𝜎
where U is the p × p prior information covariance matrix, p is the number of param-
eters in the model used in D-optimality design, 𝜎 2 is the error variance, X is a n × p
matrix of n optimal experiments, and I is the n × n identity matrix.
Regression analysis is a statistical technique for investigating and modeling the rela-
tionship between variables. A regression model that involves more than one regres-
sor variable is called a multiple regression model.
Suppose that the antimicrobial activity of a disinfectant (in log reduction) depends
on raw material concentrations in the formulation. A multiple linear regression
model that might describe the relationship is
y = 𝛽0 + 𝛽1 x1 + 𝛽2 x2 + … + 𝛽k xk + 𝜀 (3.8)
𝛽̂ = (X ′ X)−1 X ′ y (3.9)
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
28 3 Mathematical Principles for Chemical Product Design
where
⎡1 x11 x12 x1k ⎤
⎢ ··· ⎥
1 x x x2k ⎥
x = ⎢ 21 22
⎢ ⋮ ⋱ ⋮ ⎥
⎢1 x x · · · xnk ⎥⎦
⎣ n1 n2
⎡y1 ⎤ ⎡𝛽̂0 ⎤ ⎡ 𝜀1 ⎤
⎢ ⎥ ⎢̂ ⎥ ⎢ ⎥
y 𝛽 𝜀
y = ⎢ 2 ⎥ , 𝛽̂ = ⎢ 1 ⎥ , and 𝜀 = ⎢ 2 ⎥
⎢⋮⎥ ⎢⋮⎥ ⎢⋮⎥
⎢y ⎥ ⎢̂ ⎥ ⎢𝜀 ⎥
⎣ n⎦ ⎣ 𝛽n ⎦ ⎣ n⎦
′
provided that the inverse of (X X) exists. This matrix will always exist if the
regressors are linearly independent, that is, if no column of the X matrix is a linear
combination of the other columns.
Since 𝛽̂ is an estimator of 𝛽, therefore, the estimate of the vector of y can be
calculated as
y = X 𝛽̂
̂ (3.10)
where
⎡̂
y1 ⎤ ⎡𝜀̂1 ⎤
⎢̂ ⎥ ⎢ ⎥
y 𝜀̂
y = ⎢ 2 ⎥ 𝜀̂ = ⎢ 2 ⎥
̂
⎢⋮⎥ ⎢⋮⎥
⎢̂ ⎥ ⎢ 𝜀̂ ⎥
⎣ yn ⎦ ⎣ n⎦
has been given to analyze. This expert can then be used to provide projections given
new situations of interest and answer “what if” questions.
In ANN, neurons are arranged in parallel layers, with each neuron forming
weighted connections to all layers. In between lies at least one layer, termed the
hidden layer. Hidden layers act as feature detectors, and according to universal
approximation theory, a network with a single hidden layer with a sufficiently large
number of neurons can map any input to any output, with an arbitrary degree of
accuracy. In addition, there is a “bias” neuron that is connected to all neurons in the
hidden layer and the output layer. The bias neuron accepts no inputs and supplies
an invariant output. Bias neurons are added to the hidden and output layers to
preserve the universal approximation of the network. Figure 3.3 shows a typical
three-layered ANN with “I” input neurons, “H” hidden neurons, and “O” output
neurons.
The number of input and output neurons is determined by the nature of the prob-
lem under study. The number of neurons in the hidden layer is often determined by
the required degree of accuracy and is therefore a parameter in formulating an ANN
model. The input neurons transmit their inputs to the neurons in the hidden layer,
thus acting as distribution channels. The neurons in the hidden and output layers
calculate their inputs by performing a weighted summation of all the outputs they
receive from the preceding layer. Their outputs, on the other hand, are calculated by
transforming their inputs using a nonlinear transfer function. The most widely used
transfer function is the “S”-shaped logistic sigmoid transfer function. This transfer
1 2
2 3 1
k m n
xˆ k yˆ on
where ̂ ynj and y0nj are the jth measured and the predicted network outputs that cor-
respond to the jth input, respectively. Therefore, the training process requires a for-
ward pass to calculate an output and a backward pass to update the weights. Min-
imizing Ej implies that for each forward pass made, a backward pass should be
carried out.
The development of an artificial neural network model usually consists of two
steps. The first step is a training stage, where the network is subjected to a training set
Table 3.1 Input and output neurons for a network architecture containing
input, output, and one hidden layer.
Input k ̂
xk yon = ̂
̂ xk
∑I
1
Hidden m h
xm = yik wih
km
+ bm yhm = ( h)
k=0 1 + exp −xm
∑H
Output n xno = yhm who
mn + bn yon = xno
m=0
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3.7 Mixture Design of Experiments 31
of input/output patterns. The second step is a testing stage, where the performance
of the network is tested on patterns that have not been “seen” by the network during
the training stage.
0.8
0.6
2
0.4
0.
0.
2
0.
4
0.
4
6
0.
0.
6
8
0.
0.
8
X2 X3
x1 = 1 x1 = 1
x1 = 2
3
, x3 = 1
3
, x2 = 0
x 1 = x3 = 1
2
, x2 = 0 x1 = x3 = 1
2
, x2 = 0
x1 = 1 x1 = 1
x4 = 1 x4 = 1
x2 = 1 x2 = 1
x3 = 1 x3 = 1
A (4, 2) Lattice A (4, 3) Lattice
Because the independent variables x are not unique, the standard polynomial
model has been modified to become a canonical mixture polynomial model. In
general, the canonical or Scheffe forms of the mixture models are as follows:
∑
q
Linear∶ E(y) = 𝛽i xi (3.17)
i=1
∑
q
∑∑
q
Quadratic∶ E(y) = 𝛽i xi + 𝛽ij xi xj (3.18)
i=1 i < j=2
∑
q
∑∑
q
∑∑
q
Full cubic∶ E(y) = 𝛽i xi + 𝛽ij xi xj + 𝛿ij xi xj (xi − xj )
i=1 i < j=2 i < j=2
∑∑∑ q
+ 𝛽ijk 𝛽i 𝛽j 𝛽k (3.19)
i < j < k=3
∑
q
∑∑
q
∑∑
q
Special cubic∶ E(y) = 𝛽i xi + 𝛽ij xi xj + 𝛿ij xi xj
i=1 i < j =2 i < j =2
∑∑∑ q
+ 𝛽ijk 𝛽i 𝛽j 𝛽k (3.20)
i < j< k=3
The models can be used to construct a response surface and a contour plot as illus-
trated in Figure 3.6.
The response surface of a linear polynomial model is a planar surface over the sim-
plex region. The quadratic polynomial model consists of linear blending terms and
quadratic blending terms. Therefore, the response surface of a quadratic polynomial
model is a curved surface, which consists of a planar surface plus a curvature sur-
face representing the quadratic term. The response surfaces of linear and quadratic
x3 (1.00)
x1 (0.00)
x2 (0.00)
x2 (1.00)
x3 (0.00)
x1 (1.00)
Downloaded from https://onlinelibrary.wiley.com/doi/ by ibrahim ragab - Cochrane Germany , Wiley Online Library on [14/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
34 3 Mathematical Principles for Chemical Product Design
2 x Plasticizer x1 Plasticizer
1.00 1 z2 1.00
2
(−, +) 5
(+, +)
2 2 2 2
14 12 0.00 0.00
0.00 13 0.00
6
11 10
8 7
2 2 9 6 2 2 2 2
7
1.00 1.00 1.00 1.00
0.00 x2 Plasticizer 0.00 x3 Plasticizer
x2 Plasticizer x3 Plasticizer Thickness
Thickness
z1
1.00 x1 Plasticizer 1.00 x1 Plasticizer
2 2
(−, −) (+, −)
2 8 9 2 2 14 12 2
0.00 0.00 0.00 13 0.00
11
7
6 10
5 9 2
2 2 2 2 2
8
1.00 1.00 1.00 1.00
x2 Plasticizer 0.00 x2 Plasticizer 0.00 x3 Plasticizer
Thickness Thickness
III
As it happened that night, there was a thin new moon, and the
youth slept but little because of the croaking noise made by the
frogs. Presently, full awake, he sat up, and it seemed to him that the
air was full of noise, not only of frogs but of the hooting of owls and
the whirring of bats, and looking he saw the strange sight of a great
white toad with two heads, and presently about that fearful thing
other things gathered. From rock and hole came unclean creatures,
abominable serpents and centipedes and great gray spiders, and all
these gathered in a circle, the two-headed toad in the centre. With
wide-open eyes Huathia watched, although the sight of so much that
was noisome came near to benumbing and stupefying him, and
incomplete shapes seemed to be looking at him with evil eyes from
the black depth of the forest.
Soon the owl began to mourn and the song fell into words and
the youth heard this:
And so it went on, a mad and horrible concert, with bat and owl
and great ghost-moth whirling about on silent wings, until sickened of
it all Huathia rose up and clapping his hands to his ears fled from the
place. And when he had gained a quiet and lonely spot he sat down,
but in his ears rang what he had heard:
It had meant little in the night, but in a flash he saw that the
grindstone was a stone wrought by men. So fitting an arrow to his
bow he handed the weapon to Suso, telling her to shoot whatever
evil thing was discovered when he lifted the stone. With a great effort
he raised the stone suddenly, heavy though it was, lifting it high
above his head, and there, in a hollow place where the stone had
been, sat a large, white, double-headed toad.
“Shoot, Suso, shoot!” commanded the father. “Let not that evil
thing escape. It is the creature that torments me at night.”
Swift flew the arrow and it pierced the body of the toad. At the
same moment there fell from the roof of the house two monstrous
white serpents where they had lain hidden. Like lightning Huathia,
having seized the bow, sent two arrows flying, and each serpent was
cut into halves. In less than three moments three evil things died,
and it was like the sun coming from a cloud-veil, the way in which joy
came to that place. The weakness of the father fell from him like a
cloak. The bodies of the toad and the snakes withered and
shrivelled, and as a light breeze sprang up, what was left of them
was blown away as dust. There were soft stirrings in the thicket and
the whole world burst into song. So both father and daughter knew
then that the witcheries were gone and the evil creatures vanished
for ever, and that all the trouble that had been upon that place came
from the wicked stepmother.
So youth and maiden were married, and the father soon regained
his health and strength, and in all the world there were no happier
people than they.
THE TALE OF THE LAZY PEOPLE
Holi! Holi!
Huqui! Huqui!
Grind, grind, grind.
So the metates turned and turned, going round and round without
hands, and presently an Elbow-room-er that was struggling with a
corn-grinder stumbled, and both fell between the grinding stones and
in a moment were crushed to powder. In a flash house utensils and
animals learned the new trick, and in every house manikins were
pushed into the grinding stones. Then sparks began to fly and roofs
to catch on fire and manikins bolted here and there in confusion,
sometimes jamming in doorways, there were so many and all in such
disorder. Then came dazzling, flickering lightning and a great rain, so
that for very safety the manikins fled to the forest and climbed the
trees. And there they have lived ever since, for they grew hair and
became monkeys. But the remembrance of all that passed stayed
with them, and in their hearts to this very day is no love for man, and
for that very reason when a Christian passes through a forest he
must look well to himself, lest the manikins in revenge try to hurt him
by casting nuts and branches at his head.