Available online at www.sciencedirect.

com

ScienceDirect

A review of computer-aided design of paints and

coatings
Spardha Jhamb1,2, Markus Enekvist1,2, Xiaodong Liang1,2,
Xiangping Zhang3, Kim Dam-Johansen2 and
Georgios M Kontogeorgis1,2

There is an immense potential for the computer-aided tools in Introduction

the design of paints and coatings. Significant advances have Design of complex (formulated, multicomponent) chem-
been made, involving also the use of thermodynamic and in ical products is now considered to be a widely accepted
general property models for the study and theoretical area of teaching [1–4] and research [5] in chemical (and
formulation of these products. Algorithms and tools based on biochemical) engineering, as it has been previously the
such models enable the formulation chemist to speed up the case for other engineering disciplines like mechanical
design process, by allowing them to focus their experimental engineering. Some researchers have claimed that this
efforts on a selected number of reliable constituents for the field can be a third or emerging paradigm in chemical
coating formulation. engineering [6–10], but of course in order for this to be the
case it is necessary to have a systematic methodology and
Even though model-based methods and tools can save accompanied tools for the design of complex (bio-)
resources and time required for the design, service life chemical products. Many have worked in this area
prediction and formulation of new products, the experimental [11,12,13

,14–17]. One pioneer in this field, has demon-
validation cannot be done away with; as certain interactions in strated how a systematic computer-aided design (CAD)
these complex systems can be accounted for only by using methodology, supplemented by a wide range of
practical design procedures. predictive tools for estimating the necessary properties,
can be used for a wide range of formulated products, also
Machine learning algorithms can, however, be used to improve those which include polymers, surfactants, emulsifiers,
the accuracy of predictive methods, if sufficient data on solvents and other compounds [18–20].
observed anomalies from physicochemical based theoretical
predictions, is available. A particular case of complex formulated products, with a
very long history of production and use, is the paints and
Addresses
1
coatings. Paints and coatings consist of a wide range of
CERE, Department of Chemical and Biochemical Engineering, Tech- components (polymers, pigments, solvents, additives), as
nical University of Denmark, Building 229, Søltofts Plads 229, DK – 2800,
Kgs. Lyngby, Denmark
shown in Figure 1. Solvents make up sometimes over 50%
2 (by volume) of the paint formulation, which, after drying
CoaST, Department of Chemical and Biochemical Engineering,
Technical University of Denmark, Building 229, Søltofts Plads 229, DK – and curing, consist mostly of the polymer(s) and
2800, Kgs. Lyngby, Denmark pigments.
3
Beijing Key Laboratory of Ionic Liquids Clean Process, CAS Key
Laboratory of Green Process and Engineering, State Key Laboratory of The properties of paints and coatings depend, often in a
Multiphase Complex Systems, Institute of Process Engineering, Chinese
Academy of Sciences, Beijing 100190, China
rather complex way, on the interactions between the
different components. These properties can be catego-
Corresponding author: Kontogeorgis, Georgios M (gk@kt.dtu.dk) rized in more than one ways for example, non-equilibrium
or transport properties (like chemical reactivity,
flow behavior – viscosity and diffusion) that are rate-
Current Opinion in Chemical Engineering 2020, 27:107–120
dependent, equilibrium or material properties (like vapor
This review comes from a themed issue on Frontiers ofchemical pressure, liquid-liquid phase separation, solubility,
engineering
wettability and surface tension — adsorption) that are
Edited by Rafiqul Gani and Lei Zhang rate-independent, and combined properties like the
For a complete overview see the Issue and the Editorial evaporation rate or the adhesion [22].
Available online 12th February 2020
The equilibrium and non-equilibrium properties can be
https://doi.org/10.1016/j.coche.2019.12.005
further classified into bulk-phase properties and surface-
2211-3398/ã 2020 Elsevier Ltd. All rights reserved. phase properties. The bulk-phase properties are those
which are influenced only by other molecules or material
within the formulation, while the surface properties are
those which are influenced by the other material which

www.sciencedirect.com Current Opinion in Chemical Engineering 2020, 27:107–120

108 Frontiers ofchemical engineering

Figure 1

Paint

Pigment Varnish

Coloured Film Liquid Additives

Extender
Pigment Former
1. Thickener
Polymer 2. Flowing Agent
(or reactive Platiciser Solvent Diluent 3. Matting Agent
components)
4. Catalyst
5. Accelator
6. Inhibitor
7. Wetting Agents
8. Dyestuffs

Current Opinion in Chemical Engineering

The main ingredients of a paint formulation (adapted from Ref. [21]).

forms the interface with the formulation [23]. the solvent and the mass-transfer process in solvent
Alternatively, the paint properties can be classified evaluation and drying depends on both the vapor-liquid
according to the type of the key ingredients in paint equilibrium (VLE) and liquid-liquid equilibrium (LLE)
and are summarized in Figure 2. Moreover, it is important properties.
to note the distinction between the ‘bulk’ properties of
the liquid paint and the properties of the final dry paint. As the well-defined bulk-phase properties (equilibrium
The properties of the final dry paint are also affected by and non-equilibrium) of polymer solutions and of the
different drying mechanisms and polymer selection. components of paints have this far-reaching significance
in coatings technology, this work focuses on them and
An important aspect in the design of paints and coatings is describes progress made in their understanding,
to formulate products with the desired characteristics and prediction and use in a computer-aided framework.
while the experimental approach has by far dominated in
the coatings industry, the collection of experimental data, Modelling and prediction of thermodynamic
the development of predictive property estimation and combined properties
methods and the use of computer-aided techniques can Thermodynamic properties are crucial in coatings
be highly useful in this context. technology, both in the formulation business (‘the like
dissolves like’ rule is well-accepted) and, directly or
As mentioned, paints are complex products and the link indirectly, in a wide range of other phenomena such as
between the bulk-phase properties and the final ‘dry stability of paint dispersions, swelling, rheology, surface
paint’ properties, is a key aspect in the aforementioned adsorption and drying-evaporation. Besides formulation
design process. However, the focus of this review is on the (finding suitable solvents for binders), the importance of
‘bulk-phase’ properties. As mentioned often in literature phase equilibria can be felt in many coatings-related
[22,24,25], knowledge of the bulk-phase properties is situations. For example, during solvent evaporation in
often indicative of many surface phenomena and largely the drying process the stability of the system may change,
responsible for the final performance of the paint. The and problems sometimes reported about drying are
same can be said of the equilibrium and non-equilibrium related to the instability, which leads to the formation
properties and a clear link between them often exists. For of two liquid phases with different overall rheology [24].
example, the preferential adsorption of substances onto Another situation is when adhesion problems reported are
solid surfaces from a liquid solution depends to a large associated with instabilities caused by the migration of
extent on their solution properties. Surface tension, non-evaporating species like plasticizers, thickeners and
wettability and adhesion are related. The viscosity of surfactants to the interface. Finally, in order to estimate
a polymer solution is sensitive to the quality of the protective properties of different coatings, it is also

Current Opinion in Chemical Engineering 2020, 27:107–120 www.sciencedirect.com

 A review of computer-aided design of paints and coatings Jhamb et al. 109

Figure 2

Properties of Dry
Paints and Coatings
• Adhesion
• Hardness, Toughness
• Durabiity
• Flexibility
• Solvent Resistance
Paint Ingredients Paint Types
• Ease to Repair
• Polymers
according to
• Loss due to Weathering
• Solvents
Polymer Selected
• Pigments • Alkyd, Polyester,
• Additives Acrylic
• Epoxy, PU

Additive Properties
• Types (Surfactants, Paints Drying Mechanisms
Plasticizers,
Dispersants,...)
and • Evaporation
• Oxidation / Reaction
• Viscosity, Coatings
Appearance

Pigment Properties Links between

Properies
• CPVC eg. Rheoglogy and
• Colour Solvent Properties Solubility
• Solvency–
Miscibility
• Cost
• Toxicity Key Ingredients and
• Viscosity Their Properties
• Flash Point Final Paint Properties
• Evaporation Rate Types of Paints
Drying Mechanisms
Current Opinion in Chemical Engineering

Key ingredients, their properties, drying mechanisms and types of paints that play an important role in determining the final properties of paints
and coatings.

important to know the interactions between cross-linked of the method in a CAD environment, as discussed later.
polymers and other species like additives in the coatings, There are no predictive methods for ‘Radius of
external solvents and vapors. Solubility’, which needs to be supplied separately.
The existing methods do not perform very satisfactorily
Considering the complexity and multicomponent nature for pigments, as shown in Figure 3, which calls for the
of coating formulations, it is crucial to have relatively development of improved methods for HSP of pigments.
simple and predictive models when such properties are In most cases, HSP are used ‘as it is’ without
concerned, especially if the results are to be implemen- temperature corrections; but they can also be incorpo-
ted in a CAD system. Over many years and still today, rated in a thermodynamic model framework and the
the solubility parameters especially the Hansen Solubil- Flory-Huggins/HSP method developed by Lindvig
ity Parameters (HSP) have been widely used [21,26]. et al. [28] is one way to do it.
This tool is simple and can be used not only to determine
the compatibility of solvents and polymers but also for GC methods have been developed, not just for HSP
solvents and pigments. Moreover, predictive group (dd, dh, dp), but for numerous other properties and
contribution (GC) methods have been developed over this has facilitated the applications of CAD enormously.
the years for HSP [27], which greatly facilitates the use A review on the topic has been recently published [29
].

www.sciencedirect.com Current Opinion in Chemical Engineering 2020, 27:107–120

110 Frontiers ofchemical engineering

Figure 3

40

35

Estimated values of δd, MPa1/2

30

25

20

15

10

0
0 5 10 15 20 25 30 35 40
Experimental values of δd, MPa1/2

22 22

20 20

18 18
Estimated values of δh, MPa1/2

Estimated values of δd, MPa1/2

16 16

14 14

12 12

10 10

8 8

6 6

4 4

2 2

0 0
0 2 4 6 8 10 12 14 16 18 20 22 0 2 4 6 8 10 12 14 16 18 20 22

Experimental values of δh, MPa1/2 Experimental values of δp, MPa1/2

Current Opinion in Chemical Engineering

Prediction of Hansen Solubility Parameters (HSP) for organic pigments using a second-order group contribution (GC) method [27,30].

Despite the fact that such methods based on HSP do This statement holds even today. Even though Holten-
include some approximations and assumptions in the Andersen proceeds to state “for quantitative purposes,
experimental determination of the values, extrapolation GC methods are necessary”, there are many issues. For
to other conditions can be hazardous and GC methods are example, of crucial importance in coatings science is the
approximate, they are still of wide use in coatings prediction of LLE of systems containing multiple poly-
industry, as for the very complex systems involved a mers and solvents. This is a largely unresolved problem
compromise between simplicity and accuracy is often even today. There are many thermodynamic models
needed. today for polymers (see for a review Refs [31–34]) in
form of GC activity coefficient models and equations of
Holten-Andersen [25] wrote more than 30 years ago “For state with good or rather good performance for activity
qualitative purposes the three-dimensional solubility coefficients/VLE but their performance for LLE is
parameter method is excellent for evaluating the solubil- unreliable for coatings applications. Even some of the
ity of polymers in solvents. It is the belief of the author most advanced equations of state of our times like GC
that no other present methods offer the same combination versions of the SAFT theory have limited predictive
of rational foundation, simplicity and predictive power”. applicability when it comes to multicomponent

Current Opinion in Chemical Engineering 2020, 27:107–120 www.sciencedirect.com

 A review of computer-aided design of paints and coatings Jhamb et al. 111

polymer-solvent LLE [35–37]. Thus, these models are The four polymers considered are Polyisobutylene (PIB),
not suitable for CAD purposes. This is not to say they Polystyrene (PS), Polyethyl methacrylate (PEMA) and
are not useful; GC versions of the PC-SAFT model have Polyvinyl acetate (PVAC). For each of the four polymers,
been applied to LLE for a wide range of polymer-solvent the solvents: heptane, benzene, ethyl acetate and tetra-
systems (for a review see Refs. [31,37]) and the same can chloromethane, are considered to estimate the V1 1 for the
be said for models like Entropic-FV [38,39], a polymer system. The calculations are made with the software
extension of the well-known UNIFAC method, but the SPECS developed at the Technical University of
predictive results for LLE are not consistent or reliable Denmark (CERE) [40]. There is a good agreement for
in all applications and much more work is needed, most cases in the methods and experimental values and
especially for complex polymers and solvents and for assessments of miscibility indicating that, both the rule of
multicomponent systems, which are of importance in thumb and combined use with GC methods, are working
coating applications. well. However, in the case where the repeat unit of the
polymer has a similar molecular structure to that of the
An alternative to the use of full thermodynamic models, solvent, the UNIFAC – FV model performs better than
which can be easily implemented in CAD systems, is the the Entropic – FV model. This can be seen from
use of the so-called infinite dilution activity coefficients the predicted values of V1 1 of the systems ‘polyvinyl
(on weight-basis) for polymer-solvent systems (V1 1 ). acetate — ethyl acetate’, ‘polystyrene — benzene’ and
This approach has been shown in literature ‘polyisobutylene — heptane’.
[22,24,28,32,34,35] to have a wide applicability and
predictive value for fast estimation of miscibility. In these This tool is particularly useful in coatings applications as
references are mentioned several simple rules of thumb in many practical problems we are dealing with the dilute
for example, V1 1 values less than 6 indicate good concentration regime for example, during the last stages
solubility in polymer-solvent systems (and above 8 or of the solvent evaporation and in water-based coatings
10 immiscibility over the entire concentration range), the organic components are present in rather small
while for organic solvent-water systems the limits are concentrations.
below 5 for miscibility and above 7 for immiscibility.
These rules of thumb are derived based on the Flory- In the aforementioned references are also provided
Huggins equation. Moreover, such V1 1 values can equations for a simple dependence of V1 1 values with
be easily calculated by predictive GC methods like temperature, so that it is always used at the temperature
UNIFAC – FV and Entropic – FV and the results of of interest. The temperature-effect is significant and as
calculations for 16 systems is shown in Table 1. often experimental V1 1 values are at high temperatures

Table 1
1
Omega (V1 ) values (weight-based infinite dilution activity coefficients) and S/NS (solvent/non-solvent) estimations using group-
contribution activity coefficient models available in the software SPECS [40].
Polymer Solvent Experimental V1
1 ([22]; Predicted V1
1 (with percentage deviation from S/NS
DECHEMA 1992) experimental data) Estimation
UNIFAC – FV Entropic – FV
(Using SPECS) (Using SPECS)
1 Polyisobutylene Heptane 6.0 6.18 (3.0 %) 5.12 (14.7 %) S
(PIB) Benzene 7.0 7.37 (5.3 %) 5.89 (15.9 %) S
Ethyl acetate – 15.41 (–) 12.33 (–) NS
Trichloromethane 6.0 (Extrapolateda) 4.43 (26.2 %) 3.76 (37.3 %) S
2 Polystyrene Heptane – 11.55 (–) 7.95 (–) S
(PS) Benzene 4.4 (Extrapolateda) 4.29 (7.25 %) 4.30 (7.5 %) S
4.0
Ethyl acetate – 5.96 (–) 3.92 (–) S
Trichloromethane 2.1 1.98 (5.7 %) 1.69 (19.5 %) S
3 Polyethyl methacrylate Heptane 29.0 (Extrapolateda) 452.19 (1459.3 %) 18.19 (37.3 %) NS
(PEMA) Benzene 6.4 (Extrapolateda) 167.37 (2515.2 %) 6.49 (1.4 %) S
Ethyl acetate – 180.21 (–) 6.78 (–) S
Trichloromethane 2.1 (Extrapolateda) 48.63 (2215.7 %) 2.05 (2.4 %) S
4 Polyvinyl acetate Heptane 39.0 (Extrapolateda) 50.87 (30.4 %) 45.80 (17.4 %) NS
(PVAC) Benzene 5.0 4.27 (14.6 %) 4.24 (15.2 %) S
Ethyl acetate 5.3 4.71 (11.1 %) 3.95 (25.5 %) S
Trichloromethane 1.6 (Extrapolateda) 1.05 (34.4 %) 0.98 (38.7 %) S
h iT 0 =T
V1
1 ðT 0 Þ
a
In this case, V1  1
1 is experimentally obtained at another temperature and extrapolated to 25 C using the equation, V1 ðTÞ ¼ 3:699 3:669

www.sciencedirect.com Current Opinion in Chemical Engineering 2020, 27:107–120

112 Frontiers ofchemical engineering

extrapolation methods or predictive models are needed in bonding compounds. In the latter case, the association/
order to use the approach at the temperature range of SAFT theories mentioned above have been developed
interest to coatings. It is also shown that the concept is and are useful; but even in their GC form, they lack, as
useful for the estimation of the organic solvent- previously stated, the wide applicability of activity
distribution in two-phase systems (e.g. water-based coefficient models, at least for applications in the coatings
latex-type coatings) and thus a rather complex LLE industry. This situation may change in the future.
problem is very much facilitated by the use of V1 1
concept. The V1 1 concept is also very useful in the Earlier studies in CAD of coatings
analysis of the adsorption and solution tendencies of In the previous section, we have identified several
the A and B parts of A-B block co-copolymers often used thermodynamic and combined properties of relevance
for the steric stabilization of two-phase coating systems to the design of coatings, formulation and beyond. We
[22,24]. Finally, an important application is in connection mentioned rules of thumb and the value of GC estimation
to the drying of coatings which may be often divided into methods. However, a comprehensive design with help of
two mechanisms; A (controlled by diffusion in the vapor computers of paints and coatings requires at least three
phase above the liquid) and B (solvent diffusion in the elements: i) identification of the important properties and
polymer film). The evaporation rate under mechanism A predictive methods for their estimation, ii) a systematic
depends on the solvent concentration and vapor pressure, methodology and iii) computational algorithms.
temperature, air velocity and non-ideality (just the
solvent mixture as polymer concentration is low). As One of the early pioneers in the area is Dr. Dao Tsing Wu,
evaporation proceeds, and for polymer concentrations first associated with DuPont and later as consultant and
over 60% the solvent activities decline, the evaporation founder of DTW Associates Inc. Even though, to the best
rates are reduced and at the later stages the evaporation is of our knowledge, these activities have not been
determined by the solvent-polymer V1 1 . As this discus- continued, Wu and co-workers have made a significant
sion indicates, this parameter is of wide applicability in contribution to the field with pioneering developments
coating applications and can be directly used in CAD on the use of computational tools for coatings design
systems. The case of the drying/evaporation rate is also of [41–44]. In these publications he presents the overall
interest for another reason; it shows that in some (but not methodology as well as examples from his work in
all!) cases in coatings applications, a combined or com- DuPont where CAD has made a difference.
posite property can be predicted by a theory — equation
which, in turns, depends on several well-defined pure He recognizes that the availability of well-designed and
compound and/or mixture properties (in this case vapor user-oriented software is of fundamental importance for
pressure and activity coefficients) which can all be esti- the acceptance of computer-aided methods by industry,
mated from predictive GC methods. Another combined but the lack of standardization within the coatings
property is the swelling of cross-linked polymer networks industry makes it difficult to develop such software.
(as many polymers in coatings often are in their final He stated this in 1987 [42] and in many respects it is
state). This can be estimated by using an extension of the true today 30+ years later. The interested reader can find
Flory-Huggins equation by adding a term for accounting in these publications several case studies on how an
for the elastic effects caused by the deformation during acrylic polymerization model assisted in developing a
the swelling of the cross-linked polymer network [24]. procedure for the synthesis of the polymers with 5 days
The final equation includes parameters related to the of computation efforts and 2–5 experimental runs, while
structure of the network, which can be obtained from without the use of the model a 2–3 months effort was
mechanical properties (elastic modulus) of the polymer expected. In this review, we focus only on the solvent
and a thermodynamic interaction parameter, which can formulation methodology. This is crucial as solvent raw
be estimated from V1 1 of the solvent in the uncrosslinked material cost can be as much as one-third of the total raw
primary polymer molecule. material cost for coating manufacturers [43] and in many
coating formulations, more than 50% of the coating is
It should be emphasized that the models typically used for made of solvents [44]. Moreover, solvent formulation
predicting V11 mentioned above for example, Entropic- affects many end properties like aesthetics, coverage
FV and UNIFAC-FV are based on well-accepted and functional properties, often in opposing ways, which
principles such as the free-volume concept important is challenging for the formulator. According to Wu, by
for polymer solutions combined with a local-composition relating in a qualitative way the key solvent (blend)
model for the energetic interactions expressed in a group- properties to end use properties and balancing the
contribution form so that the final model is predictive. key properties can result to a computerized solvent
These concepts are all useful and often highly applicable formulation, which can be effective and efficient.
but they are not perfect. There are deviations from
experimental data, even for activity coefficients, The first step of Wu’s approach is to identify the key
especially for systems with highly polar and/or hydrogen properties and develop predictive methods of models for

Current Opinion in Chemical Engineering 2020, 27:107–120 www.sciencedirect.com

 A review of computer-aided design of paints and coatings Jhamb et al. 113

their estimation. Due to the complexity and many 6–8 are repeated until a satisfactory performance of the
components involved, only GC models are used (often coating system is obtained.
UNIFAC as is or in diverse frameworks) or other simple
ready-to-use approaches like HSP. The properties The methodology mentioned above has been incorpo-
identified are: rated in a software tool which was marketed for a period
by Wu and co-workers [45]. A short course for the use of

the solvent evaporation model to predict evaporation software was also available at the time.
and solvent composition as function of time under
specified ambient conditions (UNIFAC is used to The advantages of the CAD approach are in solvent
estimate the activity coefficients needed in the model; replacement, cost reduction and improvement of the
polymer effects are neglected due to the dilute application properties. According to Wu, using the
polymer concentration range) CAD system they have proposed, the formulations

the polymer solubility envelope using the HSP (which developed were lower in cost and a 5% reduction in
also allows an approximate assessment of whether raw material could easily be realized [43]. In the examples
phase separation will take place or not during solvent shown in his 1987a,b and 1992 publication, it was possible
evaporation) to achieve a 30% reduction in raw material cost without

the flash point of a solvent blend estimated using the affecting the application properties of the formulation.
UNIFAC-based GC method (polymer and pigments- The advantages do not stop there. According to Refs.
additives effect neglected) [41–43], the company formulators found that the

surface tension using a UNIFAC-based GC method evaporation curves can be manipulated to improve
(this made is rather out-of-date today) different application properties like sagging, application

the solvent blend viscosity (again ignoring the polymer) latitude and popping. Wu [42] says that correlation
using Eyring’s model and a GC method based on between model predictions and experimental observa-
UNIFAC for estimating the excess free energy of tions appear to exist for most product lines.
activation

the boiling point of solvent mixture (also without Wu concludes from his work on a number of important
considering polymers) advantages of the CAD approach for paints and coatings:

diverse pure solvent and polymer physical properties routine application of complex theories which can be used
by non-experts, gaining insights into problems difficult to
assess experimentally, minimizing (but not eliminating)
experimental work, examining more options and identify-
ing optimum ones and finally having a good training and
In the 1992 publication, Wu presented also an 8-step also dissemination tool for technical information. Wu also
formulation CAD methodology which makes use of the recognizes several limitations or problems such as the need
properties and methods mentioned above. He states that for multi-disciplinary approach and group work in an indus-
an effective use of his system requires some prior trial environment, time-consuming process to develop the
knowledge of the desirable range of solvent blend necessary models and even lack of data and predictive
properties and some experimentation in order to refine methods for some of the properties and of course the
the formulation. The proposed methodology starts with approximate nature of the results. We have to say that,
establishing, for each polymer, the solubility parameter while much progress has been made, see also in later
ranges for the solvents/solvent blends that dissolve the sections, many of these problems must still occur today
polymer (step 1) and then the overlap miscibility region as despite being 30+ years old these statements — we still
for the multi-polymer binder (based on the mutual see resistance in the coatings industry to adopt these
solubility regions of all polymers) (step 2). In step 3, methods or simply several of the problems mentioned
the overlap miscibility region is refined using single are still actual today. This will be discussed further when
solvents and the film formation process of ineptest and we will analyze more modern approaches presented next.
then (step 4) the solvent evaporation rate of the blend is
specified (based on previous experience or experiments). Indeed, while the approach of Wu includes many of the
Viscosity and surface tension of blends are specified elements of a modern CAD methodology, it involves several
including how they will be changed during evaporation steps and is not entirely computerized. Moreover, the num-
(step 5). In the crucial step 6, the CAD system is used to ber of properties is limited. Dr. Wu together with professor
select solvent blends that during evaporation have Gani and co-workers have subsequently [46,47] presented
solubility parameters within the overlap miscibility region computer-aided mixture design methods with pre-defined
(step 3) and the percentage evaporation versus time property constraints using optimization algorithms and both
curves march those specified in step 4. In step linear and non-linear constraints for the physical properties.
7 the formulation is checked experimentally and in step The newly then- developed second-order GC methods were
8 the constraints of steps 4 and 5 are readjusted and steps used.

www.sciencedirect.com Current Opinion in Chemical Engineering 2020, 27:107–120

114 Frontiers ofchemical engineering

These methods, which in these publications were used ingredients. Finally, the additives are selected to correct
for solvent design with cost as the objective function, the properties of the formulation.
were later extended and enhanced by Gani and co-
workers for complex formulated products, as discussed Solvent mixture design for formulation
in the next section. Solvent selection and mixture design, an essential part of
CAD, has been studied extensively over the last years.
Successful adaption of these methods for paints and
A modern approach to CAD of coating coatings could help to significantly reduce cost and waste
formulations generation while improving the performance and
The systematic CAD methodology by Constantinou, reducing the environmental impact of coatings, as a
Gani and co-workers [33,47,48] served as a basis for the significant part of a coating formulation may consist of
later work to extend the methodology to include more organic solvents [50].
complex products, such as liquid formulations of organic
coatings [11]. This computer-aided workflow allows for The framework for the design of liquid based formula-
the screening of potential candidates for active tions developed by Conte et al. [11], which is integrated in
ingredients, solvent selection, and additives. Later stages a computer-aided framework called the virtual Product
involve the necessary experimental verification of the Process Design Laboratory, or the VPPD-Lab [51], has
chosen candidates. been successfully applied to TiO2 - based coating
formulation. The computer algorithm, MIXD developed
The process of designing a chemical product must be in the aforementioned work for the systematic design of
based on a fundamental understanding of the relevant thermodynamically stable binary solvent mixtures, is an
interactions and properties. Figure 4 illustrates the efficient and reliable method. When a sensitivity analysis
methodology developed by [49]. The first task is to define is performed on the constraints chosen for the design of
the consumer-specified needs of the product, and to solvent mixture, it is found that the optimum solvent is
translate the needs into target physicochemical properties most sensitive to the constraints set on the property,
of the raw materials. For the second task, the ingredients evaporation time, T90. However, other needs of an exte-
performing the main function, which are the pigments rior coating formulation are required to be explored for
and polymers in a coating, are selected. The necessary instance, inflammability, solvency (for polymer resin)
properties are found either in databases or through GC and biodegradability. Corresponding to these needs,
models when the measured data for the property of the target properties would be flash point, HSP and
interest is not available. The third task is to design a biological oxidation parameter respectively. Additionally,
solvent mixture that both matches the given property the possibility of modelling the sensory properties like
constraints and is compatible with or dissolve the active smell / odor of the solvent mixture can be useful.

Figure 4

Step 1 Step 2 Step 3 Step 4 Step 5

Translation of
Active Ingredient Solvent Design Additive Formulation
Needs to Target
Selection and Selection Selection Evaluation
Properties

Quantify Users’ Needs

Design Solvent/Mixture Select Additive to
Select Enhance or Correct Experimental
Flow easily Viscosity
Pigment & Binder for the Selected
Pigment & Binder Formulation Properties Verification
Waterproof Water-insoluble

Current Opinion in Chemical Engineering

Framework for the CAD of coatings as implemented in VPPD-Lab.

Current Opinion in Chemical Engineering 2020, 27:107–120 www.sciencedirect.com

 A review of computer-aided design of paints and coatings Jhamb et al. 115

Venceslau et al. focused on a CAD system for optimizing Current efforts are focused on substitution of hazardous
the cost, evaporation rate, and solubility constraints for chemicals in coatings by including constraints on Envi-
solvents in coatings [50]. A case study applying the ronmental-(EH&S), Health-(EH&S) and Safety-
framework presents an optimal coating formulation for (EH&S) related properties [12,13] and on extending
a given relative evaporation rate and HSP, optimized GC methods to include estimation of pure component
based on the cost of solvent mixture. properties for pigments. The latter is approached by
inclusion of new functional groups in a multi-order GC
While solvent selection programs have been available for method to better describe the molecules, as organic
many years, Venceslau et al. [50] state that trial-and-error pigments are complex multifunctional chemicals and
experiments to find working solvents mixtures which therefore require more structural information than cur-
fulfill target properties is still common in the industry. rent GC methods provide. High accuracy for the estima-
This implies that the software tools need to be adequately tion of solubility parameters of raw materials, including
developed in order to provide a significant contribution to pigments and polymers, is essential to correctly predict
the coatings field. the various interactions and performance of the final
product. Several theses at DTU Chemical Engineering
Although Conte et al. [11] developed a generic formula- make use of this modern CAD approach. A summary is
tion methodology, it may not be comprehensive enough shown in Table 2.
for the design of coating formulations. Therefore, in the
work by Jhamb et al. [52
], an adaptation of this model- The efforts of these researchers are not the only
based framework is developed wherein the third step is current attempt to use a computational approach to
specifically tailored for the design of solvents for these coating formulations. In recent years, several discus-
formulations. The proposed framework takes into sions and articles by Sander van Loon, CEO of the
account the effect of properties of polymers and pigments company Van Loon Chemical Innovations (VLCI)
on the choice of solvents for a specified system. Moreover, have been published regarding the use of HSP for
both the MIXD algorithm from the work by Conte et al. paints, coatings, and ink formulations. In addition to
[11] for binary solvent mixture design and the computer- employing the thermodynamic-based HSP for formu-
aided molecular design (CAMD) algorithm developed in lations and homogeneous solution design as well as
the work by Gani et al. [53] for the design of single Hydrophilic Lipophilic Difference – Net Average
molecular organic solvents, are used to design solvents Curvature for emollient design, VLCI utilizes high-
for a system containing acrylic polymers. throughput experimental methods for property deter-
mination [54]. Information published by the company
The compatibility between solvents and active ingredi- indicates advances in the methodology for raw mate-
ents is screened by solubility parameters, and either rial selection, including evidence for the selection of
databases or pure component property prediction by dispersants with overlapping solubility spheres
GC methods is used to find the desired properties. A between the pigments or fillers [55
].
computer algorithm, mixture design (MIXD), then
screens the candidate compounds using linear and rig-
orous property models. The remaining candidates can
then be optimized by performance, such as a desired Figure 5
property or cost. An illustration of this algorithm applied
to binary solvent mixture design for a TiO2 based
formulation is shown in Figure 5. A second algorithm,
STABILITY, rejects any candidates that indicate a Linear Design
phase separation at their given compositions. The final
465 s
task in the computer framework is to identify which Non-linear Design
ure
qualities can be enhanced with additives, such as 59 i xt
f m
increased dispersion and spreadability or reduced micro- Stability Check 11 ro
bial growth. be
11
n um
Verification
The above-mentioned approach does not only present 8
new opportunities for design of organic coating formula- Optimal
tions but also includes significant challenges as the com- search 1
plete product involves complex interactions between the
different polymers, pigments, solvents, and additives. Current Opinion in Chemical Engineering
Further improvement for databases, predictive methods,
and the framework can extend the usefulness of the Illustration of the Binary Solvent Mixture Selection Framework for a
product design system. TiO2 based Coating Formulation [11].

www.sciencedirect.com Current Opinion in Chemical Engineering 2020, 27:107–120

116 Frontiers ofchemical engineering

Table 2 Computer-aided tools for formulation design

PhD Theses at DTU chemical engineering that make use of the Several tools based on property models and fundamentals
modern CAD approach in thermodynamics have been developed for coating
Year PhD Thesis Products
formulation design. The CAMD and MIXD algorithms,
mentioned in the previous section, are available via the
2000 Peter Harper Mixture/Solvent design
2009 Kavitha Polymer Design
ProCAMD tool of ICAS [56] and the VPPD-Lab tool [51]
Satyanarayana respectively, both developed at DTU Chemical Engi-
2010 Elisa Conte Solvent formulations-lotions neering. The generic architecture of the VPPD-Lab tool
(hairspray, insect repellent, paint, is shown in Figure 6.
sunscreen)
2014 Michele Mattei Emulsified products
(hand-wash, tank-cleaning, HSPiP, an eBook software, contains HSP and other data
sunscreen) on 10,000 chemicals in the current edition. Besides, the
2014 Nor Alafiza Blends (gasoline, lubricants) software’s capability to calculate and visualize HSP in
Yunus 3D, it has a solvent optimizer (up to 8 solvents, plus
2016 Sawitree Kalakul Jet-fuel blends
Diesel blends
graphing of solvent mixture changes depending on
2019 Spardha Jhamb Chemical Substitution, Paints relative evaporation rates, activity coefficients and tem-
2021 Markus Enekvist Paints perature), a polymer calculator and a diffusion modeler
(ongoing) [57].

Figure 6

STAGE 1

Framework
Problem Definition
Stage

Property
Models

STAGE 2
Knowledge
Model-based Base
Design Stage

Databases

STAGE 3

Algorithm
Experiment-based
Verification Stage

Current Opinion in Chemical Engineering

Generic Architecture for CAPD used in VPPD – Lab tool.

Current Opinion in Chemical Engineering 2020, 27:107–120 www.sciencedirect.com

 A review of computer-aided design of paints and coatings Jhamb et al. 117

Steven Abbott’s company, Steven Abbott TCNF Ltd, has that the appearance design of coatings is still performed
developed apps based on colloid and interfacial science by formulators based on visual inspection and experience;
through the updated Kirkwood-Buff’s 1950s theory and so, designing photorealistic predictions of coating appear-
HSP technique, in order to aid practical formulators in ance by their optical properties coupled with formulation
their experimental activities [58

]. parameters, is still a field in need of exploration. The
complexity of pigmented coatings still provide significant
Machine learning algorithms for formulation challenges for production of coatings samples from theo-
Another program for coating formulation, currently under retical models, but the proposed appearance design
development by Evonik Industries is COATINOTM, framework can provide significant value for many indus-
claims to be the first voice-controlled digital assistant trial applications.
aimed at the coatings industry [59]. While, as far as the
authors of this article are aware, few details regarding the Prediction of service life
selection principles are available, Oliver Kröhl, who is Computational prediction of service life for coatings is a
responsible for the project, states that the program will research area of high interest as experimental tests
utilize human knowledge, high throughput experimenta- require data collection over a very long period of time
tion hardware, and software based on data and machine and natural factors and differences in surface treatment
learning to find suitable raw materials, including addi- and application lead to low reproducibility. The current
tives, for the formulator [60]. method of predicting service life of industrial coatings is
usually through controlled accelerated testing. The deg-
Other notable applications of CAD for radation of organic coatings requires an understanding of
coatings the complex relationships between exposure factors,
Computer-aided coating appearance design material properties, and system performance, and can
Algorithms and models for color prediction have been be measured by factors such as loss of functionality,
available for many decades, as far back as 1944 [61]. It is a adhesion, or gloss [66]. Several computer methods and
field that has been approached from many sides, and for software tools for lifetime prediction have been proposed
many purposes. Available color matching methods for [67–71], using both statistical methods and neural
paints and coatings often utilize spectral methods like networks.
colorimetry, standard tables like those provided by Inter-
national Commission of Illumination (CIE) and statistical Design of self-stratifying coatings
Artificial Intelligence methods [62]. Today, many differ- Self-stratifying coatings is a field of coating formulation
ent systems are available to the coating formulator for that progressed significantly during the 90’s to both
both color matching and prediction such as LargoMatch understand the underlying mechanism, and to create
[63]. systematic computer tools to optimize the formulation.
These systems provide an interesting challenge for CAD,
Advances in computer graphics allowed for prediction of as the multi-resinous formulation must spontaneously
optical behavior and direct feedback to the user. For the stratify into separate layers after application, and provide
design of automotive coatings, measurements of metallic improved properties such as appearance, barriers, or
flake finishes were used to predict hue, saturation and adhesion. The different layers are usually separated by
brightness through an interactive graphics program [64]. a gradient of the resins and can, in addition to the
Though differences were noted for all experimental tests improved properties, provide a significant reduction in
of the developed software, the appearance matches were cost and time of application compared to other multi-coat
determined to be adequate, and required no prior trial- systems.
and-error experimentation. A more recent comprehensive
computer-aided appearance design system for coating An early effort to gather data on self-stratifying coatings
formulation, based on bidirectional reflectance distribu- was performed for the European Commission in 1996 to
tion function (BRDF) measurements, has been proposed build a computer model to predict stratification by statis-
by Kim and Lee [65]. BRDF defines how light is reflected tical analysis. The investigated properties were solubility
at an opaque surface and is a function of illumination parameters, solubility radius, surface tension and molec-
geometry and viewing geometry. The function takes a ular weights [72]. Notable limitations were a lack of data
negative incoming light direction, and outgoing direction, for additives, pigments, and surface-treatments resulting
both defined with respect to the surface normal, and that the study only encompassed the properties of poly-
returns the ratio of reflected radiance exiting along out- mers and solvents. Kinetic variables, which can prevent a
going direction to the irradiance incident on the surface thermodynamically favored self-stratification from taking
from incoming light direction. Each direction is itself place, were also omitted while building the model.
parameterized by the two position coordinates, azimuth
angle, zenith angle, wavelength and time; making the The design requirements have been discussed over the
BRDF a twelve-dimensional function. The authors state last decades, but some fundamental requirements are that

www.sciencedirect.com Current Opinion in Chemical Engineering 2020, 27:107–120

118 Frontiers ofchemical engineering

of the included polymers, which must be fully compatible 4. Cussler EL, Moggridge GD: Second Edition. Chemical Product
Design 2011, Vol. 9780521168 http://dx.doi.org/10.1017/
with the choice of solvent mixture. The compatibility has CBO9781139035132.
been estimated through both the HSP distance and by the 5. Hill M: Product and process design for structured products.
overlap volume of the solubility spheres [73

]. Other AIChE J 2004, 50:1656-1661 http://dx.doi.org/10.1002/aic.10293.
aspects such as the evaporation rate of solvents and the 6. Wei J: A century of changing paradigms. ChemTech 1996:16-
surface tensions of the polymers, are also taken into 18.
account. 7. Costa R, Moggridge GD, Saraiva PM: Chemical product
engineering: an emerging paradigm within chemical
engineering. AIChE J 2006, 52:1976-1986 http://dx.doi.org/
10.1002/aic.10880.
Prediction of coating functionality from formulation
8. Cussler EL, Wei J: Chemical product engineering. AIChE J 2003,
variables 49:1072-1075 http://dx.doi.org/10.1002/aic.690490502.
Various attempts have been made to predict coating 9. Hill M: Chemical product engineering-the third paradigm.
functionality from formulation variables in an effort to Comput Chem Eng 2009, 33:947-953 http://dx.doi.org/10.1016/j.
optimize the recipe. These efforts extend over more than compchemeng.2008.11.013.

two decades and use polynomial models and neural net- 10. Woinaroschy A: A paradigm-based evolution of chemical
engineering. Chin J Chem Eng 2016, 24:553-557 http://dx.doi.
works to predict selected functional properties such as org/10.1016/j.cjche.2016.01.019.
elasticity, hardness and barrier properties as a function of
11. Conte E, Gani R, Ng KM: Design of formulated products: a
formulation parameters [74]. There are now several com- systematic methodology. AIChE J 2011, 57:2431-2449 http://dx.
mercial software packages that are used to optimize the doi.org/10.1002/aic.12458.
formulation of coatings [75]. 12. Jhamb S, Liang X, Gani R, Kontogeorgis GM: Systematic model-
based methodology for substitution of hazardous chemicals.
ACS Sustain Chem Eng 2019, 7:7652-7666 http://dx.doi.org/
Conclusion 10.1021/acssuschemeng.8b06064.
CAD is still in an early stage in the paints and coatings 13. Kalakul S, Zhang L, Fang Z, Choudhury HA, Intikhab S, Elbashir N,
industrial sector and its potential is yet to be utilised to

Eden MR, Gani R: Computer aided chemical product design –
ProCAPD and tailor-made blended products. Comput Chem
the maximum. It can prove to be extremely beneficial in Eng 2018, 116:37-55 http://dx.doi.org/10.1016/j.
terms of saving time and resources required for trial and compchemeng.2018.03.029
Extends computer-aided framework and its implementation in a product
error experimentation. Several PC-based tools and appli- design software tool– ProCAPD and particularly discusses case studies
cations have been developed, which are based either on involving blend design.
thermodynamics and physicochemical property models or 14. Karunanithi AT, Achenie LEK, Gani R: Optimal (solvent) mixture
on machine learning algorithms and high-throughput design through a decomposition based CAMD methodology.
Comput Aided Chem Eng 2004, 18:217-222 http://dx.doi.org/
experimentation. However, a hybrid approach that com- 10.1016/S1570-7946(04)80102-0.
bines the advantages of fundamentals as well as data
15. Mattei M, Kontogeorgis GM, Gani R: A systematic methodology
science could potentially improve the predictability of for design of emulsion based chemical products. Comput
these tools. Aided Chem Eng 2012, 31:220-224 http://dx.doi.org/10.1016/
B978-0-444-59507-2.50036-6.
16. Ng KM: A multiscale-multifaceted approach to process
Conflict of interest statement synthesis and development. Comput Chem Eng 2001, 9:41-54
Nothing declared. http://dx.doi.org/10.1016/S1570-7946(01)80005-5.
17. Yunus NAB, Gernaey KV, Woodley JM, Gani R: A systematic
methodology for design of tailor-made blended products.
Acknowledgements Comput Chem Eng 2014, 66:201-213 http://dx.doi.org/10.1016/j.
Financial support from DTU Chemical Engineering and the Hempel compchemeng.2013.12.011.
Foundation to CoaST (The Hempel Foundation Coatings Science and
18. Gani R: Chemical product design: challenges and
Technology Centre) is acknowledged.
opportunities. Comput Chem Eng 2004, 28:2441-2457 http://dx.
doi.org/10.1016/j.compchemeng.2004.08.010.
References and recommended reading 19. Gani R, Ng KM: Product design - molecules, devices, functional
Papers of particular interest, published within the period of review, products, and formulated products. Comput Chem Eng 2015,
have been highlighted as: 81:70-79 http://dx.doi.org/10.1016/j.compchemeng.2015.04.013.

of special interest 20. Zhang L, Kalakul S, Liu L, Elbashir NO, Du J, Gani R: A computer-


of outstanding interest aided methodology for mixture-blend design. Applications to
tailor-made design of surrogate fuels. Ind Eng Chem Res 2018,
57:7008-7020 http://dx.doi.org/10.1021/acs.iecr.8b00775.
1. Seider W, Seader JD, Lewin D: Product and process design
principles. Synthesis, analysis, and evaluation. J Chem Inf 21. Turner GPA, B. J: Introduction to paint chemistry and principles
Model 2003, 53 http://dx.doi.org/10.1017/ of paint technology. Introduction to Paint Chemistry and
CBO9781107415324.004. Principles of Paint Technology. 1988 http://dx.doi.org/10.1007/
978-94-009-1209-0.
2. Wesselingh JA, Kiil S, Vigild ME: Design and Development of
Biological, Chemical, Food and Pharmaceutical Products. John 22. Holten-Andersen J, Eng K: Activity coefficients in polymer
Wiley and Sons Ltd; 2007. solutions. Progr Org Coat 1988, 16:77-97 http://dx.doi.org/
10.1016/0033-0655(88)80006-2.
3. Biegler LT, Grossmann IE, Westerberg AW: Issues and trends in
the teaching of process and product design. AIChE J 2010, 23. Buckton G: Interfacial phenomena, surface tension and liquid/
56:1120-1125 http://dx.doi.org/10.1002/aic.12256. liquid interfaces. Interfacial Phenomena in Drug Delivery and

Current Opinion in Chemical Engineering 2020, 27:107–120 www.sciencedirect.com

 A review of computer-aided design of paints and coatings Jhamb et al. 119

Targeting. CRC Press; 2000:1-25 http://dx.doi.org/10.1201/ 42. Wu DT: Modeling and simulation in the coatings industry.
9781482296785. Chemtech 1987, 17.
24. Holten-Andersen J, Eng K: Solvent activity coefficients in 43. Wu DT: Applications of computers in coatings research and
polymer solutions. Scand Paint Print Ink Res Inst 1987. Report development. J Ind Text 1987, 17:22-40 http://dx.doi.org/
T2-87. 10.1177/152808378701700104.
25. Holten-Andersen J: Physical models and coatings technology. 44. Wu DT: Computer aided application of chemical engineering
Scand Paint Print Ink Res Inst 1987. Report T12-87. principles in the development of coatings. Comput Chem Eng
1992, 16:S31-S42 http://dx.doi.org/10.1016/S0098-1354(09)
26. Hansen CM: Applications — coatings and other filled polymer 80007-0.
systems. Hansen Solubility Parameters: A User’s Handbook.
2007:137-150 http://dx.doi.org/10.5860/choice.37-3905. 45. Life Science Connect: DTW Associates, Inc. . (accessed Oct 21,
27. Stefanis E, Constantinou L, Panayiotou C: A group-contribution 2019) 2000.
method for predicting pure component properties of 46. Klein JA, Wu DT, Gani R: Computer aided mixture design with
biochemical and safety interest. Ind Eng Chem Res 2004, specified property constraints. Comput Chem Eng 1992, 16:
43:6253-6261. S229-S236 http://dx.doi.org/10.1016/S0098-1354(09)80027-6.
28. Lindvig T, Michelsen ML, Kontogeorgis GM: A Flory-Huggins
47. Constantinou L, Bagherpour K, Gani R, Klein JA, Wu DT:
model based on the Hansen solubility parameters. Fluid Phase
Computer aided product design: problem formulations,
Equilibria 2002, 203:247-260 http://dx.doi.org/10.1016/S0378-
methodology and applications. Comput Chem Eng 1996,
3812(02)00184-X.
20:685-702.
29. Gani R: Group contribution-based property estimation

methods: advances and perspectives. Curr Opin Chem Eng 48. Gani R: Computer-aided methods and tools for chemical
2019, 3:184-196 http://dx.doi.org/10.1016/j.coche.2019.04.007 product design. Chem Eng Res Des 2004, 82:1494-1504 http://
An overview on the state of the art in group-contribution-based property dx.doi.org/10.1205/cerd.82.11.1494.52032.
estimation methods and their further development and use in model-
49. Conte E: Innovation in Integrated Chemical Product- Process
based chemical product– process design.
Design -Development through a Model-based System Approach,
30. Stefanis E, Panayiotou C: Prediction of hansen solubility PhD Thesis. Technical University of Denmark; 2010.
parameters with a new group-contribution method. Int J
Thermophys 2008, 29:568-585 http://dx.doi.org/10.1007/s10765- 50. Venceslau EB, Pessoa Filho PA, Le Roux GAC: Application of
008-0415-z. computer aided mixture design in paints and coatings. Comput
Aided Chem Eng 2012, 30:587-591 http://dx.doi.org/10.1016/
31. Kontogeorgis GM, Folas GK: Thermodynamic Models for Industrial B978-0-444-59519-5.50118-0.
Applications: From Classical and Advanced Mixing Rules to
Association Theories. Wiley; 2010 http://dx.doi.org/10.1002/ 51. Kalakul S, Cignitti S, Zhang L, Gani R: VPPD-Lab: the chemical
9780470747537. product simulator. Comput Aided Chem Eng 2016 http://dx.doi.
org/10.1016/B978-0-444-63683-6.00003-4.
32. Kontogeorgis GM: The Hansen solubility parameters (HSP) in
thermodynamic models for polymer solutions. In Hansen 52. Jhamb S, Liang X, Dam-Johansen K, Kontogeorgis GM: A model-
Solubility Parameters: A Users Handbook, Second Edition. Edited
based solvent selection and design framework for organic
by Hansen CM. CRC Press; 2007 http://dx.doi.org/10.1201/ coating formulations. Prog Org Coat 2020:140 http://dx.doi.org/
9781420006834. 10.1016/j.porgcoat.2019.105471
Discusses a model-based framework for the design of solvents for
33. Gani R, Abildskov J, Kontogeorgis GM: In Computer Aided organic coating formulations.
Chemical Engineering. Edited by Kontogeorgis G. Elsevier; 2004
http://dx.doi.org/10.1016/S1570-7946(04)80016-6. 53. Gani R, Nielsen B, Fredenslund A: A group contribution
approach to computer-aided molecular design. AIChE J 1991,
34. Kontogeorgis GM, von Solms N: Thermodynamics of Polymer 37:1318-1332 http://dx.doi.org/10.1002/aic.690370905.
Solutions. In Handbook of Colloid and Surface Chemistry, 4th
ed.. Edited by Birdi K. CRC Press; 2014. 54. Gesthuizen J: Formulators can move away from trial-and-error. .
(accessed Oct 21, 2019) 2019.
35. Lindvig T, Michelsen ML, Kontogeorgis GM: Liquid-liquid
equilibria for binary and ternary polymer solutions with PC- 55. Van Loon S, Fricker B: Using HSP to Improve the Dispersibility of
SAFT. Ind Eng Chem Res 2004, 43:1125-1132 http://dx.doi.org/
Pigments and Fillers TAGS: Dispersing Agents. 2018
10.1021/ie030443m. Discusses an effective method to turn the complex task of selecting
dispersing agents into a predictive, data driven test method, by applying
36. Tihic A, Kontogeorgis GM, Von Solms N, Michelsen ML: A Hansen Solubility Parameters (HSP) to paint, coating or ink formulation.
predictive group-contribution simplified PC-SAFT equation of
state: application to polymer systems. Ind Eng Chem Res 2008, 56. Gani R, Hytoft G, Jaksland C, Jensen AK: An integrated
47:5092-5101 http://dx.doi.org/10.1021/ie0710768. computer aided system for integrated design of chemical
processes. Comput Chem Eng 1997, 21:1135-1146 http://dx.doi.
37. Tihic A, von Solms N, Michelsen ML, Kontogeorgis GM,
org/10.1016/S0098-1354(96)00324-9.
Constantinou L: Application of sPC-SAFT and group
contribution sPC-SAFT to polymer systems-capabilities and 57. Hansen C, Abbott S, Hiroshi Y: HSPiP | Hansen Solubility
limitations. Fluid Phase Equilib 2009, 281:70-77 http://dx.doi.org/ Parameters. 2019. (accessed Oct 22, 2019).
10.1016/j.fluid.2009.04.002.
58. Abbott S: Practical formulation science for particle-based
38. Kontogeorgis GM, Saraiva A, Fredenslund A, Tassios DP:

inks. Colloids Interfaces 2019, 3:1-16
Prediction of liquid-liquid equilibrium for binary polymer Brings together the most relevant colloid and interface science with the
solutions with simple activity coefficient models. Ind Eng Chem
capabilities of modern apps to bridge the gap between theory and reality
Res 1995, 34:1823-1834 http://dx.doi.org/10.1021/ie00044a033. and benefit both the colloid science community and those who need the
39. Pappa GD, Kontogeorgis GM, Tassios DP: Prediction of ternary science to improve their formulations.
liquid-liquid equilibria in polymer-solvent-solvent systems. Ind
Eng Chem Res 1997, 36:5461-5466 http://dx.doi.org/10.1021/ 59. Scott A: One firm’s vision of the future. C&EN Global Enterp
ie970191n. 2019, 97:24-25 http://dx.doi.org/10.1021/cen-09728-feature3.

40. DTU Chemical Engineering: SPECS - Center for Energy Resources 60. Thiel R: Digital assistant shines with paints and coatings expertise.
Engineering. . (accessed Oct 22, 2019) 2013. 2019.

41. Wu DT: Modeling and simulation activities in a large research 61. Park RH, Stearns EI: Spectrophotometric formulation. J Opt Soc
and development laboratory for coatings. ACS Symp Ser 1986. Am 1944, 34:112-113 http://dx.doi.org/10.1364/josa.34.000112.

www.sciencedirect.com Current Opinion in Chemical Engineering 2020, 27:107–120

120 Frontiers ofchemical engineering

62. Rich DC: Computer-aided design and manufacturing of the 70. Dias JL, Silva A, Chai C, Gaspar PL, De Brito J: Neural networks
color of decorative and protective coatings. J Coat Technol applied to service life prediction of exterior painted surfaces.
1995, 67:53-60. Build Res Inf 2014, 42:371-380 http://dx.doi.org/10.1080/
09613218.2013.819551.
63. Largo Innova AB: Products - Software - Largo. . (accessed Oct 21,
2019) 2019. 71. Chai C, De Brito J, Gaspar PL, Silva A: Statistical modelling of
the service life prediction of painted surfaces. Int J Strateg Prop
64. Meyer G, Shimizu C, Eggly A, Fischer D, King J, Rodriguez A: Manag 2015, 19:173-185 http://dx.doi.org/10.3846/
Computer aided design of automotive finishes. 10th Congr Int 1648715X.2015.1031853.
Colour Assoc 2005:685-688.
72. Joly AM: Statistical treatment of data from the European
65. Kim DB, Lee KH: Computer-aided appearance design based on community research on self-stratification. Progr Org Coat
BRDF measurements. CAD Comput Aided Des 2011, 43:1181- 1996, 28:209-222 http://dx.doi.org/10.1016/0300-9440(95)00598-
1193 http://dx.doi.org/10.1016/j.cad.2011.04.015. 6.
66. Carew JA: Modeling and prediction of service life of coating
systems for industrial environments. Jocca-Surf Coat Int 1994, 73. Beaugendre A, Degoutin S, Bellayer S, Pierlot C, Duquesne S,
77:515.

Casetta M, Jimenez M: Self-stratifying coatings: a review. Progr
Org Coat 2017, 110:210-241 http://dx.doi.org/10.1016/j.
67. Sakhnenko ND: Imitation model for predicting the lifetime of porgcoat.2017.03.011
paint coatings. Prot Met 1997, 33:393-397. Gives a historical background and comprehensively reviews models to
predict polymer compatability and the systems developed for self-stra-
68. Zhou W, Zhao YG, Li W, Qin QD, Hu SW, Tian B: Degradation tifying coatings.
formula and working lifetime prediction for high-temperature
coating. Appl Surf Sci 2006, 253:2565-2571 http://dx.doi.org/ 74. Tusar L, Tusar M, Leskovsek N: A comparative study of
10.1016/j.apsusc.2006.05.027. polynomial and neural-network modeling for the optimization
of clear-coat formulations. Surf Coat Int 1995, 78:427.
69. Busso EP, Evans HE, Wright L, McCartney LN, Nunn J, Osgerby S:
A software tool for lifetime prediction of thermal barrier 75. Lin L, Guthrie JT, He WD: Computer-aided modelling of anti-
coating systems. Mater Corros 2008, 59:556-565 http://dx.doi. corrosion coatings formulations. J Mater Sci 2003, 38:1097-
org/10.1002/maco.200804138. 1104 http://dx.doi.org/10.1023/A:1022366317952.

Current Opinion in Chemical Engineering 2020, 27:107–120 www.sciencedirect.com