Industrial Requirements for Thermodynamics and Transport Properties

This work reports the results of an investigation on industrial requirements for thermodynamic and
transport properties carried out by the Working Party on Thermodynamic and Transport properties
(http://www. wp-ttp.dk/) of the European Federation of Chemical Engineering, EFCE
(http://www.efce.info/). A carefully designed questionnaire was sent to a number of key technical
people in companies in the oil and gas, chemicals, and pharmaceutical/biotechnology sectors.
Twenty-eight companies have provided answers which formed the basis for the analysis presented
here. A number of previous reviews, specifically addressed to or written by industrial colleagues,
are discussed initially. This provides the context of the survey and material with which the results of
the survey can be compared. The results of the survey have been divided into the themes: data,
models, systems, properties, education, and collaboration. The main results are as follows. There is
(still) an acute need for accurate, reliable, and thermodynamically consistent experimental data.
Quality is more important than quantity. Similarly, there is a great need for reliable predictive, rather
than correlative, models covering a wide range of compositions, temperatures, and pressures and
capable of predicting primary (phase equilibrium) and secondary (enthalpy, heat capacity, etc.)
properties. It is clear that the ideal of a single model covering all requirements is not achievable, but
there is a consensus that this ideal should still provide the direction for future development. The
use of new methods, such as SAFT, is increasing, but they are not yet in position to replace
traditional methods such as cubic equations of state (especially in oil and gas industry) and the
UNIFAC group contribution approach. A common problem with novel methods is lack of
standardization, reference data, and correct and transparent implementations, especially in
commercially available simulation programs. The survey indicates a great variety of systems where
further work is required. For instance, for electrolyte systems better models are needed, capable of
describing all types of phase behavior and mixtures with other types of components. There is also a
lack of data and methods for larger complex molecules. Compared with the previous reviews,
complex mixtures containing carbon dioxide associated with a wide range of applications, such as
capture, transport, and storage are becoming interesting to a number of survey participants. Despite
the academic success of molecular simulation techniques, the survey does not indicate great
interest in it or its future development. Algorithms appear to be a neglected area, but improvements
are still needed especially for multiphase reactive systems (simultaneous chemical and physical
equilibrium). Education in thermodynamics is perceived as key, for the future application of
thermodynamics in the industry. A number of suggestions for improvement were made at all three
levels (undergraduate, postgraduate, and professional development) indicating that the education
is correctly perceived as an ongoing process.

1. Introduction

Accurate knowledge of thermophysical properties of fluids plays an important role in cost-effective

design and operation of chemical and biochemical plants. Broadly speaking, the thermodynamic
properties determine the feasibility of a given process, while the transport properties have a major
impact on sizing of the equipment. The requirement for accuracy and reliability varies depending on
the application. For instance, for the design of separation processes the issue is acute, as often more
than 40% of the cost of the process is related to the separation units.1 As well as having impact in
the design and operation of processes,1 the importance of thermodynamics in product design,2-
5 green and environ- mental engineering,6-8 applied nanotechnology/material science,9,10
and biotechnology11-14 was recently elucidated. Due to the diversity of products and applications,
the need for accurate and reliable thermodynamic and transport property data, over a wide range
of mixtures and conditions, is apparent.

These observations led in 2007 to the creation of a new Working Party (WP) within the European
Federation of Chemical Engineers (EFCE), with the following objectives:

1. To promote thermodynamic, transport, and physical property measurements, correlations,

predictions, and simulations in the academic and industrial chemical engineering communities, the
national societies, as well as among the national and international funding agencies;

2. To promote concerted education activities to Chemical Engineers in the areas of

thermodynamics and transport properties;

3. To offer a suitable forum to Chemical Engineers for exchange of ideas, the development of
new basic and applied research projects, and collaboration through the organization of conferences,
workshops, and related activities;

4. To promote and coordinate research activities at the European level by identifying chemical
systems and conditions that are important for novel chemical processes and for which data and/or
correlations are nonexistent or inaccurate;

5. To publish quarterly an electronic newsletter summarizing recent developments in the field.

So far, the new Working Party (WP), entitled “Working Party on Thermodynamics and Transport
Properties” has published a review on the role of thermodynamics in the chemical engineering
education in Europe and the USA.15 The objective of the current paper is to report on a second
activity that was initiated by the WP, that is, to investigate the industrial needs for thermodynamic
and transport properties. For this purpose, a WP task group on Industrial Perspectives prepared a
questionnaire that was sent to a number of key technical people in major oil and gas, chemical, and
pharmaceutical/biotechnology companies. The responses received to this questionnaire are
analyzed and discussed in Section 3.

To put the analysis in its historical perspective, we briefly summarize a number of previous reviews
that have been published in the last 20 years. Reviews emanating from both academic and industrial
sources are presented and analyzed.

2. Literature Studies on Industrial Needs

Some of the early review articles from industry including those of Tsonopoulos and Heidman16
(Exxon) and Zeck17,18 (BASF) are almost 20 years old. They focus on the needs of oil and gas and
chemical industries, which have been tradition- ally the first industries to make wide use of
thermodynamic data and models. Tsonopoulos and Heidman16 emphasized the then very dominant
role of cubic equations of state which they consider to be largely adequate for the needs of the oil
and gas industry, due to their applicability to high pressures and hydrocarbon-gas mixtures. The
authors did not discourage the use of higher-order equations of state (EoS), which were appearing
at the time (such as PHCT), but they were of the opinion that the petroleum industry would still use
cubic EoS for years to come. Twenty-five years later, the cubic EoS remain a ubiquitous working tool
in the oil and gas industry. However, association models such as SAFT and CPA are gaining ground,
as the evidence is growing that they are successful for many petroleum engineering applications
featuring water-hydrocarbon mixtures, especially in the presence of (antihydrate) cosolvents such
as alcohols, as shown in the review of De Hemptinne et al.19 (IFP).

In the early 90s, when interest in thermodynamics was regenerated by the new industrial sectors
(materials, nanotechnology, and biotechnology), Zeck17 wrote: “There is still considerable potential
for improvement for phase equilibrium thermodynamics eVen in long-established areas of the
chemical industry. With all the enthusiasm about the possibilities of thermodynamics in new areas,
it is necessary first to concentrate on performing the basic tasks”. The deficiencies of the
thermodynamic models, as Zeck saw them, were often related to insufficient precision in the
description of vapor-liquid equilibria (VLE) and liquid-liquid equilibria (LLE), as well as
multicomponent LLE, using a single model with a single parameter set. Many of the basic tasks noted
by Zeck are related to traditional, but still difficult, separation processes like complex (azeotropic,
extractive) distillations, adsorption, or extraction. About the same time, Mathias and Klotz20 (Air
Products) also considered that main-stream traditional issues such as multi- component liquid-liquid
equilibria are far from solved. They stated: “Relative merits of these (classical, etc.) mixing rules are
uncertain at present because research is in progress, but it is clear that the methods should be
judged by whether they can provide good description of multi-component LLE e.g. forwater-alcohol-
hydrocarbons”.

Almost ten years after Zeck’s review, Dohrn and Pfohl21 (Bayer) emphasized that the general needs
for thermodynamic data may differ substantially even within the same industrial sector, in their
case, the chemical industry. The answer to their question about the availability of thermophysical
data elicited responses ranging from “we don’t have enough data” to “we have enough data”
including even “we have too many data”. Each of these statements was “justified” based on the
avail- ability, or not, of relevant databases, of suitable models in process simulators or on the level
of difficulty of a particular separation process. Data especially for multicomponent mixtures can be
scarce and costly even for well-defined mixtures of industrial importance such as water-
hydrocarbons-alcohols or glycols. Dohrn and Pfohl21 illustrated moreover, via examples, how
similar models may yield different design when used in process simulators even for rather simple
mixtures. For this purpose, they modeled a mixture of ethyl benzene-styrene with the Soave-
Redlich-Kwong (SRK) EoS. More examples were given recently by Dohrn et al.22 showing that results
may be sensitive to the specific parametrization used in models. So the evaluation of thermophysical
property data should play an increasingly important role in industrial thermodynamics.

Agarwal et al.23 (Virtual Materials Inc.) in their review

focused their attention on what they called “less attractive” (or often neglected) properties such as
heats of mixing, entropies, excess volumes, and heat capacities. Such properties could be used for
additional validation and model development, and the authors investigated how cubic EoS (using
either classical or advanced EoS/GE mixing rules) perform for these properties. They conclude:

“Traditional activity coefficient models such as NRTL have limitations on the magnitude of the excess
enthalpies they can predict. How these models behave inside cubic EOS for the calculation of
derived properties is not well known and is disappointing sometimes. So, if you are modeling a
system with significant excess enthalpies using a Gibbs free-energy-based EOS and you have a good
VLE fit, this does not automatically ensure you have a good overall model from an energy balance
point of View.”

About the same time, articles on industrial needs specifically addressing polymer properties (from
ASPEN24 and Bayer25) appeared, illustrating that a large variety of models may be required for
designing polymer processes.

Simulator developers and providers have also frequently published their opinions on
thermodynamic requirements. Carl- son26 (ASPEN) discusses the importance of using the so-called
decision or selection trees for selecting appropriate models suitable for specific applications. He
further discusses various techniques, such as family plots, for estimating missing parameters. In
general, commercial simulators offer a lot of versatility to the end user, but this comes at a price of
sometimes bewildering choice of models, model varieties, mixing rules, and parameter selection.

Chen and Mathias,27 also from ASPEN, stated that in many industrial systems polymers,
electrolytes, sometimes surfactants, and mixed solvents coexist, and this gives rise to multiple
simultaneous interactions. One of the major messages of their investigation is that many existing
models have been developed for each subsystem independently (for example, only for polymeric or
electrolyte mixtures), and such independent developments should be abandoned if future models
are to have wide applicability. They further emphasize that there is a great demand for
thermodynamic models for complex products, such as consumer goods and bioprocesses. Another
important issue that is discussed is the industrial response to the developments in the field. Chen
and Mathias27 argue that industry rarely updates or replaces its thermodynamic models with newer
and better correlations unless a clear advantage is evident. It often takes a long time, of the order
of 10 years or more, for a new model to be conceived, developed, validated, and accepted by the
industry.

One of the most recent and extensive industrial articles on thermodynamic requirements is that by
Gupta and Olson28 (Dow Chemical). In particular, they state:

“Most past work focused on extensions of cubic VdW- type EoS ... This has occurred despite the now
famous recommendation of Henderson. No matter how sophisticated a mixing rule, the use of Van
der Waals-type cubic equations of state forces their inherent limitations on the users. These are the
ability to reasonably predict only the Vapour pressure of a select series of components and only an
approximate modelling of the effect of liquid density and compressibility. Van der Waals-type cubic
equations are unable to accurately model other liquid phase properties, for example, enthalpy and
heat capacity also phase equilibria at high pressures, particularly the mixture critical locus.”

Gupta and Olson illustrate, via examples (simultaneous VLE and LLE of alcohol-alkanes), that SAFT-
type models should be preferred over cubic EoS for phase equilibria of polar mixtures.

A report for the Eureka project written by Moorwood29 (InfoChem) discussed status and
requirements for thermodynamic data for a very broad range of applications including metals and
agrochemicals. In addition, Bruin30 (Unilever) discussed some thermodynamic needs specifically
related to food industry.
In the most recent joint industrial-academic work, O’Connell et al.31 analyzed in detail a number of
specific cases from a thermodynamic perspective. Special emphasis was given to the need for
simultaneous physical and chemical equilibria data, for solvent selection for pharmaceuticals, and
for formaldehyde-water and CO2-water-alkanolamine mixtures.

During the past decade, several articles have appeared from industrial sources (Astra Zeneza,
Mitsubishi, ASPEN, Merck, etc.) on thermodynamic requirements of the pharmaceutical sector,32-
37 particularly the need for accurate and reliable models for solvent screening. A variety of models
have been tested, ranging from modifications of UNIFAC and solubility parameters to modern
versions of local composition models, such as NRTL-SAC,38 and methods based on quantum
chemistry like COSMO-RS39,40 and COSMO-SAC.38 The use of quantum chemistry methods in
pharmaceutical industry has been also highlighted in industrial lectures (for example, by Natori,
2001, at the PPEPPD conference in Kurashiki, Japan).

A systematic investigation on industrial needs for both thermodynamic and transport properties has
been conducted under the so-called Forum 2000 project, and the results are documented in a
detailed report41 and several publications.42,43 One important comment from Forum 2000, by Jim
Olson (Dow Chemical), focuses on the need for reliable experimental data for families of compounds
which have not been systematically studied:

“We can never conduct all the necessary experiments - but many data measurements are missing.
For instance, multi-functional chemicals such as alkanolamines, glycol ethers, excess properties like
excess Volumes at Various temperatures, amides, liquid phase heat capacities, environmental phase
equilibrium, safety data e.g. flash- point and auto ignition temperatures, simultaneous physical and
chemical equilibrium, ...”

The need for reliable experimental data is the message that is also strongly supported by other
industrial-oriented reviews, e.g., by Rhodes44 (Union Carbide) and Richon.45

Other important comments from the Forum 2000 work are related to the need for accurate data
and models for transport properties. For example, Bill Wakeham stated: “Thermal diffusion,
although an abstruse transport property, will turn out to be extremely important in oil reservoirs,
due to the large T-differences”. He later added:

“What’s the data in tissue engineering that we need? If you look at the way cancer tumours are
sometimes removed, which is ultrasonic or microwaves, this is all done by heating. The thermal
conductivity of the tissue you heat, the heat capacity of that tissue, matters and they don’t know it.
Sometimes it is measured in Vitro never in Vivo.”

P. Mathias stated: “Transport properties are becoming increasingly important, but little attention
has been devoted to them”. He further made a point on solids: “When one actually does modelling,
with particle sizes, the issues of nucleation and growth are important. An area I think chemical
engineers did many things 30 or 40 years ago, but there’s not been a lot recently”.

So far, we have summarized some comments and opinions based on the reviews of articles from
industrial sources. Several reviews emanating from academia have also been published. In
particular, the review by Prausnitz and Tavares46 and more recently a book by Kontogeorgis and
Folas38 review a large number of thermodynamic models and present an outlook for future
developments. According to most of the academic reviews, important advancements in
thermodynamics are needed to address issues related to “post-modernism”,47,48 life sciences/bio-
technology,11-14,49,50 and chemical product development.2-5,9,10 Moreover, there is a need to
connect closely with other disciplines of chemical engineering, such as mass and heat transfer,
nucleation, and chemical kinetics. One example where such an interdisciplinary approach would pay
great dividends would be in the design of controlled drug release systems.

The postmodernism trends in chemical engineering encompass the need for more “human aspects”
in the curriculum. They emphasize the role that chemical engineering plays in society and provide
ample opportunity for thermodynamics. The near- future focus should be on addressing different
environmental and safety related aspects, for example, the use of supercritical fluids, the
optimization of processes related to CO2 capture and storage, as well as modeling of loss-prevention
scenarios.6-8,47,48

Many recent investigations39,40,51-53 point to the increasing importance of quantum chemistry

especially in the form of models like COSMO-RS and COSMO-SAC for thermodynamic calculations
and especially screening purposes for a wide range of situations, mostly at low pressures. The role
that the molecular description increasingly plays in modern thermodynamics has been emphasized
by Prausnitz:9

“Thermodynamics is often most useful when directed at developing model for new situations,
where our quantitative understanding is seriously inadequate. Pioneering molecular
thermodynamics is concerned not with improvements where primary understanding has already
been achieved, but with shedding light on situations where as yet, we know little. For many practical
problems, old theories are sufficient provided that they are used with courage and imagination.”

3. Questionnaire and the Survey

The task group Industrial Perspectives appointed by the EFCE Working Party on Thermodynamics
and Transport Properties (EFCE WP) prepared a questionnaire comprised of seven questions that
are presented in Appendix 1. The questions ranged from simple yes/no answers to the ones that
were open- ended where any issue related to thermodynamic and transport property requirements
could be addressed.

The overall purpose of the survey was to:

• Review the situation in industry with respect to thermodynamics and transport properties
requirements. The emphasis was primarily to collect information on European industry, although
some companies from the USA, Japan, Thailand, Brazil, and South Africa were also included in the
survey;

• Identify limitations in current approaches;

• Provide specific recommendations on R&D work (to be initiated by the EFCE WP and
individual researchers in this area).

A list of possible companies that could take part in our survey was created with the contact names
collected by the coauthors of this paper, and the invitations were sent out. Fifty-nine companies in
total from sixteen different countries were contacted, and the reminders were sent out to increase
the number of responses. In total, 28 replies were received, which represents a 47% response rate.
Most responses were from European but globally operating industries. Figure 1 illustrates the
industrial sectors of the companies that have responded to the survey.

4. Results

Although it is rather difficult to categorize and summarize an overwhelming number of answers

received, an attempt is made to summarize the essential features by means of Figure 2.

Figure 2 shows on the left the industrial requirements regarding thermodynamics and transport
properties, summarized by a few keywords. The requirements are usually short-term, and this is
probably why there is a resistance in trying out recently developed advanced tools, such as SAFT
EoS and molecular simulation (“first concentrating on basic tasks”). Yet, the industrial requirements
often relate to complex thermo- physical systems, hence the reliance in some instances on simple
“rule of thumb” approaches, which are believed by industry to give a good indication of trends.
Finally, industrial needs require either (i) the knowledge of thermophysical properties at given
operating conditions or (ii) in product design the ability to design the right product with specific
thermophysical properties (i.e., solvent design).

Figure 1. Responses to the questionnaire classified by industrial sector. Some responders work in
more than one category.

The “supply” from the thermodynamic community is presented in Figure 2 as a triangle that should
respond to these needs: the response is the “package” that includes models, parameters, and
algorithms. Ideally, this model should be unique, independent of the type of system or operating
conditions. To construct this ideal package, the three vertices of the triangle must be considered:

Upper left: what is the molecular system under investigation? There is a plethora of mixtures of
industrial interest, and it is more or less impossible to include every conceivable combination. A
reasonable approach would be to rely on good description of thermophysical properties of pure
species and use modeling tools to obtain the properties of the mixtures. The industrial interest
changes with time, as a result of new developments, new processes, and new products. Currently,
there is an emphasis on CO2, electrolytes, and polyfunctional molecules. Many molecular systems
are compositionally well- defined, but this is not always the case. For instance, for reservoir fluids
the detailed molecular composition is usually not available, hence some additional characterization
is needed. Upper right: what is the property required? Although VLE remains the most common,
other types of phase equilibria (LLE, SLE) are nowadays also in demand. Furthermore, there is an
increasing need for enthalpic properties (heat of mixing, heat capacities, etc.) and transport
properties. The traditional cubic EoS is of limited use for this purpose,28 and one requires more
sophisticated models that are not necessarily available in every package.

Lower corner: data. The most traditional approach is to claim that there are not enough data or to
use the available data without any analysis of their accuracy or reliability. Although it is true that for
a large number of systems the data are scarce, when the data are available one should ask the
question according to Dohrn and Pfohl:21 “do we have the correct data (that means useful in view
of the applications)?”

At the very right of Figure 2, the various state-of-the-art predictive tools that the thermophysical
property community can use, to respond to the industrial needs, are illustrated. Molecular
simulation plays an essential role, as well as quantum chemistry, as both have shown great advances
in terms of both development and applications. In the near future, other computational techniques
might become available including, for example, the ones dealing with nonequilibrium
thermodynamics or calculation of transport properties directly from intermolecular potential.
Industry is not yet fully aware of the applicability of these computational techniques to address their
requirements, which is why very few industrial comments on this issue were found in our survey. It
is thus the role of the academic community to introduce the use of advanced computational
techniques, by both further refining theoretical models and showing that the results can be used as
alternative, pseudoexperimental data, or parameter provider.

In the next set of subsections, we present and analyze the results obtained in the current survey.
For ease of use, the main conclusions from the survey are presented in the bold font in the first
paragraph of each section.

Figure 2. Schematic illustration of the position of thermodynamic tools for the oil and gas, chemical,
and process industries.

4.1. Data. There is an acute need for high-quality experimental data for both thermodynamic
and transport proper- ties, including the phase and chemical equilibrium data.
The high-quality data can only be obtained in a well- characterized experimental apparatus with a
well-defined un- certainty level, which cannot be demonstrated to be inconsistent with other data
or with theory. Although such instruments do exist, their use requires knowledge of their full-
working equation, and in some cases, it requires a number of corrections, to account for the
nonideality of the apparatus, to be made to experimental data. Thus, producing high-quality data is
expensive and time-consuming, and not surprisingly, such data are scarce. It is important to
emphasize that rather than focusing on the amount of data available for a particular system one
should focus on gathering reliable data with well-defined uncertainty levels that cover the phase
space of interest. Such sets of high-quality data, that are mutually consistent, can then be used as a
benchmark data set for a particular system essential for validating or developing predictive models
and tools.

More specific comments that were made in the survey were as follows:

(i) For complicated molecules, containing multiple functional groups (e.g., hyper-branched
polymers, dendrimers, micelles) that are used for fine chemicals, crop protection, and
bioapplications, very few phase equilibrium data exist in the open literature. So, group contribution
methods are less accurate for these molecules, as there are no data to fit the required parameters.

(ii) For many traditional/bulk chemicals, a large number of data sets exist that do not agree
with each other. Most of the data have been measured a long time ago in the apparatus whose
description does not allow for their proper levels of accuracy to be established. There is a lack of
high-quality reference data to which model parameters can be fitted for process simulation. Simple
examples mentioned in the survey include the inorganic acids, like sulfuric acid, hydrochloric acid,
and nitrous acid.

(iii) The participants of the survey felt that the emphasis should be given to more measurements
on: (i) gas solubilities;

(ii) high-pressure and high-temperature phase equilibrium data of nonideal mixtures; (iii) dilute
conditions; (iv) solubilities and diffusion in polymers; (v) reactive multiphase systems (CO2, H2S-
water-alkanolamines); and (vi) electrolytes. Essentially, there is a need for high-quality data for a
wide range of systems and conditions.

(iv) Data are also lacking in petroleum-related applications for: (i) gas hydrate systems, e.g.,
hydrate formation with CO2 and H2, with additives, and with loss of hydrate inhibitor such as
methanol and glycols; (ii) properties of heavy oils; (iii) partition coefficient of gases in
hydrotreatment conditions, especially in the presence of oxygenated fuels; (iv) heats of reaction and
of adsorption; and (v) VLLE of complex natural gas mixtures including water and electrolytes.

(v) There is an acute need for transport property data that include viscosity, thermal
conductivity, and diffusion coefficient data. Especially, emphasis was put on the lack of data for
electrolytes and ionic liquids. It was noted that even when some transport property data do exist,
they do not cover an extensive temperature and pressure range.

(vi) The lack of data is made even more acute by the observation that there are fewer research
groups (at least in Europe) that have well-equipped laboratories and have funding to devote time
to measurements. The disappearance over the years of a number of research groups that were
producing high- quality data is a worrying trend, as without the high-quality data the existing
databases will become less useful as the tendency to target quantity rather than quality will
intensify.

(vii) A number of participants indicated that when the data need is clearly identified in view of
specific industrial requirement, companies do not hesitate to spend money to measure the relevant
properties. However, as these data are often considered strategic, they are not always published.
Note that policy of expecting to find a laboratory, when a particular need arises, that can perform
the high-quality measurements is relatively short sighted. Producing high-quality data requires
continuity in terms of both maintaining apparatus and even more importantly maintaining research
expertise, as the high-quality measurements cannot be performed on a short-term basis by
inexperienced researchers.

(viii) The use of advanced computational tools, such as molecular simulation or quantum
chemistry, is sometimes mentioned as an alternative to “producing” experimental data, but this
approach is currently not commonly accepted by industry.

4.2. Models. 4.2.1. Scope of the Models. There is a need for reliable thermodynamic and
transport property models for a Wery wide range of systems.

The participants of the survey put the special emphasis on:

(i) Need for models for phase equilibria and transport properties, preferably within a single
framework, that cover all the components and conditions of interest and with good inter- and
extrapolation capabilities in temperature, pressure, and composition.

(ii) The framework should be targeted toward making use of generalized and simple equations
of state for nonpolar, polar, associating, and high boiling substances and their mixtures, with
sufficient accuracy for industrial purposes.

(iii) The number of available models in the literature should be reduced by recommending either
a unique model or a small number of models. Today, specialists in industry are required to
recommend the best modeling approach for a specific application, and the number of available
models or versions of the same model are sometimes bewildering.

More specifically, the survey results indicated:

(i) The models should cover the requirements of very diverse industrial fields: (a) oil and gas
including heavy oil, mercury compounds, syngas, asphaltenes (currently poorly characterized), and
organic sulfur compounds; (b) chemicals including complex oxygenated chemicals related to
biofuels, ionic liquids, surfactants, and polymers; (c) biotech including biopolymers,
pharmaceuticals, and building blocks; (d) colloids including emulsions, surfactants, food additives,
emulsifiers, adhesives, and gels; and (e) metal and mining. The applications were by and large
dependent on the industrial sector, but the following processes were particularly emphasized as in
need of good models: flow assurance (gas hydrates, asphaltenes, and wax); separation methods
other than distillation/extraction, e.g., adsorption and chromatography; and crystallization.

(ii) As can be seen from the above list, requirements and expectations for models are not only
diverse but also very extensive. Clearly current models cannot cover or meet all areas of interest
and expectations. However, an interesting issue is that there is a feeling that models are incomplete
or lacking even in areas which might have been considered “mature”. For instance, for bulk
chemicals it was felt that a very important system of ammonia-water-sulfuric acid is poorly
addressed by any of the available models. In the oil and gas area, the presence of mercury and sulfur
compounds was equally badly catered for by the modeling community. There is a feeling that an
analysis of the different requirements and expectations carried out for each industry sector would
be useful.

(iii) A lack of high-precision EoS for a number of systems was noted.

4.2.2. Theoretical Foundation of the Models. Traditional and novel models are equally important,
and developments in both should continue.

Model approaches used in industry range from traditional group-contribution (GC) models
especially UNIFAC to modern approaches such as COSMO-RS and SAFT. The general consensus is
that:

(i) The development of both UNIFAC and COSMO-RS should continue as they are both very
useful for industrial applications. There was a suggestion that some work should be carried out to
combine them, by for instance making use of group contributions for the sigma profiles of COSMO-
RS. The prediction of UNIFAC and COSMO-RS should be compared as the actual range of applicability
and accuracy appears to be very system-dependent. Molecular modeling should be further
developed to gain better understanding, and molecular simulations are suggested to improve the
prediction of thermophysical properties and phase equilibrium.

(ii) The GC models in general are used and are needed in many contexts that do not necessarily
involve predicting phase behavior, for instance, for computer-aided product design, properties of
heavy compounds (even for pure species: melting temperature, vapor pressures, etc.), solvent
screening, emulsions, and oxygenated compounds. Thus, the full phasing out of GC models in favor
of modern models is not yet imminent.

(iii) SAFT-based models are promising tools, but the requests from the industrial community for
these models are multiple:

• Characterization methods for oils, asphaltenes, and similar complex mixtures;

• Speed must be improved for routine applications (see also next point on standardization);

• Recommended pure-component and binary parameters should be collected and made

available;

• True improvements over classical cubic EoS should be demonstrated, especially for
multifunctional, complex, and associating compounds like oxygenated compounds (biofuels);

• The model should be applied to truly important new areas including polymers and
electrolytes;

• Standardization is needed. This is further discussed in section 4.2.3.

4.2.3. Availability of the Models. Standards are very important for thermodynamic models.
The term “Standards” has been used and understood in different (sometimes complementary)
contexts, both in implementation and model validation:

(i) Models should be present in simulators for active use by industry, otherwise, they may be
forgotten. CAPE-OPEN is a useful (and standard) tool for integrating user developed unit operations
and thermodynamics packages in commercial process simulation programs, but it is still not fully
developed and supported. One of the comments in the survey was that the analytical derivatives
are needed. The tool should be improved further, allowing exchange of thermodynamic packages
belonging to different process simulators.

(ii) The number of new models is enormous, even within the same family. For instance, there
are numerous versions of SAFT. Sometimes different results are obtained with what should be the
same model. Proper coding and validation are extremely important to ensure full reproducibility of
the results. This should also be performed by other researchers rather than just by those involved
in the model development.

The question arises how can this be accomplished? It is worth investigating some process that would
lead to a proper comparison of models. A first step in such a process could be the agreement to use
some standard well-accepted databases for validation purposes. A second step would be the
agreement on criteria for model validation which could be used by all developers. In particular, the
model developers should try to present, in a few comprehensive articles, all equations of their
models including a thorough evaluation. The codes can be made publicly available, but a good
accompanying documentation would be necessary.

4.2.4. Algorithms. Implementation of the thermodynamic models requires reliable flash and
related algorithms, and although significant developments have been made in the past (e.g.,
Michelsen and Mollerup54), new challenges are appearing, in particular, the need for very fast and
robust calculations for flashes, the presence of multiple phases, and reactive systems. Further, there
is a need to improve the stability test, the most time-consuming element of the multiphase
calculations. Un- fortunately, the field does not seem to be of active interest to the thermodynamic
community.

4.3. Systems. The most popular systems appear to be electrolytes, CO2, and larger,
polyfunctional molecules.

These systems appeared repeatedly in a survey in different contexts from a number of companies
belonging to different sectors.

In particular:

(i) There is a need for improved description for electrolyte systems that go beyond e-NRTL.38
The emphasis should be on describing all types of phase behavior (VLE/LLE/SLE) covering also
electrolyte mixtures with water, other organic chemicals (i.e., mixed solvents), and hydrocarbons
(natural gas), over extensive temperature and pressure ranges, up to high salt concentrations.
Electrolyte models are multiparameter models, and modern e-EoS38 may be a way to go; however,
much remains to be done. There is a need to bring electrolyte thermodynamic models to a level
comparable to other fields. The need for improved electrolyte models is common among a wide
range of industries.
(ii) There is a lack of data and models for larger molecules, complex molecules, biomolecules
(biopolymers, proteins, pharmaceuticals, gels), or molecules that contain multiple functional groups
(polar and associating ones) that are used as food additives.

(iii) CO2 mixtures with water, polar chemicals (methanol, glycols) including mixed solvents, H2S,
N2O, O2, hydrocarbons, and (alkanol) amines appear of importance to a wide range of applications
(gas hydrates, oil and gas, CO2 capture, transport, and storage). Of special importance to a number
of participants was that the description of the system covers very extensive temperature and
pressure ranges. The CO2- containing systems also include electrolytes, as CO2-brine (+ rock
mineral) systems are of great interest in CCS (CO2 capture and storage) applications. A number of
the systems described are chemically reactive which further complicates their thermophysical
description.

4.4. Properties. Multiple thermodynamic properties are of importance.

It was frequently stated in the survey that simultaneous description of different thermodynamic
properties and phase equilibrium types is of great importance. Internal thermodynamic consistency
of the generated data by a given model is nowadays as essential as accuracy and reliability. This is
especially important as the thermodynamic models are often used as part of the simulator by a
different set of software developers who are not necessarily aware of all the nuances of a particular
thermodynamic model.

Hence, thermodynamic modelers should no longer be satisfied with models that for instance give
only good VLE results at low pressures but are either untested or fail for other properties and regions
of the phase space.

In particular, the results of the survey indicated that for equilibrium properties:

(i) Simultaneous description of VLE and LLE (+ VLLE) is of interest with the same model and the
same parameter set, especially for complex multicomponent mixtures.

(ii) So is the simultaneous description of the phase behavior and derivative properties (e.g.,
heats of mixing).

(iii) Special attention should be given to dilute systems (infinite dilution), and the description of
the system should be mutually consistent with the description at higher concentrations.

(iv) Extrapolations to high temperatures and pressures are often needed, for instance, for gas
solubilities in water- hydrocarbon-CO2 mixtures that could involve salts and other chemicals. The
proposed correlations should have sensible extrapolation behavior.

A number of other properties were mentioned as important for specific applications:

(i) SLE is crucial for a variety of separation processes, e.g., crystallization, as well as for mineral
stability and hydrate formation in the oilfield.

(ii) Surface phenomena (adsorption) are typically equilibrium phenomena that play an
important role in a number of applications.

(iii) Chemical reactions (formation properties).

None of these properties are currently well catered for by the thermodynamic community.

Nonequilibrium properties: combined thermodynamic/ transport effects and appreciation of

nonequilibrium effects are important.

(i) There is a need for combined models for predicting thermodynamic and transport
properties, reaction rates, and diffusivities. There is a need for a good procedure to disentangle
reaction kinetics and mass transfer where both processes contribute.

(ii) Is it possible to predict transport properties from thermodynamic data alone? Can we
reliably estimate thermal and transport properties from limited vapor pressure and density data?

(iii) There is a need for models for predicting properties, especially transport properties of
complex products present in many fields of process industry, such as viscosity index of lubricant
base oils from composition and octane/cetane numbers.

(iv) What is the effect of nonequilibrium behavior? It is not always clear whether a system is in
thermodynamic equilibrium or not.

(v) There are cases where thermodynamic predictions may be quite misleading because of slow
reaction kinetics, or poor mass transfer, e.g., predicting inorganic scale deposition.

(vi) In many processes, the nonequilibrium part can be of importance, e.g., nucleation in
crystallization and gas hydrate formation.

(vii) It would be of great industrial interest to develop fundamental nonequilibrium

thermodynamic approaches for predicting transfer across boundaries (for replacing the empirical
“transfer efficiency” equations).

4.5. Education. What is the role of education in thermodynamic modeling for process and
product design? There is a clear agreement among the participants in the survey that education is
the key for an appropriate use of thermodynamics in industry. Three different levels of education
are relevant in this context, each with different requirements:

(i) Undergraduate education: it is essential that engineers in operating companies know some
basic thermodynamics but even more importantly that they are aware of the limitations of their
understanding. This is especially relevant as black-box type simulators are routinely used these days.

(ii) Professional development: with the advent of new technologies (simulators) or as a result
of job changes, people may be confronted with new challenges after several years of professional
experience. The field of thermophysical properties is evolving rapidly, and there is a need for
industry professionals to keep up with the latest developments.

(iii) Postgraduate education: the complexity and the wide applicability of thermodynamics leads
to a need for specialists within a given company, typically educated to a PhD level, to act as an
internal expert on thermodynamic and transport property issues.

4.6. Collaborations. Data collected revealed that:

(i) The R&D organization in the various companies is generally rather strong: 25 out of 28
companies have a research lab, and 9 of them have a research subsidiary.
(ii) Most of the companies that took part in the survey have bilateral collaborations with
universities. The average level of collaboration fulfills close to 10% of the total research need. Some
companies rely quite heavily on it.

(iii) 25 out of 28 companies participate in consortia. Yet, the willingness to participate in joint
R&D programs is surprisingly conservative with only 14 out of 28 indicating a clear yes.

(iv) The area for joint collaboration is rather diverse: from gathering new data to developing
parameter tables (UNIFAC consortium is often cited as an example).

5. Conclusion: Discussion and Comparison to Previous Studies

A short discussion is provided in this section contrasting the results of our industrial survey and
those from the literature that were presented in section 2.

An important common feature of current and previous investigations (see for example a recent
discussion45) is the need for experimental work which will result in high-quality measurements that
are accurate, reliable, and thermodynamically consistent. In many cases, the real accuracy of
published experimental data is far lower than the accuracy given by the authors, as discussed by
Peper et al.56 The importance of practical expertise of the experimentalists in obtaining high-
quality data has been highlighted by Dohrn et al.55 in their recent review on experimental methods
for high-pressure phase equilibria. Often, the qualification of the staff and their familiarization with
the technique have a stronger influence on the accuracy of experimental results than the choice of
the experimental method or the equipment. Our survey shows that there is a clear need for qualified
laboratories with experienced staff to provide experimental measurements. As many companies
seek to reduce in-house facilities for routine, nonexploratory type of measurements, this need will
likely increase. Universities have also reduced their experimental activities. The key problem is that
performing high-quality measurements requires accumulation of many years of experience for
which funding has to be secured. The different funding bodies in many instances do not see this
type of activity as leading-edge research and in the present climate decline to fund it. Furthermore,
the pain-staking effort leads to a rather limited publication output. Nevertheless, data are lacking
for a wide range of systems from “simple” common mixtures up to multifunctional chemicals,
complex com- pounds originating from biomass, ionic liquids, polymers, and pharmaceuticals.
Surprisingly, the data are still very scarce for solubility of water in CO2 at high pressures and for
mixtures that involve inorganic acids or ammonia.

Another common feature of the current survey and the previous ones is that reliable models are
needed and that the predictive character is essential, but many existing models are essentially
correlative. Thus, SAFT, COSMO-RS, and UNIFAC (GC) approaches should be further developed, and
it appears unlikely that a single tool will be sufficient in all cases. Most models have serious problems
when applied to (poly)electrolyte systems, mixed solvents relevant for gas treatment using amines,
and new solvents like ionic liquids. The simultaneous presence of polymers, electrolytes, and
biomolecules is also an area where none of the models perform adequately. These are areas where
important revisions and improvements are needed. It is particularly interesting that deficiencies in
current electrolyte models are apparent even for models used in commercial simulators, as
illustrated in a recent joint academia-industry project.57 Emphasis on electrolytes and polymers is
found in both previous and the current investigation. New in our survey is the special interest in
predicting the thermophysical properties of mixtures containing CO2. This is a direct result of
climate change concerns and the development of sustainable technologies, in particular, the CCS.

Regarding the capabilities of thermodynamic models, we can state that the advent of SAFT-family
models has certainly contributed to better representation of a wide range of systems (nonpolar,
polar, associating, polymers, etc.) and the simultaneous representation of various phase equilibrium
types, e.g., VLE and LLE. On the other hand, there is a clear impression from the survey that certain
important issues, also emphasized in the earlier industrial studies, have been overlooked. Good
examples are the multicomponent multiphase equilibria especially for prediction purposes and the
accurate description of derivative properties (e.g., heat capacity, speed of sound, enthalpy of
mixing) together with phase behavior. Clearly, these issues need to be considered and included in
future development-model validations.

Some of the previously published industrial reviews emphasized molecular simulation techniques
as enabling technology which can supplement the experimental data or act as pseudoexperimental
data at temperatures and pressures where experiments are difficult. However, the participants of
the current survey were not particularly encouraging the further developments of molecular
simulation, which is a rather disappointing state of affairs.

Algorithms appear to be a neglected area in many of the previous publications, but the results of
the current survey illustrate that the improvements are clearly needed, especially for multiphase
and/or reacting systems (simultaneous physical and chemical equilibria).

According to the current survey, implementation is an important issue to the industrial users. In the
previously published reviews, this aspect was not emphasized greatly, although the importance of
CAPE-OPEN58 standard has been mentioned. The issue of having standardized and validated models
is particularly highlighted in the current survey. The new models are published very frequently, but
unfortunately most, if not all, of these models are tested against only a few data sets. A large,
standardized set of data against which all models may be tested will be of great benefit, as it will
allow models to be compared on equal footing. It will also be of benefit to those developing the
models, as the validation against agreed industry benchmarks could lead to the reduction of cycle
time from development to industrial use.

A final original issue raised by the current survey is the need for education and the possible role the
EFCE WP can play in this. A number of suggestions of improving the undergraduate level courses
were made. Thermodynamics of process simulation should be taught with the emphasis on how
generic simulators work, without necessarily going into peculiarities of a particular simulator, unless
a particular simulator is used in class. It is important to get across how to distinguish between
reliable and unreliable thermodynamic data and practices. However, the main issue raised is how
to make engineers more aware of the practical consequences of “right and wrong”
thermodynamics. In particular what type of problems require initiation of proactive research and
under what circumstances would quick empirical fixes suffice.

Overall, the gap between the industrial requirements and the data availability has not appreciably
shrunk over the years. It is not immediately clear how this gap can be closed especially in the present
economic climate. The first step is to once again try to bring the relevant stakeholders closer
together.

As an example of possible actions, the French working party Thermodynamique des proce´de´s
organizes small scale one-day meetings on specific topics such as education, molecular
thermodynamics, and use of databases. The objective is to create learning and exchange
opportunities between the academic and the industrial communities. Specific potential actions from
these meetings are the installation of a Web site for exchange between educators and a Web-based
information system for molecular thermodynamics issues.

Other examples include the organization of workshops such as the one recently organized between
our working party and the German ProcesNet working party. Such workshops should be tailored so
that they address some of the numerous questions raised above. At the same time, they can act as
a forward-looking forum where the industry can articulate their medium- and long-term needs and
open a dialogue with data and model generators.

The issue of having large, standardized set of data against which all models may be tested is also a
very pertinent one that needs addressing. Although, there are databases (NIST- TRC,
DechemaDetherm, Danner and Gess59) that are used by individuals, their use is still not widespread
enough.

Of equal importance is the definition and availability of standardized models. In our view, this
requires for each such model the publication of a paper, perhaps in a dedicated journal, containing
all that is needed to perform the calculations and reproduce results. This includes all the equations
and parameter values needed to achieve this and results for a number of defined input cases.
However, discussions on the merits of the model and comparison to experimental data should be
left out. The model can then receive a unique identifier which together with this definition paper
can serve as the reference. It would also be very useful to publish the implementation of this model
including the parameters as open source code, preferably together with the definition. The vendors
of commercial software should refer to these standard models in their documentation and clearly
state when and where they deviate from them. The same holds for authors discussing a comparison
of their results against those obtained with standard models.

We see this paper as the first step in reinitiating the dialogue between the interested parties which
can lead to a more effective collaboration that can close the current gap between the requirements
and the availability of thermodynamic and transport property data and models. We feel that the
EFCE WP should play an important role in this process, acting as a conduit between the academia,
industry, and decision makers.

Acknowledgment

We acknowledge the contribution from all industry experts who replied to the survey and thank
them for their effort, without which this paper could not have been written.

Appendix 1. The Survey Questionnaire

Below are listed the questions asked in the EFCE Working Party Task group on Industrial Perspectives
Questionnaire. All questions are to be considered in relation to physical properties,
thermodynamics, and transport properties.

Question 1. Your company’s business

What are the main businesses of your company? (Put Yes or No)

Exploration and production Refining

Bulk chemicals Fine chemicals Natural gas Power generation Pharmaceuticals Agrochemicals

Question 2. Limitations in thermodynamics and transport properties

What do you consider as the limitations in thermodynamics and transport properties in addressing
your company’s challenges?

Products Processes Data Models

Process simulation Reservoir simulation CO2 sequestration Enhanced oil recovery

Extraction and processing of discounted (opportunity) crudes (heavy oil, high sulfur crudes, high acid
crudes...)

Gas processing

Long-range transport

Add as many as you wish

Question 3. R&D in thermodynamics and transport properties

How is the R&D organized in your company? Do you have your own research laboratories? Do you
have a research subsidiary?

Do you subcontract your research?

What proportion of the work is subcontracted to academic teams?

Question 4. Collaboration

Consortia or other collaborative work in thermodynamics and transport properties.

Do you participate in consortia? If so, in which areas?

Is your company willing to sponsor research work that is to be shared with the rest of the
thermodynamic community?

If so, in which areas?

Question 5. Benefit

Can you provide examples of achievements in Thermodynamic R&D where you can give an estimate
of potential business or avoided risk?

Question 6. Areas

What are your ideas and opinions concerning the development in the following areas?

Suggestions:

Try to be concrete, so refer explicitly to elements such as components (carbon dioxide, sulfur),
models (SAFT), phenomena (gas hydrates), experimental quantities (SLE, speed of sound), standards
(CAPE OPEN), algorithms (multiphase flash, speed improvement), conditions (T 500 K),
applications (heavy oil, reinjection of hydrogen sulfide), procedures (characterization of well fluids),
and so on wherever this is possible.

Look at competing or complementary approaches, such as Group Contribution methods versus

CosmoLogic. Is there a need for both? Do we have a preference?

Look at the need for standardization versus flexibility. For example, do we want yet another
equation of state, or should we devote all effort to extending the traditional ones now present in
most process simulation programs? And if the latter holds true, which ones would we single out in
this stage?

Physical Data Phenomena

Conditions and environments Models

Algorithms Procedures Implementation Applications Process simulation Standards

Add your own item(s)

Question 7. Other thoughts

Do you have any other thoughts on thermodynamics and transport properties that could be relevant
to us?