arXiv subjects

Hans Hasse

Publications and source records attributed to Hans Hasse.

At least 37 records · Page 2Linked to original sources

arXiv2024

Advancing Thermodynamic Group-Contribution Methods by Machine Learning: UNIFAC 2.0

Accurate prediction of thermodynamic properties is pivotal in chemical engineering for optimizing process efficiency and sustainability. Physical group-contribution (GC) methods are widely employed for this purpose but suffer from historically grown, incomplete parameterizations, limiting their applicability and accuracy. In this work, we overcome these limitations by combining GC with matrix completion methods (MCM) from machine learning. We use the novel approach to predict a complete set of pair-interaction parameters for the most successful GC method: UNIFAC, the workhorse for predicting activity coefficients in liquid mixtures. The resulting new method, UNIFAC 2.0, is trained and validated on more than 224,000 experimental data points, showcasing significantly enhanced prediction accuracy (e.g., nearly halving the mean squared error) and increased scope by eliminating gaps in the original model's parameter table. Moreover, the generic nature of the approach facilitates updating the method with new data or tailoring it to specific applications.

physics.chem-ph↗

arXiv2024

HANNA: Hard-constraint Neural Network for Consistent Activity Coefficient Prediction

We present the first hard-constraint neural network for predicting activity coefficients (HANNA), a thermodynamic mixture property that is the basis for many applications in science and engineering. Unlike traditional neural networks, which ignore physical laws and result in inconsistent predictions, our model is designed to strictly adhere to all thermodynamic consistency criteria. By leveraging deep-set neural networks, HANNA maintains symmetry under the permutation of the components. Furthermore, by hard-coding physical constraints in the network architecture, we ensure consistency with the Gibbs-Duhem equation and in modeling the pure components. The model was trained and evaluated on 317,421 data points for activity coefficients in binary mixtures from the Dortmund Data Bank, achieving significantly higher prediction accuracies than the current state-of-the-art model UNIFAC. Moreover, HANNA only requires the SMILES of the components as input, making it applicable to any binary mixture of interest. HANNA is fully open-source and available for free use.

cs.LG↗

arXiv2024

KnowTD-An Actionable Knowledge Representation System for Thermodynamics

We demonstrate that thermodynamic knowledge acquired by humans can be transferred to computers so that the machine can use it to solve thermodynamic problems and produce explainable solutions with a guarantee of correctness. The actionable knowledge representation system that we have created for this purpose is called KnowTD. It is based on an ontology of thermodynamics that represents knowledge of thermodynamic theory, material properties, and thermodynamic problems. The ontology is coupled with a reasoner that sets up the problem to be solved based on user input, extracts the correct, pertinent equations from the ontology, solves the resulting mathematical problem, and returns the solution to the user, together with an explanation of how it was obtained. KnowTD is presently limited to simple thermodynamic problems, similar to those discussed in an introductory course in Engineering Thermodynamics. This covers the basic theory and working principles of thermodynamics. KnowTD is designed in a modular way and is easily extendable.

cs.CE↗

arXiv2023

Deep Anomaly Detection on Tennessee Eastman Process Data

This paper provides the first comprehensive evaluation and analysis of modern (deep-learning) unsupervised anomaly detection methods for chemical process data. We focus on the Tennessee Eastman process dataset, which has been a standard litmus test to benchmark anomaly detection methods for nearly three decades. Our extensive study will facilitate choosing appropriate anomaly detection methods in industrial applications.

cs.LG↗

arXiv2022

Making Thermodynamic Models of Mixtures Predictive by Machine Learning: Matrix Completion of Pair Interactions

Predictive models of thermodynamic properties of mixtures are paramount in chemical engineering and chemistry. Classical thermodynamic models are successful in generalizing over (continuous) conditions like temperature and concentration. On the other hand, matrix completion methods (MCMs) from machine learning successfully generalize over (discrete) binary systems; these MCMs can make predictions without any data for a given binary system by implicitly learning commonalities across systems. In the present work, we combine the strengths of both worlds in a hybrid approach. The underlying idea is to predict the pair-interaction energies, as they are used in basically all physical models of liquid mixtures, by an MCM. As an example, we embed an MCM into UNIQUAC, a widely-used physical model for the Gibbs excess energy. We train the resulting hybrid model in a Bayesian machine-learning framework on experimental data for activity coefficients in binary systems of 1146 components from the Dortmund Data Bank. We thereby obtain, for the first time, a complete set of UNIQUAC parameters for all binary systems of these components, which allows us to predict, in principle, activity coefficients at arbitrary temperature and composition for any combination of these components, not only for binary but also for multicomponent systems. The hybrid model even outperforms the best available physical model for predicting activity coefficients, the modified UNIFAC (Dortmund) model.

physics.chem-ph↗

arXiv2021

Attribute-based Explanations of Non-Linear Embeddings of High-Dimensional Data

Embeddings of high-dimensional data are widely used to explore data, to verify analysis results, and to communicate information. Their explanation, in particular with respect to the input attributes, is often difficult. With linear projects like PCA the axes can still be annotated meaningfully. With non-linear projections this is no longer possible and alternative strategies such as attribute-based color coding are required. In this paper, we review existing augmentation techniques and discuss their limitations. We present the Non-Linear Embeddings Surveyor (NoLiES) that combines a novel augmentation strategy for projected data (rangesets) with interactive analysis in a small multiples setting. Rangesets use a set-based visualization approach for binned attribute values that enable the user to quickly observe structure and detect outliers. We detail the link between algebraic topology and rangesets and demonstrate the utility of NoLiES in case studies with various challenges (complex attribute value distribution, many attributes, many data points) and a real-world application to understand latent features of matrix completion in thermodynamics.

cs.LG↗

arXiv2020

Machine Learning in Thermodynamics: Prediction of Activity Coefficients by Matrix Completion

Activity coefficients, which are a measure of the non-ideality of liquid mixtures, are a key property in chemical engineering with relevance to modeling chemical and phase equilibria as well as transport processes. Although experimental data on thousands of binary mixtures are available, prediction methods are needed to calculate the activity coefficients in many relevant mixtures that have not been explored to-date. In this report, we propose a probabilistic matrix factorization model for predicting the activity coefficients in arbitrary binary mixtures. Although no physical descriptors for the considered components were used, our method outperforms the state-of-the-art method that has been refined over three decades while requiring much less training effort. This opens perspectives to novel methods for predicting physico-chemical properties of binary mixtures with the potential to revolutionize modeling and simulation in chemical engineering.

physics.chem-ph↗

arXiv2019

Reexamination of Tolman's law and the Gibbs adsorption equation for curved interfaces

In manuscript arXiv:1703.08719 [cond-mat.soft], it was claimed that the well-known deduction of Tolman's law is not rigorous, since Tolman's argument implies that two different definitions of the surface tension, called $γ$ and $\barγ$ in the manuscript, coincide. This claim is retracted as it can be shown by free-energy minimization that $γ= \barγ$ indeed holds for the Laplace radius. Joachim Groß, Philipp Rehner, Carlos Vega, Øivind Wilhelmsen, and the anonymous reviewers of The Journal of Chemical Physics contributed to finding the mistake in the manuscript.

cond-mat.soft↗

arXiv2019

MolMod: An open access database of force fields for molecular simulations of fluids

The MolMod database is presented, which is openly accessible at http://molmod.boltzmann-zuse.de/ and contains presently intermolecular force fields for over 150 pure fluids. It was developed and is maintained by the Boltzmann-Zuse Society for Computational Molecular Engineering (BZS). The set of molecular models in the MolMod database provides a coherent framework for molecular simulations of fluids. The molecular models in the MolMod database consist of Lennard-Jones interaction sites, point charges, and point dipoles and quadrupoles, which can be equivalently represented by multiple point charges. The force fields can be exported as input files for the simulation programs ms2 and ls1 mardyn, Gromacs, and LAMMPS. To characterise the semantics associated with the numerical database content, a force-field nomenclature is introduced that can also be used in other contexts in materials modelling at the atomistic and mesoscopic levels. The models of the pure substances that are included in the data base were generally optimised such as to yield good representations of experimental data of the vapour-liquid equilibrium with a focus on the vapour pressure and the saturated liquid density. In many cases, the models also yield good predictions of caloric, transport, and interfacial properties of the pure fluids. For all models, references to the original works in which they were developed are provided. The models can be used straightforwardly for predictions of properties of fluid mixtures using established combination rules. Input errors are a major source of errors in simulations. The MolMod database contributes to reducing such errors.

physics.comp-ph↗

arXiv2017

Comparison of predictions of the PC-SAFT equation of state and molecular simulations for the metastable region of binary mixtures

The metastable region of binary mixtures is examined with the PC-SAFT equa-tion of state (EOS) and compared to molecular simulations. The studied mixtures are Methane + Ethane, Methanol + Ethanol, Carbon Dioxide + Toluene, Carbon Dioxide + Hydrogen Chloride and Hydrogen Chloride + Toluene. In order to cal-culate the spinodal, the second partial derivatives with respect to the components molarities are analytically determined for the PC-SAFT EOS. Thermal properties as well as the spinodal points determined using PC-SAFT and molecular simula-tion are compared. For the studied metastable gas phases the results obtained with PC-SAFT and the molecular simulations agree well. For liquid phases the meta-stable region calculated with the PC-SAFT EOS is typically broader than that ob-tained from the molecular simulations. No unphysical results were obtained with PC-SAFT for metastable states. Hence, PC-SAFT EOS is found to be principally suitable for predicting metastable states and spinodals of mixtures.

physics.comp-ph↗

arXiv2016

Simultaneous description of bulk and interfacial properties of fluids by the Mie potential

The vapor-liquid equilibrium (VLE) of the Mie potential, where the dispersive exponent is constant (m = 6) while the repulsive exponent n is varied between 9 and 48, is systematically investigated by molecular simulation. For systems with planar vapor-liquid interfaces, long-range correction expressions are derived, so that interfacial and bulk properties can be computed accurately. The present simulation results are found to be consistent with the available body of literature on the Mie fluid which is substantially extended. On the basis of correlations for the considered thermodynamic properties, a multicriteria optimization becomes viable. Thereby, users can adjust the three parameters of the Mie potential to the properties of real fluids, weighting different thermodynamic properties according to their importance for a particular application scenario. In the present work, this is demonstrated for carbon dioxide for which different competing objective functions are studied which describe the accuracy of the model for representing the saturated liquid density, the vapor pressure and the surface tension. It is shown that models can be found which describe simultaneously the saturated liquid density and vapor pressure with good accuracy, and it is discussed to what extent this accuracy can be upheld as the model accuracy for the surface tension is further improved.

cond-mat.soft↗

arXiv2016

Molecular simulation of the surface tension of real fluids

Molecular models of real fluids are validated by comparing the vapor-liquid surface tension from molecular dynamics (MD) simulation to correlations of experimental data. The considered molecular models consist of up to 28 interaction sites, including Lennard-Jones sites, point charges, dipoles and quadrupoles. They represent 38 real fluids, such as ethylene oxide, sulfur dioxide, phosgene, benzene, ammonia, formaldehyde, methanol and water, and were adjusted to reproduce the saturated liquid density, vapor pressure and enthalpy of vaporization. The models were not adjusted to interfacial properties, however, so that the present MD simulations are a test of model predictions. It is found that all of the considered models overestimate the surface tension. In most cases, however, the relative deviation between the simulation results and correlations to experimental data is smaller than 20 %. This observation corroborates the outcome of our previous studies on the surface tension of 2CLJQ and 2CLJD fluids where an overestimation of the order of 10 to 20 % was found.

physics.comp-ph↗

arXiv2015

Contact angle of sessile drops in Lennard-Jones systems

Molecular dynamics simulation is used for studying the contact angle of nanoscale sessile drops on a planar solid wall in a system interacting via the truncated and shifted Lennard-Jones potential. The entire range between total wetting and dewetting is investigated by varying the solid--fluid dispersive interaction energy. The temperature is varied between the triple point and the critical temperature. A correlation is obtained for the contact angle in dependence of the temperature and the dispersive interaction energy. Size effects are studied by varying the number of fluid particles at otherwise constant conditions, using up to 150 000 particles. For particle numbers below 10 000, a decrease of the contact angle is found. This is attributed to a dependence of the solid-liquid surface tension on the droplet size. A convergence to a constant contact angle is observed for larger system sizes. The influence of the wall model is studied by varying the density of the wall. The effective solid-fluid dispersive interaction energy at a contact angle of 90 degrees is found to be independent of temperature and to decrease linearly with the solid density. A correlation is developed which describes the contact angle as a function of the dispersive interaction, the temperature and the solid density. The density profile of the sessile drop and the surrounding vapor phase is described by a correlation combining a sigmoidal function and an oscillation term.

cond-mat.mes-hall↗

arXiv2015

ms2: A molecular simulation tool for thermodynamic properties, new version release

A new version release (2.0) of the molecular simulation tool ms2 [S. Deublein et al., Comput. Phys. Commun. 182 (2011) 2350] is presented. Version 2.0 of ms2 features a hybrid parallelization based on MPI and OpenMP for molecular dynamics simulation to achieve higher scalability. Furthermore, the formalism by Lustig [R. Lustig, Mol. Phys. 110 (2012) 3041] is implemented, allowing for a systematic sampling of Massieu potential derivatives in a single simulation run. Moreover, the Green-Kubo formalism is extended for the sampling of the electric conductivity and the residence time. To remove the restriction of the preceding version to electro-neutral molecules, Ewald summation is implemented to consider ionic long range interactions. Finally, the sampling of the radial distribution function is added.

cs.CE↗

arXiv2015

Long range correction for multi-site Lennard-Jones models and planar interfaces

A slab based long range correction approach for multi-site Lennard-Jones models is presented for systems with a planar film geometry that is based on the work by Janecek, J. Phys. Chem. B 110: 6264 (2006). It is efficient because it relies on a center-of-mass cutoff scheme and scales in terms of numerics almost perfectly with the molecule number. For validation, a series of simulations with the two-center Lennard-Jones model fluid, carbon dioxide and cyclohexane is carried out. The results of the present approach, a site-based long range correction and simulations without any long range correction are compared with respect to the saturated liquid density and the surface tension. The present simulation results exhibit only a weak dependence on the cutoff radius, indicating a high accuracy of the implemented long range correction.

physics.comp-ph↗

arXiv2014

Molecular modelling and simulation of the surface tension of real quadrupolar fluids

Molecular modelling and simulation of the surface tension of fluids with force fields is discussed. 29 real fluids are studied, including nitrogen, oxygen, carbon dioxide, carbon monoxide, fluorine, chlorine, bromine, iodine, ethane, ethylene, acetylene, propyne, propylene, propadiene, carbon disulfide, sulfur hexafluoride, and many refrigerants. The fluids are represented by two-centre Lennard-Jones plus point quadrupole models from the literature. These models were adjusted only to experimental data of the vapour pressure and saturated liquid density so that the results for the surface tension are predictions. The deviations between the predictions and experimental data for the surface tension are of the order of 20 percent. The surface tension is usually overestimated by the models. For further improvements, data on the surface tension can be included in the model development. A suitable strategy for this is multi-criteria optimization based on Pareto sets. This is demonstrated using the model for carbon dioxide as an example.

physics.comp-ph↗

arXiv2014

ls1 mardyn: The massively parallel molecular dynamics code for large systems

The molecular dynamics simulation code ls1 mardyn is presented. It is a highly scalable code, optimized for massively parallel execution on supercomputing architectures, and currently holds the world record for the largest molecular simulation with over four trillion particles. It enables the application of pair potentials to length and time scales which were previously out of scope for molecular dynamics simulation. With an efficient dynamic load balancing scheme, it delivers high scalability even for challenging heterogeneous configurations. Presently, multi-center rigid potential models based on Lennard-Jones sites, point charges and higher-order polarities are supported. Due to its modular design, ls1 mardyn can be extended to new physical models, methods, and algorithms, allowing future users to tailor it to suit their respective needs. Possible applications include scenarios with complex geometries, e.g. for fluids at interfaces, as well as non-equilibrium molecular dynamics simulation of heat and mass transfer.

cs.CE↗

arXiv2013

Molecular simulation of nano-dispersed fluid phases

Fluid phase equilibria involving nano-dispersed phases, where at least one of the coexisting phases is confined to a small volume, are investigated by molecular dynamics simulation. Complementing previous studies on nanoscopic droplets, simulation volumes containing a nanoscopic gas bubble surrounded by a subsaturated liquid phase under tension, i.e. at negative pressure, are conducted in the canonical ensemble. The boundary conditions are chosen such that the phase equilibrium at the curved interface is thermodynamically stable. Two distinct size-dependent effects are found: Curvature induces a subsaturation of the system, leading to a smaller liquid density. For the gas in the centre of the bubble, the small diameter has an additional obverse effect, increasing its density. The curvature dependence of the surface tension is discussed by evaluating average radial density profiles to obtain the excess equimolar radius, which is found to be positive, corresponding to a negative Tolman length.

cond-mat.mes-hall↗