MSO4SC: D5.1 Case study extended design and evaluation strategy

image

Project Acronym MSO4SC

Project Title

Mathematical Modelling, Simulation and Optimization for Societal Challenges with Scientific Computing

Project Number

731063

Instrument

Collaborative Project

Start Date

01/10/2016

Duration

25 months (1+24)

Thematic Priority

H2020-EINFRA-2016-1

Dissemination level: Public

Work Package WP5 END-USER APPLICATIONS DEVELOPMENT

Due Date:

M6 (PROJECT MONTH)

Submission Date:

30/04/2017

Version:

01

Status

Draft

Author(s):

Johan Hoffman (KTH); Marcus Weber (ZIB); Vedat Durmaz (ZIB); Tania Bakhos (BCAM); Johan Jansson (BCAM); Aitgeirr Rasmussen (Sintef); Christophe Trophime (UNISTRA); Christophe Prud’homme (UNISTRA)

Reviewer(s)

Javier Francisco Nieto De Santos (ATOS);
Carlos Fernandez (CESGA);

image

The MSO4SC Project is funded by the European Commission through the H2020 Programme under Grant Agreement 731063

Version History

Version

Date

Comments, Changes, Status

Authors, contributors, reviewers

01

14/04/2017

Initial version

Johan Hoffman; Marcus Weber; Vedat Durmaz; Tania Bakhos; Johan Jansson; Aitgeirr Rasmussen; Christophe Trophime; Christophe Prud’homme

02

29/04/2017

Review

Carlos Fernández

03

29/04/2017

Review

Javier Francisco Nieto De Santos

04

29/04/2017

Final

Johan Hoffman

List of figures

List of tables

Executive Summary

The high level objective of WP5 is to demonstrate and showcase the validity and effectiveness of MSO4SC by defining and implementing a set of pilot scenarios. WP5 will also provide a comprehensive overview of the innovation benefits, the problems that are solved, and how these can link to commercial viability. This document represents deliverable D5.1 that provides a description of the pilot scenarios and an evaluation strategy of the e-infrastructure, in particular with respect to evaluation of the Technology Readiness Level (TRL) that will reach TRL8 as part of the project.

Each MADF has at least one associated pilot, which will demonstrate TRL8 of its associated MADF. A short description of the TRL concept is given, and evaluation criterions are formulated for the pilots based on this concept. We describe the pilots to be implemented, and the features to be evaluated against the TRL criterions, formulated in the form of test cases defined for each pilot. This report will be updated once at a later stage of the project as part of D5.1.

1. Introduction

1.1. Purpose

The high level objective of WP5 is to demonstrate and showcase the validity and effectiveness of MSO4SC by defining and implementing a set of pilot scenarios. WP5 will also provide a comprehensive overview of the innovation benefits, the problems that are solved, and how these link to commercial viability. This goal is decomposed into the following objectives: (i) define and coordinate the design of the pilot scenarios; (ii) provide the implementation of pilot scenarios according to the requirements specified in WP2 and the test cases defined for each pilot; (iii) evaluate MSO4SC by verifying the degree of achievements in fulfilling the requirements defined in WP2, including functional and non-functional aspects, and in the form of a target Technology Readiness Level (TRL). The results of the evaluation will be the input for the revision of the research activities of WP3 and WP4.

1.2. Overview

This document provides a description of the pilot scenarios. It also describes the overall evaluation strategy, detailing the protocol and procedures that should be followed during the evaluation of the pilots, in particular with respect to evaluation of the TRL level. The pilots and the evaluation strategy are based on the requirements identified in D2.1[1], together with direct input from the pilot coordinators and end-users. This is the first D5.1 report, which will be updated once during the project taking into account results and experiences up to that point in the project.

The evaluation plan formulates a set of features for each pilot to be validated and goals to be achieved, in the form of test cases. The results of the evaluation protocol will be reported twice during the project in the form of evaluation reports (D5.4), to function as input to the successive releases of the MSO4SC platform, and to the research community. In line with the evaluation strategy, the pilots have been detailed, specifying end-users, a set of test cases, and how the MSO4SC e-infrastructure will be used, in order to facilitate their implementation in D5.3.

In D2.1 the pilots were divided into four groups: three groups of pilots based on the MSO4SC MADFs: FEniCS, Feel++ and OPM, respectively, and one group of pilots based on other applications. The functional requirements identified in D2.1 of the envisioned infrastructure were: (i) high performance of the applications; (ii) efficient data flow between the application domain and the e-infrastructure; (iii) fast post-processing including visualization. The main non-functional requirement was (iv) usability of services with one-click deployment from the marketplace, which is of particular importance for non-technical users like authorities applying an end-user application from MSO4SC for a certain addressed societal challenge.

1.3. Glossary of Acronyms

Acronym Definition

CFD

Computational Fluid Dynamics

D

Deliverable

EC

European Commission

ESA

European Space Agency

FEM

Finite Element Method

MADF

Mathematical Development Framework

MPI

Message Passing Interface

NASA

National Aeronautics and Space Administration

RANS

Reynolds Averaged Navier-Stokes equations

TRL

Technology Readiness Level

WP

Work Package

Table . Acronyms.

2. Evaluation strategy

In this section we describe the evaluation strategy, in the form of an evaluation plan detailing the protocol to be followed during the evaluation of the pilots. The evaluation criteria are formulated to demonstrate the progress to TRL8 of the e-infrastructure as defined in D2.2 [2], in particular the Mathematical Development Frameworks (MADFs) and the MSO Portal. Each MADF has at least one associated pilot, which will demonstrate TRL8 of its associated MADF and the MSO Portal. We start by a short description of the TRL concept, and then describe how we will evaluate the pilots with respect to this concept.

2.1. Technology Readiness Level (TRL)

Technology Readiness Level (TRL) is a method to estimate the technology maturity of a component or product during the development process. TRL is based on a scale from 1 to 9, with 9 being the most mature technology. The TRL concept provides a framework that enables consistent and uniform assessment of technical maturity across different technology fields.

Although the TRL scale is conceptually universal, the precise definition of the different levels differs between agencies such as NASA, ESA and EC. We will here adopt the EC definition[multiblock footnote omitted] of TRL6 to TRL8, outlined in Table 1. All pilots satisfy TRL6 at the start of the project. The main difference between TRL7 and TRL8 is that at TRL8 the pilots have reached a level of maturity that allows the end-users to use the service independently from the developers of the service, and whereas TRL7 verifies the functional requirements identified in D2.1, TRL8 in addition verifies the non-functional requirements.

TRL EC definition Pilot evaluation criterion

TRL6

Technology demonstrated in relevant environment (industrially relevant environment in the case of key enabling technologies).

All pilots satisfy TRL6 at the start of the MSO4SC project.

TRL7

System prototype demonstration in operational environment.

A prototype is demonstrated for pilot test cases representative of the operational environment of the pilot end-user.

TRL8

System complete and qualified.

The pilot end-user can independently use the service.

Table : TRL definition by EC2, and the associated MSO4SC pilot evaluation criteria.

2.2. Evaluation plan

To apply the TRL scale to the MADFs, we need to adapt the EC definitions to the context of the pilots that will serve as evaluation criteria for the MADFs.

We outline the MSO4SC interpretation of the TRL criteria in Table 2 for TRL7 and TRL8, and we recall that all MADFs and pilots already satisfy TRL6 at the outset of the project. Pilots are formulated together with end-users of the MSO4SC technology. The TRL operational environment is interpreted as the operational environment of the end-user, with the TRL7 criterion defined as a prototype of the pilot being demonstrated in a test case representative for the end-user environment. The criterion for TRL8 is defined as a service that can be used independently by the end-user through the MSO Portal.

The pilot evaluation protocol is described below, based on the functional and non-functional requirements identified in D2.1.

TRL Evaluation protocol

TRL7

  • Run pilot test cases on the MSO4SC e-infrastructure.

    • Verify efficient data flow between application domain and the e-infrastructure.

    • Verify high performance of the application.

    • Verify fast post-processing including visualization.

  • Summarize the findings for the evaluation report (D5.4).

TRL8

  • Verify independent end-user usability of service, including one-click deployment from the MSO Portal.

  • Summarize the findings for the evaluation report (D5.4), including end-user certification of usability.

Table : Pilot evaluation protocol.

3. Definition of pilots

In this section we describe the pilots to be implemented in D5.3. The features to be evaluated correspond to the features listed in the development roadmap in D2.2, which will be evaluated in through test cases defined for each pilot. Over the course of the project, the test cases may be modified or new test cases may be added. First, the pilots associated to the three MADFs are defined (FEniCS, Feel++, OPM), and then the remaining pilots.

3.1. FEniCS pilots

Two pilots are based on the FEniCS MADF, with the common requirements that the pilots together with the FEniCS MADF should be deployed at the MSO4SC e-infrastructure, including support for post-processing and visualization as part of the MSO Portal, e.g. by ParaViewWeb. The input is in the form standard CAD data or an STL surface, whereas the output is a time series of sampled solutions over a tetrahedral volume mesh.

Unresolved include directive in modules/ROOT/pages/d5.1/README.adoc - include::../../pilots/fenics/floatingwindturbine/README.adoc[]

Unresolved include directive in modules/ROOT/pages/d5.1/README.adoc - include::../../pilots/fenics/3dairqualitypredictioncfd/README.adoc[]

3.2. Feel++ pilots

The common requirements of the Feel++ pilots are that the pilots should be deployed at the MSO4SC e-infrastructure, including support for post-processing and visualization as part of the MSO Portal, e.g. by ParaViewWeb. The input is in the form standard CAD data or an STL surface, whereas the output is a time series of sampled solutions over a volume mesh.

Unresolved include directive in modules/ROOT/pages/d5.1/README.adoc - include::../../pilots/feelpp/hifimagnet/README.adoc[]

Unresolved include directive in modules/ROOT/pages/d5.1/README.adoc - include::../../pilots/feelpp/eye2brain/README.adoc[]

3.3. OPM pilots

The requirement of the OPM pilot is that the pilot should be deployed at the MSO4SC infrastructure, with support for parallel MPI and efficient ensemble simulations. The input is in the form standard CAD data or an STL surface, whereas the output is a time series of sampled solutions over a volume mesh.

Unresolved include directive in modules/ROOT/pages/d5.1/README.adoc - include::../../pilots/opm/opm-flow/README.adoc[]

3.4. Other pilots

3.4.1. ZibAffinity

End-user: Pharmaceutical companies.

The requirement of the pilot is that the pilot should be deployed at the MSO4SC infrastructure, with support for parallel MPI.

ZIBAffinity [15] uses molecular dynamics (MD) simulations and methods of statistical thermodynamics in order to estimate binding affinities for biological host–guest systems (HGS). Having uploaded a small drug-like molecule under observation as input, the user selects one or more protein target structures from a database of force field-parameterized models and submits one job per target-ligand combination to the queue of the CESGA high performance computer. After job processing, the results are made available to the user.

The affinity is estimated as a linear combination of averages of molecular observables according to a linear interaction energy [16] model. Ensuing from the uploaded small molecule, GROMACS MD simulations, with at most 61 different starting positions, are performed in parallel. The optimal binding position (binding mode) is then extracted from that data and provided as a 3D molecular structure serving (Figure 6), along with thermo-statistical data (Figure 6) as the basis for absolute or relative binding affinity estimation.

urface with electrostatic information of a protein in complex with a transformation product of the antibiotic sulfamethoxazole.

Figure 12: Preferential host–guest binding model (left), and conformational entropy (flexibility) during molecular simulation (right).

For each available target molecule separately, the linear combination factors of the free energy equation above must have been learned in advance during a training phase and must be stored in the target database together with the target’s initial configuration and further required information. At runtime, they are used for the calculation of absolute binding affinities (test case 1). If, in contrast, no training data is available for a particular protein target, ZIBAffinity can use standard weights or estimate relative binding affinities (test case 2).

Pilot features to be validated are correct predictions of the binding affinities (inside the statistical range), run-time on the cloud system compared to running ZIBAffinity on HLRN (North-German Supercomputing Alliance), and usability of the data storage and data management concepts of MSO4SC.

Use Case 1: ZIBAffinity will be used for a particular test case that is extensively described in a recent article [17]. In this test case several small drug-like molecules are tested against the alpha-estrogen receptor as target molecule. Laboratory data for binding free energies ΔGExp are available (x-axis in Figure 7) and have been used for parameter estimation regarding the linear interaction model presented above. After that training phase, the ZIBAffinity software predicts binding energies ΔGComp of all given small molecules (y-axis in Figure 7).

plot-py_lie2011_fit-loo.png

Figure 12: Predicted vs. experimental binding free energies of the alpha-estrogen receptor and a set of 31 ligands using the software ZIBAffinity.

With this approach it is possible to estimate the estrogenicity of small-molecules, which is important for prediction of effects of trace pollutants as endocrine disruptors, a major challenge for water cleaning plants [18] and a severe problem of industrialized countries.

If laboratory data is not available, then ZIBAffinity only provides a priority list of trace pollutants for further biochemical experiments. ZIBAffinity predicts their potential toxicological risk (see Figure 8) on the basis of “adverse” thermo-statistical data.

ransrisk priority table

*Figure 13: Potential toxicological risk on the basis of “adverse” thermo-statistical data.   *

ZIBAffinity can be used for the prediction of several host-guest-affinities (not only to proteins). One example is the prediction of the elution order of Liquid Chromatographic Separations [19]. This prediction is important for the choice of the column material of separation columns. Thus, it provides important data for the design of chemical processes. A well-suitable column separation material is very important for producing pharmaceuticals with fewer side effects (due to pollutants). In terms of drug delivery, ZIBAffinity can help to design drug-carriers with certain drug-release profiles [20].

Use Case 2: ZIBAffinity is used to design small drug-like molecules with a high affinity for special pharmaceutical targets (pain relief drugs [21] or estrogen receptor inhibitors [22]). This test case is important for the planning of the data storage and data management in MSO4SC. The data that is created in these projects is extremely valuable for pharmaceutical companies and, thus, needs a very high level of protection against data loss and data security.

3.4.2. 3DAirQualityPrediction

The requirement of the pilots is that the pilots should be deployed at the MSO4SC infrastructure, with statistical data.

End-user: OMSZ (The Hungarian Meteorological Service)

Urban citizens are exposed to air pollution at an increased level, which causes many premature deaths [23]. In cities, the main producer of the most relevant pollutants, the nitrogen oxides (NOx) is the traffic, which is responsible for emitting more than 40% of this contaminants. In order to support city authorities and policy makers in their job for reducing air pollution arising from traffic, computational models have been used for running scenarios for some decades. However, accurate models that take into account real 3D geometry of cities including buildings with high spatial resolution and are easy-to-use at the same time seem to be lacking. The main difficulty of matching accuracy and usability is that accuracy needs supercomputers, which is traditionally of difficult use. Bringing the existing 3DAirQualityPrediction framework to the MSO4SC infrastructure and using the fast and accurate FEniCS-HPC application 3DAirQualityPrediction-CFD as module of the framework 3DAirQualityPrediction the project will match accuracy and usability.

The framework is composed of modules, namely traffic, emission, meteorology, dispersion and the core module, which is either for evaluating assessments or performing optimization or control. An overview of the 3DAirQC workflow is seen on Figure 14.

Macintosh HD:Users:jhoffman:Library:Containers:com.apple.mail:Data:Library:Mail Downloads:E7B589A8-16C5-4E09-B96F-65D1EDCA2C50:Workflow_health_3_fenics.png

Figure 14. An overview of the 3DAirQualityPrediction workflow for running scenarios for health indicators depending on traffic, fleet and meteorology data or traffic and meteorology measurements and simulations.

Preprocessing of the data

The preprocessing steps of the simulation modules are based on several toolkits consisting of Blender (see www.blender.org/) tools for 3D modelling, in-house 3D meshing tools and some commercial tools of ANSYS. All of these steps need normally special and time consuming work, which is done mainly automatically using our tools. For illustration of the tools for geometry preprocessing and meshing see Figure 15 and Figure 16, respectively.

Full_Gyor_01

Figure 15. Example of CAD geometry of the town resulted from city geoinformatic data base using Blender scripts

octree_high_6_6_2060K_no_logo Polyhedra_halo

Figure 16 An overview of the meshes: octree mesh generated by in-house multi thread Java code (left) and polyhedral mesh resulted from using ANSYS Fluent (right).

The traffic module

For modelling the urban traffic we have been using macroscopic and microscopic models. These are based on

  • historical traffic count data of a big campaign and/or

  • monitoring data arising from operational data collected by city and national road authorities.

Based on these data, the user has an option to choose PTV VISSIM or VISUM for microscopic or macroscopic simulation of the traffic or either use just statistical or measurement data.

map4 traffic detectors

Figure 17. An overview of traffic sensor network of the city (on the courtesy of Hungarian National Roads Nonprofit Ltd. -Magyar Közút Nonprofit Zrt.

The emission module

For modelling the emission of the vehicles in the traffic the European standard emission model, the COPERT model is used. Actually it is a script that computes emissions of the pollutants at street segments that serve as forcing terms in the dispersion module.

The meteorology module

For boundary conditions of the dispersion module we used meteorology data from the national official operational data of the Hungarian Meteorology Service (OMSZ), which uses the AROME (Application of Research to Operations at Mesoscale) non-hydrostatic numerical weather prediction model

The dispersion module

The project will use 3DAirQualityPrediction-CFD application for simulation of the emitted pollutants. There are two options of running the CFD solver. The first one is the frozen flow field model where the wind field is precomputed with a RANS or an adaptive LES model of 3DAirQuallityPrediction-CFD according to the setting given in a configuration file. Then dispersion of the pollutants is computed with the simple advection-diffusion(-reaction) module of 3DAirQuallityPrediction-CFD. In the second option the wind field and dispersion of the pollutants is computed simultaneously. These options are governed by the configuration of the 3DAirQuallityPrediction-CFD.

kep3

Figure 18. NOx concentrations at 1.5m height according to the simulation results with RANS simulation with Ansys Fluent. Note that in the project the high quality 3DAirQuallityPrediction-HPC adaptive LES module will provide even higher accuracy.

Test cases

The pilot test cases will be performed in city of Győr where all data including traffic monitoring and also historic data are available.

Test case 1: In this pilot traffic and meteorology data are taken from historical data. Thus running the 3DAirQualityPrediction framework all data and codes can be ported and run at the infrastructure of the MSO4SC infrastructure. Several meteorology and traffic scenarios will be run based on - precomputed or statistical – meteorological and traffic data.

Test case 2: In this pilot traffic and meteorology data are taken from measurements and, optionally, traffic is simulated. Thus running the 3DAirQualityPrediction framework data transfer from the server collecting measurements data from meteorology and traffic has to be incorporated, Data transfer a dedicated server of SZE that runs the propriety software PTVVISSIM/VISUM will be solved, or alternatively on MSO4SC infrastructure, using the license provided by SZE.

4. Conclusions

In this report we have presented the pilots to be evaluated in WP6. We provide a description of the pilot scenarios and an evaluation strategy for the e-infrastructure, in particular with respect to evaluation of the Technology Readiness Level (TRL) that will reach TRL8 as part of the project. The focus is the MSO Portal and the MADFs, where each MADF has at least one associated pilot.

5. References

  1. MSO4SC D2.1 End Users’ Requirements Report, 2017.

  2. MSO4SC D2.2 MSO4SC e-infrastructure Definition, 2017.

  3. K. Kleefsman, G. Fekken, A. Veldman, B. Iwanowski and B. Buchner, “A volume-of-fluid based simulation method for wave impact problems”, Journal of computational physics, 2005.

  4. J. Jansson, V. Nava, G. Aguirre, R. Vilela de Abreu, M. Sanchez , G. Perez, J. Hoffman and J. L. Villate, “Estimation of hydrodynamic viscous characteristics of an offshore wind platform using adaptive finite element simulations”, 1st prize for best poster at Bilbao Marine Energy Week, 2015.

  5. J. Jansson, V. Nava, M. Sanchez, G. Aguirre, R. Vilela de Abreu, J. Hoffman and J. L. Villate, “Adaptive simulation of unsteady flow past the submerged part of a floating wind turbine platform”, Proceedings ECCOMAS VI International conference on computational methods in marine engineering, 2015.

  6. J. Hoffman, J. Jansson, R. Vilela de Abreu, N. C. Degirmenci, N. Jansson, K. Müller, M. Nazarov and J. H. Spühler, “Unicorn: Parallel adaptive finite element simulation of turbulent flow and fluid-structure interaction for deforming domains and complex geometry”, Computers and Fluids, 2013.

  7. J. Hoffman, J. Jansson, N. Jansson and R. Vilela De Abreu, “Towards a parameter-free method for high Reynolds number turbulent flow simulation based on adaptive finite element approximation. Computer Methods in Applied Mechanics and Engineering, 2015.

  8. V. Nava, G. Aguirre, J. Galvan, M. Sanchez-Lara, I. Mendikoa and G. Perez-Moran, MARINET: “Identification and validation of hydrodynamic characteristics of a Semi-Submersible Offshore Wind Platform through tank test”, under peer review.

  9. V. Nava, G. Aguirre, J. Galvan, M. Sanchez-Lara, I. Mendikoa and G. Perez-Moran, “Experimental Studies on the Hydrodynamic Behavior of a Semi-Submersible Offshore Wind Platform”, Proceedings of 1st International Conference on Renewable Energies Offshore, 2014.

  10. J. Jansson, V. D. Nguyen, M. M. Ginard, D. Castanon Quiroz, L. Saavedra, E.Krishnasamy, A. Goude and J. Hoffman, “Direct finite element simulation of turbulent flow for marine based renewable energy”, under peer review.

  11. S.M. Salim, et al., Numerical simulation of atmospheric pollutant dispersion in an urban street canyon: Comparison between RANS and LES, Journal of Wind Engineering and Industrial Aerodynamics, Vol.99(2-3), pp.103-113, 2011.

  12. J. Béard and F. Debray, The French High Magnetic Field Facility, Journal of Low Temperature Physics, 2013, 170(5-6), pp541-552.

  13. C. Daversin, C. Prudhomme, C. Trophime. Full 3D MultiPhysics Model of High Field PolyHelices Magnets. IEEE Transactions on Applied Superconductivity, Institute of Electrical and Electronics Engineers, 2016, 26 (4), pp.1-4.

  14. C. Daversin-Catty. Reduced basis method applied to large non-linear multi-physics problems. Application to high field magnets design. Analysis of PDEs [math.AP]. IRMA (UMR 7501), 2016.

  15. V. Durmaz, FU Berlin, Atomistic Binding Free Energy Estimations for Biological Host–Guest Systems, Doctoral Thesis, 2016.

  16. J. Åqvist, C. Medina, J. E. Samuelsson: A new method for predicting binding affinity in computer-aided drug design. Protein Eng, 7:385−391, 1994.

  17. V. Durmaz, S. Schmidt, P. Sabri, C. Piechotta, M. Weber: A hands-off linear interaction energy approach to binding mode and affinity estimation of estrogens. J. Chem. Inf. Model, 53(10):2681–2688, 2013.

  18. V. Durmaz, M. Weber, J. Meyer, H. Mückter: Computergestützte Simulationen zur Abschätzung gesundheitlicher Risiken durch anthropogene Spurenstoffe in der Wassermatrix. KA Korrespondenz Abwasser, Abfall, 3/15:264-267, 2015.

  19. V. Durmaz, M. Weber, R. Becker: How to Simulate Affinities for Host-Guest Systems Lacking Binding Mode Information: Application in the Liquid Chromatographic Separation of Hexabromocyclododecane Stereoisomers. Journal of Molecular Modeling, 18(6):2399-2408, 2012.

  20. M. Weber, C. Zoschke, A. Sedighi, E. Fleige, R. Haag, M. Schäfer-Korting: Free Energy Simulations of Drug loading for Core-Multishell Nanotransporters. J Nanomed Nanotechnol, 5(5):234, 2014.

  21. V. Spahn, G. Del Vecchio, D. Labuz, A. Rodriguez-Gaztelumendi, N. Massaly, J. Temp, V. Durmaz, P. Sabri, M. Reidelbach, H. Machelska, M. Weber, C. Stein: A nontoxic pain killer designed by modeling of pathological receptor conformations. Science, 355(6328):966-969, March 2017.

  22. F. Abendroth, M. Solleder, D. Mangoldt, P. Welker, K. Licha, M. Weber, O. Seitz: High affinity fluorescence labelled ligands for the estrogen receptor. Eur. J. Org. Chem., 2015(10):2157-2166, 2015.

  23. WHO, 2014: Burden of disease from ambient and household air pollution. Report. www.who.int/phe/health_topics/outdoorair/databases/en/