=Paper=
{{Paper
|id=Vol-1490/paper43
|storemode=property
|title=Development of the requirements template for the information support system in the context of developing new materials involving Big Data
|pdfUrl=https://ceur-ws.org/Vol-1490/paper43.pdf
|volume=Vol-1490
}}
==Development of the requirements template for the information support system in the context of developing new materials involving Big Data==
https://ceur-ws.org/Vol-1490/paper43.pdf
Data Mining and Big Data
Development of the requirements template for the
information support system in the context of developing
new materials involving Big Data
Grechnikov F.V., Khaimovich A.I.
Samara State Aerospace University
Abstract. We consider a concept of databases for the numerical and physical
experiments in the information management system used for new materials
development. For this purpose, brief analysis of information systems and
databases for large-scale experiments has been performed. Special features and
general approach to the templates of the functional requirements are suggested.
Keywords: big data technology, the genome of materials innovation
infrastructure materials, prediction of new materials, the pattern of
requirements
Citation: Grechnikov F.V., Khaimovich A.I. Development of the requirements
template for the information support system in the context of developing new
materials involving Big Data. Proceedings of Information Technology and
Nanotechnology (ITNT-2015), CEUR Workshop Proceedings, 2015; 1490:
364-375. DOI: 10.18287/1613-0073-2015-1490-364-375
One of the major issues tackled by the leading engineering departments and
individual production engineers developing products for the aerospace branch with
the use of new materials involves making all the strength calculations (structural
analysis) with regard of such materials under consideration of mechanical,
thermophysical and other features at the macro-level. At the micro-level though (the
microstructure field) materials pertain to anisotropy of structure due to availability of
heterogeneous phase composition, point-type and other localized defects, non-
uniformity of crystal composition, foreign inclusions, etc. These localized defects as
well as external micro-mechanical damages of the parts’ surface give rise to crack
formation and development, which in their turn lead to off-design cases and
breakdowns of parts in the process of their operational use and maintenance. It is
especially true with regard to composite materials with mechanical alloying,
compounds of the ‘binding-reinforcing fiber’ type and others. Well-known are cases
(e.g., at the AIRBUS and BOEING companies), when compound parts were
withdrawn from manufacturing final products due to the mentioned reasons.
364
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
The material’s life cycle consists of several stages:
1. development of the material’s new composition;
2. optimization of its properties;
3. designing and developing products made of the material;
4. testing and certification;
5. production;
6. operation including repair and recovery;
7. disposal.
The innovations infrastructure in developing new materials is shown in Fig. 1 [1].
Designing of new materials requires creation of the existing materials’ database:
mathematical models of the materials, experimental data as to materials in form of
photographic images and data based on vibroacoustic emission as well as digital data
on the processes of generating the products’ samples made of the respective materials.
Fig. 1. – Innovations infrastructure in developing new materials [1]
Integration of the following methods is adapted in development of the materials
database:
1. Tools of computer-based simulation (physical models of nano- and micro-level,
numerical models and systems of simulation-type modeling of the macro-level
continuum mechanics)
2. Toolware and experimental analysis methods
3. Digital data exchange and digital information storage formats
To enable works as to designing new material, it is necessary to create integrated
medium in the BIG DATA system; the reason for that is that this database shall
integrate sound data, experimental data photographs and existing materials
mathematical models.
365
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
The input data (material database) for analysis are as below:
1. Results of simulation-type modeling of the structure – software modeling the
material’s structure (phase constitution, interrelationship, parts by volume, three-
dimensional distribution and phases positioning)
2. Full-scale experiment results (standard tests) – microstructure pieces and material’s
submicrostructure obtained by means of light and electronic microscopy, results of
the X-ray-structural analysis (describing material’s structure and structural
imperfections), and other analysis types. Final models as to the material’s structure
obtained as a result of the full-scale experiment data capture (development of
special-purpose technique and software is required) linked to the models of P.1
3. Mechanical and other properties of the material (based on the experiment outcome,
P. 2) – in strict compliance with the models of the structure.
The output information in the system of modeling new materials shall be viewed
as the damage evaluation service based on the neural network. Principle of the service
performance: on the basis of numerous selections of fragments of structures with
defects as well as other parameters characterizing the material’s macro- and micro-
condition, the part’s geometry parameters, operation conditions and calculated or
experimentally obtained variants of damage or pathology development, training of the
network is conducted. Flowsheet of the services’ performance with regard to
diagnostics and damage evaluation is presented in Fig. 2.
Fig. 2. – Flowsheet of the services performance with regard to diagnostics and damage
evaluation. Arrows indicate the forecast service blocks.
366
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
The forecast is made in the following way: selected parameters are delivered at the
input, which is similar to the elements of the training set of selected stuff. Selected
parameters are obtained for the part material under testing, which in its turn has been
selected from the industrial lot of the parts. For the required (possibly critical)
working conditions, the grid generates a forecast of material’s response.
In the event of unfavorable forecast, series of refined calculations and experiments
are made, expert opinion is given as to ultimate solution and corrective actions are
undertaken, which prevent failure of the product.
Training and clarification of the grid-related forecast is undertaken periodically by
virtue of increase of the training set in size.
Some stages while creating new materials are implemented with the aid of
hardware and special-purpose software. This being so, each stage can be implemented
within special application package, which implies access to the complete package on
the part of the material scientists. Moreover, scientific applications often require
special skills for their installation, adjustment and startup, which the majority of
researchers lack.
It is proposed to develop methods of producing cloud-based system of information
support aimed at developing new materials, which would enable integration of the
complete set of software tools into a unified medium and provide the researchers with
the opportunity of operating it with the help of user-friendly web-interface having
Internet access via web-browser. The web-interface as well as the computer modeling
with the aid of application packages will be implemented in the cloud medium.
Start of computations in the cloud is proposed in the following way. In the cloud
web-interface of the system, the user is expected to chose the system-supported
application package from the list and assign the set of package- and task-specific
input parameters (including the number of cores and RAM necessary for making
relevant calculations) and in case of necessity – to indicate the file or directory with
initial mega-data way, which is equally necessary to computations. After initiating the
task start from the web-interface, the system will check availability of all the
obligatory data and correctness of their input. In the event of successful check and
availability of uncommitted resources, the system initiates startup of the
corresponding number of virtual machines with preinstalled application software
required for making calculations set by the user. After finalizing the calculations, the
system will preserve results via the user-set way in the cloud storage.
The system based on the developed techniques will also enable storage of various
information on properties of already known materials, experimental data linked to
previously conducted tests as well as performance search by these data with the
purpose of forecasting properties of new materials without undertaking any
complementary full-scale experiments.
In the USA and Europe, this direction is marked by similar works within the
framework of developing material genome under the use of the Big Data approach. In
Russia, elaboration of methods with the aim of further creation on their basis a portal
for researchers in the field of material science is just at the initial stage.
June 24, 2011, the US President announced the Materials Genome Initiative – in
order to double the speed, at which development and production of new materials is
367
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
done. Acceleration of developing advanced materials is of fundamental importance
for achieving global competitiveness (http://www.whitehouse.gov/sites/default
/files/microsites/ostp/materials_genome_initiative-final.pdf).
New materials are of composite nature – they have multi-component and multi-
phase structure: without adequate simulation modeling, information resource and data
exchange development of new-generation materials (given the existing empirical
approach) is either impossible of too time-consuming due to their complexity. The
Materials Genome Initiative is aimed at promoting development of new materials; it
accelerates the process and makes it more cost efficient. December 4, 2014, strategic
plan of the Initiative was published (http://www.nist.gov/mpi/uploacl/MGI-
StrateQicPlan-2014.pdf) NSTC (National Science and Technology Council). The
subcommittee for genome includes NIST, Department of Energy, Department of
Defense, NSF, NASA, NIH, USGS, DARPA in coordination with the infrastructure
of Nanotechnologies Knowledge.
In much the same way that the Human Genome Project accelerated development of
biological sciences owing to identification and decoding of basic construction blocks
of human genetic code, the Materials Genome Initiative (MGI) will accelerate
understanding of the material science fundamentals, providing information ensuring
creation of new products and processes. MGI will require unprecedented level of
cooperation of participants including government, industry, academic entities,
professional communities, national laboratories, thus resulting in rebirth of the
respective American industrial sectors. In order to integrate results of experiments,
numerical methods (finite elements method, etc.), theoretical approach, the strategic
plan envisages creation of a network of the MGI (Material Genome Initiative)
resources – with the purpose of developing reliable techniques of precise simulation
modeling, improving tools of performing and processing experiments’ outcomes.
Directional effect of these integrated data – discovery of new materials, development
of analytical information for increasing value of the obtained experimental and
computation data. Other objectives of the plan, linked to intensive data processing,
include creation of means for materials’ information-specific infrastructure
implementation as well as development of the first rate techniques of maintaining
proper storages of databases as per materials.
In order to start a dialogue within the framework of the MSE community
(Engineering of Development and Study of Materials), the National Institute for
Standards and Technologies (NIST, USA) conducted a workshop on materials’ digital
data in May 2012 under the aegis of MGI (Materials Genome Initiative). The
workshop determined a number of issues that have to be settled in the process of
creating data-specific strategy for materials; these include: schemes/ontologies of
materials, standards of presenting data and metadata, data repositories/data achieves,
data quality, incentives for joint use of data, intellectual systems and tools for data
search [2].
The European Union is involved in elaborating standards of data exchange as for
engineering materials within the framework of the European Committee for
Standardization [3]. These standards mainly emphasize engineering materials for the
aerospace branch. The European Commission finances activity of the expert group
368
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
called Integrated Computational Materials Engineering (ICMEg); the Group has been
established as an entity for integrated materials computation engineering – with the
purpose of developing standards and protocols required for supporting digital data
exchange on materials [4].
In Russia, it is necessary to launch developments as to creating genome of
materials under utilization of the Big Data technology. For this, it is necessary to form
requirements with regard to large-volume digital data covering the stage of collecting
and processing of experiments’ outcomes as for mechanical properties studies (in
dynamics, for example) or microstructure fragments, stage of the numerical
simulation modeling of the process of materials’ responses – with the view of further
use of the collected experimental data for shaping an integrated model of materials-
specific knowledge. The integrated knowledge model includes information required
for concluding forecast related to the materials’ response (including those with
noncritical defects) – in the course of maintenance or in the technological process.
Testing of the end-to-end information support and shaping the list of requirements as
to materials’ genome (for the damage evaluation case) could be performed as shown
in Table 1.
Table 1. Application examples (use cases) of the materials information support services as per
damage evaluation
Experimental data on Simulation model of Response
pathology (defects, the material’s response prediction model
heterogeneities) in the (computational model).
material’s structure Technology of identifying
imperfections by indirect
information
Unstructured data Modeling of crack Forecast
storage on microstructure development in CAE – (training) model of
of material with micro- system with preparing material’s strength
cracks finite element pattern on variation on the
the basis of selected basis of neural
micro-structure fragments networks
with the aid of cloud
service
Fragment (sample) of Spectral analysis Forecast
vibroacoustic emission technology as per (training) system of
signal in the event of emergencies identification monitoring
critical material load (e.g., on the basis of identifying materials’
in static test, deviation of emission parameters responses in
the operational conditions emergencies
and geometry from
nominal ones (mechanical
treatment, sheet
pressworking, etc. for the
know and new materials))
369
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
Currently, there are precedents in Russia of global distributed information systems
supporting experiments operating large data volumes. For illustrative purposes, one
can mention the Central Information-Computer Complex of the Joint Institute for
Nuclear Research (JINR) (in the town of Dubna), which develops as a multifunctional
center for storing, processing and analyzing data and which is designed for providing
a broad spectrum of opportunities to users on the basis of components integrated into
it: grid-infrastructure of Tier-1 and Tier-2 – with the purpose of supporting
experiments on LHC (ATLAS, ALICE, CMS, LHCb), FAIR (CBM, PANDA) and
large-scale experiments; general-purpose computation cluster; cloud computation
infrastructure; computation heterogenic cluster HybriLIT; training-and-research entity
for distributed and parallel calculations [5].
Based on the experience of designing projects information support systems
ATLAS, CMS of the JINR employees [6-10], it is possible to define a number of
requirements concerning the databases conceptual structure with regard to
information support of experiments within the genome of material project.
For full CMS databases’ structure and hierarchy see Table 2, Appendix A [6].
Conclusion
High competition under the global market conditions requires continuous
improvement of consumer properties in the output products’, the major part of which
depends on materials used in manufacturing. In this regard, rapid development of new
materials with required properties is critically important; moreover, it is important
often not just for individual companies but for branches of nation’s economy.
Development of methods of organizing interrelations of heterogeneous virtual
services and applications in Big Data will enable a system of digital support of new
materials development in Russia. The distributed system will make it possible to
integrate on a uniform (from the uses’ point of view) platform the whole set of
diversified data required by the material scientists – with the purpose of forecasting
defects development in the process of creating new materials.
Worthwhile mentioning is availability of commercial software in form of
interfaced modules, designed for solving adjoint problems of the structural analysis,
in which the composite material microstructure’s influence on the developed structure
is taken into account, e.g., Digimat, MultiMech. However, there is no available
forecast service (e.g., solution on the neural networks basis). This enables rapid
forecast solution on the basis of pre-conducted experiments and calculations with the
aid of Big Data.
Recommendations on the information services assisting experiments have been
validated and are more evident. For the dynamically resizable data, nested and user-
defined arbitrary structure (multimedia, for example) NoSQL warehouses are
preferable. Thus, the usage of the object model is recommended for distributed
databases with a large number of complex relationships: cross-reference links, many-
to-many relationships between objects.
370
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
A flexible object database can be a better choice for Configuration, but DB
Conditions, Component and Event Data tag databases with slowly changing structures
and simple relationship can use RDBMS for their implementation.
Appendix A
The Configuration DB will support only the data that is directly required to
start and support the DAQ (Data AcQuisition) system in an efficient operating
condition. This database should be elastic because not only data but the structures
of data can vary. This means that the Configuration DB must be developed in close
collaboration with the DAQ group.
Table 2. CMS databases structure
Database level Database purpose
Equipment management Holds structured data about all
database detectors parts as equipment elements.
Construction database Holds all information about relations
between different equipment elements.
Conditions database Holds all information about detector
conditions (data on operating conditions).
Configuration database Holds all information required to bring
the detector in any running mode.
Typical use cases of DBs for physical experiments [6].
371
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
372
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
The analysis of databases of physical experiments, such as ATLAS and CMS,
shows that they have similar use cases and close structure of corresponding databases.
Below the typical use cases for Configuration (Fig. 3), Conditions (Fig. 4),
Component and Event (tag) (Fig. 5), Geometry (Fig. 6) databases is presents [6-10].
Fig. 3. – Use cases for Configuration DB
373
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
Fig. 4. – Use cases for Condition DB
Fig. 5. – Use cases for Event DB
374
Information Technology and Nanotechnology (ITNT-2015)
Data Mining and Big Data Grechnikov F.V., Khaimovich A.I. Development of...
Fig. 6. – Use cases for Geometry DB
References
1. Materials Genome Initiative National Science and Technology Council Committee on
Technology Subcommittee on the Materials Genome Initiative
http://ssau.ru/files/science/conferences/itnt2015/itnt_2015_46.pdf
2. Ward CH, Warren JA, Hanisch RJ. Making materials science and engineering data
more valuable research products, 2014. http://www.immijournal.com/content/3/1/22
3. Austin T, Bullough C, Gagliardi D, Leal D, Loveday M. Prenormative research into
standard messaging formats for engineering materials data. Int J Dig Curation, 2013; 8: 5-
13. doi:10.2218/ijdc.v8i1.245.
4. Schmitz GJ, Prahl U. ICMEg, the integrated computational materials engineering expert
group a new European coordination action. Integr Mater Manuf Innov, 2014; 3:2.
doi:10.1186/2193.
5. Joint Institute for Nuclear Research, 2014. Annual Report, 2015; 14.
6. Akishina EP et al. Conceptual Considerations for CBM Databases. Communications of
the Joint Institute for Nuclear Research, 2014. E10-2014-103: 25.
7. Gallas EJ et al. An integrated overview of metadata in ATLAS. CHEP 2009 Conference,
Prague, Czech Republic, J March 23, 2009.
8. Miotto GL, Aleksandrov I, Amorim A, Avolio G et. al. Configuration and control of the
ATLAS trigger and data acquisition. Nuclear Instruments and Methods in Physics
Research Section, 2010, A, 623(1): 549:551.
9. Almeida J, Dobson M, Kazarov A et. al. The ATLAS DAQ system online configurations
database service challenge. J. Phys. Conf. Ser., 2008; 119(2): 022004.
10. Vaniachine AV, von der Schmitt JG. Development, deployment and operations of
ATLAS databases. J. Phys. Conf. Ser., 2008; 119(2): 072031.
375
Information Technology and Nanotechnology (ITNT-2015)