APC 2015 Annual Report Download - page 31

2015 REGISTRATION DOCUMENT SCHNEIDER ELECTRIC 29
OVERVIEW OF THE GROUP’S STRATEGY, MARKETS AND BUSINESSES
1
RESEARCH & DEVELOPMENT
fi nd its way to the person needing it most, depending on where he/
she is. Once advised, another set of mobile based applications,
including augmented reality, further guide the user to transform
these insights into action.
By combining three platforms : the Smart Sensor Platform, the
Embedded Control Platform and the Digital Services Platform
Schneider Electrics provide a consistent and interoperable value
chain starting from data acquisition, continuing with data processing
and transmission and up to data storage and analytics to present
actionable dashboards to our customers.
Optimization and Analytics
In 2015, major evolutions occurred in Schneider Electric in the
Optimization and Analytics domain: exchanges within the Group
enabled a good characterization of the potential for future connected
offers and related analytics, as well as a complementary view of
relevant analytics for non-connected (or less connected) offers; the
Digital Services Platform (DSP) reached a level of maturity which
enables its use in applications managing signifi cant amounts of
data and integrating analytics.
Technology anticipation addressed different types of actions:
•Exploration of analytics for connected offers in a variety of
contexts: homes, senior residences, commercial buildings, and
industrial plants. In most cases, the exploration of new use cases
builds on the premise that the analytics will be implemented
on top of the Digital Services Platform and will enable (i) better
understanding of correlations between data and (ii) use of
understanding to improve operational and/or investment
decisions. In some cases, this exploration led to a prototype
of a brick that could be reused for other applications. External
collaborations has been important, e.g., the Tribute European
project, a PhD partnership with the University of Grenoble Alpes,
on machine learning for virtual sensors, and joint work with Duke
University on the use of a machine learning algorithm to optimize
temperature control.
•Investigation of condition monitoring and diagnosis
analytics for asset performance management, often but not
necessarily, in the context of a connected offer. Both data-
oriented (machine learning) models and explicit (more or less
complex) physical models can be used for this purpose, inducing
very different constraints on the data requirements and global
solution architecture. The analytics are used to improve the
reliability, availability, maintainability and safety of devices and
systems. Important partners in 2015 included Uppsala University
(IT Business), and the University of Grenoble Alpes.
•Analytics for planning and control for electrical networks
(Infrastructure Business), HVAC in buildings and data centers
(Buildings & Partner and IT Businesses) or industrial systems such
as mines, cement plants, water networks, pipelines, refi neries,
food and beverage plants (Industry Businesses). Following the
acquisitions of the previous years, Schneider Electric already has
a signifi cant offer in this domain, but the frontiers continuously
move as technology progresses and with the increasing
availability of more and more data. In particular, progress on
energy optimization in residential districts and in manufacturing
plants has been enabled in the context of the Ambassador, Hyllie
Smart Grid, and Arrowhead European cooperative projects, as
well as with two PhD partnerships with the GIPSA and LIRMM
laboratories. For electrical networks, our partnership with Mines
ParisTech has been complemented by a new PhD collaboration
with G2ELab and INRIA.
Modeling and simulation
Regarding lifetime cost, the design phase of industrial projects plays
a critical role in reducing both time and cost of system deployment.
Invensys, acquired by Schneider Electric in January 2014, is a
leader in the simulation of complex continuous processes, such as
refi neries or chemical plants . During the design phase, the plant
can be simulated to optimize its design, validate its performance
and start operator training before it is even built. The IT division
has developed a similar set of tools for data centers, including 3D
thermal simulations to validate the design of the cooling system.
R&D teams are working to generalize this offering to any kind
of industrial system, including large and complex buildings like
hospitals, in partnership with the leading CAD/CAE suppliers in
these domains. Filling the gap between design and operational
systems will not only decrease design-and-build costs but also
those linked to maintaining and developing systems over the 30+
years’ life expectancy that is common in some industries.
Within fi ve years, one can expect that industrial systems will be
developed like modern software, starting from a model of the
process, followed by a simulation based on this model, developed
and tested « against » the model and fi nally deployed on totally
standardized hardware.