Systems’ optimization and data analysis

Maximising systems’ global performance thanks to acknowledgement of multi-physics and inter-components interactions, optimising business responsiveness thanks to flexible and agile processes : knowledge of technical and economic trade-offs makes strategy safer and frees innovation.

System optimization: synergy of skills in the sake of overall performance

How to quickly assess the overall impact of a system’s parameter change, implying to involve experts of every physics on every subs-system? How to make development agile to keep up with changes in requirements and uncertainties along the system lifecycle?

Bertin Technologies supports industry in the search of design and manufacturing optimal solution.

On one hand every system is subject to  various physics such as aerodynamics, heat transfer and structural strength, on the other hand it must target goals such as maximizing performance, controlling risk, minimizing cost, often conflicting.

The stake is to ensure to make the right decisions in the arbitrage of significant parameters for the sole benefit of the comprehensive system, and not of a part of it.

Schema_plateforme_optimisation_0.jpg

Illustration Principle of model-based optimisation

A unifying and effective approach

Bertin developed a methodology to connect the different behavioral models of affecting disciplines and leverage optimization techniques suitable to objectives and constraints specific to each system. Assessment campaigns (usually digital) at the whole system level can be carried out on wide parameters variation ranges; besides quantifying multidisciplinary trade-offs and helping in their understanding, the information coming from these campaigns provide a database reusable thanks to analysis techniques in order to prioritize significant factors, to fit predictive models, to find correlations, etc.

The client gains a key benefit at the earliest automation stages (models coupling, chaining): the various experts spend less time to integrate system evolutions to their specialty, for the benefit of reflection and creativity in interaction with the project.

Schema_boucle_MSO_0.jpg

Illustration Main steps of system-level Modelling, Simulation, Optimisation methodology

State-of-the-art technologies to innovate and capitalise

Bertin monitors the state-of-the-art and invests in R&D of skills and technologies to support systems’ optimization, such as:

  • Model reduction, in order to drastically shorten simulation budgets while preserving essential characteristics of the model
  • Uncertainties quantification, in order to consider results’ accurateness in decision-making
  • Open Source components integration, in order to take advantage of existing community developments and validations.

Optimisation_chauffage_port_arctique.jpg

Illustration Sizing of heating system ensuring ship access in Arctic by means of Multiphysics modelling (thermal and fluid)

System optimization implies speeding (up to near real time) overall simulation process, also critical to:

  • Any multi-query approach: parametric, statistical, stochastic studies
  • Real time applications (digital twin) useful for training, user interfaces design/test, operation diagnosis
  • The use of light platforms like industrial controllers, web-applications, mobile phones.

Optimisation_PriseAir_Formule1_v2.jpg

Illustration Optimization of critical component of Formula 1 engine

A motivated team willing to take on challenges

Bertin gathers engineers and PhDs with complementary skills in physics, mathematics and computer science, used to working together and attentive to clients. Our team focuses on value to the customer, when selecting and developing the most relevant commercial, open source or home-made software components.

Bertin has its own resources exclusively devoted to its clients computations.

Bertin thus gets the best value out of entrusted concerns, whether in terms of:

  • analysis:  elements for decision-making
  • process (in design, validation):  changes in practices.

Traceability of assessed configurations and choices made participate in projects quality.

Couplage_HF.jpg

Illustration Combustion/acoustics HF coupling

Consistently harnessing results for everyone's profit

Techniques for data analysis then data visualization are all the more powerful as used in close link with physical phenomena:

  • in order to envisage design: market research, technological survey, competitive analysis
  • in order to plan/exploit simulations and tests: experiments designing, system understanding, screening leading parameters, classifying operational points or configurations, ...
  • in order to process results and present findings: correlation analysis, sensitivity to the uncertainties.

Compo_exploitation_données.jpg

Illustration Data processing examples

Examples of optimization activities combining physical meaning with mathematical/numerical techniques

Global optimization of turbulent wake interactions between wind turbines within a farm

Increasing complexity and reliability requirements for systems force to include in design processes increasingly expensive modeling, whether in computation time or capacity. To enhance performance and responsiveness throughout the design cycle, Bertin has committed alongside partners in a project of complex phenomena model reduction: MECASIF project.

Compo_eolienne_MECASIF.jpg

Illustration Wind turbine modelled within MECASIF

This approach has been used on behalf of a wind turbine manufacturer. The purpose was to achieve an aerodynamic turbulent multi-scale modeling of the wake downstream each wind turbine, in order to optimize interactions with its neighboring turbines, and thus to maximize power generation.

Bertin developed a novel multi-level reduction strategy which makes possible overall optimization of a wind-farm by combining:

  1. a physical basis/trade expertise: factor 10 between reduced-order model and full-order model
  2. a mathematical basis: factor 50.000 between reduced-order model and full-order model.

Reduction_modele_eolienne.jpg

Illustration Multi-level model reduction strategy

Weather data synthesis for decision support

Within the framework of techno-operational experiments of decision-support tools, Bertin suggested and investigated the feasibility of methodological guidelines in synthetizing meteorological data, towards their operational use (reduced computing resources and time-limit) to compute their impact on the weapons systems: Principal Component Analysis and Clustering.

Compo_synthese_données.jpg

Illustration Impact of data synthesis and operational results

Contact us

For more information, contact us!

* Required fields