ESTECO Unveils modeFRONTIER ® 2014
A powerful, self-adapting optimization algorithm, enhanced data mining and database handling and three new direct integration nodes: the latest release of modeFRONTIER reaches new performance levels by streamlining the management of highly complex design scenarios.
ESTECO SpA, a leading provider of numerical optimization solutions for engineering design, has announced the release of modeFRONTIER ® 2014 - the latest version of its multi-disciplinary and multi-objective optimization and integration platform that comes with a range of innovative functionalities and improvements aimed at making life easier for designers.
modeFRONTIER 2014 introduces the powerful multi-purpose self-adapting algorithm - pilOPT - created by the ESTECO numerical team to offer a “one-click” optimization experience. By combining global and local search and balancing real and RSM-based optimization, pilOPT achieves good coverage of the Pareto front in a very short time. It requires the configuration of only one parameter – the number of design evaluations - and ensures substantial time savings through a broad exploration of the optimal design region and the efficient use of computational resources.
Powell’s method has been included in the optimization portfolio as a valid and efficient alternative to gradient-based algorithms, requiring considerably less computational effort due to its derivative-free approach. This single-objective optimization algorithm is particularly useful for calculating the local minimum of continuous complex functions - especially those without an underlying mathematical definition.
The Stepwise Regression RSM algorithm provides an automatic procedure for selecting the polynomial regression model with the optimal number and combination of predictor terms, yielding the lowest approximation error and highest prediction capability. The response surface is iteratively adapted to problem complexity until the best possible value of the selected quality criterion is achieved. The resulting polynomial is accurate, yet simple and requires no effort on behalf of the user when training and testing the fitness of different candidate models.
The new DOE Designer Node completes the set of native modeFRONTIER nodes, along with the SubProcess and Scheduling Project nodes, aimed at creating re-usable modular and discipline-specific workflows. The DOE Designer Node generates DOE configurations at runtime, allowing the user to expand the coverage of the design space and to focus on precise regions in order to fine-tune the exploration phase in accordance with problem characteristics.
On the integration side, three new direct nodes - NX CAE by Siemens, Solidworks Flow Simulation and PowerFLOW suite by EXA –developed to automate data extraction and simulation execution and fully exploit modeFRONTIER optimization capabilities.
Date: October 20, 2014