Electrical energy generating (fuel cell) or energy storage devices (batteries) are commonly used to provide the propulsion energy required in electrical vehicles. The batteries are also commonly used for portable electronics and are growing in popularity for military and aerospace applications.
Design of these energy generating and storing components in terms of efficiency, lifetime, cost, size, weight, and also fuel economy is a complex task. It is challenging due to the multitude of physical phenomena that need to be simultaneously optimized in order to achieve proper fuel cell operation, the multitude of design objectives that need to be optimized, and the large amount of computational resources necessary to solve the governing equations of a fuel cell.
Multidisciplinary automation and multi-objective optimization
Battery and fuel cells design process might require multiple software solutions covering all involved disciplines including chemistry & materials, fluid dynamic, electronics, thermal, costing etc. Therefore, engineers need generic methods able to adapt to any software and data environment. pSeven provides the advanced automation & integration capabilities to integrate any conventional CAD/CAE software in an optimization workflow to ensure that the tools used by different disciplines can share data efficiently, which is necessary for the efficient multi-Objective optimization.
Design strategy is usually based on multilevel analysis: fuel cell, or single component level, and then system level involving all disciplines and components. The behavior of the component can be captured in predictive models which can subsequently be integrated at a system level. Predictive modeling techniques based on Machine Learning and AI become a key enabler, because they produce fast mathematical representation of a component behavior. Such techniques must be efficient to produce good quality model in a potentially challenging, hence expensive nonlinear design space. High cost is related to the heavy simulations in the high dimensional space or to the expensive physical experiments on the testbed.
Adaptive DoE
To generate samples on which further predictive models are built, an advanced Design of Experiments technique, called Adaptive DoE, is available in pSeven. DoE techniques allow to select inputs at which outputs of the model are measured to explore design space or to get as much information as possible about the model behavior using a small number of observations. Adaptive DoE considers model behavior before adding new points and takes into account linear and non-linear constraints of the model.
Adaptive DoE allows building of accurate predictive model for a minimal number of simulation or physical experiment.
Predictive models
Once the adaptive DoE has been performed, a predictive model can be built using advanced Machine Learning and Artificial Intelligence algorithms available in pSeven. Such algorithms are based on the proprietary advanced technique and made affordable to any engineer thanks to a special abstraction layer called “SmartSelection”. That layer will free engineers from selection and tuning of algorithms, resulting in the development cycle reduction.
pSeven can build predictive model even by combining data samples from different sources thanks to the embedded Data Fusion (DF) technique. DF can in ex. leverage data sample coming from High Fidelity simulation, with Low Fidelity simulation and with experiments datasets. So, all data sources representing the component behavior, whatever fidelity level, are leveraged.
System Simulation
When predictive models are ready, they can be assembled at a higher level for System Simulation. The assembly of models can be achieved within pSeven in a native format, or models can be exported in a neutral format, like: FMI, C / C#, Excel, Matlab-Octave, for multidisciplinary analysis outside of pSeven using Systems Engineering frameworks.