A sustainable and integrated computational pipeline to win complexity
by Gianluigi Rozza and Andrea Martini
Sustainability is nowadays a crucial topic related to several research and development areas and it gets a main role to inspire new innovative technologies, digitals and not, to ensure an economic growth both in the short and long term. Computational sciences play a fundamental role in this context and the current aim is to combine the use of data science with numerical analysis tools, to achieve a reliable numerical simulation within complex systems. This approach moves itself also from a sustainable point of view, regarding the computational costs related to simulations.
High performance computing, especially related to the industrial environment, is increasing and this means an exponential raise of time and computing allocations, and even more energy consumption. Supercomputing become so an essential tool to allow performances and quickness to guarantee the increasing computational demand. Computing centres, with a peta flop power, are spreading around but it means also a high energy power consumption in terms of Mega Watt(s).
A possibility that allows this perspective to be more sustainable, especially in terms of energy consumption, is the development of reduced order models, which merged to modern algorithms based on augmented intelligence and data science, can open doors to new perspectives, while reducing the number of operations, memory allocation, and computational time. This approach contributes to export complex computational environments to portable devices, like laptop, tablet and even smartphone, thus enabling real-time computing.
It works on pre-computed solutions database by separating computational steps in offline (supercomputer) and online (laptop-tablet) ones: in this way, thanks to model order reduction, new innovative computational technology could contribute in exporting numerical simulation in fields with currently limited exploitation. Thus, we can imagine which innovation this approach can bring within crucial sectors as biomedical, energy and logistics ones, for example.
In this era, characterised by an increasing of data availability and their usage, our goal is to move to big models which represent another important key to guarantee a fitted integration in complex systems framework, by considering a sustainable environment in terms of computational efforts.
So, the direction to follow is the one in which big data are often no more enough: we need to develop big models which allow to refine mathematical models, integrate complex systems, quantify uncertainty to reduce the gap between reality and models, thus boosting product and process innovation, as well as integration between complex systems.
In this scenario, the digital twins are ready to play a big role as a disruptive and innovative tool: in the building phase they give the possibility to simulate unlimited versions of the same prototype by saving time and realisation costs.
Moreover, they allow a continuous information sharing process both of historical data and actual data, to enhance the monitoring of the entire life-cycle of a product or a process.
So, digital twins represent big models which match model order reduction, IoT, augmented intelligence, virtual reality, data, by providing useful economic and sustainable benefits. In future they will open doors to astonishing innovations as the human digital twin: let us think about early diagnosis and medical actions customised for one single patient thanks to the integrated simulations pipeline. Even if it is a complex system, the digital twin represents the new goal to achieve medical successes, better welfare and quality of life for everyone.
Researchers at SISSA mathLab are collaborating in the creation of the Live Demo Odyssea for digital twin(s) developments in the framework of SMACT national competence center for Industry 4.0 in the North-Est of Italy (https://www.smact.cc/) thanks to important industrial cooperation with companies like Electrolux, Wartsila, Danieli.
