The e-Foan iD project focuses on the development of sustainable drive systems for vehicles and mobile devices that are intended for special applications.

Sustainable drive systems include battery electric drives, fuel cells and hydrogen drives. However, their range and operating time is very limited. Special applications, which include mining, large construction sites and slope grooming, are characterised by the fact that they have high load requirements and long operating times and are often used in demanding environmental conditions. However, many special applications are characterised by a limited area of use, trained operating personnel and central coordination. The aforementioned boundary conditions are particularly advantageous for battery electric drives, as they can significantly minimise their biggest disadvantage, their limited range.

Based on these criteria, this project is developing a method for vehicle management that enables real-time range prediction and the coordination of individual vehicles. The aim of this project is to develop a simulation model for a single vehicle. The focus is on the conception of a vehicle that is primarily designed for battery-electric drive and for interaction with fleet management.

The vehicle model is based on common physical approaches to energy and heat management. It is a time-resolved, backward-computing and zero-dimensional model that functions in real time and is in constant data exchange with the real vehicle and fleet management (digital twin). The model presented below includes a representation of the entire vehicle, with a particular focus on the energy storage system for developing the operating strategy. The holistic approach is contrasted with a comparatively simple mapping of the components and subsystems. The following aspects are mapped in the model: Efficiency, storage level, performance of the electrical components, thermal management and HVAC, where required for energy management.

Due to the different subsoil conditions in the respective application areas (gravel, sand, snow, etc.), physical modelling for special applications in particular proves to be challenging. At the same time, however, precise data is available and the system as a whole is relatively clearly defined, which allows conclusions to be drawn about the use of data-based methods. With the help of machine learning, a data-driven model is to be developed that is able to fulfil these complex tasks embedded in the physical model of the overall vehicle (hybrid model architecture) and thus enable real-time predictions of performance requirements.