CELL4LIFE - Reversible SOCs as a link between electricity, heat and gas networks to increase the self-sufficiency and resilience of neighbourhoods

A system consisting of a solid oxide fuel cell and a Machine Learning-based control system for increasing efficiency and minimizing degradation is being developed. As a link between all energy supply networks, the system is intended to increase the self-sufficiency and resilience of plus-energy districts.

Short Description

Starting point / motivation

The use of reversibly operated high-temperature fuel cell CHP-modules (rSOC-CHP) enables in the fuel cell mode 

  • a highly efficient production of electricity and heat directly at the location of consumption and
  • also offers the possibility to locally produce and store green hydrogen if the system is operating in electrolysis mode.

Operating the rSOC (reversible solid oxide cell)-CHP (combined heat and power) module in the fuel cell mode enables the conversion of the chemical energy of a gaseous fuel (e.g. natural gas) and an oxidizer directly into electrical energy by only producing very low emissions. Heat and water are gained as by-products. The resulting heat can be used for hot water preparation and for room heating. The fed-in of surplus heat into the district heating grid is also an option.

In electrolysis mode, steam is converted into green hydrogen with the help of electricity generated from renewable energy sources and stored for later use. This operation is particularly favored by the low tariff periods resulting from the fluctuating generation of wind and photovoltaic systems.

The stored green hydrogen can be used in fuel cell mode as a completely emission-free alternative to natural gas. The project also aims to investigate the addition of hydrogen to natural gas (feed into the natural gas grid) or the partial substitution of fossil fuels with hydrogen and their influence on the behavior of the entire infrastructure.

Due to the simple scalability of the rSOC-CHP technology, not only the sustainable supply of single-family and multi-party houses, but also of larger office complexes up to entire city districts could be realized. In the project CELL4LIFE, this technology is to be examined primarily in the context of increasing the autarky rate and the resilience of plus-energy districts.

Contents and goals

Both reliability and durability must be increased to accelerate the commercialization of solid oxide fuel cells. A time-efficient and accurate prediction of system performance as a function of the operating environment could reduce the time required to find the operating optimum within a wide range of parameters. In order to predict the performance of the rSOC-CHP technology, a forecasting method based on a neural network is being developed in CELL4LIFE.

The three main goals of the project are:

  1. The optimal use of the existing infrastructure on the supplier (district heating, gas & electricity grid) and on the consumer side (single-family and multi-family houses, plus energy quarters).
  2. The optimized operation of the rSOC-CHP technology to increase the reliability and durability of the modules, taking into account the actual goals of increasing the self-sufficiency and resilience of plus-energy quarters.
  3. The demonstration of the system with a laboratory prototype.


The project pursues a control strategy of the system based on machine learning (ML). First, the multiphysics model (heat transfer, mass transport, chemical reaction, impulse transport, electrochemical reaction) is developed according to the experimental parameters and validated based on the experimental results. Parametric studies are then carried out to examine the effects of various parameters on the SOFC/SOEC.

In the meantime, a data set of inputs and outputs is generated from multiphysics models. The neural network (NN) is trained with this data set to map the relationships between the operating conditions and the performance parameters. The NN model is then used to generate new outputs and the results are compared with those of the multiphysics model to validate the accuracy.

Finally, the trained NN model is used to optimize the SOFC/SOEC (solid oxide electrolyzer cell/solid oxide fuel cell) through a combination with a genetic algorithm. The genetic algorithm seeks a solution in the form of binary digits that generate offspring through selection, crossing and mutation. These candidates are scored using a fitness function, with the iteration continuing until the convergence criterion is met. The genetic algorithm is suitable as a heuristic method for optimization problems, provided a suitable database is used.

Expected results

The overriding goal of CELL4LIFE is to develop economic operating and business models for the use of rSOC technology in positive energy districts and to conceptualize them accordingly so that the requirements for a PED can be met.

The following results are to be achieved:

  • The requirements for rSOC technology for use in PEDs are known. This primarily includes questions about the dynamic operation of the rSOC system in combination with the existing storage and renewables in PEDs and in the context of the higher-level supply systems.
  • The optimal mode of operation of the rSOC technology (low degradation, high efficiency) under the framework conditions of PEDs is known.
  • The control strategy should recognize degradation mechanisms early using artificial intelligence and adjust the operating parameters of the overall system accordingly. This regeneration technique is intended to reverse changes that have already occurred and thus increase the life of the cells or at least slow down or prevent degradation.
  • The existing infrastructure on the supply (district heating network, gas network) and on the consumer side should be used in the best possible way, so that the integration of the rSOC technology into PEDs can be as simple as possible.
  • The substitution of natural gas with hydrogen for decentralized feed-in is technically, legally and economically possible.
  • Demonstration of the system in the form of a laboratory prototype.

The results of this project are intended to minimize the laboratory effort when examining various application scenarios for rSOCs.

Project Partners

Project management

4ward Energy Research GmbH

Project or cooperation partners

  • Institut für Wärmetechnik, Technische Universität Graz
  • Kristl, Seibt & Co. Gesellschaft m.b.H.

Contact Address

4ward Energy Research GmbH
Reininghausstraße 13A
A-8020 Graz
Tel.: +43 (664) 882 518 30
E-mail: markus.rabensteiner@4wardenergy.at
Web: www.4wardenergy.at