VAlidation driven DEvelopment of Modern and Efficient COMbustion technologies

Project Information

VADEMECOM

Grant agreement ID: 714605

DOI

10.3030/714605

Project closed

EC signature date 9 November 2016

Start date 1 April 2017

End date 31 March 2022

Funded under

• EXCELLENT SCIENCE - European Research Council (ERC)

Total cost

€ 1 499 110,00

EU contribution

€ 1 499 110,00

Coordinated by

UNIVERSITE LIBRE DE BRUXELLES
Belgium

Periodic Reporting for period 4 - VADEMECOM (VAlidation driven DEvelopment of Modern and Efficient COMbustion technologies)

Reporting period: 2021-10-01 to 2022-03-31

Summary of the context and overall objectives of the project

Combustion science will continue to play a major role in energy production, particularly for sectors where electrification is more challenging. Nevertheless, profound innovation is needed to develop combustion technologies ensuring fuel flexibility, high energy efficiency and very low pollutant emissions. To meet this goal, the combustion community needs to face a grand-challenge, namely the development of a validated, predictive, multi-scale modelling capability, to optimize the design and operation of evolving fuels in advanced technologies. While the use and value of turbulent combustion models continue to increase across combustion industries, current capabilities fall well short of what is needed in reliable design tools, especially when compared to the achievements in related fields such as solid mechanics and fluid dynamics. The objective of VADEMECOM is to drive the development of modern and efficient combustion technologies, by means of accurate and adaptive models, allowing a detailed description of the phenomena only where necessary. This is the key to transition from case-specific to generally applicable modelling approaches and to promote innovation in the energy sector. Indeed, combustion may appear as a mature technology, from the perspective of the old energy scenario. In today’s world, however, we must rethink the combustion field in the perspective of energy efficiency, fuel flexibility and environmental impact, to ensure future generations with affordable and sustainable energy and healthy environment.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

The significant achievements of VADEMECOM can be summarised as follows:
WP1. Experiments on the ULB semi-industrial MILD furnace were carried out to generate data that bridge the gap between the laboratory and the industrial scale. The investigation focused on identifying the critical features of such a combustion regime. It produced valuable validation experimental data (in-flame temperatures, flame chemiluminescence and chemical composition) to validate numerical simulation approaches. We focused on fuel flexibility and investigated the feasibility of MILD combustion with fuels and fuel mixtures representative of future energy carriers. In particular, methane-hydrogen and ammonia-hydrogen mixtures were targeted, in the presence of diluents, to investigate the potential and limitations of existing MILD combustion technologies and identify new technological developments. A particular focus was put on identifying measures to limit NO emissions in the presence of hydrogen- and ammonia-based mixtures.
WP2. The predictivity of available chemical mechanisms for hydrogen, methane and their mixtures was assessed at the conditions met in MILD combustion (relatively low-temperature increase and high dilution). This allowed identifying critical steps in existing chemical mechanisms for hydrogen, syngas, methane, ammonia, and mixtures. A novel strategy for optimising detailed chemical mechanisms was developed, integrating the uncertainty of kinetic parameters, a curve matching-based objective function and a methodology for optimising pressure-dependent reactions.
WP3. Several data-driven approaches were developed to reduce the number of transport equations to be solved in numerical simulations of combustion processes and to allow using optimally reduced chemical mechanisms in different regions of the flow, respectively. Both approaches showed great potential in reducing the computational cost while accurately accounting for finite-rate chemistry effects. Moreover, new turbulent combustion closures based on reactor concepts were developed and validated in MILD combustion for various fuels and systems in the context of both Reynolds-Averaged Navier Stokes and Large Eddy Simulations.
WP4. In the context of surrogate models, a novel approach combining size reduction, via Principal Component Analysis, with advanced regression methods was proposed to accurately predict the behaviours of combustion systems, characterised by high dimensionality, both in input and outputs. It was shown that the approach could deliver physics-based, reduced-order models thanks to the inclusion of physical constraints during the size reduction process. The first-of-its-kind digital twin of a combustion furnace working in MILD conditions was also developed, opening the way to accurate reduced-order models in the industry.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

We have made significant progress beyond the state of the art in several areas of the project:

- Novel data-driven approach for detailed kinetic mechanisms optimization. The approach is founded on a curve matching-based objective function and includes, for the first time, a methodology for the optimisation of pressure-dependent reactions via logarithmic interpolation (PLOG format).

- Data-driven approaches for the reduction of the computational burden associated to detailed chemical mechanisms. In particular, we have demonstrated the potential of unsupervised learning algorithm to identify the most relevant features of reacting systems and develop locally optimal approaches for the simulation of complex reacting flows.

- Turbulent combustion closures that allow to efficiently manage finite-rate chemistry and complex kinetic schemes. We have demonstrated the superior performances of these models with respect to the currently used ones, in the framework of novel combustion technologies, such as MILD combustion, where the interactions between chemical kinetics and turbulent mixing is of paramount important and sub-grid closure should account for that accurately.

- Nvel approaches for the development of accurate reduced-order models, that can be used for optimisation and uncertainty quantification with confidence and without the computational burden associated to full model. In particular, we have combined size reduction, via Principal Component Analysis, with advanced regression methods, to that allows to accurately predict the behaviours of combustion systems, characterised by high dimensionality, both in input and outputs. The first-of-its-kind digital twin of a combustion furnace has been demonstrated ans awarded the Distinguished paper award 2021 by the Combustion Institute.

In terms of perspectives, we maintain our overarching goal which consists in proposing a unified modelling approach that can predict the behaviour of advanced combustion technologies and be used, with confidence, in optimisation, new design and decision making. This implies:

- Demonstrating the feasibility of MILD combustion with a variety of fuels (hydrogen, ammonia, methane and their mixtures).
- Developing optimised and validated comprehensive chemical mechanisms for MILD combustion conditions and develop strategies for their reduction and inclusion in large scale simulation.
- Developing models that can include realistic chemistry in the simulation of combustion systems, combining sate-space parameterisation, mechanism reduction, machine learning, efficient chemistry management.
- Developing approaches to assess the confidence in the predictions from computational modelling, expanding our approaches for uncertainty quantification to realistic combustion systems.

Integration of experiments and simulation with machine learning, to develop data-driven approaches