Citation: Procacci, A.; Cafiero, M.; Sharma, S.; Kamal , M.M.; Coussement, A.; Parente, A. Digital Twin for Experimental Data Fusion Applied to a Semi-Industrial Furnace Fed with H 2 -Rich Fuel Mixtures. Energies 2023, 16, 662. https:// doi.org/10.3390/en16020662 Academic Editor: Giancarlo Sorrentino Received: 30 November 2022 Revised: 21 December 2022 Accepted: 29 December 2022 Published: 5 January 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). energies Article Digital Twin for Experimental Data Fusion Applied to a Semi-Industrial Furnace Fed with H 2 -Rich Fuel Mixtures Alberto Procacci 1,2, * , Marianna Cafiero 1,2,3 , Saurabh Sharma 1,2 , Muhammad Mustafa Kamal 1,2 , Axel Coussement 1,2 and Alessandro Parente 1,2 1 Aero-Thermo-Mechanics Laboratory, École Polytechnique de Bruxelles, Université Libre de Bruxelles, 1050 Brussels, Belgium 2 Brussels Institute for Thermal-Fluid Systems and Clean Energy (BRITE), Université Libre de Bruxelles and Vrije Universiteit Brussel, 1050 Brussels, Belgium 3 Institute of Mechanics, Materials, and Civil Engineering, Université Catholique de Louvain, Place du Levant 2, 1348 Louvain-la-Neuve, Belgium * Correspondence: alberto.procacci@ulb.be Abstract: The objective of this work is to build a Digital Twin of a semi-industrial furnace using Gaussian Process Regression coupled with dimensionality reduction via Proper Orthogonal De- composition. The Digital Twin is capable of integrating different sources of information, such as temperature, chemiluminescence intensity and species concentration at the outlet. The parameters selected to build the design space are the equivalence ratio and the benzene concentration in the fuel stream. The fuel consists of a H 2 /CH 4 /CO blend, doped with a progressive addition of C 6 H 6 . It is an H 2 -rich fuel mixture, representing a surrogate of a more complex Coke Oven Gas industrial mixture. The experimental measurements include the flame temperature distribution, measured on a 6 × 8 grid using an air-cooled suction pyrometer, spatially resolved chemiluminescence measurements of OH ∗ and CH ∗ , and the species concentration (i.e., NO, NO 2 , CO, H 2 O, CO 2 ,O 2 ) measured in the exhaust gases. The GPR-based Digital Twin approach has already been successfully applied on numerical datasets coming from CFD simulations. In this work, we demonstrate that the same approach can be applied on heterogeneous datasets, obtained from experimental measurements. Keywords: digital twin; data fusion; dimensionality reduction 1. Introduction The need for the rapid decarbonization of the global economy requires the develop- ment of new combustion systems that are both efficient and flexible, to allow the use of new zero-carbon fuels such as hydrogen and ammonia [1]. These requirements, coupled with the necessity of limiting the production of harmful pollutants, impose strict operating conditions on the combustion systems. This means that the design and control of these systems is crucial, and the margin of error is limited. For this reason, the model of trial-and-error employed for the design of a traditional combustion system is both too time-consuming and error prone to be applied on the development of new combustion designs. Luckily, the recent developments in machine-learning techniques and the increasing availability of data offer various tools that can be exploited in the design and operation of combustors. In particular, the development of Digital Twins (DTs) has been increasingly regarded as a way to substantially improve both the knowledge of industrial systems and their control [2–4]. The DT is defined as a digital representation of a physical object that can closely simulate its behavior in the real environment [2,4]. The DT can help in the design, production and service phases of a product. In the design phase, the DT is useful in the iterative optimization and the virtual evaluation, while in the production phase it can be Energies 2023, 16, 662. https://doi.org/10.3390/en16020662 https://www.mdpi.com/journal/energies