Context and objectives
In a context of sustainable development, clean energy is strongly promoted in the European energy mix. Among the various solar energy technologies, Concentrating Solar Power (CSP) will play a key role in the future: its share of global electricity is envisioned to reach 11% in 2050 . The main drawback of CSP technologies is their costs, that drop slower than those of photovoltaics. This PhD is part of the SFERA-III project (see https: //sfera3.sollab.eu), funded by the European Union’s Horizon 2020 research and innovation programme. Two main topics will be addressed during this PhD: (a) the short-term (up to 30 min) forecast of direct normal irradiance (DNI), which can be defined as the direct irradiance received on a plane normal to the Sun, using a sky imager, and (b) the model-based predictive control (MPC) of a solar fuel reactor using the aforementioned forecasts.
Users of solar concentrating research infrastructures (RIs) are hosted for a short period of time to carry out their experiments. That is why an efficient use of the solar resource during this period is critical for obtaining valuable and exploitable experimental results. One way to increase the useful on-sun experimental time available to RI users during their stay is to provide them accurate intrahour forecasts of DNI, both before the experiments and, through a continuous update, while they are being made.
Because solar energy is inherently variable and intermittent, dynamic control and automation tools are needed to ensure continuous processes. In this sense, the SFERA-III project adresses, among other things, the developement of a MPC-based controller dedicated to solar fuel production under varying solar conditions. Such a tool has to offer highly-valuable know-how and software assets for the further development of solar fuel production technologies, for both near- and long-term concerns.
Short-term forecast of DNI
DNI is the only component of solar irradiance that is truly intermittent, since it can vary from its maximum value to zero in seconds. Because DNI is affected by changeable factors, such as position, optical depth and speed of clouds, it is extremely difficult to forecast accurately. This task cannot be achieved using satellite imagery or numerical weather predictions, due to their limited spatial and temporal resolutions . The use of sky imagers has thus been emerging in order to capture the high spatial and temporal variability of cloud cover. In the last few years several studies on DNI forecasting using sky imagers have been made (see e.g. [3, 4, 5, 6, 7, 8, 9]). However, to accurately apprehend the high variability of DNI, the results obtained so far still need to be improved.
DNI, hereafter denoted as I, can be divided into two multiplicative terms, I = Ics ⋅ kc, where Ics is the clear-sky DNI and kc is the clear-sky index. Ics corresponds to the direct normal irradiance obtained when no cloud is occulting the Sun. The clear-sky index kc reflects the attenuation of Ics due to clouds: it spans from 0 (when a thick cloud is occulting the Sun) to 1 (when there is no cloud in front of the Sun). DNI forecasting can thus be achieved by forecasting Ics and kc.
To this end, a sky imager providing high-quality high dynamic range images has been developed and installed at the PROMES-CNRS laboratory (see Figure 1). Various models and algorithms have also been derived in order to (a) forecast in real time the atmospheric turbidity and the clear-sky DNI, (b) detect the clouds and estimate their motion, and finally (c) forecast the DNI in vicinity of the sky imager [6, 7, 8, 9, 10,11, 12, 13, 14]. These results need to be improved, especially regarding the estimation of cloud motion and clouds’ optical depth. To this end, the PhD candidate will use a database of sky images and various measurements (global horizontal irradiance, direct normal irradiance, diffuse horizontal irradiance, temperature, etc.) to improve existing algorithms and create new developments.
Predictive control of a solar fuel reactor
The combination of biomass and solar energy in a solar-driven thermochemical gasification process is of particular interest to convert solid carbonaceous materials to gaseous carbon neutral solar fuels , consisting primarily of CO and H2 commonly called “synthesis gas” , thereby offering an efficient means of storing intermittent solar energy.
An example of reactor where solar gasification of carbonaceous feedstock takes place  is shown in Figure 2.
Several research institutes involved in the SFERA-III project have shown interest in solar fuel reactors. In France, the LITEN laboratory of CEA (the French Alternative Energies and Atomic Energy Comission) is collaborating with the PROMES-CNRS laboratory on this scientific topic [19, 20], thanks to the reactor developped in Odeillo, at the 1 MWth solar furnace. In Almería (Spain), the CIEMAT laboratory has also developed a solar fuel reactor . There are a number of strategies to tackle the variable and intermittent nature of solar energy, including heliostat control systems to
regulate input power, or changing reactor operation to adapt to varying input power. In case of solar gasification processes, it is sensible to investigate hybridization to ensure continuous production. Gasification or reforming processes are especially suited to hybridization, because
they can operate in allothemal (solar heat) and/or autothermal (partial combustion of feedstocks) mode. To fully realize the potential of such strategies, the development of autonomous control systems is required. Some works have already been completed on the dynamic control of solar fuel reactors , but they focus on PI control and robust control, so they do not take into account the predictability of DNI.
The PhD candidate will be in charge of the development and application of dynamic control and automation tools for solar fuel reactors and heliostat fields working in tandem under varying solar conditions. Based on short-term forecasts of DNI, a MPC-based controller  is likely to improve the operation (efficiency) of solar fuel reactors. The dynamic model of the considered reactor, needed for implementing such a control strategy, will be conjointly developed by PROMES-CNRS and CEA using data coming from experimental campaigns as well as the most recent state-of-the-art in this field.
Prerequisite knowledge and skills
In order to apply, the candidate should have the following:
• Academic degree: Master’s degree or equivalent
• Scientific skills: signal and image processing, model-based predictive control, optimization, system identification, time series analysis and forecasting, artificial intelligence (machine learning, deep learning)
• Software skills: Matlab, oriented-object programming (C++, Python)
• Language skills: English and French (oral and written)
PROMES-CNRS (see http://www.promes.cnrs.fr) is a research unit belonging to the Institute for Engineering and Systems Sciences (INSIS). The laboratory is under contract with the University of Perpignan Via Domitia (UPVD). It is located in three sites: Odeillo-Font Romeu (1 MWth solar furnace), Targassonne (5 MWth Thémis solar tower power plant), and Perpignan-Tecnosud. PROMES-CNRS leads the laboratory of excellence SOLSTICE (Solar Energy: Science, Technology and Innovation for Energy Conversion) and has a high level of expertise in all aspects of solar engineering. PROMES-CNRS is composed of about 170 people, partitioned into 8 research groups. The main research fields are “materials under extreme conditions” and “conversion, storage, and transport of solar energy”.
The PhD will start in September or October 2019, with a remuneration of about 1700 € per month. The PhD candidate will be located in Perpignan-Tecnosud, at PROMES-CNRS, and will work under the supervision of:
• Julien Eynard (Associate Professor, )
• Stéphane Thil (Associate Professor, )
• Stéphane Grieu (Full Professor, )
 IEA. Technology Roadmap: Solar Thermal Electricity – 2014 edition. International Energy Agency, 2014.
 R. Chauvin. “Évaluation de la ressource solaire pour la gestion optimisée de centrales CSP”. PhD thesis.
Université de Perpignan Via Domitia, Apr. 2016.
 S. Quesada-Ruiz et al. “Cloud-tracking methodology for intra-hour DNI forecasting”. In: Solar Energy 102 (2014), pp. 267–275.
 D. Bernecker et al. “Continuous short-term irradiance forecasts using sky images”. In: Solar Energy 110 (Dec. 2014), pp. 303–315.
 Y. Chu et al. “Real-time forecasting of solar irradiance ramps with smart image processing”. In: Solar Energy114 (2015), pp. 91–104.
 R. Chauvin et al. “Short-term DNI Forecasting Using a Sky Imager”. In: 22nd International Conference on Concentrating Solar Power and Chemical Energy Systems (SolarPACES 2016). Abu Dhabi, United Arab Emirates, Oct. 2016.
 R. Chauvin et al. “Generating high dynamic range images using a sky imager”. In: 20th IFAC World Congress on Automatic Control (World IFAC 2017). Toulouse, France, July 2017.
 R. Chauvin et al. “A new approach to the real-time assessment and intraday forecasting of clear-sky direct normal irradiance”. In: Solar Energy 167 (June 2018), pp. 35–51.
 J. Nou et al. “Towards the intrahour forecasting of direct normal irradiance using sky-imaging data”. In:Heliyon 4.4 (2018), e00598.
 J. Nou et al. “Atmospheric Turbidity Forecasting using Side-by-side ANFIS”. In: Energy Procedia 49 (2014), pp. 2387–2397.
 R. Chauvin et al. “Cloud detection methodology based on a sky-imaging system”. In: Energy Procedia 69 (2015), pp. 1970–1980.
 R. Chauvin et al. “Modelling the clear-sky intensity distribution using a sky imager”. In: Solar Energy 119 (2015), pp. 1–17.
 J. Nou et al. “Clear-sky Irradiance Model for Real-time Sky Imager Application”. In: Energy Procedia 69 (2015), pp. 1999–2008.
 J. Nou et al. “A new approach to the real-time assessment of the clear-sky direct normal irradiance”. In:Applied Mathematical Modelling 40.15-16 (Aug. 2016), pp. 7245–7264.
 R. Taylor, R. Berjoan, and J. Coutures. “Solar gasification of carbonaceous materials”. In: Solar Energy 30.6 (1983), pp. 513–525.
 D. Gregg et al. “Solar coal gasification”. In: Solar Energy 24.3 (1980), pp. 313–321.
 M. Puig-Arnavat et al. “State of the art on reactor designs for solar gasification of carbonaceous feedstock”. In: Solar Energy 97 (2013), pp. 67–84.
 A. Muroyama et al. “Modeling of a Dynamically-Controlled Hybrid Solar/Autothermal Steam Gasification Reactor”. In: Energy Fuels 28.10 (Sept. 2014), pp. 6520–6530.
 H. Boujjat et al. “Simulation of biomass gasification in a novel high-temperature directly irradiated hybrid solar/combustion reactor”. In: 24th International Conference on Concentrating Solar Power and Chemical Energy Systems (SolarPACES 2018). Casablanca, Morocco, Oct. 2018.
 S. Chuayboon, S. Abanades, and S. Rodat. “Insights into the influence of biomass feedstock type, particle size and feeding rate on thermochemical performances of a continuous solar gasification reactor”. In: Renewable Energy 130 (2019), pp. 360–370.
 H. Villafán-Vidales et al. “An overview of the solar thermochemical processes for hydrogen and syngas production: Reactors, and facilities”. In: Renewable and Sustainable Energy Reviews 75 (2017), pp. 894–908.
 E. F. Camacho and C. B. Alba. Model Predictive Control. Second edition. Springer-Verlag London, 2007.