Innovative Modelling Approaches for the Design, Operation and Control of Complex Energy Systems with Application to Underground Infrastructures

The ventilations systems play a key role in underground infrastructures for health and safety of occupants during normal operation as well as during accidents. Their performances are affected by selection of the optimal design, operation and control that is investigated by predicting air flow. The calculation of ventilation flows and their interaction with fires can be done with different modelling approaches that differ in the accuracy and in the required resources. The 3D computational fluid dynamics (CFD) tools approximate the flow behaviour with a great accuracy but they require high computational resources. The one dimensional (1D) models allow a compact description of the system with a low computational time but they are unsuitable to simulate thermal fluid-dynamic scenarios characterized by turbulence and gradients. Innovative tools are necessary in order to make the analysis and optimization of these systems possible and accurate in a reasonable time. This can be achieved both with appropriate numerical approaches to the full domain as the model order reduction techniques and with the domain decompositions methods as the multiscale physical decomposition technique. The reduced order mode techniques as the proper orthogonal decomposition (POD) is based on the snapshots method provides an optimal linear basis for the reconstruction of multidimensional data. This technique has been applied to non-dimensional equations in order to produce a reduced model not depending on the geometry, source terms, boundary conditions and initial conditions. This type of modelling is adapted to the optimization strategies of the design and operation allowing to explore several configuration in reduced times, and for the real time simulation in the control algorithms. The physical decomposition achieved through multiscale approaches uses the accuracy of the CFD code in the near field e.g. the region close to the fire source, and takes advantage of the low computational cost of the 1-D model in the region where gradients in the transversal direction are negligible. In last years, the multiscale approach has been proposed for the analysis of tunnel ventilation. Among the several CFD codes used in this field, the Fire Dynamic Simulator (FDS) is suitable for the multiscale modelling. This is an open source CFD package developed by NIST and VTT and presents the HVAC routine in which the conservation equations of mass, energy and momentum are implemented. Currently, the HVAC module does not allow one to consider heat and mass transfer, which significanltly limits the applications. For these reasons a multiscale simulator has been created through the fully integration of a 1D continuity, momentum, energy and mass transport equation in FDS modifying its source codes. The multiscale simulator thus obtained, is based on a direct coupling by means of a Dirichlet-Neumann strategy. At each 1-D-CFD interface, the exchange flow information occurs prescribing thermo-fluid dynamic boundary conditions. The 1-D mass transport equation computes the diffusion of the exhaust gas from the CFD domain and the relative concentration that is particularly interesting in the case of back layering of smoke. The global convergence of the boundary conditions at each 1-D-CFD interface has been analyzed by monitoring the evolution of thermo-fluid dynamic variables (temperature, velocity, pressure and concentration. The multiscale simulator is suitable for parametric and sensitivity studies of the design and the operation ventilation and fire safety systems. This new tool will be available for all the scientific community. In this thesis, Chapter 1 provides a general introduction to the role of the system ventilation in underground infrastructures and to the innovative modelling strategies proposed for these systems. Chapter 2 offers a description of the 1D network modelling, its fluid-dynamic application to the Frejus tunnel and its thermal application to ground heat exchangers. In Chapter 3, the proper orthogonal decomposition method is presented and its application to the optimal control of the sanitary ventilation for the Padornelo Tunnel is discussed. To demonstrate the applicability of POD method in other fields, boreholes thermal energy storage systems have been considered in same chapter. In particular, a multi-objective optimization strategy is applied to investigate the optimal design of these system and an optimization algorithm for the operation is proposed. Chapter 4 describes the multiscale approach and the relative simulator. The new open tool is used for modeling the ventilation system of the Monte Cuneo road tunnel in case of fire. Results show that in the case of the current configuration of the ventilation system, depending on the atmospheric conditions at portals, smoke might not be fully confined. Significant improvements in terms of safety conditions can be achieved through increase of in smoke extraction, which requires the installation of large dumpers and of deflectors on the jet fans. The developed tool shows to be particularly effective in such analysis, also concerning the evaluation of local conditions for people evacuation and fire-brigades operation.