Modelling and identification of anaerobic digestion processes
The current state-of-the-art model for describing anaerobic digestion processes is the Anaerobic Digestion Model No. 1 (ADM1). It includes 19 physical/chemical/biological processes which occur in series and in parallel, more than 30 state variables and about 60 parameters. Indeed, the wide range of variability of a large number of stoichiometric and kinetic parameters, requires the estimation of the most sensitive ones for accurate model predictive ability. The purpose of this thesis is the calibration and validation of the ADM1 using data collected from pilot- and full-scale digesters fed with municipal waste sludge and expired food from dairy industries. Two/three series of data available from previous operation and monitoring of digesters will be analyzed and used for parameters identification and model validation. Furthermore, an iterative procedure exploiting data from batch laboratory-scale tests will be automated and applied in view of optimal model calibration.
LFT modelling and identification of BMP tests
The Biochemical Methane Potential (BMP) test is an essential tool to derive practical knowledge for optimizing and operating full-scale anaerobic digesters, monitoring, modelling, and evaluating process performances, or when a scenario analysis is being developed. Despite its usefulness is undoubted, long duration of BMP tests is problematic for many of its applications, especially when prompt results to take decisions are required. Over the last decades, several scientific contributions demonstrated that a reduction of BMP test duration is possible by predicting the final gas production. The purpose of this thesis work is the development of a new procedure/algorithm to obtain a preliminary estimation of the experimental BMP result by using a linear fractional transformation (LFT) formulation of the original nonlinear model for the identification of its parameters. The efficacy and the efficiency of the algorithm developed will be verified using experimental data of BMP tests performed on different substrates commonly fed to anaerobic digesters.
Fault analysis of a machine tool based on a Digital Twin
The availability of data for artificial intelligence and the simulation of anomalous conditions are two critical challenges for contemporary industry. It is already possible to acquire and analyze large sets of databases, but the availability of data on the state of failure requires long operating times of the machines (months or in some cases years) and even when this data is available it is not guaranteed that a tendency to the fault state in order to be able to identify it again later. The thesis work aims to improve and validate a multi-physical model – already partially developed in Modelica – of a CNC machine installed at the MADE competence center (www.made-cc.eu). Through appropriate simulation campaigns of the implemented model it will be possible to generate fault data, on which the identification of the machine fault states will be based. The thesis work finally includes the development of a demo showing the identification of faults for different configurations of the machine model.
Development of a Modelica library for the simulation of the biochemical processes of anaerobic digestion
The environmental emergency resulting from global warming is multiplying efforts to reduce carbon dioxide emissions on the one hand and scientific research on the sequestration techniques of these emissions and on the production of renewable energy on the other. In particular, in recent years there has been a growing diffusion of biochemical digestion plants for agricultural and urban waste for the production of biofuels. The thesis work will aim at the creation of a Modelica library for the simulation of the systems of the biochemical processes of anaerobic digestion. Validation of the library based on data from pilot and full-scale plants is envisaged, and possibly the synthesis of optimal strategies for commissioning the plants.
LFT-based identification of index 1 nonlinear DAE systems
Parameter identification of nonlinear systems, based on a reformulation of the model in LFT form, has already been successfully applied in various application cases, from biochemical processes to vehicle dynamics. For this purpose, a MATLAB toolbox was created. The primary goal of the thesis will be the extension of the toolbox to the case of DAE systems (systems of algebraic-differential equations) of index 1. As an application case, the parameter identification of an innovative process for the production of biogas will be considered. Furthermore, the method will be compared with other tools for parametric identification not based on the use of derivatives for the calculation of sensitivities.
Online identification of cornering stiffnesses
In a previous thesis work, the problem of offline estimation of a vehicle’s cornering stiffnesses was tackled by an LFT (Linear Fractional Transform) reformulation of the single track model and the subsequent application of a maximum likelihood method. The good results obtained suggested the extension of the approach to an online estimate. The thesis work will therefore have as objective the implementation of the online estimation method, considering both an experimental type vehicle and, possibly, a suitably instrumented real vehicle.