The aim of PhD project is to conceive improved bone implants and prosthesis based on the gyroid structure. Gyroid foams allow to obtain suitable mechanical properties to mimic human bone contributing to the performance of the implant in the patient. To do so, innovative tools such as machine learning will be used. Moreover, it is important to mention that the gyroid structure can be used for structural purposes, and thus, build lighter yet structurally sound parts.

The potential of Human-Robot Collaboration (HRC) is far from exhausted. Currently, Collaborative robots (COBOTs) are advertised for ease of commissioning and the capacity to operate without fencing, and thus share the workspace with humans. Even when the COBOT is programmed to admit some degree of interaction with the worker, the robot system is “socially blind” to Ergonomics & Human Factors (E&HF). The lack of a valid and sound ergonomics framework cognizant of the implications of HRC, as well as the inexistence of a correlation between risk assessment/prevention and complementary robot behaviors, hinders the progress in this direction. To evolve the state of the art of current collaborative robots in the manufacturing industry, this project propose the creation of a real-time assessment framework appropriate to the HRC tasks. This expected outcome will support the next step towards collaborative robots capable of understanding and adapting to improve the physical and cognitive ergonomics of human coworkers.

The Industry 4.0 revolution has grown the necessity of the Facility Management (FM) discipline to deal with live data. Therefore, the Digital Twin (DT) technology can significantly support the FM sector and its management processes. DTs can be essential for operating “dynamic facilities” ,understood as those undergoing recurrent changes in their layout. For the BIM models as the primary source of information and graphical representation in a DT, open formats such as IFC are encouraged (open BIM). However, because of the high complexity of the IFC schema, the object-based real-time layout modification of IFC DTs is a challenge to overcome. In this context, FM lacks guidance on implementing an efficient integrated management system necessary for the current industry demands. Therefore, this work proposes an approach to implementing a new integrated management system and the method to guarantee a full DT to reflect real-time objects’ location using the IFC schema.

As a disruptive technology, additive manufacturing (AM) is changing the state and structural dynamics of the supply chain (SCs), and their ability to cope with SC vulnerabilities and disruptions, i.e., supply chain resilience (SCR). SCR is vital for business continuity when facing unforeseen disruptions such as the recent COVID-19 pandemic. However, to date, no research has investigated the implications of AM adoption for SCR, and therefore, this PhD project aims to fill this knowledge gap. Drawing on a mixed-method research approach, this PhD project will investigate how AM adoption impacts the state of SCs that consequently influence certain SC capabilities and vulnerabilities, thereby affecting SCR. Subsequently, an AM-SCR framework will be developed to aid decision-makers in the optimal (re)design of SCs by determining which resilience practices to capitalize on when using AM technology so as to develop the necessary SC capabilities to deal with SC vulnerabilities more effectively.

Modern power systems are characterized by an increasing need of Ancillary Services (AS) due to forecasting errors, unpredicted contingencies in the internal system and unforeseen disturbances in neighboring systems. After decades of independent evolution of transmission and distribution grids, recent research projects have been addressing TSO-DSO cooperation in terms of shared big data platforms of both networks’ real-time measurements. In this context, we propose a Model Predictive Control (MPC) for the real-time activation of Ancillary Services supporting the transmission network operation, allowing for the exploitation of resources present on the distribution side. The model is intended for use either post fault or when online security assessment detects an insecurity state, and guarantees that the activation of resources does not violate the distribution grids constraints. The amount of data and model running time are challenges to be addressed through a linearization process. The methodology will be applied using to real world data from European networks.

Assessing and predicting structural health is one of the most important aspects of an infrastructure’s life cycle. The devices that are used in the modern era have enabled us to predict the future of the structure using various monitoring techniques and also various maintenance and management methodologies to prolong the life span of the structure and to have a healthy structure to be utilized by the user. The most contemporary system for monitoring with low installation and maintenance cost with high reliability is Micro-Electro-Mechanical Systems (MEMS) technology and Internet of Things network using LTE (4G Long Term Evolution). The management aspect can be developed using computerized systems such as Building Information Modelling (BIM) that help in the planning, designing, implementation and maintenance of the building. Present research work will use open source software and a standard file transfer data format to transfer information.

This PhD project proposes an innovative Market Model suited to the concept of a renewable energy community, in which its members can interact through Energy Sharing and share, among themselves, the storage capacity of their stationary batteries or electric vehicles, Storage Sharing. In order to obtain an optimum energy planning for the community, the proposed multi-agent Market Model will be based on Artificial Intelligence (AI), with a hierarchical control structure, which will base their decisions on technical and economic aspects, considering the community load and renewable production forecasts and the restrictions, imposed by each member, for the community participation. To ensure the applicability of the proposed Market Model, a long-range wireless communication module will be developed, based on the Internet of Things (IoT), which will allow the monitoring and control, in real time, of the parameters of each member of the community.

Scientific Machine Learning (SML) is a recent trend involving the application of physically or mathematically-informed Machine Learning (ML) techniques to speedup classical numerical computations that are central to many fields of engineering and science. The application of SML techniques to Structural Design (SD) allows designers to explore and assess different solutions quickly, avoiding the need to use traditional Finite Element Models, which, despite being very precise, are computationally expensive and quickly become the bottleneck for experimentation.

My Ph.D. project will contribute to real-time 3D simulations in SD, which will facilitate the exploration of new designs and find optimal solutions without the need to have an expensive computational infrastructure. Furthermore, with this project, we will contribute to the connection of the Mechanical Engineering and ML community. These future cooperations will allow accelerating the digital transformation in manufacturing.

This PhD project aims to develop a computational inverse Quantitative Structure Activity Relationship model, to optimize industrial catalytic epoxidation of small alcohols and alkenes. In turn, it will address sustainable manufacturing solutions towards this high economic value chemical reaction. The model aims to target the chemical space by implementing recent data science methodologies, which in turn feed and train machine learning models with data obtained from available experimental results. The project intends to embolden chemists and industries with a predictive model that eases reaction conditions, avoiding high cost quantum mechanics calculations and catalyst design by trial-and-error. It will also allow to find optimal epoxidation catalysts using biomass-based feedstock, thus avoiding the need of raw fossil-fuel commodities. This model will allow to reduce the environmental impact of these industries, lowering waste production, while generating higher product yields.

As the nature of markets changes over time, so must the industry, in order to adapt and meet the increased consumer demands. Perfume design is a complex task that currently relies on expertise from perfumers, which affects its ability to adapt to paradigm changes. Thus, a novel smart Cyber-Physical System (CPS) is proposed as an innovative solution. The main goal of this project is to build a CPS intended for on-demand optimal design and production of perfumes. The CPS will be powered by Deep Neural Network (DNN) surrogate models. The components of the CPS together will constitute a disruptive systemwith self-managing and cognitive abilities. These components will be concisely interconnected through the Internet of Things (IoT) and use Cloud Computing as engine.

In the process of digital transformation in Industry – a sector that employs around 25% of active workers in Portugal – there is a need to evaluate workers’ well-being, as both musculoskeletal and mental disorders affect a great part of them due to lack of effective methodologies for guaranteeing their occupational health. In the proposed PhD project – integrated in the MIT-Portugal project Operator in a consortium of several acknowledged research groups and corporations – it is aimed to extract, study and understand the occupational factors that affect the well-being of the operator and design solutions to track job quality and prevent risk exposure. Henceforth, sensors will be used to monitor physiological and motion signals in work context and a software that from the signals assesses risk exposure will be built based on novel deep learning architectures. Finally, the operator will be provided feedback with relevant information, allowing a self-conscious change of conduct.

Building Information Modelling (BIM) objects are at the heart of the transition of the Architecture, Engineering and Construction industry toward digitalization. BIM object creators are expected to comply with BIM object standards, taking into account compliance to Construction Product Regulations, standardization and interoperability requirements. This heavy task is limiting the widespread worldwide adoption of BIM processes. This work addresses this issue by (i) applying a systemized science-based approach for standardizing construction products data using the concept of Product Data Templates, (ii) creating a tool for the automation BIM objects data compliance to the latest standards and regulations, and (iii) setting the base for the creation of an open source BIM object repository.

This research is related with the ongoing Greenshoes 4.0 mobilizer project in which the required computational modelling tools will be developed. The suggested approach in this project will improve the modelling capability, by combining the available codes with AI methods, thus, noticeably minimizing the dependence on the persons involved in the design processes. This approach is mandatory due to the inability to process systematically, efficiently and in a timely manner, the large amount of available information.

Additive manufacturing (AM) technology is a key driver for the digital transformation, as it promises to dramatically change business and innovation models, by shrinking supply chains and altering the global economy. However, findings reveal a lack of research addressing the most prominent trends currently driving AM: such as decentralized production of customized products and small-to-medium volumes transition, thus highlighting the disruptive nature of the expected changes in the industry. Aligned with the MPP2030 funded seed project – Introducing 3D Printing into the Production Chain – the goal of this Ph.D. is to use data-driven techniques and optimization approaches to provide decision-makers with information on how AM techniques will impact processes. The successful accomplishment of this goal will have a positive impact on the automotive, furniture, and semiconductor sectors, by achieving a framework to provide practical guidance on the steps needed to optimize AM processes.

Machine learning (ML) is often criticized for its lack of interpretability and transparency. In order to tackle these problems, the field of explainable artificial intelligence (XAI) emerged in recent years. While XAI offers strategies to gain insight into how models learn and their decision-making processes it lacks the concept of acceptance of AI. Interpretability and transparency play a role in acceptance but do not fully encapsulate all factors that contribute to it. Thus, research into what constitutes acceptance needs to be made.
As part of the MIT-Portugal project OPERATOR, the proposed PhD project will contribute to XAI and acceptance of AI systems within the industrial workforce to produce recommendations based on cognitive and biosignals monitoring and by developing quantifiable measures of acceptance and trust for AI systems, depict understandable and interactive visualization schemes while also designing ML models that incorporate these measures.

Pacifiers are excellent to calm babies, but there are numerous risks about its prolonged use. The current approaches used to design pacifiers are strongly based on accumulated experience. In a previous work of the candidate, it was shown that realistic computational models allow obtaining concrete data on the effects of pacifiers, opening the route to the digital transformation of the pacifier design and manufacturing methods. The work proposed in this project aims at developing a novel pacifier design methodology together with an innovate adaptive pacifier, both conceived with the support of computational modelling tools. The adaptive pacifier will comprise an intelligent sensing system that will allow monitoring its operation in real-time. Data mining will be applied to identify useful patterns in the collected data, which will be used to parameterize the pacifier usage limits, taking into consideration each user suction pattern, thus allowing continuous pacifier usage supervision.

Current plantwide statistical process monitoring (SPM) methods face problems in detecting and diagnosing localized faults in high-dimensional complex processes, compromising personal safety, quality, efficiency and the environment. They are also unable to handle the multiplicity of time scales that are relevant for fault detection and to conduct degradation prognosis activities. This project proposes a novel framework that brings causality concepts to plantwide distributed SPM, improving fault diagnosis, and defines the optimal aggregation level of variables, for improving monitoring performance. Furthermore, a multiscale decomposition of the time series will be conducted to separate the time scales for monitoring (high frequency bands) from those that represent slow degradation modes (low frequency bands). The proposed methodology is called MAG-NET (Multilevel and multiscale AGgregation of causal NETworks) and it unifies plantwide detection, diagnosis and prognosis.

Wire-Arc Additive Manufacturing (WAAM) is an emerging additive manufacturing (AM) process that uses wire as the feed-stock and an electric arc to melt it. WAAM has a promising sustainability potential due to its better material efficiency, higher build rates, and low cost. However, research on its sustainability from a Triple-Bottom-Line perspective i.e., environmental, economic, and social impacts is still new. This PhD plan proposes to evaluate the WAAM´s environmental, economic, and social sustainability using Life Cycle Assessment (LCA), Life Cycle Costing (LCC), and Social Life Cycle Assessment (S-LCA) methodologies, respectively. Based on these assessments, process parameters, variables, and production scenarios ensuring sustainable production by WAAM will be recommended. Furthermore, an online platform will be created to quickly calculate and compare the environmental, economic, and social impacts of WAAM and other manufacturing processes to select the most sustainable approach.

This proposal aims at the development of an industrial Human-Robot Collaboration (HRC) framework, within the scope of industry 5.0 and the digital transformation in manufacturing. It intends to address the growing number of Work-Related Musculoskeletal Disorders (WRMSDs),bringing innovation in ergonomic interventions by providing valuable collaborative robot assistance according to the worker needs through the development of human-aware and human-care control strategies, powered by Artificial Intelligence and interoperable wearable technology. Novelty includes bi-directional interaction between human and robot by identifying and monitoring in real-time the worker’s motion and fatigue during a collaborative task and tailoring the robot physical-collaboration accordingly. Expected benefits include the operator’s well-being, healthier and safer working environments, increased productivity and reduction on the costs derived from WRMSDs, while fostering companies’ industrial competitiveness.

Digital twin (DT) is one of the most modern and promising technologies in realizing smart manufacturing and implementing Industry 4.0. DT offers opportunity to integrate the physical world with digital world with a seamless data source. Civil engineering industry, in general, is facing many challenges in the process of digital transformation to improve efficiency and technology to meet the current growth rate of the economy. DT technology has the potential to transform and improve the exploitation and management of infrastructure in civil engineering, especially in the service phase. Based on DT model, managers and maintenance operators can test different scenarios, improve efficiency, and make accurate decisions in maintenance of the structure, leading to reduction of management and other regular monitoring costs, as well as accurate prediction of risks during the lifespan of the infrastructure. This research focuses on building a digital twinning model for the continuous structural health monitoring of critical infrastructure in the service phase.

This project aims to design and implement a cognitive two-layer Cyber-Physical System (CPS) process to separate methane from CO2. The CPS-PSA will be powered with Artificial Intelligence (AI) and the automated smart system with Internet of Things (IoT) that will yield autonomy and self-management capabilities which, in turn, will lead to high-efficiency and sustainable operation. Altogether, the project is aligned with Portuguese plans for science development and the United Nations Sustainable Development Goal 9.

Ni alloys, particularly INCONEL alloys, present very low machinability, making them a popular subject for machining studies. Generally, using high amounts of cutting fluid is needed to machine INCONEL alloys, however, the use of these coolants is extremely hazardous for the environment and worker safety, while increasing the overall process cost. To minimize these issues, many solutions can be adopted, such as the use of coated tools, lubrication techniques or even by trying to optimize the process itself. In this work, different sustainable INCONEL machining techniques will be tested, evaluating tool-wear, process vibrations and cutting forces, as well as the machined surface integrety of the workpiece material. Data obtained from these tests will be used to develop and validate process optimization models, with a focus on improving the process’s sustainability while retaining good production quality, as well as predicting tool-wear and produced surface quality.