Energy & Infrastructure Systems: Modeling, Computing, and Control (EIMC2)

Subgroups: 

Resilient Infrastructure Networks Lab (RESIL): PI  Saurabh Amin  https://resil.mit.edu/saurabh-amin
The RESIL group focuses on the design of information and control systems for cyber-physical-human infrastructures using stochastic control, game theory, statistical learning, and optimization in networks. The group works on four main problem areas: (1) Resilient network monitoring and control; (2) Information systems and incentive design; (3) Optimal resource allocation for disaster response and recovery; and (4) Sustainability and decarbonization of supply chains. By focusing on important questions in the domains of electric power, logistics and transportation systems, we develop theory and design tools for improving the performance of these infrastructure systems in the face of disruptions, both stochastic and adversarial. A unique aspect of our agenda is to integrate the design network monitoring and control algorithms with economic incentive schemes to help infrastructure users and operators make optimal decisions in the presence of uncertainties. This agenda is supported by our approach to: (i) model the cyber-physical-human interactions in infrastructures and evaluate key risks; (ii) develop tools to detect and respond to both local and network-level failures; and (iii) design incentive schemes that improve the aggregate levels of public good, while accounting for the dependencies and private information among strategic entities. 

Anuradha Annaswamy Group,  PI Anuradha Annaswamy, https://meche-dev.mit.edu/people/faculty/aanna[at]mit[dot]edu
Environmental and sustainability concerns are resulting in a rapid integration of distributed energy resources (DER) such as solar PV, batteries, and EVs into the power grid. Coupled with the recent advancements in Internet-of-Things (IoT) technology, this is steering the electric grid away from the traditional centralized structure with unidirectional pathways towards a distributed, interconnected web with edge computing capabilities. Our lab focuses on distributed decision-making for ensuring a reliable and resilient power grid, especially in the face of large penetration of DERs. Of equal interest is decision making at fast time scales. Tools of distributed optimization, distributed control, adaptation and learning, combined realization of stability, optimality, and safety, behavioral model of consumers, considerations of equity, and design of cyber-physical-human systems represent current research directions. Validation of these tools using domain-specific high fidelity models is of interest. Where possible, collaborations with practitioners towards field implementation of these concepts are being continuously explored.

Azizan Lab,  PI  Navid Azizan , website  https://azizan.mit.edu
Azizan Lab develops the theoretical foundations and practical methodologies for supercharging societal-scale intelligent systems that can operate reliably, efficiently, and autonomously. Our research lies in the span of machine learning, systems and control theory, optimization, network science, and economics. Our applications include energy systems and infrastructure.

Energy Analytics Group, PI  Audun Botterud, https://botterud.mit.edu/
The main goal of our research is to improve the understanding of the complex interactions between engineering, economics, and policy in electricity markets and power systems, and ultimately enable the transition towards a cost efficient and reliable zero-carbon energy system. We are particularly interested challenges related to the integration of renewable energy and energy storage into a smarter electricity grid. Towards this end, we use methods from operations research and decision science combined with fundamental principles of electrical power engineering and energy economics. At a more general level, our research focuses on decision making under uncertainty in complex systems. We work on operational and planning problems within electricity markets and energy systems at local, regional, and global scales.

Munther Dahleh Group, PI Munther Dahleh, https://dahleh.lids.mit.edu/
https://dahleh.lids.mit.edu/networked-decisions/

The group works on markets for future power grids. Representative publications:
https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9690618&tag=1
https://arxiv.org/abs/2203.10206 

Donti Group, PI Priya Donti, https://priyadonti.com/
Our group's research focuses on developing robust machine learning methods for power grid operations, planning, and control. Methodologically, this entails exploring ways to incorporate relevant physics, hard constraints, and decision-making procedures into deep learning workflows, or otherwise combining approaches from electrical engineering, operations research, and control theory with approaches from deep learning and reinforcement learning. We are particularly interested in challenges pertaining to the integration of time-varying renewable energy, distributed devices, and flexible demand into electric power systems.

Chuchu Fan Lab, PI Chuchu Fan, https://aeroastro.mit.edu/realm/
The realm group works on developing methods that accelerate statistical safety verification, testing, and design automation of energy and other autonomous systems. The methods automatically search for design parameters that achieve good performance of systems, characterize the robustness of a design, predict corner cases where it is likely to fail, and guide future design iterations using the predicted corner cases.

Electric Energy Systems Group (EESG@MIT), PI Marija Ilic,   https://eesg.mit.edu/
The Electric Energy Systems Group (EESG@MIT) researches topics related to modeling, and cyber (control, communications) design of the rapidly changing electric energy systems. Theoretical areas underlying this work are: the unified modeling of social-ecological energy systems (SEES) as cyber-physical systems (CPS); and the related cyber design called “Dynamic Monitoring and Decision Systems” (DyMonDS).  

The key idea being pursued by EESG@MIT rests on the recognition of the multi-layered nature of these energy systems; the modules are highly heterogeneous and technology-specific, and their physical models are derived by domain experts. The interconnected system model is based fundamentally on general conservation laws, which, in turn allows the existence of aggregate interaction variables and their use for representing higher-level system dynamics in transparent ways.   

This modeling can make use of many techniques developed in the Laboratory for Information and Decision Systems (LIDS) at MIT to solve difficult domain application problems, which, jointly, would provide reliable, resilient, sustainable and cost-effective electric energy service at value. It enables convergence of physical, cyber, economic and policy models and their inter-dependencies. 

We are in the process of a DyMonDS Digital Twin (DDT) that will be used for both education and research in these domain applications. Such a facility is expected to greatly help with emulation of complex SEES and, in particular, with the demonstrating and assessing of the effects of cyber designs on their performance. Applications of modeling and cyber design include large-scale terrestrial electric power systems; local terrestrial power grids (utility distribution; civil, military and naval microgrids), and, most recently, turboelectric distributed propulsion for future aircraft systems.

Kalyan Veeramachaneni: Director, MIT Data to AI Lab. https://dai.lids.mit.edu/
MIT’s Data-to-AI Lab, founded in 2015, is a team of like-minded scientists who combine Big Data + Human Interactions + Impactful Domains (machine + human + positive societal impact). In 2017. The group focuses on building large-scale AI systems that work alongside humans, continuously learning from data that generate and integrate predictions into “augmented” human decision-making. The algorithms, systems and open-source software developed by the MIT Data-to-AI (DAI) Lab are deployed for applications in the financial, healthcare, educational and energy sectors.

A recent project at the MIT Data to AI Lab focuses on creating AI-driven generative models for customer load data. The group is working alongside a teams from universities across Ohio, Pennsylvania, West Virginia, and Tennessee, to develop and deploy smart grid modeling services. These generative models have far-reaching applications, including grid modeling and training algorithms for energy tech startups. When the models are trained on existing data, they create additional, realistic data that can augment limited datasets or stand in for sensitive ones. Stakeholders can then use these models to understand and plan for specific what-if scenarios far beyond what could be achieved with existing data alone. For example, generated data can predict the potential load on the grid if an additional 1,000 households were to adopt solar technologies, how that load might change throughout the day, and similar contingencies vital to future planning.

Zardini Lab: PI  Gioele Zardini https://zardini.mit.edu/people/prof-gioele-zardini/
Driven by societal challenges, the goal of the Zardini Lab at MIT is to develop efficient computational tools and algorithmic approaches to formulate and solve complex, interconnected system design and autonomous decision-making problems. The complexity has at least two origins. First, the co-design of complex systems (e.g., large networks of cyber-physical systems) involves the simultaneous choice of components arising from heterogeneous natures (e.g., hardware vs. software parts), while satisfying systemic constraints and accounting for multiple objectives. Second, different components are interconnected via interactions between different stakeholders, which often feature multiple, conflicting objectives (e.g., within an intermodal mobility system). 

Typically, solving this kind of problem is attempted by employing multi-objective optmization techniques, which tend to lack computational scalability, modularity, compositionality, and, importantly, interpretability and interdisciplinarity. 

Equipped with strong mathematical skills, and a broad and diverse engineering background, we solve such problems by employing and enhancing techniques from optimization, control theory, game theory, domain theory, and applied category theory, and we apply results to society-critical problems in mobility, logistics, autonomy, automotive, aerospace, energy, and complex systems in general.