Closing workshop: Looking forward to 2050
Monday 29th April 2019 to Friday 3rd May 2019
09:30 to 09:50  Registration  
09:50 to 10:00  Welcome from David Abrahams (Isaac Newton Institute)  
10:00 to 11:00 
Janusz Bialek (Skolkovo Institute of Science and Technology) Mathematics for energy systems: an engineer’s view
The presentation will contain a personal view of an engineer on the complex interactions between mathematicians and engineers, exploring the myths, tensions, and modes of collaboration. It will also present a personal view on the trends and themes shaping power system research over the last 40 years and how they involved mathematicians. The presentation will end up by showing an example of unresolved research problems that call for mathematics: making use of wealth of information provided by Phasor Measurement Units.

INI 1  
11:00 to 11:30  Morning Coffee  
11:30 to 12:30 
Adilson Motter (Northwestern University) Power Grids: Failures and Opportunities
A fundamental problem in electrical grids is
that the same connections that give a network its functionality can promote the
spread of failures that would otherwise remain confined. Understanding the
resulting cascading failures has been hindered by the lack of realistic
largescale modeling that can account for variable system conditions. In this
presentation, I will discuss our fullscale modeling of the U.S.South Canada
powergrid network, which allows us to characterize the set of network
components that are vulnerable to cascading failures. I will show that the
vulnerable set consists of a small but topologically central portion of the
network and that large cascades are disproportionately more likely to be
triggered by initial failures close to this set. These results elucidate
aspects of the origins and causes of cascading failures, and point to
opportunities for failure mitigation in connection with grid design and
operation.

INI 1  
12:30 to 13:30  Lunch at Westminster  
13:30 to 14:30 
Anupama Kowli (Indian Institute of Technology) SMART Planning and Operations of Grids With Renewables and Storage a.k.a. SPOReS
This talk will introduce the SPOReS project started under the Mission Innovation Smart Grids challenge in India. The goal of SPOReS is to facilitate planning for reliability and affordability along with balancing of grid supply with storage, flexible loads and power electronic controllers. This talk will present two specific deliverables of SPOReS. On the planning front, a graphbased approach for enhancing the reliability of distribution systems by installing new tielines will be introduced. The problem is cast as an edgeconnectivity augmentation problem. We will discuss the associated solution approaches, ways to factor in practical considerations and metrics based on network topology that capture the reliability impacts of the specific interventions. On the operational front, solutions for load management will be discussed to facilitate demand response and community energy sharing.

INI 1  
14:30 to 15:30 
Veronika Grimm (FriedrichAlexanderUniversität ErlangenNürnberg) Generation and network investment in electricity markets: using multilevel optimization to assess the role of the market design The talk will present a multilevel optimization approach that allows to analyze the longrun impact of electricity market design on investment and production decisions. The presentation will give an overview over various recent contributions that focus on electricity markets that operate under a zonal pricing market design, as it is established in Europe. We apply the approach to the German electricity market as an example to demonstrate the impact of (i) the establishment of additional price zones, (ii) regionally differentiated network fees (iii) marketdriven RES curtailment on key performance indicators such as welfare, economic rents, generation mix and locations, network investment, or electricity prices in the long run. We also show how the model can be used to endogenously determine an optimal configuration of price zones on a network, or to assess the long run impact of structural (regulatory) uncertainty about the future market design. The talk will present work with various coauthors which will be mentioned along the presentation. 
INI 1  
15:30 to 16:00  Afternoon Tea  
16:00 to 17:00 
Eddie Anderson (University of Sydney) Rothschild Lecture: What do we agree on when we disagree? Forward contracts with private forecasts
Forward
contracts are used for hedging purposes when firms operate in a spot market.
What will happen when firms have different views on the future distribution of
prices and are risk averse? We discuss different ways in which two firms may
agree on a bilateral forward contract: either through direct negotiation using
the ideas of a Nash bargaining solution, or through a broker. We discuss a type
of equilibrium in which each firm offers a supply function linking quantities
and prices, and the clearing price and quantity for the forward contracts are
determined from the intersection. Each firm may also be able to use the offer
of the other firm to augment its own information about the future price. 
INI 1  
17:00 to 18:00  Welcome Wine Reception at INI 
09:00 to 10:00 
Shmuel Oren (University of California, Berkeley) Mobilizing Grid Flexibility for Renewables Integration through Topology Control and Dynamic Thermal Ratings
(Joint work with Jiaying Shi) The rapid penetration of renewable resources into the electricity supply mix poses challenges for the efficient dispatch of resources due to the inherent uncertainty and variability of such resources. Hence, in order to accommodate large amounts of renewables in is necessary to account for their output uncertainty and mobilize the flexibility of the system embedded in conventional generation, demand side resources and the transmission grid. In this talk we formulate a stochastic unit commitment optimization in which we expand the traditional recourse actions that are available to mitigate the adverse effect of renewables variability. In particular we include in these recourse action, topology control through transmission switching and dynamic line ratings that account for the heating and cooling of transmission lines. We will demonstrate the potential gains from such recourse actions through test cases and discuss heuristic approaches for alleviating the computational burden resulting from such a formulation. 
INI 1  
10:00 to 11:00 
John SimpsonPorco (University of Waterloo) Optimal SteadyState Control with Application to Secondary Frequency Control of Power Systems
The optimal steadystate control problem is that of designing a feedback controller for a multivariable system which regulates selected system variables to the solution of a constrained optimization problem, despite parametric modelling uncertainty and unmeasured exogenous disturbances. For example, an instance of this problem occurs in the design of secondary frequency control systems, where control resources should be optimally used subject to the elimination of frequency deviations and restoration of interarea tieline flows. We present a constructive design framework that reduces the OSS control problem to a more classical control problem of output regulation (tracking and disturbance rejection). Robustness issues which arise in the constructive procedure are discussed, leading to a solution of the OSS control problem for the case where the plant is LTI with structured parametric uncertainty and unknown disturbances are constant in time. We apply the results to frequency control of power systems show that the OSS control framework recovers several recent decentralized and distributed secondary frequency controllers from the literature. Collaborators: Liam S. P. Lawrence (UWaterloo) and Enrique Mallada (JHU) 
INI 1  
11:00 to 11:30  Morning Coffee  
11:30 to 12:30 
Johanna Mathieu (University of Michigan) Optimal Power Flow with Stochastic Reserves
My research seeks to improve the sustainability of electric power systems by developing methods to support the integration of renewable energy sources while maintaining system reliability. In this talk I will describe our efforts to develop approaches to dispatch power systems with reserves provided by stochastic resources like residential air conditioners. We formulate the problem as a chance constrained optimal power flow problem with uncertainty in renewable energy production, load consumption, the capacity of loadbased reserves. Solving such problems is challenging because we do not have good models of the uncertainties. I will describe our work exploring of a variety of solution approaches including distributionally robust optimization, which gives us reliable but costly solutions. I will also describe a novel distributionally robust approach that allow us to leverage structural information about the uncertainty to reduce the conservatism of standard approaches.

INI 1  
12:30 to 13:30  Lunch at Westminster College  
13:30 to 14:30 
Claudia D’Ambrosio (CNRS (Centre national de la recherche scientifique)) Smart Grids Observability using Bilevel Programming
Monitoring an electrical network is an important and challenging task. Phasor measurement units (PMU) are devices that can be used for state estimation of this network. We consider a PMU placement problem and propose two new approaches to model this problem, which take into account a propagation rule based on Ohm’s and Kirchoff’s laws. First, we describe the natural binary linear programming model based on an iterative observability process. Then, we remove the iteration by reformulating its fixed point conditions to a bilevel program. We propose two methods to solve such a problem. The first is based on the observation that the integrality constraint of lower level problem can be relaxed so as to derive a single level reformulation by replacing such a problem with its dual. The second is a tailored cutting plane algorithm. We show through computational results that the tailored cutting plane method is much more effective than the others on a set of instances taken from the literature. Joint work with Sonia Toubaline, PierreLouis Poirion, and Leo Liberti 
INI 1  
14:30 to 15:30 
Claudia Sagastizabal (Universidade Estadual de Campinas) Nonsmooth Optimization put to good use in energy problems
We examine problems in managing a mix of power plants generating electricity, using optimization modeling. The modeling includes definition of goals, such as minimizing carbon footprints, maximizing revenue, or reducing risk of blackouts. We discuss how to exploit structural properties of nonsmooth objective functions.

INI 1  
15:30 to 16:00  Afternoon Tea  
16:00 to 17:00 
Marija Ilic (Massachusetts Institute of Technology); (Carnegie Mellon University) Dynamic monitoring and decision systems (dymonds) framework for dataenabled integration in complex electric energy systems
In this talk we introduce a unifying Dynamic
Monitoring and Decision Systems (DyMonDS) framework that is based on a multilayered
modeling for aggregation and minimal coordination of interactions between the
layers of complex electric energy systems. Using this approach, distributed
control and optimization problems are formulated so that: (1) the lowlevel
decision makers optimize cost of local interactions while accounting for their
heterogeneous technologies, as well as for their social and risk preferences;
and, (2) the higher layer aggregators and coordinators optimize the cost of all
interactions at their levels to enable cooperative control. The interactions of
each layer are abstracted by using unifying energy state space and the Lagrange
coefficients associated with the general physical laws. This sets the bases for
both nonlinear control of power electronicallyswitched automation and for
market design formulation. Potential benefits (such as enhanced reliability,
resiliency, and efficiency) from integrating flexible technologies, storage,
and control, in particular, are illustrated on simple IEEE test systems.

INI 1 
10:00 to 18:00  Managing Next Generation Energy Systems  
19:00 to 22:00 
Formal Dinner at Trinity College (Old Kitchen) LOCATION Trinity College (Old Kitchen) Cambridge, CB2 1TQ, U.K.x MENU Starters Salad of Madgett’s Duck Confit  Hazelnuts, Gésiers, Duck Liver Mousse, Rhubarb and Sweet Pickled Shallot (V) Cream of Watercress Soup  Grain Mustard Sabayon and Cheddar Straw Pastry Crumb (Milk, Egg, Sulphites, Mustard, Wheat) Main course Cambridgeshire Lamb Loin  Samphire Grass, Parmesan and Artichoke Risotto and Mustard Glazed Crôutons (V) Slow Cooked Cumin Bell Peppers, Caper Butter Sauce, Beetroot Pearl Barley, Lemon Kohlrabi + Broad Beans (Milk, Wheat, Sulphites, Celery) Pudding Vanilla & Star Anise Panna Cotta  Lemon Sherbet and Grape Molasses DRESS CODE Smart casual 
09:00 to 10:00 
Mahnoosh Alizadeh (University of California, Santa Barbara) MobilityAware Load Management Algorithms for Electric Vehicles
In this talk, we will discuss pricing mechanisms and/or routing algorithms that can be used to manage the charging demand of a population of electric vehicles. After presenting an overview of some recent results, two main papers will be discussed. In the first part of the talk, I will present on the design of optimal pricing and routing policies to provide differentiated services in a network of fast charging stations, where customers have different values of time and energy demands. The second part of the talk is focused on residential EV demand management under realtime pricing programs. We will discuss the application of multiarmed bandit based algorithms for price design, where customers' price response is modeled as a stochastic system with unknown parameters. Given the presence of reliability constraints in power systems, we will discuss the application of wellknown bandit heuristics in constrained environments and discuss the effects of reliability constraints on the regret guarantees of the algorithms.

INI 1  
10:00 to 11:00 
Na Li (Harvard University) The Role of Prediction And Realtime Learning in Online Optimization and Control
Uncertainties place a key challenge in transforming the grid into one with a large renewable energy integration and active users' participation. The volatile renewable generation constantly introduces disturbance to the grid operation; while the stochastic user behavior makes it difficult to use demand as a reliable resource for balancing realtime supplydemand. This talk will present two lines of work to address these issues. Firstly, we will study how to use a short term prediction, e.g., renewable energy, to improve the performance for online optimization and control, e.g, economic dispatch with ramping cost. Then, we will study how to use realtime learning to design residential demand response programs for achieving grid reliability while balancing the exploitation and exploration of the uncertain user behavior.

INI 1  
11:00 to 11:30  Morning Coffee  
11:30 to 12:30 
David Hill (University of Hong Kong); (University of Sydney) Exploring fundamental limits in future grids with mathematics, computation and databased methods The talk will present results for computing stability limits using a variety of approaches each with advantages and disadvantages arising from the models used. Mathematics favours simplified equationbased models; computation favours detailed physicsbased models and newer databased methods sometimes claim to be actually ‘modelfree’. As systems become more complex in likely scenarios for grids out to 2050and become more everchanging there should be a new discussion about the merits of the associated methods. Results obtained in future grid and high renewable projects in Sydney and Hong Kong respectively will be used for illustration. 
INI 1  
12:30 to 13:30  Lunch at Westminster College  
13:30 to 14:30 
Enrique Mallada (Johns Hopkins University) Embracing Low Inertia for Power System Frequency Control: A Dynamic Droop Approach
CoAuthors: Yan Jiang, Richard Pates, and Fernando Paganini Abstract: The transition into renewable energy sources with limited or no inertia is seen as potentially threatening to classical methods for achieving grid synchronization. A widely embraced approach to mitigate this problem is to mimic inertial response using gridconnected inverters. That is, introduce virtual inertia to restore the stiffness that the system used to enjoy. In this talk, we seek to challenge this approach and advocate towards taking advantage of the system’s low inertia to restore frequency steady state without incurring in excessive control efforts. With this aim in mind, we develop an analysis and design framework for inverterbased frequency control. We define several performance metrics of practical relevance for power engineers that contemplate system disturbances and measurement noise, and systematically evaluate the performance of standard control strategies, such as virtual inertia and droop control. Our analysis unveils the relatively limited role of inertia on improving performance as well as the inability of droop control to enhance performance without incurring in considerable steadystate control efforts. To solve this problem, we propose a novel dynamic droop control (iDroop) for gridconnected inverters exploiting classical lead/lag compensation from control theory that can significantly outperform existing solutions with comparable and in many cases significantly smaller control efforts. 
INI 1  
14:30 to 15:30 
Ana Busic (INRIA); (CNRS  Ecole Normale Superieure Paris) Balancing the Grid through Distributed Control of Flexible Loads
A control scheme that unifies feedforward and feedback
control for a large collection of heterogeneous loads is proposed. The
feedforward part capitalizes on the knowledge of gridevent forecasts, such as
predicted ramps in solar and wind energy and the approximate periodicity of the
demand cycle.
The slower timescale feedforward control strategy
utilizes a model predictive control (MPC) framework designed on the basis of
two pieces of
information: aggregate power consumption from each class
of loads and the state of charge surrogate that is a part of the leaky battery
model. Feedback control is performed at a faster time scale in order to reject
realtime disturbances. The control architecture incorporates the decentralized
loadlevel stochastic control design, which ensures that consumer QoS constraints
are respected.

INI 1  
15:30 to 16:00  Afternoon Tea  
16:00 to 17:00 
Steven Low (CALTECH (California Institute of Technology)) Timevarying nonconvex optimization with application to OPF
Optimal power flow (OPF) problems are fundamental for power system operations. They are nonconvex and, in future applications, timevarying. We present a firstorder proximal primaldual algorithm and a secondorder algorithm for general timevarying nonconvex optimization and bound their tracking performance. We incorporate realtime feedback in our algorithms for applications to timevarying OPF problems, and illustrate their tracking performance numerically. (Joint work with Yujie Tang, Caltech, Emiliano Dall’Anese, U of Colorado, Andrey Berstein, NREL) 
INI 1 
09:00 to 10:00 
Laura Diaz Anadon (University of Cambridge); (Harvard University) Innovation in energy technologies and the role of governments
Over the past 20 years there has been a growing amount of empirical research trying to understand the extent to which technology innovation in energy technologies can contribute to mitigating climate change. This presentation will focus on what we know regarding the role of governments in promoting such innovation in lowcarbon energy technologies, with a particular emphasis on the United States and China, the largest greenhouse gas emitters. She will discuss some of the main challenges of building and resourcing effective, empowered institutions able to tackle climate change more rapidly. Her presentation will also discuss how research could help overcome some of these challenges.

INI 1  
10:00 to 11:00 
Chris Dent (University of Edinburgh); (The Alan Turing Institute) Statistical modelling for planning and policy applications
This presents joint thinking with a number of colleagues, key among them being Amy Wilson, Stan Zachary, Michael Goldstein, Henry Wynn, Jim Smith and Peter Challenor. The contributions of various students and postdocs are also acknowledged, and credited in the presentation where relevant.

INI 1  
11:00 to 11:30  Morning Coffee  
11:30 to 12:30 
Florian Doerfler (ETH Zürich) DataEnabled Predictive Control of Autonomous Energy Systems
We consider the problem of optimal and constrained control for unknown systems. A novel dataenabled predictive control (DeePC) algorithm is presented that computes optimal and safe control policies using realtime feedback driving the unknown system along a desired trajectory while satisfying system constraints. Using a finite number of data samples from the unknown system, our proposed algorithm uses a behavioral systems theory approach to learn a nonparametric system model used to predict future trajectories. We show that, in the case of deterministic linear timeinvariant systems, the DeePC algorithm is equivalent to the widely adopted Model Predictive Control (MPC), but it generally outperforms subsequent system identification and modelbased control. To cope with nonlinear and stochastic systems, we propose salient regularizations to the DeePC algorithm. Using techniques from distributionally robust stochastic optimization, we prove that these regularization indeed robustify DeePC against corrupted data. We illustrate our results with nonlinear and noisy simulation case studies from aerial robotics, power electronics, and power systems.

INI 1  
12:30 to 13:30  Lunch at Westminster College  
13:30 to 14:30 
Sean Meyn (University of Florida); (University of Illinois) State Space Collapse in Resource Allocation for Demand Dispatch
The term demand dispatch refers to the creation of virtual energy storage from deferrable loads. The key to success is automation: an appropriate distributed control architecture ensures that bounds on quality of service (QoS) are met and simultaneously ensures that the loads provide aggregate grid services comparable to a large battery system. A question addressed in our 2018 CDC paper is how to control a large collection of heterogeneous loads. This is in part a resource allocation problem, since different classes of loads are more valuable for different services. The evolution of QoS for each class of loads is modeled via a state of charge surrogate, which is a part of the leaky battery model for the load classes. The goal of this paper is to unveil the structure of the optimal solution and investigate short term market implications. The following conclusions are obtained: (i) Optimal power deviation for each of the M 2 load classes evolves in a twodimensional manifold. (ii) Marginal cost for each load class evolves in a twodimensional subspace: spanned by a costate process and its derivative. (iii) The preceding conclusions are applied to construct a dynamic competitive equilibrium model, in which the consumer utility is the negative of the cost of deviation from ideal QoS. It is found that a competitive equilibrium exists, and that the resulting price signals are very different than what would be obtained based on the standard assumption that the utility is with respect to power consumption. It is argued that price signals are not useful for control of the grid since they are inherently open loop. However, the analysis may inform the creation of heuristics for payments within the context of contracts for services with consumers. 
INI 1  
14:30 to 15:30  Panel  INI 1 