Monday, June 22, 2020
Tutorial: Optimal Power Flow
AEST Mirror 0:00-4:00 London | 7:00-11:00 HKT | 16:00-20:00 PDT | 19:00-23:00 EDT | 9:00-13:00 AEST
Tutorial: Blockchain and Energy
AEST Mirror 9:00-13:00 London | 16:00-20:00 HKT | 1:00-5:00 PDT | 4:00-8:00 EDT | 18:00-22:00 AEST
Tuesday, June 23, 2020
ACM e-Energy Opening
London Mirror 8:15-9:00 London | 15:15-16:00 HKT | 0:15-1:00 PDT | 3:15-4:00 EDT | 17:15-18:00 AEST
PDT Mirror 8:15-9:00 PDT | 16:15-17:00 London | 23:15-0:00 HKT | 11:15-12:00 EDT | 1:15-2:00 AEST
Session 1: Energy Markets
Session Chair: Chenye Wu (Tsinghua University)
London Mirror 9:00-10:30 London | 16:00-17:30 HKT | 1:00-2:30 PDT | 4:00-5:30 EDT | 18:00-19:30 AEST
PDT Mirror 9:00-10:30 PDT | 17:00-18:30 London | 0:00-1:30 HKT | 12:00-13:30 EDT | 2:00-3:30 AEST
Alexander J. M. Kell (Newcastle University); Matthew Forshaw (Newcastle University); A. Stephen McGough (Newcastle University)
paper
Show Abstract
Abstract: Electricity market modelling is often used by governments, industry and agencies to explore the development of scenarios over differing timeframes. For example, how would the reduction in cost of renewable energy impact investments in gas power plants or what would be an optimum strategy for carbon tax or subsidies?
Cost optimization based solutions are the dominant approach for understanding different long-term energy scenarios. However, these types of models have certain limitations such as the need to be interpreted in a normative manner, and the assumption that the electricity market remains in equilibrium throughout. Through this work, we show that agent-based models are a viable technique to simulate decentralised electricity markets. The aim of this paper is to validate an agent-based modelling framework to increase confidence in its ability to be used in policy and decision making.
Our framework can model heterogeneous agents with imperfect information. The model uses a rules-based approach to approximate the underlying dynamics of a real world, decentralised electricity market. We use the UK as a case study, however, our framework is generalisable to other countries. We increase the temporal granularity of the model by selecting representative days of electricity demand and weather using a $k$-means clustering approach.
We show that our framework can model the transition from coal to gas observed in the UK between 2013 and 2018. We are also able to simulate a future scenario to 2035 which is similar to the UK Government, Department for Business and Industrial Strategy (BEIS) projections. We show a more realistic increase in nuclear power over this time period. This is due to the fact that with current nuclear technology, electricity is generated almost instantaneously and has a low short-run marginal cost \cite{Department2016}.
Ahmad Attarha (The Australian National University); Paul Scott (The Australian National University); Sylvie Thiebaux (The Australian National University)
paper
Show Abstract
Abstract: The integration of distributed energy resources (DER) has created a demand-side flexibility which can be traded in the electricity market by aggregators. However, generating bids that accurately represent the flexibility of consumers while maintaining the network limits is a challenging task—especially since the aggregators typically do not have access to the network data nor the bids of other aggregators. To overcome these challenges, we propose a price-generating bidding strategy enabling aggregators that share the same distribution network to participate in the energy and FCAS (frequency control ancillary service) markets. Complying with the Australian National Electricity Market (NEM), we develop energy-FCAS trapeziums that represent aggregators’ energy and FCAS bid interdependency across their fleet of flexible consumers. We also obtain the prices at which the aggregators should submit their energy and FCAS bids. Moreover, to ensure network feasibility for any market clearing output, we obtain the network feasible region using three sets of optimal power flows (OPFs). Aggregators’ trapeziums are then restricted to be within the network feasible region, making them ready to submit to the NEM. We illustrate the effectiveness of our proposed approach using 207 consumers being served by three aggregators in a 69-bus distribution network. The results show that our approach could increase aggregators’ benefits by 18\%, on average, compared to a price-taking approach.
Vivek Deulkar (IIT Bombay); Jayakrishnan Nair (IIT Bombay)
paper
Show Abstract
Abstract: The inherent intermittency of renewable sources like wind and solar
has resulted in a bundling of renewable generators with storage
resources (batteries) for increased reliability. In this paper, we
consider the problem of energy sharing between two such bundles,
each associated with their own demand profiles. The demand profiles
might, for example, correspond to commitments made by the bundle to
the grid. With each bundle seeking to minimize its loss of load
rate, we explore the possibility that one bundle can supply energy
to the other from its battery at times of deficit, in return for a
reciprocal supply from the other when it faces a deficit itself. We
show that there always exist \emph{mutually beneficial} energy
sharing arrangements between the two bundles. Moreover, we show that
Pareto-optimal arrangements involve at least one bundle transferring
energy to the other at the maximum feasible rate at times of
deficit. We illustrate the potential gains from such dynamic energy
sharing via an extensive case study.
Romaric Duvignau (Chalmers University of Technology); Verena Heinisch (Chalmers University of Technology); Lisa Göransson (Chalmers University of Technology); Vincenzo Gulisano (Chalmers University of Technology); Marina Papatriantafilou (Chalmers University of Technology)
paper
Show Abstract
Abstract: Due to ever lower cost, investments in renewable electricity generation and storage have become more attractive to electricity consumers in recent years. At the same time, electricity generation and storage have become something to share or trade locally in energy communities or microgrid systems. In this context, peer-to-peer (P2P) sharing has gained attention, since it offers a way to optimize the cost-benefits from distributed resources, making them financially more attractive. However, it is not yet clear in which situations consumers do have interests to team up and how much cost is saved through cooperation in practical instances. While introducing realistic continuous decisions, through detailed analysis based on large-scale measured household data, we show that the financial benefit of cooperation does not require an accurate forecasting. Furthermore, we provide strong evidence, based on analysis of the same data, that even P2P networks with only 2-5 participants can reach a high fraction (96% in our study) of the potential gain, i.e., of the ideal offline (i.e., non-continuous) achievable gain. Maintaining such small communities results in much lower associated costs and better privacy, as each participant only needs to share its data with 1-4 other peers. These findings shed new light and motivate requirements for distributed, continuous and dynamic P2P matching algorithms for energy trading and sharing.
Session 2: Smart Buildings
Session Chair: Romaric Duvignau (Chalmers University of Technology)
London Mirror 11:00-12:00 London | 18:00-19:00 HKT | 3:00-4:00 PDT | 6:00-7:00 EDT | 20:00-21:00 AEST
PDT Mirror 11:00-12:00 PDT | 19:00-20:00 London | 2:00-3:00 HKT | 14:00-15:00 EDT | 4:00-5:00 AEST
Zimu Zheng (Huawei Technologies Co., Ltd); Daqi Xie (Huawei Technologies Co., Ltd); Jie Pu (Huawei Technologies Co., Ltd); Feng Wang (Huawei Technologies Co., Ltd)
paper
Show Abstract
Abstract: It is well-known that the HVAC (heating, ventilation and air conditioning) dominates electricity consumption in commercial buildings. Existing study on HVAC has shown that it is important to accurately quantify the performance profile of a chiller, namely coefficient of performance (COP), and data-driven COP prediction has been recently proposed. However, the task definition for COP prediction, e.g., the number of needed models and the context when the model should be used, is left as an open question. We propose a framework of \underline{Me}tadata-driven Mu\underline{l}ti-task C\underline{O}P Prediction with Adaptive Task \underline{D}efinition Methodolog\underline{y} (MELODY) which defines and learns multiple COP tasks. To the best of our knowledge, this is the first method that adaptively defines COP prediction tasks according to various datasets. As such, this method can select specific COP models under varied contexts to estimate COP. A key idea is to use metadata to dynamically define multiple tasks. We provide a formal definition of metadata and two sources and methods to extract metadata. We evaluate the performance of our scheme by applying it to real-world data, spanning four months obtained from multiple chillers across eight buildings in two large industrial parks in an international metropolis. The results show that our solution outperforms state-of-the-art COP prediction methods and is able to save on 252 MWh of electricity consumption for one month in each of the eight buildings, which is an improvement of over 35\% compared to the current mode of operation of the chillers in the buildings.
Srinarayana Nagarathinam (Tata Consultancy Services); Vishnu Menon (Tata Consultancy Services); Arunchandar Vasan (Tata Consultancy Services); Anand Sivasubramaniam (Penn State University)
paper
Show Abstract
Abstract: Optimal control of building heating, ventilation, air-conditioning (HVAC) equipment has typically been based on rules and model-based predictive control (MPC). Challenges in developing accurate models of buildings render these approaches sub-optimal and unstable in real-life operations. Model-free Deep Reinforcement Learning (DRL) approaches have been proposed very recently to address this. However, existing works on DRL for HVAC suffer from some limitations. First, they consider buildings with few HVAC units, thus leaving open the question of scale. Second, they consider only air-side control of air-handling-units (AHUs) without taking into the water-side chiller control, though chillers account for a significant portion of HVAC energy. Third, they use a single learning agent that adjusts multiple set-points of the HVAC system.
We present MARCO - Multi-Agent Reinforcement learning COntrol for HVACs that addresses these challenges. Our approach achieves scale by transfer of learning across HVAC sub-systems. MARCO uses separate DRL agents that control both the AHUs and chillers to jointly optimize HVAC operations. We train and evaluate MARCO on a simulation environment with real-world configurations. We show that MARCO performs better than the as-is HVAC control strategy. We find that MARCO achieves performance comparable to an MPC Oracle that has perfect system knowledge; and better than MPC suffering from systemic calibration uncertainties. Other key findings from our evaluation studies include the following: 1) distributed agents perform significantly better than a central agent for HVAC control; 2) cooperative agents improve over competing agents; and 3) domain knowledge can be exploited to reduce the training time significantly.
Yunzhe Guo (Tsinghua University & The Hong Kong Polytechnic University); Dan Wang (The Hong Kong Polytechnic University); Arun Vishwanath (IBM Research Australia); Cheng Xu (The Hong Kong Polytechnic University); Qi Li (Tsinghua University)
note
Show Abstract
Abstract: In recent years, many machine learning (ML) models have been developed for enhancing the performance of heating, ventilation and air conditioning (HVAC) systems. In all these studies, it is commonly assumed that building data is collected and stored at a central location, usually a cloud server, where the ML models are trained. Collecting data in a centralized location introduces privacy concerns since building data can reveal sensitive information such as the arrival and departure patterns of occupants. In this paper, we advocate federated learning (FL), a new distributed learning paradigm, where an overall ML model is trained without the need for exchanging raw data between the data source and the cloud.
It has been noted that model training through FL can compromise the accuracy of ML models. In this paper, we study the question: what is the impact of FL on HVAC model accuracy? As there is no FL platform readily applicable for HVAC analytics, we first develop BuildFL, an open-source platform that is specifically designed for FL of HVAC models. We then present a comprehensive measurement study using five HVAC ML models applied to three building data sets. We analyze the impact of different factors on the model accuracy and set the stage for a deeper study of FL to enable enhanced privacy-preserving HVAC models.
Keynote 1 Anuradha Annaswamy, MIT
London Mirror 13:30-14:30 London | 20:30-21:30 HKT | 5:30-6:30 PDT | 8:30-9:30 EDT | 22:30-23:30 AEST
PDT Mirror 13:30-14:30 PDT | 21:30-22:30 London | 4:30-5:30 HKT | 16:30-17:30 EDT | 6:30-7:30 AEST
Significant changes have occurred all over the world even over the past decade in the energy landscape. Globally, there’s a big push towards a 100% incorporation of wind and solar
power for electricity production, with synergistic support from various technologies. For example, in the US, natural gas prices have declined, costs of renewable energy technologies have come down,
and large-scale battery energy storage technologies have advanced rapidly. There are however a host of challenges, most of which are due to the intermittency and unpredictability of the renewable
energy resources. This talk will focus on some of the solutions for the deep integration of these renewable resources for electricity production that are control-centric. A distributed optimization
approach that judiciously combines renewable generation with storage and flexible loads has the possibility for ensuring power balance even with growing penetration of renewables. Flexibility in other
interdependent infrastructures such as train networks can be integrated with solar and wind power generation nodes, storage sites, and flexible consumption can lead to real-time power balance and
optimal power flow. This presentation will cover some of these challenges, highlights of the current research in distributed optimization, and use-case studies that illustrate the role of distributed
and dynamic optimization in renewable-rich power grids.
Session 3: The Power Grid
Session Chair: Omid Ardakanian (University of Alberta)
London Mirror 15:00-16:30 London | 22:00-23:30 HKT | 7:00-8:30 PDT | 10:00-11:30 EDT | 0:00-1:30 AEST
PDT Mirror 15:00-16:30 PDT | 23:00-0:30 London | 6:00-7:30 HKT | 18:00-19:30 EDT | 8:00-9:30 AEST
Jimmy Horn (Horn Wind LLC); Yutong Wu (University of Texas at Austin); Ali Khodabakhsh (University of Texas at Austin); Evdokia Nikolova (University of Texas at Austin); Emmanouil Pountourakis (Drexel University)
paper
Show Abstract
Abstract: We propose a method to minimize the long-term cost of energy generation while improving grid stability. Currently, the cost of energy generation is minimized myopically (day by day) via the economic dispatch problem, which i) does not internalize the effects of generation variability, ii) does not account for the long-term effects of losing too many existing (paid off) conventional plants, and iii) has the detrimental impact of not systematically maintaining grid inertia. The current dispatch solution favors low cost but inherently more variable renewables, which require intermittent back-up from either conventionals or expensive peakers.
We first propose our Augmented Dispatch for Inertia method which incorporates the cost of maintaining grid inertia stability directly in the economic dispatch selection, thus more accurately capturing the impact of renewable energy growth and conventional plant retirements.
Second, to address the long-term loss of conventional plants due to their underuse, we propose our Balanced Dispatch algorithm that selects key, future-needed conventional generators with enough frequency to maintain their viability. We show via simulation that our methods result in substantially lower long-term generation cost and a notable increase in grid resilience.
Liuzixuan Lin (University of Chicago); Andrew A. Chien (University of Chicago & Argonne National Lab)
paper
Show Abstract
Abstract: Generation type of power plant (e.g. steam, wind) is an important attribute in power grid and energy market studies such as bidding strategy, audit of generation mix, and accounting for load-generation matching. Recently, regional transmission organizations (RTOs) and independent system operators (ISOs) are increasingly redacting a wide range of grid and market data attributes to protect their participants’ business interests. Lack of this information can prevent important power grid research.
We propose techniques to infer power plant generation types based on publicly-available market data. We develop and evaluate these techniques on data available from the Midcontinent Independent System Operator (MISO). Evaluation shows successful classification of
power plants, achieving 100% precision and 99.5% recall for wind plants, and 91.7% overall accuracy. On the basis of generated power, our classification shows 100% precision and 99.8\% recall for wind plants and 93.2% overall accuracy.
Our ultimate goal is to generalize to a wide range of RTOs/ISOs. We explore three feature types (bid pattern, capability, and operation), and evaluate their classification value for MISO. We also assess applicability to other RTOs/ISOs based on available market data. These studies inform the efficacy of the features for generation-type inference in other RTOs/ISOs.
Allan Almeida Santos (Technical University of Darmstadt); Amr Rizk (Ulm University); Florian Steinke (Technische Universität Darmstadt)
paper
Show Abstract
Abstract: Embedding applications consisting of interconnected logical function blocks, denoted Virtual Network Requests (VNR), onto physical compute and communication networks, denoted Substrate Networks (SN), allows for the automatic generation of a variable degree of redundancy.
The need for this feature arises for instance in smart power distribution grids with many decentral devices.
Their heterogeneous communication interconnections often feature low reliability and face frequently changing conditions.
At the same time, high service reliability is required for critical applications such as voltage control.
In this work, we show how to detect potential voltage violations in a medium voltage feeder
with high probability at low monitoring cost.
We employ a probabilistic power flow model to determine the time-dependent required reliability for the links of voltage monitoring VNRs and embed it onto a SN consisting of a mix of plausible smart grid communication technologies.
We use a novel approach based on chance-constrained mixed integer linear programming to generate a minimal, but sufficient degree of redundancy.
This allows for optimal resource usage of the SN, flexibility to adapt the embedding in case of changes of the VNR or SN parameters, and reduced design efforts in comparison to manual redundancy planning.
Compared with a static redundancy setup, the operational communication costs can be more than halved in our simulation experiments.
Brinn Hekkelman (Centrum Wiskunde & Informatica (CWI)); Han La Poutré (Centrum Wiskunde & Informatica (CWI))
paper
Show Abstract
Abstract: The problem of network flow congestion occurring in power networks is increasing in severity. Especially in low-voltage networks this is a novel development. The congestion is caused for a large part by distributed and renewable energy sources introducing a complex blend of prosumers to the network. Since congestion management solutions may require individual prosumers to alter their prosumption, the concept of fairness has become a crucial topic of attention. This paper presents a concept of fairness for low-voltage networks that prioritizes local, outer matching and allocates grid access through fair division of available capacity. Specifically, this paper discusses three distinct principal notions of fair division; proportional, egalitarian, and nondiscriminatory division. In addition, this paper devises an efficient algorithmic mechanism that computes such fair allocations in limited computational time, and proves that only egalitarian division results in incentive compatibility of the mechanism.
Session 4: Electric Vehicles
Session Chair: Mohammad Hajiesmaili (UMass Amherst)
London Mirror 17:00-18:30 London | 0:00-1:30 HKT | 9:00-10:30 PDT | 12:00-13:30 EDT | 2:00-3:30 AEST
PDT Mirror 17:00-18:30 PDT | 1:00-2:30 London | 8:00-9:30 HKT | 20:00-21:30 EDT | 10:00-11:30 AEST
Abdullah Al Zishan (University of Alberta); Moosa Moghimi Haji (University of Alberta); Omid Ardakanian (University of Alberta)
note
Show Abstract
Abstract: This paper proposes an adaptive additive-increase multiplicative-decrease (AIMD)-like algorithm for controlled charging of plug-in electric vehicles in a power system. The proposed algorithm is decentralized and model-free, and relies on congestion signals received from sensors deployed across the network to avoid congestion. We use multi-agent reinforcement learning to dynamically adjust the parameters of the adaptive AIMD algorithm assuming that charging points are independent agents. We adopt imitation learning to pre-train these agents and an off-policy actor-critic deep reinforcement learning algorithm to determine the optimal control in the online setting. Simulation results obtained in a parking station with several charging points corroborate that the proposed algorithm closely tracks the available capacity of the network while avoiding line or transformer overloading, and outperforms the AIMD algorithm and other baselines in terms of utilization.
Jonas Schlund (Friedrich-Alexander-University Erlangen-Nürnberg); Marco Pruckner (Friedrich-Alexander-University Erlangen-Nürnberg); Reinhard German (Friedrich-Alexander-University Erlangen-Nürnberg)
paper
Show Abstract
Abstract: We propose a new methodology for modeling flexibility availability (FlexAbility) of decentralized electric loads, e.g., electric vehicle charging, with an intuitive visualization method. The approach includes a novel method for aggregating and disaggregating flexibility that is more accurate and less complex than existing approaches. In addition, it is suitable for online flexibility determination and dispatch. It is the first which enables to consider a total energy constraint per individual load. We enable the determination of guaranteed aggregated FlexAbility over a time horizon by means of calculating flexibility dispatch paths. We then propose a method for maximizing the bidirectional power flexibility of unidirectional charging for generic applications in the power grid. We combine both new methods in a simulation model of electric vehicles with realistic mobility behavior. We are the first to provide an evaluation of the bidirectional power flexibility from unidirectional charging of electric vehicles, which is found to be bounded by the minimal capability to decrease charging power. We show that there is a trade-off between power and energy flexibility. Today, 20 thousand of the typical electric vehicles in Germany are able to keep bidirectional power flexibility of at least 1.3 MW available during a whole year. The general modeling approach is applicable for other flexible loads with flexible profiles and a total energy constraint as well.
Ye Tian (Iowa State University); Jia Liu (Iowa State University); Cathy Xia (OSU)
paper
Show Abstract
Abstract: Spurred by increasing fuel shortage and environmental concerns, ride-sharing systems have attracted a great amount of attention in recent years. Lying at the heart of most ride-sharing systems is the problem of joint trip-vehicle matching and routing optimization, which is highly challenging and results in this area remain rather limited. This motivates us to fill this gap in this paper. Our contributions in this work are three-fold: i) We propose a new analytical framework that jointly considers trip-vehicle matching and optimal routing; ii) We propose a linearization reformulation that transforms the problem into a mixed-integer linear program, for which moderate-sized instances can be solved by global optimization methods; and iii) We develop a memory-augmented time-expansion (MATE) approach for solving large-sized problem instances, which leverages the special problem structure to facilitate approximate (or even exact) algorithm designs. Collectively, our results advance the state-of-the-art of intelligent ride-sharing and contribute to the field of sharing economy.
Bo Sun (The Hong Kong University of Science and Technology); Tongxin Li (California Institute of Technology); Steven Low (Caltech); Danny H.K. Tsang (The Hong Kong University of Science and Technology)
paper
Show Abstract
Abstract: There is an increasing need for spatial and temporal schedule tailored to the requests and preferences of electric vehicles (EVs) in a network of charging stations. From the perspective of a charging network operator, this paper considers an online decision-making problem that recommends charging stations and the corresponding energy prices to sequential EV arrivals, and schedules the charging allocation to maximize the expected total revenue. To address the uncertainties from future EV arrivals and EVs’ choices with respective to recommendations, we propose an Online Recommendation and Charging schedule algorithm (ORC) that is parameterized by a value function for customized designs. Under the competitive analysis framework, we provide a sufficient condition on the value function that can guarantee ORC to be online competitive. Moreover, we design a customized value function based on the sufficient conditions in an asymptotic case, and then rigorously prove the competitive ratio of ORC in the general case. Through extensive experiments, we show that ORC achieves significant increments of revenues compared to benchmark online algorithms.
Aakash Krishna (TCS Research & Innovation); Ajay Narayanan (TCS Research & Innovation); Sunil Krishnakumar (TCS Research & Innovation); Prasant Misra (TCS Research & Innovation); Arunchandar Vasan (TCS Research & Innovation); Venkatesh Sarangan (TCS Research & Innovation); Anand Sivasubramaniam (Penn State University)
note
Show Abstract
Abstract: In many metropolitan cities, multi-unit residential buildings (MURB) are becoming more common than single-family independent homes due to lack of urban space. MURB residents (around 42% in Europe) are potential adopters of electric vehicles (EV), but lack a private garage for EV charging. They need to exclusively rely on public charging, which currently serves only 5% of EVs. As EVs become more prevalent, the lack of extensive public charging can create a short-term demand-supply mismatch in specific city neighbourhoods, as well as preclude long-term growth in EV adoption.
We believe that uberization of private garage chargers that are typically under-utilized during day-time can alleviate this problem. In this work, we examine how a charging service provider can match public charging demand with private suppliers while using a demand-response based pricing model. We base our study on real-world traffic patterns for the city of Luxembourg by augmenting the Luxembourg SUMO traffic scenario (LuST) simulator. Specifically, an EV’s charging demand is modeled by a state machine with charge/discharge dynamics based on Tesla Model-S. Our preliminary results suggest that the proposed uberization strategy has the potential to gracefully handle demand spikes with higher revenue yield for a charging service provider, even while handling different categories of service users.
Wednesday, June 24, 2020
Keynote 2 David Edwards, Horizon Power, Australia
London Mirror 8:30-9:30 London | 15:30-16:30 HKT | 0:30-1:30 PDT | 3:30-4:30 EDT | 17:30-18:30 AEST
PDT Mirror 8:30-9:30 PDT | 16:30-17:30 London | 23:30-0:30 HKT | 11:30-12:30 EDT | 1:30-2:30 AEST
The Carnarvon DER trials are setting a visionary destination for grid automation in high penetration DER microgrids using VPP technology to optimise network operation through
orchestration of customer rooftop solar and battery storage. This presentation covers some of the key learnings so far, highlighting how the project has employed IoT and IoE, forecasting,
predictive analytics, machine learning and customer value exchange. The goal is a new operational model using clean energy technology that will assist economic development in regional
and remote communities.
Session 5: Human and Energy
Session Chair: Yashar Ghiassi-Farrokhfal (Rotterdam School of Management)
London Mirror 10:00-12:00 London | 17:00-19:00 HKT | 2:00-4:00 PDT | 5:00-7:00 EDT | 19:00-21:00 AEST
PDT Mirror 10:00-12:00 PDT | 18:00-20:00 London | 1:00-3:00 HKT | 13:00-15:00 EDT | 3:00-5:00 AEST
Laura Fiorini (University of Groningen); Linda Steg (University of Groningen); Marco Aiello (University of Stuttgart)
note
Show Abstract
Abstract: Everyday activities requiring electrical or thermal power imply sustainability decisions. Choices for different energy sources, which equipment to use, and the timing of activities have major implications for CO$_2$ emissions. Being aware of each of them and accounting for their impact is nearly impossible. First, it is unclear how to assess the sustainability footprint of a decision; second, the complexity of the implications of all such decisions is overwhelming. To make things more concrete, we consider a simple as well as common task: cooking a dish of pasta. We measure the sustainability of the decisions involved in terms of CO$_2$ emissions and we use historical data of German CO$_2$-emission intensity calculated with both the average method and the marginal one. We find that starting from hot or cold tap water can imply up to 35% difference in emissions, depending on the timing and the chosen equipment. However, the complexity and size of information involved in such sustainability choices require the adoption of digitalized and automated systems, which, in turn, raises questions about user acceptability and (mis)trust in such technologies.
Andreas Reinhardt (TU Clausthal); Christoph Klemenjak (University of Klagenfurt)
paper
Show Abstract
Abstract: Electrical consumption data contain a wealth of information, and their collection at scale is facilitated by the deployment of smart meters. Data collected this way is an aggregation of the power demands of all appliances within a building, hence inferences on the operation of individual devices cannot be drawn directly. By using methods to disaggregate data collected from a single measurement location, however, appliance-level detail can often be reconstructed. A major impediment to the improvement of such disaggregation algorithms lies in the way they are evaluated so far: Their performance is generally assessed using a small number of publicly available electricity consumption data sets recorded from actual buildings. As a result, algorithm parameters are often tuned to produce optimal results for the used data sets, but do not necessarily generalize to different input data well. We propose to break this tradition by presenting a toolchain to create synthetic benchmarking data sets for the evaluation of disaggregation performance in this work. Generated synthetic data with a configurable amount of concurrent appliance activity is subsequently used to comparatively evaluate eight existing disaggregation algorithms. This way, we not only create a baseline for the comparison of newly developed disaggregation methods, but also point out the data characteristics that pose challenges for the state-of-the-art.
Vadim Arzamasov (Karlsruhe Institute of Technology (KIT)); Rebecca Schwerdt (Karlsruhe Institute of Technology (KIT)); Shahab Karrari (Karlsruhe Institute of Technology (KIT)); Klemens Böhm (Karlsruhe Institute of Technology (KIT)); Tien Bach Nguyen (Karlsruhe Institute of Technology (KIT))
paper
Show Abstract
Abstract: Large-scale smart meter roll-outs all over the world are one effect of the ongoing energy transition. This poses a significant risk to consumer’s privacy. Battery based load hiding (BBLH)—where an energy storage system is employed to obscure actual demand patterns—is one possibility to still retain privacy. In recent years many different BBLH algorithms have been proposed. But although most of them were assessed with some formally defined privacy measure, the current state of the art sorely lacks any comparability.
We give an overview of privacy measures proposed for this scenario, available storage technologies, and datasets used for the assessment of BBLH. Furthermore, we conduct a study of how these factors influence the ratings of several state-of-the-art BBLH algorithms. Our results illustrate the need for standardization as well as further research into meaningful privacy measures. Achieving this is necessary for private households to make an informed decision on which BBLH algorithm is best for their specific situation.
Laura Fiorini (University of Groningen); Marco Aiello (University of Stuttgart)
paper
Show Abstract
Abstract: Buildings’ energy consumption accounts for approximately 35% of
emissions in most industrialized countries. In spite of several
studies on the economy of energy management in buildings, the
environmental aspect has often been overlooked. Therefore, in the
context of decarbonization, we investigate the potential for smart
homes to lower their CO$_2$ footprint while saving on their energy
bills. We model a smart home as a multi-energy system equipped
with several technologies to satisfy both electric and thermal demands.
A home energy management system (HEMS) coordinates
the supply and demand of energy carriers by shifting consumption
in time and by changing energy vectors based on dynamic energy
prices and marginal CO$_2$-emission intensities. The HEMS aims at
reducing daily CO$_2$ emissions and/or energy costs preserving user’s
satisfaction. Due to the binary nature of on-off decisions and information
uncertainty, we formulate a multi-objective mixed-integer
linear programming (MILP) problem within a model predictive control
(MPC) framework. Using prices and CO$_2$-emission intensities
of the German power grid, our approach is effective in reducing
both energy costs and CO$_2$ emissions, balancing between the two
objectives. The results show that integrating energy carriers has a
higher impact than time-flexible loads. If solar panels are available,
emissions and costs strongly depend on the importance given by
the users to the environmental and economic goals.
June Lukuyu (University of Massachusetts Amherst); Aggrey Muhebwa (University of Massachusetts Amherst); Jay Taneja (University of Massachusetts Amherst)
paper
Show Abstract
Abstract: Though electricity access remains out of reach for roughly one billion primarily rural and low-income people, crucial strides have been made in developing new pathways for connecting households and businesses to electricity supplies. Among these, decentralized minigrids — typically comprised of generation, storage, and a medium- and low-voltage distribution network — have considerable technical promise for balancing recent advances in decentralized generation as well as grid sensing and communication systems with the overwhelming economies-of-scale enjoyed by electricity grids. However, low revenues and, in response, high tariffs necessary for cost recovery stifle the widespread development of this promising pathway for electrification.
In this paper, we study techniques for addressing the principal challenge for sustainable minigrids: demand stimulation among rural customers. Specifically, we evaluate the potential for conversion of diesel-based fishing boats in Lake Victoria to electric motor and battery-based systems that can provide a crucial anchor load for a nascent 650 $kWp$ hybrid solar-battery-diesel minigrid. We conduct a survey among fishing boat operators ($n=69$) to characterize the target population and deploy a custom tracking system to measure fishing boat movement patterns. Using these primary data along with secondary data on customer consumption, we select a candidate electric mobility system, create synthetic loads of residential and business customers, and construct technical and financial models of the complete minigrid system. We then use these models to evaluate the excess capacity on the minigrid for electric boats, evaluate the tradeoffs among electric mobility and manufacturing on the minigrid, and assess the impacts of demand response capabilities for charging the boats. We find that electric boat charging contributes to at least 17% more consumption per day resulting in substantial technical as well as financial value to the minigrid system, though perhaps at the cost of additional use of the system’s backup diesel generator. We find that adding shifting capabilities to electric boat charging can save up to 6% of diesel expenditures at little to no impact on the system Net Present Value. We combine these minigrid-scale evaluations with design considerations for a future boat tracking system, providing guidance for minigrid designers and operators to incorporate the potentially attractive load class of electric mobility systems.
Day 2 Q&A Social Session (Alpha Session)
London Mirror 12:15-13:15 London | 19:15-20:15 HKT | 4:15-5:15 PDT | 7:15-8:15 EDT | 21:15-22:15 AEST
PDT Mirror 12:15-13:15 PDT | 20:15-21:15 London | 3:15-4:15 HKT | 15:15-16:15 EDT | 5:15-6:15 AEST
Poster Session
London Mirror 13:30-15:00 London | 20:30-22:00 HKT | 5:30-7:00 PDT | 8:30-10:00 EDT | 22:30-0:00 AEST
PDT Mirror 13:30-15:00 PDT | 21:30-23:00 London | 4:30-6:00 HKT | 16:30-18:00 EDT | 6:30-8:00 AEST
Best Paper Award and ACM e-Energy 2021
London Mirror 15:00-15:45 London | 22:00-22:45 HKT | 7:00-7:45 PDT | 10:00-10:45 EDT | 0:00-0:45 AEST
PDT Mirror 15:00-15:45 PDT | 23:00-23:45 London | 6:00-6:45 HKT | 18:00-18:45 EDT | 8:00-8:45 AEST
Session 6: Forecasting and Data
Session Chair: Jay Taneja (UMAss Amherst)
London Mirror 16:15-17:30 London | 23:15-0:30 HKT | 8:15-9:30 PDT | 11:15-12:30 EDT | 1:15-2:30 AEST
PDT Mirror 16:15-17:30 PDT | 0:15-1:30 London | 7:15-8:30 HKT | 19:15-20:30 EDT | 9:15-10:30 AEST
Benedikt Heidrich (Karlsruhe Institute of Technology); Marian Turowski (Karlsruhe Institute of Technology); Nicole Ludwig (Karlsruhe Institute of Technology); Ralf Mikut (Karlsruhe Institute of Technology); Veit Hagenmeyer (Karlsruhe Institute of Technology)
paper
Show Abstract
Abstract: Forecasting the energy demand is essential for network operators to balance the grid, in particular with the increasing share of renewable energy sources. Neural networks, especially deep neural networks, have shown promising results in recent forecasting tasks. However, they often struggle learning periodicities in time series efficiently. In line with the finding that deep learning can be improved with statistical information, we introduce profile neural networks based on the fast and promising convolutional neural networks. The underlying idea of profile neural networks is that decomposing periodic energy time series into a standard load profile, a trend, and a colorful noise module improves the forecasting accuracy. The proposed deep neural network architecture is applied to real-world electricity data from buildings on a university campus, more specifically of one building with strong seasonal variation and one building with weak seasonal variation. The new architecture outperforms current state-of-the-art deep learning benchmark models regarding the forecasting accuracy on forecast horizons of one day and one week-ahead, improving the mean absolute scaled error by up to 25%, as well as regarding the trade-off between training time and accuracy.
Marcus Voss (Technische Universität Berlin (DAI-Labor))
note
Show Abstract
Abstract: What makes a household-level short-term load forecast “good”? Individual household load profiles are intermittent, as distinct peaks correspond to specific activities in the household. Using traditional point-wise error metrics to assess household-level forecasts may lead to, for instance, double-digit mean absolute percentage errors. One reason is a double penalty incurred if a peak is forecasted correctly in amplitude, but with a small delay in time. An adjusted forecast error measure based on local permutations was proposed to assess household-level forecasts by optimally aligning the peaks bounded by a displacement limit. This work shows how the choice of this parameter leads to different “best” forecasts in terms of specific applications, namely the optimization objectives of an energy management system. For that, different parameterizations of the Local Permutation Invariant (LPI) distance are compared within k-Nearest Neighbors as a forecasting model for three different optimization objectives. A simulation study on 100 households of the CER dataset shows that the optimal parameterization can decrease the peak load on average by over 22.5\% compared to the Euclidean distance. However, for increasing self-sufficiency and minimizing costs, no significant improvements can be achieved. This implies that household-level forecasts should generally be evaluated in terms of their application, as traditional metrics as a proxy may not express its “goodness” adequately.
Akhil Soman (University of Massachusetts Amherst); Amee Trivedi (University of Massachusetts Amherst); David Irwin (University of Massachusetts Amherst); Beka Kosanovic (University of Massachusetts Amherst); Benjamin McDaniel (University of Massachusetts Amherst); Prashant Shenoy (University of Massachusetts Amherst)
note
Show Abstract
Abstract: Battery-based energy storage has emerged as an enabling technology for a variety of grid energy optimizations, such as peak shaving and cost arbitrage. A key component of battery-driven peak shaving optimizations is peak forecasting, which predicts the hours of the day that see the greatest demand. While there has been significant prior work on load forecasting, we argue that the problem of predicting periods where the demand peaks for individual consumers or micro-grids is more challenging than forecasting load at a grid scale. We propose a new model for peak forecasting, based on deep learning, that predicts the $k$ hours of each day with the highest and lowest demand. We evaluate our approach using a two year trace from a real micro-grid of 156 buildings and show that it outperforms the state of the art load forecasting techniques adapted for peak predictions by 11-32\%. When used for battery-based peak shaving, our model yields annual savings of \$496,320 for a 4 MWhr battery for this micro-grid.
John R. Ward (Okta, Inc.); Sean K. Barker (Bowdoin College)
paper
Show Abstract
Abstract: The proliferation of smart outlets and meters with submetering capabilities has led to an explosion in the availability of device-level energy data. The increasing volume of current and historical data presents a storage and distribution challenge, particularly for utilities and large-scale energy datasets. To address these challenges, we present Powerstrip, a fast, effective, and nearly-lossless compression algorithm for integer energy data. Powerstrip is optimized for device-level measurements and exploits common characteristics of real-world energy consumption to achieve typical compression rates over 90% on such data. We evaluate Powerstrip on real-world energy data and compare against multiple state-of-the-art compression algorithms. Our experiments show that when compared to the best reference algorithms, Powerstrip achieves the highest compression ratios (by up to 35%) as well as the fastest speeds (by up to 70%). We also present case studies demonstrating the potential of Powerstrip for large-scale energy data storage and distribution.
Day 2 Q&A Social Session (Beta Session)
London Mirror 17:30-18:30 London | 0:30-1:30 HKT | 9:30-10:30 PDT | 12:30-13:30 EDT | 2:30-3:30 AEST
PDT Mirror 17:30-18:30 PDT | 1:30-2:30 London | 8:30-9:30 HKT | 20:30-21:30 EDT | 10:30-11:30 AEST
Thursday, June 25, 2020
Keynote 3 Pierluigi Mancarella, University of Melbourne, Australia
London Mirror 8:30-9:30 London | 15:30-16:30 HKT | 0:30-1:30 PDT | 3:30-4:30 EDT | 17:30-18:30 AEST
PDT Mirror 8:30-9:30 PDT | 16:30-17:30 London | 23:30-0:30 HKT | 11:30-12:30 EDT | 1:30-2:30 AEST
Digital energy systems (DES) are highly distributed cyber-physical systems in which small-scale distributed energy resources such as solar PV,
different types of storage, controllable loads, etc., can be actively monitored and controlled via pervasive availability of ICT and smart grid technologies.
This keynote address will discuss opportunities and challenges for emerging DES platforms in the context of low-carbon electricity grids with deep penetration
of renewables. Particular focus will be put on how DES could be intelligently orchestrated to provide grid flexibility services and participate in different
energy markets, as well as autonomously controlled to respond to extreme (for example, weather-driven) events, thus enhancing grid resilience. Specific applications
that will be shown will cover technical, commercial and regulatory aspects of DES from a number of recent projects in Australia, UK, and Europe. The final aim is
to illustrate how development of DES platforms for smart buildings, smart communities, microgrids, virtual power plants, and distributed energy marketplaces can
facilitate an affordable, secure and resilient transition towards a low-carbon energy future.
Session 7: Energy Transmission and Control
Session Chair: Jayakrishnan Nair (IIT Bombay)
London Mirror 10:00-12:00 London | 17:00-19:00 HKT | 2:00-4:00 PDT | 5:00-7:00 EDT | 19:00-21:00 AEST
PDT Mirror 10:00-12:00 PDT | 18:00-20:00 London | 1:00-3:00 HKT | 13:00-15:00 EDT | 3:00-5:00 AEST
Fabian Neumann (Karlsruhe Institute of Technology); Tom Brown (Karlsruhe Institute of Technology)
paper
Show Abstract
Abstract: The common linear optimal power flow (LOPF) formulation that underlies most transmission expansion planning (TEP) formulations uses bus voltage angles as auxiliary optimization variables to describe Kirchhoff’s voltage law. As well as introducing a large number of auxiliary variables, the angle-based formulation has the disadvantage that it is not well-suited to considering the connection of multiple disconnected networks. It is, however, possible to circumvent these variables and reduce the required number of constraints by expressing Kirchhoff’s voltage law directly in terms of the power flows, based on a cycle decomposition of the network graph. For generation capacity expansion with multi-period LOPF, this equivalent reformulation was shown to reduce solving times by an order of magnitude. Allowing line capacity to be co-optimized in a discrete TEP problem makes it a non-convex mixed-integer problem. This paper develops a novel cycle-based reformulation for the TEP problem with LOPF and compares it to the standard angle-based formulation. The combinatorics of the connection of multiple disconnected networks is formalized for both formulations, a topic which has not received attention in the literature. The cycle-based formulation is shown to conveniently accommodate synchronization options. Since both formulations use the big-$M$ disjunctive relaxation, useful derivations for suitable big-$M$ values are provided. The competing formulations are benchmarked on a realistic generation and transmission expansion model of the European transmission system at varying spatial and temporal resolutions. The cycle-based formulation solves up to 31 times faster for particular cases, while averaging at a speed-up of factor 4.
Dorothea Wagner (Karlsruhe Institute of Technology); Matthias Wolf (Karlsruhe Institute of Technology)
paper
Show Abstract
Abstract: When designing an electric transmission grid, it is important to ensure that
the resulting grid is reliable. In particular, it should remain operable if
one piece of equipment fails (N-1 criterion).
In this work we focus on the failure of single transmission lines.
We consider a
criticality measure by Witthaut et al., which captures the
dynamic behavior of failing lines. The criterion itself is based on
maximum graph-theoretic flows in suitably defined residual networks. In a
first step, we
compare it to the N-1 criterion and find that networks without critical
edges tend to satisfy the N-1 criterion.
We then formulate the criticality measure as set of linear constraints, which
may form a building block in transmission network design problems.
In particular, we introduce these constraints into a basic
Transmission Network Expansion Planning (TNEP) formulation,
obtaining models for the two problems Criticality-Constrained Transmission Network Expansion Planning (CC-TNEP) and Criticality Minimal Expansion (CME).
We study the effects of adding these constraints on the time needed for
solving the models.
Moreover, we provide a simple heuristic for CME, which often finds optimal
solutions but in less time.
Tongxin Li (California Institute of Technology); Steven H. Low (California Institute of Technology); Adam Wierman (California Institute of Technology)
paper
Show Abstract
Abstract: Consider a system operator that wishes to optimize its objectives over time subject to operational constraints as well as private constraints of controllable loads managed by an aggregator. In this paper, we design a $\textit{real-time}$ feedback signal for the aggregator to quantify and communicate its available flexibility to the system operator. The proposed feedback signal at each time is the conditional probability of future feasible trajectories that will be enabled by the operator’s decision. We show that it is the unique distribution that maximizes a system capacity for flexibility. It allows the system operator to maintain feasibility and enhance future flexibility while optimizing its objectives. We illustrate how the design can be used by the system operator to perform online cost minimization and real-time capacity estimation, while provably satisfying the private constraints of the loads.
Adithya Ramanujam (Indian Institute of Technology, Bombay); Mahesh Parihar (Indian Institute of Technology,Bombay); Suchitra Swain (Indian Institute of Technology, Bombay); Krithi Ramamritham (IIT Bombay)
note
Show Abstract
Abstract: Imbalances in electricity supply and demand are a significant problem in developing countries resulting in rolling blackouts/load shedding. Due to the inconvenience caused by blackouts, several brownout (partial blackout) strategies have been proposed. In this paper, we resort to keeping the electricity demand of a household within a specified threshold. Our approach combines two distinct methods of end-point load (appliance) control: priority-based techniques and combinatorial optimization-based techniques to create a hybrid approach that gives users flexibility in configuring their preferences on-the-fly. We have quantified the user preferences using the Analytical Hierarchy Process, which is useful for solving such Multi-Criteria Decision-Making problems.
Sascha Gritzbach (Karlsruhe Institute of Technology (KIT)); Dorothea Wagner (Karlsruhe Institute of Technology (KIT)); Matthias Wolf (Karlsruhe Institute of Technology (KIT))
paper
Show Abstract
Abstract: The Wind Farm Cabling Problem (WCP) aims at finding the cost-minimal inter-array cable
routing, also known as internal cable layout,
of a wind farm so that all turbine generation
is transmitted to the substations. For each possible connection in
the wind farm, one of several cable types can be selected. Each cable
type comes with a thermal capacity and unit length costs.
WCP can be modeled as a graph theoretic minimum-cost flow problem
with a step-cost function on each edge.
We extend a deterministic “hill-climbing” heuristic from the literature.
This heuristic runs into local minima from which it is not able
to recover. We embed this algorithm into a framework which involves
strategies for escaping these minima. These escaping strategies allow
the heuristic to
descend into other, possibly better, minima.
We design three such strategies and provide
an extensive statistical evaluation comparing these strategies. The best combination of strategies is
evaluated against Gurobi 9.0.0 on a MILP formulation
and a Simulated Annealing-based heuristic from the
literature on publicly available synthetic benchmark sets.
Our simulations show that our framework works exceptionally well
on the largest benchmark instances where it provides better solution
within 15 minutes than
Gurobi within one day on 80 % of the input instances.
The simulations on the benchmark sets are complemented by
a case study on the world’s soon-to-be largest offshore wind farm: Hornsea One.
Day 3 Q&A Social Session (Alpha Session)
London Mirror 12:15-13:15 London | 19:15-20:15 HKT | 4:15-5:15 PDT | 7:15-8:15 EDT | 21:15-22:15 AEST
PDT Mirror 12:15-13:15 PDT | 20:15-21:15 London | 3:15-4:15 HKT | 15:15-16:15 EDT | 5:15-6:15 AEST
Session 8: Solar PV
Session Chair: Sean Barker (Bowdoin College)
London Mirror 13:30-15:00 London | 20:30-22:00 HKT | 5:30-7:00 PDT | 8:30-10:00 EDT | 22:30-0:00 AEST
PDT Mirror 13:30-15:00 PDT | 21:30-23:00 London | 4:30-6:00 HKT | 16:30-18:00 EDT | 6:30-8:00 AEST
Julian De Hoog (IBM Research Australia); Stefan Maetschke (IBM Research Australia); Peter Ilfrich (IBM Research Australia); Ramachandra Rao Kolluri (IBM Research Australia)
note
Show Abstract
Abstract: Solar photovoltaic (PV) is the fastest growing form of energy generation today, and many countries are seeing significant uptake of distributed solar PV on the rooftops of homes and businesses. However, many of these systems are not accurately registered, and central records of distributed solar PV are often not up-to-date. At the same time, high levels of solar PV are introducing challenges for many stakeholders in the energy sector, such as market operators and network operators, who need to forecast total rooftop solar PV generation across entire regions. One possible solution to this problem is to identify existing solar PV generation systems using overhead satellite and aerial imagery. While there have been early promising attempts in this direction, there are nevertheless many important research challenges that remain to be addressed. In this paper we survey the state of the art in this nascent area, describe the challenges that exist, and advocate for novel research questions that are worthy of further exploration. By identifying these areas of interest we aim to generate greater awareness of the potential value of satellite and aerial imagery for identification of solar PV, which will ultimately facilitate large scale uptake of solar PV and other renewable generation technologies.
Peter Lusis (Monash University); Ariel Liebman (Monash University); Lachlan L. H. Andrew (University of Melbourne); Guido Tack (Monash University)
paper
Show Abstract
Abstract: Autonomous droop control PV inverters have improved voltage regulation compared to the inverters without grid support functions, but more flexible control techniques will be required as the number of solar photovoltaic (PV) installations increases. This paper studies three inverter future deployment scenarios with droop control inverters, non-exporting inverters, and coordinated inverter control (CIC). The network operation and the interaction between various inverter control methods are studied by simulating inverter operation on two low-voltage networks. Considering 30% PV penetration as the base case, we demonstrate that coordinated inverters can mitigate overvoltages and voltage fluctuations caused by the tripping of passive inverters in 85% of PV location cases when at least as many coordinated as passive inverters are deployed on the 114-node test feeder. However, this rate reduced to 37% with the IEEE 906-node network demonstrating that the deployment of coordinated inverter control may not be able to reverse passive inverter-related voltage disturbances when the build-up of passive inverters has reached a certain threshold.
The aggregated PV output from coordinated inverters can be also used to provide grid support services. When the low-voltage networks operate close to the upper voltage limits, the change in the power output from coordinated inverters following a regulation request may be partially offset by passive inverters. Considering an equal number of passive and coordinated inverters, this paper shows that for each unit of the down-regulation request delivered by coordinated inverters, passive inverter output may increase by up to 0.2 units and decrease by up to 0.45 units during coordinated inverter up-regulation.
Vishnu Arayamparambil Vinaya Mohanan (The University of Melbourne); Robin John Evans (The University of Melbourne); Iven Mareels (IBM Research Australia); Ramachandra Rao Kolluri (IBM Research Australia)
paper
Show Abstract
Abstract: Large scale penetration of grid-following inverters into the electricity network presents various technical challenges to grid reliability. It is well-known that the ability of the grid to maintain a stable frequency is inhibited by adding traditional photovoltaic (PV) generators. In this work, a detailed model of a simplified grid is presented and it is shown that the proportion of PV generation and instability are positively correlated. A pair of eigenvalues associated with the field dynamics of the synchronous generators undergo a Hopf bifurcation can be attributed to this instability. Such a Hopf bifurcation severely constricts the feasible operating domain of the grid and may hinder normal operation. Modifying traditional grid-tied PV controllers and its impact grid stability is assessed through small-signal, bifurcation and transient numerical analysis. Traditional PV controllers that are modified to virtual synchronous machine (VSM) type controllers show improvement in system damping. Unlike traditional grid-tied inverters, VSM type inverters participate in critical modes of the synchronous generator (SG) and augments the operational domain of the SG+VSM system significantly, more importantly, almost eliminating the need for renewable energy curtailment. A case-study approach is used to present some key results on improvements in damping ratio, feasibility domain and transient stability. Finally, a feasibility domain curve is introduced and discussed in an aim to generalize the overall stability of any such system.
Yiju Ma (The University of Sydney); Daniel Gebbran (The University of Sydney); Archie Chapman (The University of Queensland); Gregor Verbic (The University of Sydney)
paper
Show Abstract
Abstract: Rapidly rising PV installations in low-voltage (LV) distribution networks means that they are likely to exceed the network hosting capacity. For this reason, distribution network service providers (DNSP) have begun to mandate connection codes, such as inverter Volt/Var control and/or PV active power curtailment, to mitigate the network problems. The latter approach allows for more PV installations which may cause an existing PV system to become inefficient as it is curtailed more often. This paper investigates the effects on overall economic efficiency and individual customer welfare of natural uncoordinated rooftop PV investment processes that arise when customers invest in PV systems independently to maximize their individual welfare. We develop a novel game theoretic framework that computes the annual payoffs to customers for different PV investment sizes, given the installations of other customers. This calculation is based on an AC power flow model that includes inverter connection standards that link customers’ annual payoffs via their effects on AC network voltages and consequent PV curtailment responses. We show that the interaction of PV investments produces a convex congestion game with continuous action sets, which has a pure Nash equilibrium that can be found using an adaptive learning process. Then, to evaluate the efficiency of the investments under the game model, we compute an optimal centrally-coordinated PV investment profile, found by solving an optimal PV sizing problem that maximizes social welfare across all customers. Comparing the value of investment patterns for the game and the centrally-coordinated optimization shows: (i) a price of anarchy of 1.5, which indicates the efficiency loss resulting from uncoordinated PV investments, and (ii) the inequity of a skewed distribution of benefits, penalising customers closer to the distribution transformer and benefiting those towards the end of the feeder. This model provides a quantitative tool for evaluating policies and regulations that improved coordination and allocation of PV (and other energy distributed energy resources) hosting capacity between customers on LV feeders.
Session 9: Energy Storage and Batteries
Session Chair: Jia Liu (Iowa State University)
London Mirror 15:30-17:00 London | 22:30-0:00 HKT | 7:30-9:00 PDT | 10:30-12:00 EDT | 0:30-2:00 AEST
PDT Mirror 15:30-17:00 PDT | 23:30-1:00 London | 6:30-8:00 HKT | 18:30-20:00 EDT | 8:30-10:00 AEST
Nasir Mehmood (Lahore University of Management Sciences); Naveed Arshad (Lahore University of Management Sciences)
note
Show Abstract
Abstract: The need for a high ramping energy resource for frequency regulation is increasing due to the high penetration of intermittent and variable renewable energy sources, such as wind and solar, in the electricity grid. Traditionally, special generators have been used for frequency regulation. These generators can provide high capacity but have a very slow response time. Battery energy storage (BES) has gotten tremendous attention due to the advancement in technology. BES has a very fast response time, which makes it suitable for frequency regulation. In this paper, we perform an economic analysis of a distributed energy storage participating in the PJM and NYISO regulation markets. The distributed storage consists of many small consumers’ installed batteries. A centralized entity at a microgrid level controls the distributed storage using our proposed algorithms. The economic analysis is performed from the perspective of individual storage owners. Our results show that the five-year net-present-value (NPV) of the consumers’ investment is positive if the utility shares 30% (or above) of the regulation revenue with the storage owners and keeps the rest of the 70%.
Diego Kiedanski (Télécom Paris); Ariel Orda (Technion); Daniel Kofman (Télécom Paris)
paper
Show Abstract
Abstract: As of today, energy storage for residential consumers represents a considerable investment that is not guaranteed to be profitable. Shared investment models in which a group of consumers jointly acquires energy storage have been proposed in the literature to increase the attractiveness of these devices. Such models naturally employ concepts of cooperative game theory.
In this paper, we extend the state-of-the-art cooperative game for modeling the shared investment in storage by adding two crucial extensions: {\em stochasticity} of the load and {\em discreetness} of the storage device capacity.
As our goal is to increase storage capacity in the grid, the number of devices that would be acquired by a group of players that cooperate according to our proposed scheme is compared to the number of devices that would be bought by consumers acting individually.
Under the same criteria of customer profitability, simulations using real data reveal that our proposed scheme can increase the deployed storage capacity between $100\%$ and $250\%$.
Rishikesh Jha (University of Massachusetts Amherst); Stephen Lee (University of Pittsburgh); Srinivasan Iyengar (Microsoft Research India); Mohammad Hajiesmaili (University of Massachusetts Amherst); David Irwin (University of Massachusetts Amherst); Prashant Shenoy (University of Massachusetts, Amherst)
paper
Show Abstract
Abstract: Reducing our reliance on carbon-intensive energy sources is vital for reducing the carbon footprint of the electric grid. Although the grid is seeing increasing deployments of clean, renewable sources of energy, a significant portion of the grid demand is still met using traditional carbon-intensive energy sources. In this paper, we study the problem of using energy storage deployed in the grid to reduce the grid’s carbon emissions. While energy storage has previously been used for grid optimizations such as peak shaving and smoothing intermittent sources, our insight is to use distributed storage to enable utilities to reduce their reliance on their less efficient and most carbon-intensive power plants and thereby reduce their overall emission footprint. We formulate the problem of emission-aware scheduling of distributed energy storage as an optimization problem, and use a robust optimization approach that is well-suited for handling the uncertainty in load predictions and intermittent renewables. We evaluate our approach using a state of the art neural network load forecasting technique and real load traces from a distribution grid. Our results show a reduction of $>$0.5 million kg in annual carbon emissions —- equivalent to a reduction of 23.3% in our electric grid emissions.
Jiasheng Zhang (Tsinghua University); Nan Gu (Tsinghua University); Chenye Wu (Tsinghua University)
paper
Show Abstract
Abstract: Energy storage has exhibited great potential in providing flexibility in power system to meet critical peak demand and thus reduce the overall generation cost, which in turn stabilizes the electricity prices. In this work, we exploit the opportunities for the independent system operator (ISO) to invest and manage storage as public asset, which could systematically provide benefits to the public. Assuming a quadratic generation cost structure, we apply parametric analysis to investigate the ISO’s problem of economic dispatch, given variant quantities of storage investment. This investment is beneficial to users on expectation. However, it may not necessarily benefit everyone. We adopt the notion of marginal system cost impact (MCI) to measure each user’s welfare and show its relationship with the conventional locational marginal price. We find interesting convergent characteristics for MCI. Furthermore, we perform $k$-means clustering to classify users for effective user profiling and conduct numerical studies on both prototype and IEEE test systems to verify our theoretical conclusions.
Day 3 Q&A Social Session (Beta Session)
London Mirror 17:30-18:30 London | 0:30-1:30 HKT | 9:30-10:30 PDT | 12:30-13:30 EDT | 2:30-3:30 AEST
PDT Mirror 17:30-18:30 PDT | 1:30-2:30 London | 8:30-9:30 HKT | 20:30-21:30 EDT | 10:30-11:30 AEST
Friday, June 26, 2020