Deva P. Seetharam

Occupancy detection in commercial buildings using opportunistic context sources

Accurate occupancy information in commercial buildings can enable several useful applications such as energy management and dynamic seat allocation. Most prior efforts in this space depend on deploying an additional network of deeply coupled sensors to gather occupancy details. This paper presents a novel approach for occupancy detection using only context sources that are commonly available in commercial buildings such as area access badges, Wi-Fi access points, Calendar and Instant Messaging clients. We present models to conduct a situation-centric profiling using such sources and evaluate results of those models. Through a pilot study of a building floor with 5 volunteers for 6 weeks, we demonstrate the potential for detecting occupancies with accuracy as high as 90%.

User-sensitive scheduling of home appliances

Demand response (DR) programs encourage end-use customers to alter their power consumption in response to DR events such as change in real-time electricity prices. Facilitating household participation in DR programs is essential as the residential sector accounts for a sizable portion of the total energy consumed. However, manually tracking energy prices and deciding on how to schedule home appliances can be a challenge for residential consumers who are accustomed to fixed price electricity taris. In this work, we present Yupik, a system that helps users respond to real-time electricity prices while being sensitive to their context and lifestyle. Yupik combines sensing, analytics, and optimization to generate appliance usage schedules that may be used by households to minimize their energy bill as well as potential lifestyle disruptions. Yupik uses jPlugs, appliance level energy metering devices, to continuously monitor the power usage by various home appliances. The consumption patterns as well as data from external sources are analyzed using data mining algorithms to infer users preferred usage profile. Using the preferred profile as a reference, Yupiks optimization engine generates multiple usage plans that attempt to minimize energy and inconvenience costs. Some of Yupiks capabilities are demonstrated with the help of preliminary data collected from a home that was instrumented with jPlugs to monitor the power usage of a few devices.

nPlug: a smart plug for alleviating peak loads

The Indian electricity sector, despite having the worlds fifth largest installed capacity, suffers from a 12.9% peaking shortage. This shortage could be alleviated, if a large number of deferrable loads, particularly the high powered ones, could be moved from on-peak to off-peak times. However, conventional DSM strategies may not be suitable for India as the local conditions usually favor only inexpensive solutions with minimal dependence on the pre-existing infrastructure. In this work, we present nPlug, a smart plug that sits between the wall socket and deferrable loads such as water heaters, washing machines, and electric vehicles. nPlugs combine real-time sensing and analytics to infer peak periods as well as supply-demand imbalance and reschedule attached appliances in a decentralized manner to alleviate peaks whenever possible. They do not require any manual intervention by the end consumer nor any enhancements to the appliances or existing infrastructure. Some of nPlugs capabilities are demonstrated using experiments on a combination of synthetic and real data collected from plug-level energy monitors. Our results indicate that nPlug can be an effective and inexpensive technology to address the peaking shortage.

Softgreen: Towards energy management of green office buildings with soft sensors

This paper describes an approach for saving energy in commercial buildings, based on the information gathered from pre-existing opportunistic context sources. Most energy management systems rely on a heavy instrumentation strategy to infer occupancies, and unfortunately ignore already available opportunistic context sources, that can provide significant information about occupancy. We present models to conduct a Context Profiling with available context sources, to infer spatial occupancy measures. Further, we model electrical loads of several types to infer potential energy savings. Through a pilot study of a building with 5 users for 30 days, we identify intra-building areas where additional instrumentation of occupancy sensors is not necessary and demonstrate potential for significant reduction in energy consumption. We believe such Context Profiling can provide insights to significantly reduce deployment and management costs for future occupancy detection and energy management systems.

Phase identification in smart grids

Electricity is distributed throughout the electrical power network in 3-phase voltage. This power reaches households as a single-phase voltage, generally 115vac or 240vac. This is achieved by allocating households with either phases A, B, or C of the final 3-phase power distributed to the street through a low voltage transformer. A present problem confronting the electrical power industry is identification of which particular phase a household is connected to. This information is often not tracked and the mechanisms for identifying phase require either manual intervention or costly signal injection technologies. Phase information is important as it is a foundation for the larger problem of balancing phase loads. Unbalanced phases lead to significant energy losses and sharply reduced asset lifetimes. In this paper we propose a new approach to compute household phase. Our techniques are novel as they are purely based upon a time series of electrical power measurements taken at the household and at the distributing transformer. Our methods involve the use of integer programming and solutions can be retrieved using branch and bound search algorithms implemented by MIP solvers such as CPLEX. Furthermore, as the number of measurements increase, continuous relaxations of integer programs may also be used to retrieve household phase efficiently. Simulation results using a combination of synthetic and real smart meter datasets demonstrate the performance of our techniques and the number of measurements needed to uniquely identify household phase.

Real-time hybrid state estimation incorporating SCADA and PMU measurements

This paper proposes a novel hybrid state estimation method using traditional SCADA (Supervisory Control And Data Acquisition) and newly deployed limited PMU (Phasor Measurement Unit) measurements. System states are estimated when a set of SCADA and/or PMU measurements come in. As PMU measurements come much faster (typically one sample in 20ms) than SCADA measurements (typically one sample in 10 seconds), in between two SCADA measurements, system states of PMU unobservable buses are interpolated using an interpolation matrix (H)live PMU measurements. In between two SCADA samples, if PMU measurements change significantly, pre-computed interpolation matrix (H) is compensated with a sensitivity change matrix (ΔH) and system states are estimated using the corrected interpolation matrix. In order to compute the ΔH, the method classified the measurement set into four sub-sets i.e. PMU measurements, SCADA measurements of PMU boundary buses with significant change, SCADA measurements adjacent to the selected boundary buses, and remaining SCADA measurements and run a modified weighted least square method with different weights corresponding to each sub-set of measurements. This compensation improves the estimation accuracy significantly. Effectiveness of the proposed scheme is evaluated on a number of IEEE benchmark test systems and evaluation results are presented in this paper.

Real time contingency analysis for power grids

Modern power grids are continuously monitored by trained system operators equipped with sophisticated monitoring and control systems. Despite such precautionary measures, large blackouts, that affect more than a million consumers, occur quite frequently. To prevent such blackouts, it is important to perform high-order contingency analysis in real time. However, contingency analysis is computationally very expensive as many different combinations of power system component failures must be analyzed. Analyzing several million such possible combinations can take inordinately long time and it is not be possible for conventional systems to predict blackouts in time to take necessary corrective actions. To address this issue, we present a scalable parallel implementation of a probabilistic contingency analysis scheme that processes only most severe and most probable contingencies. We evaluate our implementation by analyzing benchmark IEEE 300 bus and 118 bus test grids. We perform contingency analysis up to level eight (contingency chains of length eight) and can correctly predict blackouts in real time to a high degree of accuracy. To the best of our knowledge, this is the first implementation of real time contingency analysis beyond level two.

Varying bandwidth resource allocation problem with bag constraints

We consider the problem of scheduling jobs on a pool of machines. Each job requires multiple machines on which it executes in parallel. For each job, the input specifies release time, deadline, processing time, profit and the number of machines required. The total number of machines may be different at different points of time. A feasible solution is a subset of jobs and a schedule for them such that at any timeslot, the total number of machines required by the jobs active at the timeslot does not exceed the number of machines available at that timeslot. We present an O(log(Bmax/Bmin))-approximation algorithm, where Bmax and Bmin are the maximum and minimum available bandwidth (maximum and minimum number of machines available over all the timeslots). Our algorithm and the approximation ratio are applicable for more a general problem that we call the Varying bandwidth resource allocation problem with bag constraints (BagVBRap). The BagVBRap problem is a generalization of some previously studied scheduling and resource allocation problems.

Learning to be energy-wise: discriminative methods for load disaggregation

In this paper we describe an ongoing project which develops an automated residential Demand Response (DR) system that attempts to manage residential loads in accordance with DR signals. In this early stage of the project, we propose an approach for identifying individual appliance consumption from the aggregate load and discuss the effectiveness of load disaggregation techniques when total load data also includes appliances that are unmonitored even during the training phase. We show that simple discriminative methods can directly predict the appliance states (e.g. on, off, standby) and the predicted state can be used to calculate energy consumed by the appliances. We also show that these methods perform substantially better than the generative models of energy consumption that are commonly used. We evaluated the proposed approach using publicly available REDD data set, and our experimental evaluation demonstrates the improvement in accuracy.

Deep conservation in urban India and its implications for the design of conservation technologies

Rapid depletion of fossil fuels and water resources has become an international problem. Urban residential households are among the primary consumers of resources and are deeply affected by resource shortages. Despite the global nature of these problems, most of the solutions being developed to address these issues are based on studies done in the developed world. We present a study of energy, water and fuel conservation practices in urban India. Our study highlights a culture of deep conservation and the results raise questions about the viability of typical solutions such as home energy monitors. We identify new opportunities for design such as point-of-use feedback technologies, modular solutions, distributed energy storage, harnessing by-products and automated load shifting.