Geoff Lyon

Towards an understanding of campus-scale power consumption

Commercial buildings are significant consumers of electricity. In this paper, we collect and analyze six weeks of data from 39 power meters in three buildings of a campus of a large company. We use an unsupervised anomaly detection technique based on a low-dimensional embedding to identify power saving opportunities. Further, to better manage resources such as lighting and HVAC, we develop occupancy models based on readily available port-level network logs. We propose a semi-supervised approach that combines hidden Markov models (HMM) with standard classifiers such as naive Bayes and support vector machines (SVM). This two step approach simplifies the occupancy model while achieving good accuracy. The experimental results over ten office cubicles show that the maximum error is less than 15% with an average error of 9.3%. We demonstrate that using our occupancy models, we can potentially reduce the lighting load on one floor (about 45 kW) by about 9.5%.

Following the electrons: methods for power management in commercial buildings

Commercial buildings are significant consumers of electricity. The first step towards better energy management in commercial buildings is monitoring consumption. However, instrumenting every electrical panel in a large commercial building is expensive and wasteful. In this paper, we propose a greedy meter (sensor) placement algorithm based on maximization of information gained, subject to a cost constraint. The algorithm provides a near-optimal solution guarantee. Furthermore, to identify power saving opportunities, we use an unsupervised anomaly detection technique based on a low-dimensional embedding. Further, to better manage resources such as lighting and HVAC, we propose a semi-supervised approach combining hidden Markov models (HMM) and a standard classifier to model occupancy based on readily available port-level network statistics.

Real time asset tracking in the data center

The importance and difficultly of asset tracking make it worthy of attention. We focus on data centers consisting of vertical racks where each rack may accommodate a variety of equipment. We describe an asset tracking system which automatically detects and identifies equipment within rack; has “pinpoint” accuracy, i.e., location resolution equals asset size; relays this information to possibly several management back-ends; includes a back-end application that maintains a location history for all equipment; and uses a visualization tool to display both the current state and the history of deployment.The solution features a flexible architecture that simplifies the connection with both existing and future asset management applications. The architecture supports simple configuration, load balancing, and redundancy. Care has been taken to use widely recognized standards wherever possible.

Exploring Advanced Metering Infrastructure Deployments for Commercial and Industrial Sites

Smart meters have become increasingly common as an approach to benchmark and assess building energy use. In this paper, we explore what type of metering infrastructure may be required to derive value from the application of smart meters in the commercial and industrial sectors. As an example, we find that sole reliance upon a site-level smart meter—which has been the focus of most existing deployment models—provides sufficient data to extract summary statistics about how the energy use of a given site may compare to a typical ‘average’ site, but such installations fail to provide adequate detail about where the energy use is occurring or why any discrepancies might be occurring. To resolve these issues, we install a multi-tier advanced metering infrastructure (AMI) at a mixed use (industrial and commercial) campus. We use this AMI deployment to gain insight at different levels of the consumption hierarchy—from sites to buildings, panels, sub-panels, and end loads. The paper concludes by discussing the trade-offs associated with such augmented metering at each level within the hierarchy, with a view towards providing guidelines for AMI deployment at other industrial and commercial campuses.© 2012 ASME

A Self-Configuring Sensing System for Data Centers

Wiring sensors in a data center is extremely expensive in comparison to the wiring of computing equipment. This is due to the central architecture of traditional sensing systems, which requires long wires to be connected between racks and a central box. In addition, a re-layout of the racks after the sensor wires are deployed is practically impossible. We propose using a wireless self-configuring network of smart sensing nodes to alleviate these problems. We explore how to design the sensor control software to be self-reconfiguring when nodes relocate. The software is divided in three layers: network organization, data aggregation and visualization. In this paper, we identify several insights into the thermal monitoring requirements, design issues and initial design solutions for these layers.

Data Analytics for Managing Power in Commercial Buildings

Commercial buildings are significant consumers of electricity. We propose a number of methods for managing power in commercial buildings. The first step toward better energy management in commercial buildings is monitoring consumption. However, instrumenting every electrical panel in a large commercial building is an expensive proposition. In this article, we demonstrate that it is also unnecessary. Specifically, we propose a greedy meter (sensor) placement algorithm based on maximization of information gain subject to a cost constraint. The algorithm provides a near-optimal solution guarantee, and our empirical results demonstrate a 15% improvement in prediction power over conventional methods. Next, to identify power-saving opportunities, we use an unsupervised anomaly detection technique based on a low-dimensional embedding. Furthermore, to enable a building manager to effectively plan for demand response programs, we evaluate several solutions for fine-grained, short-term load forecasting. Our investigation reveals that support vector regression and an ensemble model work best overall. Finally, to better manage resources such as lighting and HVAC, we propose a semisupervised approach combining hidden Markov models (HMMs) and a standard classifier to model occupancy based on readily available port-level network statistics. We show that the proposed two-step approach simplifies the occupancy model while achieving good accuracy. The experimental results demonstrate an average occupancy estimation error of 9.3% with a potential reduction of 9.5% in lighting load using our occupancy models.

Developing Resource Consumption Insights From Campus-Scale Water Monitoring Infrastructures

Water consumption at many commercial campuses is a significant portion of resource expenditure, often with limited or no visibility into the individual branch or point of use locations, all of which summate to provide utility based reporting and invoicing, mostly on a monthly basis.In this paper, we present a case study where a commercial campus’ water distribution system is being instrumented to obtain a more granular measure of water usage. Measurement granularity is improved both in the time domain, transitioning from monthly to hourly or more frequent reporting, and in the spatial domain with all major end loads and significant branch loads being classified or monitored. Specifically, additional instrumentation is deployed in two distinct phases. The first phase added wireless transducers to the existing utility installed mechanical meters, enabling them to transmit consumption data every quarter hour. The second phase will instrument existing branch flow meters and also insert new flow meters to certain end-point loads and sub-branches. This will enable point or clustered data polling on the order of every few seconds. We also obtain additional information by polling an existing HVAC building management system for water related points of interest.We find that the collection and storage of granular water consumption information has the potential to create a detailed demand-side mapping of water usage on campus; providing data with significantly shortened time periods compared to the use of utility billing alone. We use this information to obtain hourly and daily consumption summaries at the site level and for specific end-load devices. From these results, we have created a hybrid consumption estimation of water consumption at the campus level, which contains a mixture of surveyed estimations and dynamic readings. This model provides improved accuracy and insights when compared to static site survey estimations. Due to the age and complexity of the site, primarily a result of numerous engineering changes over the site’s 60 year lifespan and a lack of detailed historical documentation, further work is ongoing to determine which additional endpoint loads or branched sub-sections we will instrument. We plan to use these additional data points to refine our water distribution model; hoping to accurately map individual buildings, floors and functional areas over time.At present, our site level instrumentation has been beneficial in revealing a number of insights regarding unexpected consumption events, most of which were attributed to scheduled maintenance activities. The ongoing monitoring of individual end-point loads has also highlighted areas of significant demand, which could be prioritized for conservation initiatives, and has shown where systemic adjustments could reduce demand peaking and flatten the flow requirements our campus places on the supplying utility.Copyright

Automating server tracking for data centers

Asset management in state of the art data centers is still a manual process. An automated system to track the location of servers and create a real-time inventory would not only improve operations management but also reduce operational costs. This demo shows a novel application of location aware wireless networks, RFID technology and visualization software integrated in end-to-end system to track servers in a data center.