Advances in machine learning and computer vision, combined with increased access to unstructured data (e.g., images and text), have created an opportunity for automated extraction of building characteristics, cost‐effectively, and at scale. These characteristics are relevant to a variety of urban and energy applications, yet are time consuming and costly to acquire with today’s manual methods. Several recent research studies have shown that in comparison to more traditional methods that are based on features engineering approach, an end‐to‐end learning approach based on deep learning algorithms significantly improved the accuracy of automatic building footprint extraction from remote sensing images. However, these studies used limited benchmark datasets that have been carefully curated and labeled. How the accuracy of these deep learning‐based approach holds when using less curated training data has not received enough attention. The aim of this work is to leverage the openly available data to automatically generate a larger training dataset with more variability in term of regions and type of cities, which can be used to build more accurate deep learning models. In contrast to most benchmark datasets, the gathered data have not been manually curated. Thus, the training dataset is not perfectly clean in terms of remote sensing images exactly matching the ground truth building’s foot‐print. A workflow that includes data pre‐processing, deep learning semantic segmentation modeling, and results post‐processing is introduced and applied to a dataset that include remote sensing images from 15 cities and five counties from various region of the USA, which include 8,607,677 buildings. The accuracy of the proposed approach was measured on an out of sample testing dataset corresponding to 364,000 buildings from three USA cities. The results favorably compared to those obtained from Microsoft’s recently released US building footprint dataset.

Our motivation in this work is to find an adequate probability distribution to fit sales volumes ofdifferent appliances. This distribution allows for the translation of sales rank into sales volume.This paper shows that the log-normal distribution and specifically the truncated version are well suited for this purpose. We demonstrate that using sales proxies derived from a calibratedtruncated log-normal distribution function can be used to produce realistic estimates of marketaverage product prices, and product attributes. We show that the market averages calculated with the sales proxies derived from the calibrated, truncated log-normal distribution provide better market average estimates than sales proxies estimated with simpler distribution functions.

Accurate savings estimations are important to promote energy efficiency projects and demonstrate their cost-effectiveness. The increasing presence of advanced metering infrastructure (AMI) in commercial buildings has resulted in a rising availability of high frequency interval data. These data can be used for a variety of energy efficiency applications such as demand response, fault detection and diagnosis, and heating, ventilation, and air conditioning (HVAC) optimization. This large amount of data has also opened the door to the use of advanced statistical learning models, which hold promise for providing accurate building baseline energy consumption predictions, and thus accurate saving estimations. The gradient boosting machine is a powerful machine learning algorithm that is gaining considerable traction in a wide range of data driven applications, such as ecology, computer vision, and biology. In the present work an energy consumption baseline modeling method based on a gradient boosting machine was proposed. To assess the performance of this method, a recently published testing procedure was used on a large dataset of 410 commercial buildings. The model training periods were varied and several prediction accuracy metrics were used to evaluate the model's performance. The results show that using the gradient boosting machine model improved the R‐squared prediction accuracy and the CV(RMSE) in more than 80 percent of the cases, when compared to an industry best practice model that is based on piecewise linear regression, and to a random forest algorithm.

This document describes a field validation and verification of the BuildingIQ Predictive Energy Optimization (PEO) technology based on a five-site study. BuildingIQ describes its PEO technology as a software-as-a-service (SaaS) platform that optimizes commercial building HVAC control for system efficiency, occupant comfort, and cost. It is targeted for use in large, complex buildings, and integrates with the building automation system (BAS) to conduct supervisory control. The PEO algorithm defines optimal space air temperature setpoints that are automatically implemented at the variable air volume (VAV) terminal units when possible, or through supply air temperature and duct static pressure setpoints at the air handling unit (AHU) level. The optimization is built upon a learned predictive model that provides a 24-hour ahead forecast of the building’s power profile, using weather forecasts and historical operational data; this model is updated every 4 to 6 hours. Demand-responsive load reductions may also be implemented.

The surge in interval meter data availability and associated activity in energy data analytics has inspired new interest in advanced methods for building efficiency savings estimation. Statistical and machine learning approaches are being explored to improve the energy baseline models used to measure and verify savings. One outstanding challenge is the ability to identify and account for operational changes that may confound savings estimates. In the measurement and verification (M&V) context, ‘non-routine events’ (NREs) cause changes in building energy use that are not attributable to installed efficiency measures, and not accounted for in the baseline model's independent variables. In the M&V process NREs must be accounted for as ‘adjustments’ to appropriately attribute the estimated energy savings to the specific efficiency interventions that were implemented. Currently this is a manual and custom process, conducted using professional judgment and engineering expertise. As such it remains a barrier in scaling and standardizing meter-based savings estimation. In this work, a data driven methodology was developed to (partially) automate, and therefore streamline the process of detecting NREs in the post-retrofit period and making associated savings adjustments. The proposed NRE detection algorithm is based on a statistical change point detection method and a dissimilarity metric. The dissimilarity metric measures the proximity between the actual time series of the post-retrofit energy consumption and the projected baseline, which is generated using a statistical baseline model. The suggested approach for NRE adjustment involves the NRE detection algorithm, the M&V practitioner, and a regression modeling algorithm. The performance of the detection and adjustment algorithm was evaluated using a simulation-generated test data set, and two benchmark algorithms. Results show a high true positive detection rate (75%-100% across the test cases), higher than ideal false positive detection rates (20%-70%), and low errors in energy adjustment (<0.7%). These results indicate that the algorithm holds for helping M&V practitioners to streamline the process of handling NREs. Moreover, the change point algorithm and underlying statistical principles could prove valuable for other building analytics applications such as anomaly detection and fault diagnostics.

As data science comes to buildings, the promise of using machine learning and novel sources of data has received much attention. Advances in machine learning and computer vision algorithms, combined with increased access to unstructured data (e.g., images and text), have created an opportunity for automated extraction of building characteristics – cost-effectively, and at scale. Acquisition of features such as footprint are time consuming and costly to acquire with today’s manual methods, but can be streamlined through intelligent software-based solutions applied to satellite images. When combined with aerial RGB and thermal images, full 3D geometries and thermal maps can be constructed to determine additional characteristics such as window to wall ratio, height, number of stories and envelope thermal characteristics. In this paper we present three contributions to accelerate these high potential opportunities: (1) a methodical analysis of how these features can be integrated into today’s simulation and data driven software tools to enhance efficiency measure identification and owner/operator decision making; (2) development and accuracy testing of open source deep neural network methods to extract building footprints from satellite imagery, including the curation and application of openly available GIS datasets for training and continued development by others; and (3) an open framework for drone-based image capture and creation of 3D building geometries. This work represents an important bridge between high-level studies that span diverse application areas and those that detail point solutions yet cannot be easily replicated or extended.

Trustworthy savings calculations are critical to convincing regulators of both the cost-effectiveness of energy efficiency program investments and their ability to defer supply-side capital investments. Today’s methods for measurement and verification (M&V) of energy savings constitute a significant portion of the total costs of energy efficiency programs. They also require time-consuming data acquisition. A spectrum of savings calculation approaches is used, with some relying more heavily on measured data and others relying more heavily on estimated, modeled, or stipulated data. The rising availability of “smart” meters and devices that report near-real time data, combined with new analytical approaches to quantifying savings, offers potential to conduct M&V more quickly and at lower cost, with comparable or improved accuracy. Commercial energy management and information systems (EMIS) technologies are beginning to offer M&V capabilities, and program administrators want to understand how they might assist programs in quickly and accurately measuring energy savings. This paper presents the results of recent testing of the ability to use automation to streamline some parts of M&V. In this paper, we detail metrics to assess the performance of these new M&V approaches, and a framework to compute the metrics. We also discuss the accuracy, cost, and time trade-offs between more traditional M&V, and these emerging streamlined methods that use high-resolution energy data and automated computational intelligence. Finally we discuss the potential evolution of M&V and early results of pilots currently underway to incorporate M&V automation into ratepayer-funded programs and professional implementation and evaluation practice.

Trustworthy savings calculations are critical to convincing regulators of both the cost-effectiveness of energy efficiency program investments and their ability to defer supply-side capital investments. Today’s methods for measurement and verification (M&V) of energy savings constitute a significant portion of the total costs of energy efficiency programs. They also require time-consuming data acquisition. A spectrum of savings calculation approaches is used, with some relying more heavily on measured data and others relying more heavily on estimated, modeled, or stipulated data. The increasing availability of “smart” meters and devices that report near-real time data, combined with new analytical approaches to quantify savings, offers the potential to conduct M&V more quickly and at lower cost, with comparable or improved accuracy. Commercial energy management and information systems (EMIS) technologies are beginning to offer these ‘M&V 2.0’ capabilities, and program administrators want to understand how they might assist programs in quickly and accurately measuring energy savings. This paper presents the results of recent testing of the ability to use automation to streamline the M&V process. In this paper, we apply an automated whole-building M&V tool to historic data sets from energy efficiency programs to begin to explore the accuracy, cost, and time trade-offs between more traditional M&V, and these emerging streamlined methods that use high-resolution energy data and automated computational intelligence. For the data sets studied we evaluate the fraction of buildings that are well suited to automated baseline characterization, the uncertainty in gross savings that is due to M&V 2.0 tools’ model error, and indications of labor time savings, and how the automated savings results compare to prior, traditionally determined savings results. The results show that 70% of the buildings were well suited to the automated approach. In a majority of the cases (80%) savings and uncertainties for each individual building were quantified to levels above the criteria in ASHRAE Guideline 14. In addition the findings suggest that M&V 2.0 methods may also offer time-savings relative to traditional approaches. Finally we discuss the implications of these findings relative to the potential evolution of M&V, and pilots currently being launched to test how M&V automation can be integrated into ratepayer-funded programs and professional implementation and evaluation practice.

Buildings account for about 40 percent of the total energy consumption in the U.S. and emit approximately one third of greenhouse gas emissions. But they also offer tremendous potential for achieving significant greenhouse gas reductions with the right savings strategies. With an increasing amount of data from buildings and advanced computational and analytical abilities, buildings can be made “smart” to optimize energy consumption and occupant comfort. Smart buildings are often characterized as having a high degree of data and system integration, connectivity and control, as well as the advanced use of data analytics. These “smarts” can enable up to 10–20% savings in a building, and help ensure that they persist over time. In 2009, Microsoft Corporation launched the Energy-Smart Buildings (ESB) program with a vision to improve building operations services, security and accessibility in services, and new tenant applications and services that improve productivity and optimize energy use. The ESB program focused on fault diagnostics, advanced analytics and new organizational processes and practices to support their operational integration. In addition to the ESB program, Microsoft undertook capital improvement projects that made effective use of a utility incentive program and lab consolidations over the same duration. The ESB program began with a pilot at Microsoft's Puget Sound campus that identified significant savings of up to 6–10% in the 13 pilot buildings. The success of the pilot led to a global deployment of the program. Between 2009 and 2015, there was a 23.7% reduction in annual electricity consumption (kWh) at the Puget Sound campus with 18.5% of that resulting from the ESB and lab consolidations. This article provides the results of research conducted to assess the best-practice strategies that Microsoft implemented to achieve these savings, including the fault diagnostic routines that are the foundation of the ESB program and organizational change management practices. It also presents the process that was adopted to scale the ESB program globally. We conclude with recommendations for how these successes can be generalized and replicated by other corporate enterprises.

In a rapidly evolving energy industry, utilities are dealing with new challenges like integrating distributed energy resources and market saturation for advanced lighting retrofits. Demand-side management programs require new approaches to meet aggressive carbon reduction goals. Advanced measurement & verification (M&V) is an energy data analysis method using smart meter data in combination with analytics to quantify energy efficiency project savings. Advanced M&V shows great promise for supporting next generation commercial programs including retro commissioning, multi-measure retrofits, and behavior change programs. Advanced M&V captures real project impacts at the meter, but sometimes non-project events can also impact consumption (so-called “non-routine events” [NREs]). Accurately detecting and accounting for NREs is important for reducing uncertainty of savings estimates and helps manage investment risk for different stakeholders (e.g., utilities, building owners, ESCOs). Recent research has shown promise in establishing data-driven techniques to identify and adjust for NREs, but fundamental questions still remain, such as: how can you distinguish NREs from acceptable noise in energy consumption profiles? What is the frequency and magnitude of NREs? Can their detection and adjustment be automated and streamlined? This paper documents the state of the art in NRE quantification and analysis. The results of research to quantify the frequency, nature and direction of NREs, and methods and metrics for determining a trigger threshold for taking action on NREs are presented. The paper also documents the latest technical guidance on application of NRE detection and adjustment methods.