Adam Roloff, Altura Associates
As HVAC and controls engineers we live and breathe building analytics data, however it’s easy to forget that for many of the folks managing these properties, there is very limited time to spare beyond executing on their current business workflows and daily duties. Understanding how existing business processes can be leveraged so that building analytics can become a natural part of the larger process is critical to its success as a tool. Our discussion will focus on some of the strategies, tools, lessons learned that have come from reaching out to the folks on the other side of the equation: including facilities O&M staff, energy managers, and capital project planners.

Derek McGarry, PointGuard
The Pathway to Value – Key Ingredients for Success in Analytics. Let’s be real, analytics can be intimidating and complicated. This discussion will explore common barriers and issues that get in the way of delivering value from analytics. We will share lessons learned from the field, and what key ingredients improve the chances for success. Discussion will include exploration of: managed services vs software-only; role of service contractors vs commissioning engineers vs property managers vs facility engineers; KPI’s that matter; people dynamics.

Cory Mosiman, Smart Building Specialist, WSP USA
Building performance metrics have traditionally been evaluated from a facility management perspective, with the overall goal of reducing energy consumption and/or peak demand. Moreover, commercial office spaces are often largely underutilized, and energy consumption isn’t considered in conjunction with how well the building is used/occupied. This presentation will offer a new method for analysis of the building energy consumption in conjunction with the objective space utilization, offering new opportunities for FDD and energy performance validation.

Celeste Cizik
1. Utilizing SkySpark analytics to detect and repair HVAC equipment in a middle school in Denver
2. Detecting and Correcting Economizer Operation

Paul Quinn
SkySpark – Core to Efficiency in a Major New Airport

Stephanie Fetchen
Why should we integrate electric rates with SkySpark? How should we best go about it? Building automation systems and smart meters play a key role in achieving energy efficiency and cost savings for facility owners and operators. Historically, the majority of the reports and quantifications related to energy use are shown in kWh and kW. As rates become more complex pure kWh and kW information is not adequate to support the optimal use of energy and its costs. For, example in some rate programs operators can achieve lower overall costs by using more kWh to help avoid kW demand peaks. They key to these types of analysis is to combine tariff rate data with consumption and demand data. The inclusion of detailed electric rate information in SkySpark enables us to translate kWh and kW into dollars. Providing saving and/or cost information in dollars is much more meaningful to our end users than seeing it in kWh or kW. SkySpark has a tariff engine, but you need rate data to take advantage of it. But rates can be complex, hard to access and presented in many different formats. Complexity in electric rates is caused by the availability of multiple rate options, as well varying levels of details within each rate schedule. So how do you gather the data you need to include electric rates in SkySpark? In this discussion, we will address all of these issues, and options to simplify the inclusion of rates in your systems.

Matthew Giannini
The SkySpark Tariff Engine and Energy App Tariff Views

Jerry Weatherhogg
The “Commissioner” a View Builder App that can be used as either a commissioning tool to trend I/O point values, or as a way to quickly build and deploy SkySpark projects in existing BACnet environments.

Jake MacArthur
Different methods of displaying KPIs around avg, min and max zone air CO2.

Jay Herron
A Utility Benchmarking app provides a high level view of building performance using only historical utility bills and weather data. Users easily can graph utility consumption and cost across time, adjust for different billing lengths, and normalize by site area and weather. Weather is normalized by calculating a baseline regression using cooling degree days, heating degree days, or both over a defined time period, and then applying the baseline to the displayed time period. Facility usage can be compared with the weather baseline, allowing the user to determine if performance has improved or degraded. Multiple sites can be graphed together to compare operation over time, determine underperforming buildings of a portfolio, and prioritize improvements.

Justin Davies
Energy Information dashboards with SkySpark View Builder

Phil Birch
Haystack Tagging Application – This View Builder application uses the project haystack tagging standard to assist in applying tags and names to points in SkySpark. This is done using the haystack tagging standard and regular expression (REGEX). The application first uses these REGEX expressions to attempt to find matches for equipment. A simple UI allows the user to apply the necessary tags with the press of a button, along with applying any other descriptive tags (like “variableVolume”). The application then moves on to tagging the points under these equipment. Using REGEX and equipment association the application attempts to find matches for each point. The UI allows the user to tag these points by a simple button press. If any points did not find a match, they are sorted into an “unknown points” list which the user can then select the point definition and tag them accordingly.

Celeste Cizik
Portfolio Analysis for a Major Resort – Combining energy data and benchmarking to create an “Opportunity Rating” for ECM’s.

Matthew Giannini
Introduction to Machine Learning functions in SkySpark

Sean Stackhouse
One Class Support Vector Machine Learning (SVM) is a type of unsupervised machine learning that allows for detection of outlying data. The benefit is that we do not have to define out what the out of range values are, the model will find them automatically. Rather the number of outliers found can be adjusted with a simple sensitivity value. Once an “ideal” model is created for a set of data, the model can be applied to similar sensors or equipment and issues identified.

We’ve integrated One Class SVM through R. Combined with viewbuilder, we can view the anomalies, see the trends, tune and save the model, and identify issues. From there, engineering expertise can input the cause of the issue. Future machine learning can utilize the data to create more powerful, supervised learning models.

Adam Wallen
Calculus and Signal Analysis Functions