1.0 Executive Summary
Meter data collection from buildings is highly variable in terms of source, format, and quality. BuildingIQ is able to receive, load, and process any data stream, whether from third-party vendors, through file transfer, or through email. Even manual loading of data is acceptable. The raw data—the data of record— is retained and stored in the cloud for reference purposes, but is also cleaned up—processed by quality-scrubbing algorithms—to smooth out the data stream. It can then be more readily processed by BuildingIQ’s Predictive Energy Optimization™ platform.
This stream of high quality data can be processed in numerous formats, from kW to kWh, and received by the platform in either cumulative form or in interval readings, ranging from minutes to days. Automation of the whole-building energy data stream is one of the keys to BuildingIQ’s success in optimizing the energy management of large commercial and public buildings.
2.0 Introduction and Background
The cloud-based Predictive Energy Optimization (PEO) platform is a paradigm-shifting technology used to reduce energy consumption in commercial buildings. The optimization process involves three basic steps: first, learning the thermal dynamics of the building; second, refining the model through the slow convergence between parameter prediction and actual readings from the building; and finally optimization, in terms of both energy efficiency and cost.
This paper focuses on the important process of collecting meter data and feeding it into the PEO platform. Raw data is stored in the cloud for possible future reference (e.g. auditing), while simultaneously being converted to a higher quality data stream readily processed by PEO.
Establishing the whole-building energy consumption profile is key. The foundation of BuildingIQ’s model involves quantifying the relationship between wholebuilding energy consumption, outside weather, and the internal temperatures of the building.
3.0 Collecting the Data
Meter data can be collected, retrieved, and loaded into the PEO system through a variety of automated data streams. It could, for example, be sent by a third-party vendor, such as Panoptix, Johnson Control, or Schneider; it could be manually loaded as comma separated values (CSV); or sent as a file transfer (FTP) or an email (simple mail transfer).
See Figure 1. BuildingIQ is able to handle all sources and forms of metering data, a critical capability since each building tends to read and record data in different ways. Thus far, there is no accepted, uniform process of meter collection for commercial and public buildings; it varies from building to building, city to city, and from country to country.
Meter data can also be collected in a variety of formats. For any given building, metering data could be sent by the BMS in cumulative form, for example, or in interval form. Intervals can range from minutes to hours to days. It could be expressed in terms of power in kilowatts (kW), or in terms of energy consumption in the form of kilowatt-hours (kWh), or even in the case of gas, in terms of BTUs. In some countries, BuildingIQ has had to install its own utility meters to obtain reliable and consistent data.
BuildingIQ’s model is able to handle all of these formats. In terms of data intervals, the system can down-sample the data storage rate —that is, go from minutes to hours. It can also up-sample the data storage rate, but only as low as the granularity of the raw data it receives.
4.0 Creating a Higher Quality Stream
A data stream can be smooth one minute, erratic the next. There can be gaps, surprising stoppages, and abnormal spikes. Meter readings that range consistently day after day from 200kW to 400kW can inexplicably jump to 700kW to 1000kW and then fall back. Such anomalies and gaps need to be filtered out or set aside for subsequent analysis.
BuildingIQ has a variety of quality-scrubbing algorithms to smooth out the data stream, making it more representative of the real energy profile of the building. Called the “quality stream,” this process assists PEO in refining its predictions and doing so faster and more accurately. The quality stream helps remove confusion from the raw data stream, which unfiltered can throw off the PEO model as it tries to optimize the building’s energy usage.
BuildingIQ goes a step further in the data screening by providing a human override capability.
Experienced operators are able to reinsert a piece of data that the machine has automatically rejected. The machine—the quality scrubbing algorithms—may say “this is a bad piece of data,” while the operator says, “No, let’s keep that one. It is telling us something important that we don’t want to get lost.” 5.0 Displaying the Data
Metering data can be displayed in a variety of formats, with optional graphics that clients can bring up instantly — on a computer— to answer questions and concerns about energy usage. They can zoom in on the data for tighter granularity if they chose, watching the flow of energy minute by minute, for example. Alternatively, they can zoom out to view the ebb and flow of energy usage on a monthly basis. Clients can also look at the conjunction of metering data with other variables in the system, such as set points, or supply-air temperature or supply-air pressure. The flexibility of displays creates a learning environment for the client to see how operational changes affect energy requirements.
Metering data can be displayed in operational terms, typically in energy units, or it can be displayed in financial terms, where energy pricing is applied to the consumption in real time. Displays of client savings—energy usage after BuildingIQ’s implementation is compared to a historical baseline—can be shown in both energy and financial terms (kilowatt-hours or dollars). Savings can be displayed in periodic terms (per day, per month), or in cumulative terms (to date).
6.0 Reporting and Audit Trail
BuildingIQ has built transparency into their platform design. Measurement & Verification (M&V) are integral to the system. Meter data at any scale over any time period are easily generated and/or displayed for monthly reporting. The availability of the raw data, which can be retrieved and overlaid on the quality data stream, can assist in performance evaluation and/or an audit of the savings.
Customers typically have their own audit team so the model’s transparency gives them a powerful tool to ensure that perceived savings are in fact real, and will hold up under scrutiny by regulators, utilities, or investors alike. A clean audit trail helps secure future financing.
7.0 The BuildingIQ Advantage
BuildingIQ brings some unique advantages to the task of building automation. First, the automation of meter data itself is streamlined and efficient, freeing clients to go about their core business. When spot checks or audits are required, the tools are at their fingertips—built in—to assist audit teams.
Second, the raw meter data is stored in the cloud so that when there is a discrepancy or contested reading or interpretation, the factual record of original data is readily available. The algorithms used to generate a quality data stream from the raw data stream are considered the best in the business.
Third, the platform is extraordinarily flexible. From data collection to data processing to data display—it can handle any meter data feed in any format, clean it up, and use it to the client’s advantage. It can then display the results in a manner tailored to the client’s interests and needs.
Fourth, there is no intrusion during implementation and the tenant’s comfort is paramount. And finally, energy savings begin to accrue quickly, usually within months following implementation. Savings average about 15-20% of total HVAC consumption. M&V is built in so that clients have solid proof that savings are as claimed.
Site Reference: https://buildingiq.com/resources/case-studies/