Data and analytics to inform energy retrofit of high performance buildings

Tianzhen Hong, Le Yang, David Hill, Wei FengApplied Energy 2014 · DOI

TL;DR

This paper introduces a holistic, data-driven approach to identifying energy retrofit opportunities in high-performance buildings (HPBs). By analyzing detailed building performance data through energy profiling, benchmarking, and diagnostics, the study demonstrates that even energy-efficient buildings have significant potential for further energy savings, offering actionable insights for facility managers.

What Problem Does This Paper Address?

Buildings are major energy consumers, and while high-performance buildings (HPBs) are designed for efficiency, there's a growing question about their actual performance and whether further energy savings are achievable. Traditional energy audit methods are often insufficient for the complex systems in HPBs. This paper addresses the challenge of identifying and quantifying energy retrofit opportunities in these advanced buildings by leveraging detailed, measured performance data and advanced analytics. The goal is to move beyond generic energy-saving measures and pinpoint specific areas for improvement in buildings that already incorporate energy-efficient technologies.

Methods

The authors propose a three-tiered analytical framework:

  • Energy Profiling: This initial step involves analyzing time-series data (hourly or sub-hourly) to understand energy use patterns across different time scales (annual, monthly, weekly, daily) for the entire building and its major end-uses. This helps identify anomalies and seasonal variations.
  • Energy Benchmarking: This level compares the building's energy performance against a peer group of similar buildings. The study utilizes the ISO Standard 12655 for a consistent energy data model, enabling comparisons of whole-building energy use intensity (EUI) and end-use specific performance.
  • Energy Diagnostics: This most detailed level dives deep into specific systems and components, using high-frequency data (e.g., 1-minute intervals) to identify operational deficiencies and potential savings. This involves analyzing metrics like part-load ratio (PLR) for chillers and fan performance curves.

The study also introduces an energy data model based on ISO Standard 12655 to standardize energy use representation across different buildings.

Datasets

The primary dataset used in this study comes from a high-performance building in California: the CalSTRS Headquarters. This data was collected from the building's Energy Management System (EMS) and Building Automation System (BAS) over a period of one year (March 2012 to February 2013). The data includes:

  • Energy Use Data: Hourly data for major end-uses like lighting, plug loads, data center (IT equipment and CRAC units), and HVAC components (chillers, fans, pumps, etc.).
  • HVAC Operating Data: 15-minute or 1-minute interval data on supply and return air/water flow rates, temperatures, damper positions, and equipment status.
  • Environmental Data: Outdoor air temperature.

For benchmarking, the study uses data from the California Commercial End Use Survey (CEUS) database, which contains information on 2790 commercial buildings in California. A peer group of 31 large office buildings constructed after 1991 was selected from this database.

Key Insights and Ideas

  • HPBs Still Have Significant Savings Potential: Contrary to the assumption that highly efficient buildings have few remaining savings, this study found substantial opportunities for further energy reduction in an already high-performing building.
  • Data Granularity is Crucial: Moving from monthly utility bills to hourly and even minute-level data unlocks the ability to identify specific operational issues and pinpoint savings opportunities that would otherwise remain hidden.
  • Data-Driven Diagnostics Reveal Hidden Issues: Even when overall performance metrics (like chiller COP) appear normal, detailed analysis can reveal detrimental operational patterns (like frequent short cycling) that impact efficiency and equipment lifespan.
  • Benchmarking Needs Standardized Data Models: The ISO 12655 standard is highlighted as essential for enabling meaningful comparisons between buildings, especially for end-use benchmarking.
  • Data Center and Air Movement are Key Targets: The study identified the data center as the largest energy consumer and air movement (AHU fans) as a significant area with potential for improvement, even in efficient systems.

Key Results

  • Estimated Energy Savings: The study estimates that 20% of the building's total electricity consumption, equivalent to 850,764 kWh per year, could be saved through identified retrofit measures. This includes approximately 696,000 kWh from the data center, 136,257 kWh from AHU fans, and 18,507 kWh from chillers.
  • Data Center PUE: The data center's average Power Usage Effectiveness (PUE) was 1.66, which is within the "Standard" to "Good" range, but could potentially be improved to 1.1 (comparable to leading data centers) for significant savings.
  • Chiller Performance: While the average chiller COP was normal, frequent cycling at low part-load ratios was identified as detrimental. Replacing one chiller with a smaller unit and optimizing control strategies could increase integrated COP by 15% and reduce energy use by 14%.
  • AHU Fan Efficiency: AHU fans were found to be highly efficient, but potential savings of up to 22% (136,257 kWh/year) were identified by optimizing the supply air static pressure setpoint reset.
  • Outdoor Air Economizer: The outdoor air economizer was found to be operating normally, contributing to cooling by utilizing outside air when conditions were favorable.
  • Building Manager Actions: Based on the findings, the building manager implemented measures such as reducing supply air flow, adjusting night purge setpoints, calibrating chiller controls, and consolidating data center server loads.

Limitations and Future Work

  • Data Gaps and Quality: The study acknowledges challenges with data availability, particularly the gas meter being out of service and data gaps in the BAS. This limited the analysis of gas heating and some end-uses.
  • Mixed-Use Building Complexity: For mixed-use buildings, some end-uses are shared, requiring approximations for analysis, which may affect absolute accuracy.
  • Labor-Intensive Data Processing: Downloading and pre-processing data from EMS and BAS systems is described as labor-intensive and not fully automated, posing a challenge for widespread adoption.
  • Future Work: The authors suggest that obtaining hourly or sub-hourly gas use data would enable further analysis of gas heating savings. They also emphasize the need for adequate and good quality data at short time intervals (1-5 minutes) for a complete year to be the cornerstone for detailed analysis and effective retrofit identification.

Summary updated

BibTeX
@inproceedings{hong2014data,
  author = {Tianzhen Hong and Le Yang and David Hill and Wei Feng},
  title = {{Data and analytics to inform energy retrofit of high performance buildings}},
  booktitle = {Applied Energy},
  year = {2014},
  doi = {10.1016/j.apenergy.2014.03.052},
}
Figure