Objective

The primary objectives of this project were to demonstrate: (1) energy efficiency gains achievable in small- to medium-sized buildings with a Model Predictive Control (MPC)-based whole-building optimal control and (2) reduction in first costs achievable with a wireless sensor network (WSN)-based building heating, ventilation and air conditioning (HVAC) control system compared to a conventional wired system. The second objective is key because first cost is a barrier to wider application of advanced HVAC control and 70% of the first cost is attributed to installation (wiring) and commissioning.

Technology Description

This project demonstrated an advanced energy management and control system in an existing building at the U.S. Army Construction and Engineering Research Laboratory (CERL) in Urbana-Champaign, Illinois. The medium-size office building underwent a retrofit of the HVAC system and controls employing a technology called optimal MPC, which offers significant potential for saving energy by providing a means to dynamically optimize various sub-systems, such as fans, cooling and heating coils, to take advantage of building utilization and weather patterns and utility rate structures.

The United Technologies Research Center (UTRC)-led team, partnered with Army-CERL facility staff and researchers at the Oak Ridge National Laboratory (ORNL) and the University of California Berkeley, tested as a proof-of-concept the on-line implementation of model-based predictive and optimal control of the HVAC system in a 7000 sq. ft. portion of the CERL building. The system was retrofitted with a commercial off-the-shelf open protocol building automation system. The existing controls operated the HVAC system continuously during the day, maintained fixed temperature set points in the Air Handling Unit (AHU) heating and cooling deck discharges, and used a fixed outdoor air fraction for ventilation in the building. The MPC approach aimed to increase system efficiency by continuous adjustment of the system schedule of operation, heating and cooling set points, and fresh air levels brought into the building, based on predicted and measured occupancy levels, internal loads, and weather forecasts. System and indoor environment measurements of supply air temperatures and airflows, occupancy, zone temperature, relative humidity, and carbon dioxide levels were used to inform construction of models for HVAC equipment, thermal factors, and occupancy models, and to configure control design. Sub-metering was used to establish baseline energy consumption and to verify performance improvements. To reduce installation cost, wireless sensors were utilized wherever possible, particularly for occupancy sensing and thermal comfort. The WSN self-configures routing of data through a gateway to a central control computer that hosts the algorithms.

Demonstration Results

A multi-variable optimization problem to minimize energy consumption and cost while guaranteeing zonal comfort over a 3-hour predictive horizon was formulated and solved periodically on-line. The algorithms were integrated with the building automation system and evaluated experimentally from July 2011 to February 2012. A 55-65% reduction in HVAC system energy use was demonstrated while improving occupant comfort. Of this, nearly 35% improvement was achieved via off-line adjustments of the system’s operation schedule and heuristic adjustments of the heating and cooling coil set points. This post-retrofit state of the HVAC system involved implementation of direct digital controls (DDCs) and a basic building automation system. The additional improvement of 60-80%, relative to the post-retrofit heuristic implementation, was accomplished by on-line dynamic optimization of the building. A 10-15% installation cost reduction was accomplished by use of a robust WSN versus a fully wired network. The advanced control system and algorithms were monitored by UTRC and CERL until April 2012. Following this testing and evaluation, the CERL facility management team reverted back to the post-retrofit mode in anticipation of further upgrades to the remainder of the facility.

It should be noted that the present implementation of optimal controls was for a specific form of central building HVAC system involving a dual deck configuration. Such systems are prevalent in older buildings, of which there are many in the Department of Defense (DoD) stock, and are more prone to energy waste from system duct losses and leakages, compared to the single deck HVAC systems that are deployed more commonly now. This could explain some of the large energy savings accomplished when going from a pneumatic control approach for 24/7 operation to a DDC mode operation (considered as a post-retrofit baseline for optimal control mode). Furthermore, a more fine-tuned DDC mode control strategy involving reset of the cooling and heating deck set points based on outside weather, rather than a seasonal setting as employed at the demonstration site, would have captured some of the savings achieved by the optimal scheme. Finally, much of the optimal control mode performance data was obtained for heating season operation, although some cooling mode data was captured between July and September 2012, primarily for pneumatic and DDC modes of operation. More detailed assessments and analysis for different variants of the central HVAC system and of baseline DDC mode control approaches are needed to ascertain the variability in the energy use and peak power reduction benefits across DoD stock.

This model-based control methodology can be extended to hydronic heating and cooling systems where variable speed technologies are becoming prevalent and robust, but multivariable optimal control methodologies are lacking. The building HVAC control technology is applicable to small- and medium-sized buildings, which represent a significant portion of the DoD building stock. The demonstrated energy savings of more than 60% reduction in HVAC system energy use is estimated to lead to nearly 20% building level energy use reduction (assuming conservatively that HVAC systems constitute 30% of total building energy use). This represents significant progress toward the 30% gains in energy efficiency beyond 2003 levels mandated by Executive Order 13423. Renovations and retrofits are driven toward a 20% savings goal relative to pre-retrofit 2003 levels, and by this measure the improvements demonstrated in this project represent the potential to meet the goal through broader scale implementation of optimal control technology alone.

Implementation Issues

Key challenges were identified in the additional cost to install the WSNs, particularly the skill level and familiarity required by the contractor to deploy them. This adversely impacted the installed cost gains that were accomplished through the use of a wireless sensor infrastructure. Further, the lack of familiarity with, and related perceived risk in, the maintenance for the optimal control platform, which utilizes Matlab and optimization toolboxes, was an impediment to longer term sustained deployment of the technology at the demonstration site. Technical challenges remain in the scalability and level of automation required to obtain relevant dynamic system models and for the configuration and commissioning of the optimal control algorithms with the building management system.

  • Building Management System (BMS),

  • Wireless,