A novel algorithm was tested against a large and diversified dataset comprising points from five buildings, two vendors, three distributors and more than 20K points. Overall the algorithm identifies about 90% of VAVs and 80% of AHUs and reaches an accuracy of about 90% in detecting the points required by a test application. The algorithm was incorporated into a VOLTTRON ready utility.
A fault detection and diagnostics system for rooftop air conditioners was developed using low-cost electronics. The system was designed to be compatible with the VOLTTRONTM platform. The underlying fault detection and diagnostics methodology utilizes virtual sensors to measure parts of the equipment operation that are sensitive to common faults. Using virtual sensors reduces costs while also providing accurate and reliable diagnostics.
This project implemented, validated and documented an automated system for training virtual refrigerant charge sensors for rooftop unit ACs. The system automatically tunes empirical parameters of a virtual sensor for estimating the amount of refrigerant in a system. The engineering time and costs associated with calibrating a virtual sensor are reduced because of the automated testing in an open laboratory and the reduced number of tests.
Whole building energy models do not always provide satisfactory predictions to facilitate decision making during design, due to large number of uncertainties in model input parameters. CBEI presents a computationally e?cient process for uncertainty quanti?cation, sensitivity analysis and automated calibration of building models. This is demonstrated using an energy simulation model of a medium sized o?ce building.
In 2015-16 United Technologies Research Center (UTRC), in collaboration with the CBEI, developed OpenStudio measures for seven systems/components used in HVAC retrofit packages identified with high energy saving potentials in BP4 for small office, medium office, stand-alone retail, and primary school buildings in certain climate zones.
Fact Sheet on OpenStudio Enhancements
This paper presents Model Predictive Control (MPC) and Fault Detection and Diagnostics (FDD) technologies, their on-line implementation, and results from several demonstrations conducted for a large-size HVAC system. The two technologies are executed at the supervisory level in a hierarchical control architecture as extensions of a baseline Building Management System (BMS). The MPC algorithm generates optimal set points for the HVAC actuator loops which minimize energy consumption while meeting equipment operational constraints and occupant comfort constraints.
A practical control algorithm for coordinating bot AC and refrigeration equipment was developed and evaluated using an energy simulation testbed for a convenience store. It was validated using actual convenience store data. The simulations allowed evaluations of savings for the unit coordinator compared to conventional control over a cooling season. The controller was designed to minimize implementation costs in that it does not require additional sensors and is self-learning.
Field demonstrations provided test and evaluation data for virtual sensor based AFDD concepts and provided a laboratory demonstration on the VOLTTRON platform.
This paper presents the implementation and experimental demonstration results of a practically effective and computationally efficient model predictive control (MPC) algorithm used to optimize the energy use of the heating, ventilation, and air-conditioning (HVAC) system in a multi-zone medium-sized commercial building.
HVAC package solutions were identified that met the stated objectives, based on 6 building types (quick service restaurant, full service restaurant, small hotel. large hotel, supermarket, and convenience store) in 6 region/climate zone combinations.
This paper presents a systematic development process of whole-building energy models as performance benchmarks for retrofit projects. Statistical regression-based models and computational performance models require utility data for calibration and validation.
A scalable low-cost optimal chiller plant control algorithm was developed and effectively demonstrated with 128 case studies covering a variety of chiller plant load variations with each case being a weekly simulation of whole-building dynamic HVAC system models with closed loop local controls and supervisory chiller plant controls. Model-in-the-loop (MiL) analysis suggests a promising average energy saving of ~15% for medium office buildings and an average energy saving of ~10% for large hotel buildings.
This project developed and demonstrated novel techniques for cost-effective AFDD for Air Handling Units for small/medium commercial buildings. The diagnostic accuracy is over 95% and the payback period is less than two years.
Widespread deployment of advanced controls and diagnostics in small and medium buildings has been held back by the cost and complexity involved in applying these solutions to individual buildings. CBEI demonstrated data?driven adaptive, self?learning control?oriented models for building HVAC sub?systems and building thermal and envelope dynamics in two medium buildings.