Ph。D Candidate, Dept。 of Civil & Environmental Engineering, Carnegie Mellon University, 5000 Forbes Ave。, Pittsburgh, PA 15213; PH (412) 953-2517; email: pine@cmu。edu

2Professor, Dept。 of Civil & Environmental Engineering, Carnegie Mellon University,76170

3Professor and Head, Dept。 of Civil & Environmental Engineering, Carnegie Mellon University, 5000 Forbes Ave。, Pittsburgh, PA 15213; email: garrett@cmu。edu 4Assistant Professor, Dept。 of Civil & Environmental Engineering, Carnegie Mellon University, 5000 Forbes Ave。, Pittsburgh, PA 15213; email: marioberges@cmu。edu

ABSTRACT Heating, ventilating and air conditioning (HVAC) systems account for about 16% of the total energy consumption in the United States。 However, research shows that 25%-40% of the energy consumed by HVAC systems is wasted because of undetected faults。 Actively detecting faults requires continuously monitoring and analyzing the status of hardware and software components that are part of HVAC systems。 With the increasing complexity in HVAC systems, fault detection that relies on manual processes becomes even more challenging and impractical。 Hence, a computerized approach is needed, which enables HVAC systems to continuously monitor, assess and configure themselves。 This paper proposes an integrated framework for developing and implementing self-configuring approaches to operate and maintain HVAC systems。 The discussions include the identification of functional requirements, a synthesis of existing self-configuring approaches, and an analysis of the requirements for developing an integrated framework using an implemented prototype system。

INTRODUCTION

Buildings account for 41% of the total energy consumption and 38%  of carbon dioxide emissions in the United States。 About 40% of the energy consumed in both residential and commercial buildings is used by HVAC systems (DoE 2008; EIA 2008)。 However, research shows that 25%-40% of the energy used by HVAC systems is wasted due to faults, such as misplaced and uncalibrated sensors, malfunctioning controllers and controlled devices, improper implementation and execution of control logic, improper integration of control software and hardware components, and sub-optimal control strategy (Mansson and McIntyre 1997; Liddament 1999; Liu et al。 2002; Roth et al。 2005)。 This waste accounts for $36 – $60 billion every year in the United States (EIA 2008)。 Indirect social and environmental impacts of the waste are

COMPUTING IN CIVIL ENGINEERING 803

beyond estimation due to the fast depleting energy resources and increasing environmental pollution (Liang and Du 2007)。

Several researchers have stated that a primary reason for the occurrence of different types of faults that result in significant waste in energy is that HVAC systems are getting increasingly complex and it is difficult for operators to manually detect and diagnose these faults (Lee et al。 2004; Katipamula and Brambley 2005a; Jagpal 2006)。 Due to increasing needs for better indoor environment control, more and more HVAC systems are equipped with software and hardware components。 To maintain the desired performance of these HVAC systems, operators need to continuously monitor and diagnose hundreds of components。 Moreover, because different faults occurring in HVAC systems can have similar symptoms, it is difficult for the operator to diagnose the root cause of the faults (Schein and Bushby 2005)。 All of these issues make it very difficult, if not impossible, to manually monitor the performance of HVAC systems and to detect possible problems resulting in inefficient operations。

Computerized approaches, such as computer-aided fault detection and diagnosis (FDD), automated commissioning and optimized operating schedule, have been studied and developed to address some of these challenges associated with manual operation and maintenance of HVAC systems。 Both laboratory and real-world experiments have been conducted to validate the energy saving capability of these approaches (Mansson and McIntyre 1997; Castro 2004; Katipamula and Brambley 2005a; Katipamula and Brambley 2005b; Schein and Bushby 2005)。