"INTRODUCTION Recent advances in sensor technology@ data-gathering hardware@ communication@ data analysis@ and modeling have led to dramatic strides being made in the following three areas: a. field measurement and verification (MV) of energy and cost savings to ensure that the installed systems meet the specified performance predictions resulting from installing energy conservation measures (ECMs); b. automated fault detection and diagnosis (FDD) of HVACR systems for maintaining optimal performance via predictive maintenance; and c. integrated automation and control of building systems and services that are meant to assist in proper facility management@ which include energy management@ comfort monitoring@ facility operation@ and services billing and communication with the energy supplier. The scope of this paper is limited to the first two areas only. An increasing number of energy performance contracts require verification by actual field monitoring of the energy and cost savings resulting from implementing energy efficiency projects. The National Association of Energy Service Contractors developed protocols for the measurement of retrofit savings in 1992@ which were followed by federal protocols@ such as FEMP (1996)@ IPMVP (1997)@ and ARI (1998)@ and@ finally@ ASHRAE Guideline 14 (ASHRAE 2002). There are also numerous refereed publications in this area@ for example@ the ASME Special Issue (Claridge 1998) or references listed in Reddy and Claridge (2000). Investigators and service companies are being required to develop custom measurement plans and analytical procedures for each project@ which increases total project costs. An important issue during the MV process is the duration over which pre-retrofit measurements need to be taken in order to identify a baseline performance model that accurately captures system behavior over the entire year. Obviously one would like to gather and analyze data for as short a period as possible (both during the pre-retrofit and post-retrofit periods)@ while meeting the verification requirements. On the other hand@ HVACR systems are strongly influenced by both diurnal cyclic operation as well as seasonal variation in operation (for example@ control setpoints such as cold deck temperatures of an air handler) and driving parameters (such as outdoor temperature)@ which would suggest that at least a whole year be used for model development even though a large fraction of such data may be superfluous. Essentially two types of options are available: (a) interrupted monitoring@ where one would monitor@ say@ over a week during each month of the year so as to capture annual variability (this may@ however@ not be a practical option)@ and (b) continuous monitoring@ where the monitoring is done as a block over a certain period of the year. Several studies (Kissock et al. 1998; Katipamula et al. 1998; Reddy et al. 1998; Reddy et al. 2002) have investigated the latter option in an empirical manner and made recommendations as to the season (or time of the year) that is likely to yield performance models of HVACR systems that provide most accurate predictions of annual performance. Though the recommendations are consistent with our physical understanding@ these are anecdotal and lack a clear scientific basis as well as the means of ascertaining whether@ and the extent to which@ incremental monitoring and the data thus collected provide ""added value or new information"" to the monitored data set already obtained. In the last few decades@ the issue of fault detection and diagnosis (FDD) in the performance of engineering systems has drawn the attention of a number of researchers. With many engineering systems growing larger and more complex@ the need to operate and control them safely and reliably has extended beyond the normally accepted safety-critical systems to being able to continuously operate them in an optimal manner (Chen and Patton 1999). As early as the 1960s@ it was realized that faults in critical systems@ such as nuclear power plants@ space exploration@ and weapon systems could have grave consequences. Even a minor malfunction may cause the failure of the whole system@ resulting in loss of time@ money@ and even life. Such considerations led to research into FDD supervisory systems in order to identify even relatively minor malfunctions as early as possible@ while emphasizing detection speed@ sensitivity@ and false alarm rate. Several textbooks have been written on this topic (for example@ Himmelblau [1978]@ Tzafestas et al. [1987]@ Pouliezos and Stravrakakis [1994]@ and Gertler [1998]). This optimum performance of the system would involve reducing the occurrence of sudden@ disruptive@ or dangerous faults@ i.e.@ minimizing system performance degradation@ product deterioration@ and equipment damage while improving human comfort and safety. FDD systems have also been studied under the terminology ""condition monitoring@"" whose purpose was to assist in the implementation of predictive maintenance as against the more common strategies of breakdown and planned maintenance (Davis 1998). Numerous studies relevant to HVACR equipment and systems are reviewed by Katipamula et al. (2001) and Comstock et al. (1999). A fundamental issue in model-based automated FDD is the identification of an accurate fault-free model of system performance as quickly as possible (i.e.@ from a monitoring period as short as possible) so as to subsequently use it for FDD purposes."