|
Description
The companion paper proposed a general methodology of calibrating detailed building energy simulation programs to performance data that also allowed the determination of the prediction uncertainty of intended energy conservation measures. The methodology strived to provide a measure of scientific rigor to the process of calibration as a whole, which has remained an art form with no clear consensus guidelines despite being followed by numerous professionals for several decades. The proposed methodology, while providing a clear structure consistent with that adopted in more mature scientific fields, also uses expert domain knowledge and is flexible enough to satisfy different users with different personal preferences and biases. This paper attests to the overall validity of the methodology by presenting the results of applying it to three case study office buildings--two synthetic and one actual. Conclusions on various variants of the overall calibration methodology are presented, along with guidelines and a summary of lessons learned on how to implement such a calibration methodology. Future research needed prior to implementation in commercial hourly detailed simulation programs is also identified.
INTRODUCTION
A previous paper (Reddy 2006) provided a pertinent and detailed literature review of calibrated simulation techniques, describing their uses, strengths, weaknesses, procedures, and tools as well as pertinent issues related to model fitting uncertainty. The companion paper (Reddy et al. 2007) proposed a general methodology of calibrating detailed building energy simulation programs to performance data that consisted of various concepts adapted from the general scientific literature. The calibration methodology involved the following five major steps:
1. An important first step is to prepare a preliminary simulation input file of the building that is as realistic and error-free as possible. This would entail making sure that the simulation program has the capability of modeling the type of building and systems present and that the inputs have been entered correctly.
2. Next, reduce the dimensionality of the parameter space by performing walk-through audits and heuristics. For a given building type, identify/define a set of influential parameters and building operating schedules along with their best-guess estimates (or preferred values) and their range of variation characterized by either the minimum-maximum range or the upper and lower 95th probability threshold values. The set of influential parameters to be selected should be such that they correspond to specific and easy-to-identify inputs to the simulation program.
3. Next, perform a "bounded" coarse grid calibration (or unstructured or blind search) using a mid-point Latin Hypercube Monte Carlo (LHMC) simulation involving numerous trials or realizations with different combinations of input parameter values. This would allow filtering or identification of the most promising solutions of input parameter vectors and also provide a means of identifying the more sensitive or strong parameters by performing a regional sensitivity analysis (RSA).
4. Subsequently, perform a guided search calibration to further refine or improve the calibrated solutions identified by the coarse grid search.
5. Rather than using only the best calibrated solution (determined on how well it fits the data) to make predictions about the effect of intended energy conservation measures (ECMs), use a small number of the top plausible solutions. Not only is one likely to obtain a more robust prediction of the energy and demand reductions, but this would allow determining their associated prediction uncertainty as well. The justification of doing this has been pointed out in Reddy et al. (2007) but is repeated below. The conventional wisdom was that once a simulation model was calibrated with actual utility bills, the effect of different intended ECMs could be predicted with some degree of confidence by making changes to one or more of the model input parameters that characterize the ECM. Such thinking is clearly erroneous since the utility billing data are the aggregate of several end-uses within the building, each of which is affected by one or more specific and interacting parameters. While performing calibration, the many degrees of freedom may produce good calibration overall even though the individual parameters may be incorrectly identified. Subsequently, altering one or more of these incorrectly identified parameters to mimic the intended ECM is very likely to yield biased predictions.
Though the methodology is applicable to any building energy simulation program, the scope of ASHRAE Research Project 1051 was restricted to the DOE-2 program, a widely used, public domain, fixed schematic hourly simulation program (Winkelmann et al. 1993) and to the widely prevalent case where year-long utility billing data are the only performance data available for calibration. It was presumed that the level and accuracy of knowledge about the building geometry, scheduling, and various system equipment would be consistent with a "detailed investment grade" audit, involving equipment nameplate information as well as some limited on-site measurements (clamp-on meters, etc.) performed during different times of the day (morning, afternoon, night) as well as over different days of the week in order to better understand variability in some of the simulation inputs. The intent of this paper is to summarize the results of applying the calibration methodology to two synthetic and one actual office buildings, as well as draw practical guidelines and conclusions on how to practically implement such a calibration methodology. Detailed descriptions of the buildings and systems (as well as the architectural and system files input to the simulation) can be found in Reddy et al. (2006).
DESCRIPTION OF BUILDINGS AND PARAMETERS
Two types of generic office buildings have been selected: (a) simpler buildings, generally small to medium in size with two or three stories, using unitary-type equipment with one source of energy, namely, electricity only, and (b) more complex buildings, generally medium to large in size with several stories, using decentralized equipment and more than one source of heat, namely, electricity and gas.
The calibration methodology was evaluated against two synthetic buildings (one simple and one complex) and a third actual building, which falls into the simple building category. Evaluation using the synthetic buildings involved selecting a building and specifying its various construction and equipment parameters as well as its operating schedules (called reference values), and using the DOE-2 simulation program to generate "electric utility bill" data for a whole year coinciding with calendar months. The utility billing data are then assumed to be the measured data against which calibration is performed. Since the "correct" or reference parameters are known beforehand, we can evaluate the accuracy and robustness of the proposed calibration methodology by determining how correctly the calibrated models can fit the utility bill data and also how accurately the effect of various ECMs can be predicted.
For the sake of simplicity, the calibration is assumed to be performed for the case when exactly one year of utility bills is available (i.e., 12 monthly bills). In reality, the analyst may use multi-year utility bills provided the building has not undergone any major changes in operation and equipment retrofits/replacements, since this is likely to reduce calibration uncertainty and result in a more robust calibration. This aspect is left for a follow-up study to investigate.
A summary of key information related to building size, geometry, and systems assumed for all three buildings is provided in Table 1. Synthetic building S1 is assumed to be a typical single-story office building of 25,000 [ft.sup.2] (2,322 [m.sup.2]) with packaged variable air volume (VAV) and electric heat representative of a simple building with one source of energy (electricity only). Typical Meteorological Year (TMY2) data from two geographic locations were selected for evaluation, Philadelphia, PA, and Dallas, TX. The synthetic complex office building selected (S2) is also simulated for two weather conditions, Atlanta, GA, and Dallas, TX, using TMY2 data. It is a class A large building (218,000 [ft.sup.2] [20,289 [m.sup.2]]) with seven floors and a penthouse with the lobby, cafeteria, service areas, and mechanical/electrical rooms on the first floor and offices on the remaining floors. Building cooling is provided by electricity, while heating is by natural gas. The actual office building selected (A1), located in Collegeville, PA (about 20 miles west of Philadelphia), is a class A building constructed in 2002 and occupied since the summer of 2003. Though a large building with four floors, it is all-electric and has four VAV rooftop units. Utility bills for a whole year (January-December 2004) are used as the basis for the calibration, along with included pertinent billing period read dates (which are around mid-calendar month), number of billing days, total electrical consumption, and demand data. The total electrical consumption was normalized to a daily basis, which is the values used during the calibration process. Weather data (temperature, solar, wind, etc.) synchronous with the utility billing periods are required by the simulation program and were picked from hourly climatic data acquired from the Climatic Services Division of the National Climatic Data Center, Ashville, NC.
Architectural details and summary sheets... |
More articles from HVAC & R Research
Validation of a coupled multizone-CFD program for building airflow and..., March 01, 2007
Lighting heat gain parameters: experimental method (RP-1282)., March 01, 2007
PCA-FDA-based fault diagnosis for sensors in VAV systems.(Principal co..., March 01, 2007
A methodology for diagnosing multiple simultaneous faults in vapor-com..., March 01, 2007
Measurement and correlation of two-phase frictional performance of ref..., March 01, 2007
Looking for additional articles?
Click here
to search our database of over 3 million articles.
|
.%20%20&position=1)