Date of Award

Fall 2014

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Mechanical Engineering

First Advisor

James E. Braun

Committee Chair

James E. Braun

Committee Member 1

Piotr Domanski

Committee Member 2

Eckhard Groll

Committee Member 3

W. Travis Horton

Abstract

This thesis describes the development of standardized methodologies and methods for evaluating the performance of fault detection and diagnostics (FDD) protocols that are intended for air-cooled vapor compression unitary air-conditioners. Although there are currently many FDD protocols available commercially and there are jurisdictions in which the building codes require FDD, there is currently no standardized method to determine how effective the FDD might be at detecting and diagnosing faults.^ An FDD evaluation method is proposed, in which the evaluator passes data from a large number of scenarios to a candidate FDD protocol, and collects the responses from each scenario. The possible results for each scenario are: (a) No Response; (b) Correct; (c) False Alarm; (d) Misdiagnosis; (e) Missed Detection; (f) No Diagnosis. The set of all responses is organized by considering the impact of the fault on the system's cooling capacity and efficiency for each scenario. This is referred to as fault-impact-based evaluation.^ Input data for the evaluation process have been collected from experiments conducted in several different laboratories in previous research projects. These data have been vetted and organized into a standardized format that was developed as part of this project. The data have also been used by a colleague, Howard Cheung, to generate gray-box models, so that evaluations based on simulation data can be conducted. This approach has several advantages over using experimental data. The most important of these advantages is that the finite set of existing experimental data are not sufficient for meaningful evaluation of FDD performance because they aren't uniformly distributed throughout the input space.^ The evaluation methods are illustrated with case studies of a six FDD protocols that are currently in widespread use. Most of these protocols are found to perform quite poorly, underscoring the importance of evaluation of FDD protocols. However, the results are complex and can be difficult to interpret. ^ To provide more meaningful results, a figure of merit (FOM) is proposed. This FOM indicates the overall utility of a protocol by analyzing economic considerations from typical applications. It considers a wide-ranging set of potential FDD deployment conditions and assigns monetary costs and benefits to them, weighting each with its probability of occurring. Several case studies are conducted on nine FDD protocols and they show that one of the protocols is far more valuable than the others

Share

COinS