1、 Zheng ONeill is an assistant professor and Yanfei Li is a PhD student and a graduate research assistant for the Department of Mechanical Engineering at the University of Alabama in Tuscaloosa, AL. An EnergyPlus/OpenStudio-based Fault Simulator for Buildings Yanfei Li Zheng ONeill, PhD, PE Student M
2、ember ASHRAE Member ASHRAE ABSTRACT Building energy systems often consume in excess of 20% more electrical energy than was the design intent largely because of equipment performance degradation (e.g., filter or heat exchanger fouling), equipment failures, or detrimental interactions among subsystems
3、 such as cooling and then reheating of conditioned air. Identifying the root causes of efficiency losses is challenging because a gradual erosion of performance can be difficult to detect. Furthermore, diagnostic algorithm performance is limited by available fault ground truth data. An analytical fr
4、amework and model-based simulation capability is desired to develop fault ground truth data that can be used to deploy robust diagnostics for building energy systems including building envelope, lighting, Heating, Ventilation and Air-conditioning (HVAC) equipment and systems, etc. Such fault simulat
5、ors can also be used for fault impact analysis for risk management. An EnergyPlus/OpenStudio-based fault simulator is being developed for such purposes. EnergyPlus is a whole building simulation free program from DOE. This paper is focusing on the faults that are implemented using OpenStudio measure
6、s. These measures are created in OpenStudio Application or the Parametric Analysis Tool, which are written in Ruby scripts. These faults related measures act like add-on macro to make changes to the existing energy model to reflect faults. This fault simulator aims to simulate a variety of faults fr
7、om building subcomponents and subsystems including building envelope insulation, occupancy schedule, air handler economizers, heating and cooling coils, fans, etc. The development of such a fault simulator using OpenStudio measures and testing results of fault impacts in terms of energy consumptions
8、 will be presented in this paper. INTRODUCTION Building heating, ventilation, and air-conditioning (HVAC) systems are very complex with many subsystems and equipment including cooling loops, heating loops, and auxiliary appliances (e.g., fans and pumps). It is not uncommon that HVAC systems and equi
9、pment fail to operate at the desired and normal conditions, leading to achieving up to 40% of the energy saving potentials (Narayanaswamy 2014). The failures, or faults, are categorized as either abrupt or degradation. (Haves, 1997). The typical abrupt faults are the sudden failure of equipment part
10、s, like broken fans or stuck outside air dampers. The degradation faults usually go unnoticed like the fouled coils or the leaking valves that occur after some period of operation. The faults also cause the discomfort and poor indoor air quality for indoor environments (Mills 2010). Since the 1980s,
11、 fault detection and diagnosis (FDD) has received a lot of attentions and has been used to solve problems causing the abnormal energy consumption in buildings. For real HVAC systems, there are a number of FDD algorithms and tools available either by stand-alone software or embedded software (Hyvrine
12、n and Krki 1996). The International Energy Agency (IEA) Annex 25 investigated the building optimization and FDD (Hyvrinen and Krki 1996), and they discussed the physical-model based FDD and data-driven based FDD (e.g., ARMAX model,). The IEA Annex 34 (Arthur and Jouko 2001) addressed the practical i
13、ssues of HVAC FDD tools implementation in real buildings. But the model-based FDD methods are relying much on the accuracy of the reference models, which are sometimes not readily available. Yu and Paassen (2002) provided a general modeling method for model-based FDD of building HVAC systems, which
14、is composed of a hierarchical modeling procedure, parameterization and tunings. But this approach needs the integration of real data from building energy management system with the model. House and Vaezi-Nejad (2001) introduced a rule-based FDD algorithm for air handling unit (AHUs) and validated it
15、s accuracy. For all these FDD algorithms, they need to be tested and validated using data that contains faults. Ideally, we would like to use the real operation data with known faults for this purpose. Functional tests in buildings could be performed to generate some faulty data. However, such a pro
16、cedure is complicated, time consuming and sometimes cost prohibitive. An analytical framework and model-based fault simulation is well positioned for generating normal and faulty ground truth data to test different FDD algorithms. On the other hand, by studying the fault simulation, we can quantify
17、fault impacts in terms of building energy consumption. This will enable facility managers to better understand the fault behaviors and the relationship between building energy consumption and faults. Unfortunately, most of existing building simulation programs such as EnergyPlus (EnergyPlus 2015), T
18、RNSYS (TRNSYS 2015), eQuest (eQuest 2015), etc. either have limited fault simulation capabilities or always assume normal status for HVAC systems and equipment. This paper presents a preliminary fault simulation using the OpenStudio platform. First, some background information about state-of-the-art
19、 fault simulation and OpenStudio will be introduced, and then fault modeling using OpenStudio Measures is illustrated using three typical faults (i.e., faulty outside air damper, faulty fan, and fouled heating coil). This will be followed by a case study of a fault impact analysis using a DOE medium
20、 office references building. BACKGROUND There are 417 building energy modeling and simulation programs listed on the DOE website (DOE 2015). However, most of them are not capable of fault simulation. Liu (1997) developed the AirModel, which is a building simulation tool for simple fault simulations
21、of an airside system. But the AirModel cannot be used for HVAC waterside fault simulations. Liu et al. (2002) reviewed and assessed the AirModel for fault simulations of AHUs. They recommended EnergyPlus in lieu of AirModel for simulating faults. Nevertheless, EnergyPlus indeed has limited capabilit
22、ies in fault simulation. Basarkar et al. (2011) identified and characterized 18 HVAC faults simulated in EnergyPlus with a goal to assess the fault impacts on building energy consumption and occupant comfort for retrofit analysis. It further demonstrated that EnergyPlus has a great potential for fau
23、lt simulations. In the building community, researchers have been developing the dynamic faults models for building systems and equipment in Modelica (ONeill and Chang 2011) as well. OpenStudio provides a graphic interface of EnergyPlus for implementing the energy model changes and analyzing the ener
24、gy influence thereafter. A “measure” in OpenStudio is a generic tool for model modification. A measure is a set of Ruby scripts (Ruby 2015) written by users for modifying specified parameters. It can be applied to individual models or generic models, which can greatly reduce the modeling time and ef
25、forts. The measures are usually used for energy efficiency and energy conservation purposes. Currently, there are a total of 185 measures for lighting, HVAC, reporting, etc. in NRELs Building Component Library (BCL) website (BCL 2015). For example, one of the measures is Enable Economizer Control. T
26、his measure can be used to control economizer (On/Off). Users can also develop their own measures for specific purposes. This feature makes the fault simulation possible through measures in OpenStudio that uses EnergyPlus as the simulation engine. Such fault simulation will only need a small portabl
27、e interface written in Ruby scripts to be read into OpenStudio. Other fault simulation approaches include directly modifying the source code of OpenStudio/EnergyPlus, which is more time-consuming. Thus, OpenStudio measures provide much flexibility to manipulate the OpenStudio/EnergyPlus parameters a
28、nd model variations. However, currently there are no fault-related measures in NRELs BCL (BCL 2015). One of the objectives for this pilot study is to create measures in OpenStudio for fault simulation in buildings. As long as the user can modify the parameters associated with building envelope, HVAC
29、 system, equipment, etc. into faulty states, a corresponding measure (if available) can be directly applied to simulate faults. For example, a fouled coil can be modeled by modifying the UA factor (defined as the product of overall heat transfer coefficient U (W/m2-K BTU/hr-ft2-F) and the coil heat
30、transfer surface area A (m2ft2) using the measure described later in this paper. METHODOLOGY There are two ways to implement measures in OpenStudio. First, users can download the available measures from the BCL website and apply them to the baseline models directly, if these measures are the appropr
31、iate ones for the selected purpose. For example, the output reporting measures can be utilized to report available output variables from EnergyPlus in OpenStudio. Another way is to create the customized measure by following the instructions of the Measure Writing Guide (MWG 2015). The Measure Writin
32、g Guide is a good start for developing measures by object-oriented programming of Ruby to select or modify the appropriate EnergyPlus/OpenStudio objects of building envelope, HVAC system, equipment, operation schedules, etc. These measures can be translated, recognized, and implemented in OpenStudio
33、. An important step is to explore the OpenStudio application and the user interface to mark down all the objects involved with the specific problem. Then following the Software Development Kit (SDK 2015) documentation, it is necessary to understand the inheritance diagram and member functions, where
34、 the user has to identify the appropriate models, functions, arguments, etc. After a customized measure is created, it is strongly recommended to debug the customized measure before any application. In this paper, three faults measures are introduced. First, a general description of the fault itself
35、 and associated energy impacts is provided, followed by how the fault measure is created in OpenStudio. Then, functions related to the fault parameters that the user needs to modify or add directly are described. An inheritance diagram from the SDK documentation to assist the measure development is
36、also provided. 1. Faulty outside air damper (stuck minimum damper position) During normal operation of an air side economizer, the outside air fraction will vary based on the comparison of dry bulb air temperature (or enthalpy) between outside air and return air to achieve as much free heating or co
37、oling as possible. During operation, the minimum damper position could become stuck at a higher percent open due to control loops or mechanical problems. The economizer would have no control of the normal minimum outside air fraction with such a fault. The outside air damper would vary between a hig
38、her minimum position and the maximum according to the control logic and operation. Bringing more outside air into the system when the system is required to be operated with a minimal outside airflow rate. This will lead to energy waste by heating the extra cold outside air in the winter and cooling
39、the extra hot outside air in the summer. In OpenStudio/EnergyPlus, the amount of outside air is decided by the outside air controller. In this study, the minimal outside air flow rate is the product of the given minimum outside air flow rate and the minimum outside air schedule. (Note: There are oth
40、er ways to define the minimal outside air flow rate in EnergyPlus.). The prerequisite is that the economizer be enabled (BCL 2015). In this fault measure, the minimal outside airflow rate can be modified by users. The measure also assigns a schedule for the minimal outside air flow rate. The minimum
41、 amount of outside air entering the HVAC systems is calculated by: 0kairVV= (1) Where is the actual (faulty) minimal outside airflow rate entering the system (m3/s CFM), is the normal minimal outside air flow rate (m3/s CFM), k is the schedule factor of the minimal outside airflow rate. As an exampl
42、e presented in the study (see details in the next section), the minimal outside airflow rate is changed from 0.72 m3/s (1,526 CFM) at the normal state to 4.32 m3/s (9,154 CFM) for the faulty state. A snapshot of this measures graphical user interface (GUI) in OpenStudio is shown in Figure 1(a). The
43、inheritance diagram (SDK 2015) is a graph illustrating the complicated inheritance relationships among different objects of object-oriented airV0Vprogramming in OpenStudio. This diagram helping helps developers to obtain proper functions for various measures development. A snapshot for the inheritan
44、ce diagram of faulty outside air damper measure is shown in Figure 1(b). (a) Measure GUI Snapshot (b) Measure Inheritance Diagram Snapshot Figure 1 Faulty Outside Air Damper Measure 2. Faulty supply fan with a stuck minimum speed During normal operation, the variable air volume fan will vary the air
45、flow rate according to the demand side airflow requirements. The airflow rate will vary from the minimum airflow rate to the maximum airflow rate. This faulty fan with a faulty minimum speed assumes the fan is out of control either due to a non-functional motor controller or damaged control communic
46、ations. The fan will move more air than is required when stuck at a higher speed, or it will move less air when stuck at a lower speed. The fan power consumption and airflow rate are related as described in Equations (2)-(4) (EnergyPlus 2015), where the fan performance curve equation (Equation (3) i
47、s crucial. (2) (3) max max/ (e ) / (6350e ) ( )p pl tot air p pl tot airQ f m P or Q f m P IP unit = =(4) Where m is the actual mass flow in kg/s (lb/min), is the maximum flow or design flow in kg/s (lb/min), are the 5 fan power coefficients, is the flow fraction, is the design pressure rise in Pa (
48、in. of water), is the fan total efficiency, is the air density at the standard condition in kg/m3 (lb/ft3), is the part load factor, is the total fan power in W (hp). When the system is required to be operated at the minimum airflow rate, a fan with a larger than the normal minimum speed will result
49、 in a higher flow fraction and a higher part load factor correspondingly. Therefore, this fault will increase the total power consumption whenever the system is operated compared to operating at the normal minimum airflow rate. When operating with an airflow rate between the faulty minimum airflow rate and the maximum airflow rate, there are no impacts on fan power consumption. An example of this measure is when the nominal fan minimum airflow rate is 0.0 m3/s (0.0 CFM), we can change it to 4.0
