1、4691 Development and Implementation of HVAC-KBCD: A Knowledge-Based Expert System for Conceptual Design of HVAC in many instances, the selection is based strictly on the lowest owning (or first) cost system. Pricing information can be obtained from sources such as R.S. Means Mechanical Cost Data (Me
2、ans 2002a), R.S. Means Maintenance and Repair Cost Data (Means 2002b), equipment representatives, and construction and l. The terms “synthesis” and “configurations” are used interchange- 2001). ably in this paper. service contractors. In many cases, the engineer and the owner use a “selection matrix
3、” (as described in ASHRAE 2000) for decision making instead of the more rigorous procedure explained previously. The final product is a set of documents that include a full description of the design criteria and the design constraints, a description of the selected systems, sizes and capacity, a pre
4、liminary sequence of control for the proposed systems, and conceptual drawings and schematics. This allows the owner (or owners representative) to identify the most appropriate design to satis both the needs and stipulated budget. PROBLEM STATEMENT AND SOLUTION APPROACH Currently there are no mechan
5、isms to automatically synthesize feasible secondary and primary systems that can then be exported and linked to the corresponding models in an hour-by-hour building energy simulation program. As explained earlier, the configurations have to be defined a- priori, resulting in a limited number of alte
6、rnatives (as shown in Figure 1) and limited system configurations. The proposed solution methodology is to automate the process of generating a set of feasible HVAC Sriram 1997; Tong and Sriram 19921). Technical papers in this area include “HI-RISE, an expert system for preliminary design of high-ri
7、se buildings (Maher et al. 1988); “SEED-Config,” which is intended for conceptual structural design (Fenves et al. 2000); “BEAD (Fazio et al. 1989, 1991) for generating design alternatives for building envelope elements, and selection of HVAC systems for small buildings (Shams et al. 1994a, 1994b).
8、A special publication by ASHRAE (1 995) and Maor and Reddy (2003) provide a comprehensive literature review on the application of AI methods in building systems. PROCESS MODEL OF PROPOSED METHODOLOGY The solution methodology proposed here will involve automating the process of synthesizing secondary
9、 and primary HVAC however, if this information is unavail- able, a typical occupancy profile must be assumed. 1. Architectural knowledge includes four main components: Building application and type. Describes the application of the building, for example, office buildings, schools, retail spaces, hot
10、els, etc. The knowledge base will address specific issues that are critical for each applica- tion. Building class. A class can be defined as a subgroup of a building type. In the case of office buildings, there are three types of offices buildings-classes A, B, and C- that differ in standards and o
11、ther amenities (Gause 1998). Building geometry and structure. The basic geometry and building elements are normally known during the conceptual design. In many cases, a typical and simple geometry can be used to describe the building for mod- eling purposes. For example, a rectangular shape is widel
12、y used for office buildings. 236 ASHRAE Transactions: Research Building thermal zoning. Typical thermal zoning can be applied in order to model the building. For buildings with two or more stories, the minimum number of zones will be ten. Thermal zoning is required to accurately represent building z
13、ones for building energy simulation. The ten zones include four perimeter zones for floors without a roof (one for each exposure), one central core zone for floors without roof, four zones for each perime- ter zone with roof (for each exposure), and one core zone for a central core with roof. In som
14、e cases, the designer elects to add five more zones to represent the first floor in order to take into account ground coupling. Operating schedule knowledge relates to how the building is occupied and operated. Further, this is broken up into occupancy, lighting, equipment, and systems. In some case
15、s, this information is available from the architect/owner during the conceptual design. However, in many cases, this information is unknown, and typical schedules for occu- pancy, lighting, equipment, and systems can be used to model the operation of the building. Site-specific knowledge represents
16、all of the knowledge that is unique to the building being analyzed. Examples are geographical location, availability of energy sources, energy costs, and architectural constraints stipulated by the user or designer. HVAC however, it can be applied to class B buildings. 2. Secondary systems-conjigura
17、tion knowledge includes two levels. Level 1: Configuration of basic components to a zone subsystem deals with synthesizing a subsystem serving an individual zone of a building from basic components such as fans, coils, humidifiers, economizers, etc., which are further subdivided by the type of energ
18、y source used (Figure 7). A decomposition tree of the sys- tem (static knowledge) is used in conjunction with a heuristic search procedure (dynamic knowledge) to syn- thesize the subsystem. Systems are configured using a WAC a R Sysiems Table Table Table Table Table Table Table 2 1 3 Equipment opera
19、tion CHW plant availability Chiller #2 1 t Svstem Tvoe I Gas-Engine Driven Water-cooled Rotarv Chiller I ASHRAE Transactions: Research 245 Table 8. Description of Chiller Plant Configuration (continued) Compressor COP Engine Idle Speed Ratio Electrical Usage Minimum Operating Point Component Chiller
20、 Number Size I System Type I Gas-Engine Driven Water-cooled Rotary Chiller 1 4.9 0.3 0.00122 0.2 Value Definition Value Number of chillers 1 Percent of total plant 40 I Engine COP Secondary Pump Heat Recovery 1.55 Min airflow ratio 0.666 Power ratio at min 0.3 Requiredhot required No Engine jacket h
21、eat recovery effectiveness 0.287 Exhaust heat recovery effectiveness 0.213 Space heating Yes I- l I I I l CaDacitv ton I 1 I Flow GPM/ton 12.4 246 ASHRAE Transactions: Research Heuristlc 1 Knowledge i Figure 6 HVAC hence, the need to understand the various manner in which knowledge can be represente
22、d or captured. Basically, knowledge can be represented either (1) in a rule-based manner or (2) in a frame-based manner. In a rule-based system, rules are used to represent the problem, with each of the rules capturing some heuristics; the collection of the rules is the experts understanding of the
23、problem. In the case of a rule-based system, the knowledge engineer has to code each of the rules and link them logically to capture the experts thinking process. A frame-based system is entirely different: any object that has properties and values (examples are chillers, boilers, pumps) is part of
24、the system. After iden- tifying these objects, the knowledge engineer has to (i) collect and organize them such that they will contain class-instance relations and (2) find a methodology that will let these objects communicate with each other in a way that will provide a solu- tion to the problem. W
25、e suggest that both the rule-based and frame-based knowledge representation be used for HVAC a VAV system supply fan varies the airflow by utilizing inlet guide vanes or variable-speed drives. A heating system can include a hot water coil or an electric heater. In order to synthe- Table 9. Instance
26、of Frame-Based Representation of a VAV System for Perimeter Zone (VAV PI) VAV System-Perimeter Zone VAV P1 Object Name: Class Name: Properties: size subsystems, the structure of every system has to be repre- sented in a form that will create each and every configuration that meets the design require
27、ments. As presented earlier, a typical structure of a VAV system can be represented as a decomposition tree, which will allow configuration of VAV systems when configuration rules are applied. This is a Level 1 configuration. The result is a set of subsystems that will occupy the last level in the d
28、ecomposition tree shown in Figure 9. (TPFC-P and FPFC-P denote two- pipe and four-pipe fan coil units, respectively, serving the perimeter zones). The same approach to knowledge represen- tation can be applied to every system, forming a complete tree of secondary and primary systems. Figure 10 illus
29、trates the synthesis of chiller plants (using centrifugal air-cooled and water-cooled, absorption and gas-driven engines) subjected to configuration rules with chiller type, percent size, and sequence being the three variables for each chiller. Rule-Based Knowledge Representation Durkin (1 994) defi
30、nes a rule-based (or production) system as a computer program that uses an inference engine to process problem-specific information contained in the work- ing memory, using a set of rules contained in the knowledge base. In order to recompose subsystems from basic compo- nents and to synthesize buil
31、ding systems from subsystems, rules must be applied that can be can be divided into three groups: Rules for subsystem synthesis (Level 1) are also divided into two main groups, each group representing different sets of rules that must be applied to recompose sub- systems. These two groups are define
32、d as subsystem assembly rules and subsystem component application rules. The first group represents rules that deal with compatibility of components within subsystems. Exam- 248 ASHRAE Transactions: Research wlPlan ts- Perimeter VAV-P2 VAV-PB VAVLP4 VAV-P5 VV-PG Constrained Congursons tsMo Figure 9
33、Hierarchical decomposition and configuration table ofperimeter VAV systems serving three distinct zones. I I Chiller Plant Cntr6UC-R4,000, perimeter sys- tems will be with radiators Gas is not available, chiller will not have any absorption or gas engine chiller Table 13. Example of Owning Cost Func
34、tion and System Factor I Cost Function Coefficients I AHU 1.5768 1.5768 1.5768 System Humidifier Zone Reheat Baseboard Heat Ductwork Diffusers Piping Electrical Control Factor 0.000 0.669 0.000 1.722 4.693 1.117 0.075 0.000 9.852 0.000 1 .O67 0.000 1.722 4.693 0.000 0.075 0.000 9.133 0.000 0.000 2.1
35、21 1.722 4.693 0.680 0.075 0.000 10.867 Owning or first cost for each configuration is calculated by using routines embedded in the DES, combined with an owning-cost library. The owning-cost library is, in essence, a set of owning-cost models structured in the same manner as the secondary and primar
36、y systems generated in the synthe- sizer module. For example, a secondary system with the desig- nation VAVP1 has an owning-cost model that represents a cost relation against system airflow CFM. A typical VAV cost model includes the following: (a) air-handling unit (AHU) with its major components su
37、ch as coils, fans, economizer, etc., (b) zone reheat and baseboard heating, (c) ductwork and duct insulation, (d) diffusers and grilles, (e) piping and piping insulation, and (f) electrical and controls. In order to quantify these values for the application under investigation, cost infor- mation fr
38、om Means (2002) is used along with estimated quan- tities of ductwork and piping for typical floor plans. These values are normalized such that a final relation of cost ($) vs. airflow rate (CFM) can be established through regression analysis. For example, a model of owning cost for a VAV system ser
39、ving a perimeter zone is of the type: System cost ($) = AHU constant + (CFW e * Area * System Factor) where the AHU constant and system factor are empirical coef- ficients whose numerical values are determined by linear regression (see Maor 2002 for complete details). Table 13 assembles a list of mo
40、del coefficients for three different VAV systems serving perimeter zones. For system VAVP1, the AHU constant is 10,310 and the system factor is 9.852. These numerical values are computed from the costs of each individ- ual component, which constitutes the complete system. The same technique can be u
41、sed for determining owning costs of primary systems. Here, the owning cost must include, not only major components such as chillers and cooling towers, but also equipment such as pumps and piping. For example, a chiller plant (see Table 14) that is to be normalized as $/ton includes the following co
42、mponents: chiller (can be from the electric chiller family or thermal), cooling tower, evaporator pump, condenser pump, expansion tanks and accessories, piping, valves and piping specialties, and electri- cal and controls. A similar component breakdown is performed for determining owning cost of var
43、ious types of boiler systems and DHW. The maintenance cost of primary systems can be calcu- lated similarly (Maor 2002). Although this approach is not extremely accurate, it is adequate since the goal is a conceptual design. This methodology can provide cost information to the owner and the designer
44、 sufficient to narrow the search space to the most promising design alternatives. ASHFtAE Transactions: Research 251 Chiller System Electric Rotary WC (ROTWC) Chiller 7 150.000 75 100 125 150 175 200 250 300 350 400 450 ton ton ton ton ton ton ton ton ton ton ton 52,900 59,800 65,838 71.013 81,219 7
45、0.725 85,963 90.850 104.363 110.975 123.625 I I I I? I? I? I? I? I I?I?I?I?I Cooling tower Evaporator pump 16,113 19,128 22,142 25,157 28,172 31,187 37,217 43,247 49,276 55,306 61,336 67,366 9,475 9,963 10,897 11,831 12,156 12,319 12,481 14,147 14,980 15,813 15,813 17,438 Total $ $Iton 1369 1155 102
46、2 922 887 724 102,638 115,493 127,742 138,330 155,297 144,822 INNER WORKINGS OF HVAC-KBCD SYNTHESIZER The purpose of this section is to explain in detail the intri- cacies of the HVAC-KBCD synthesizer, which is the primary and unique conceptual novelty of this paper. While Figure 2 showed the overal
47、l conceptual methodology of the entire process, Figure 11 details the inner working, i.e., the logical sequence of operations in the synthesizer element of the HVAC-KBES module. The synthesizer element is the major component of the KBES. It is responsible for configuring HVAC lack of method- ologies
48、 to synthesize subsystems and systems, resulting in heavy reliance on heuristics; pauciy of experienced design- ers; and increasing complexity of energy rate structures and new energy code requirements. This paper then describes a design methodology involving integrating a KBES that can automate the
49、 synthesis of the various secondary and primary systems of a building HVAC&R system and currently avail- able building simulation programs to provide an efficient solu- tion to the above problems. This paper describes the general framework, structure, capabilities, and inner working of the various elements of the KBES (which we call HVAC-KBCD). The system configurations from the HVAC-KBCD can be made to comply with current energy standards (which are partially embedded in the knowledge base), thereby ensuring compliance with energy codes. It is important to note that
