1、AN-04-3-1 Development and Testing of the Characteristic Curve Fan Model Jeff Stein, P.E. Member ASHRAE ABSTRACT This paper describes the development and testing of the characteristic curve fan model-a gray-box model. This model produces fan ejciency as a function of airflow and fan static pressure.
2、It is accurate, relatively easy to calibrate, and could be easily incorporated into commercial simulation programs. Also presented is an application of an existing model to predict fan speed from airflow and fan static pressure. These models were developed as apart of a larger research project to de
3、velop design guidelines for built-up variable air volume fan systems. The models have been successfully employed in comparative analysis of fan types, wheel diameters, fan staging, and anal- ysis of supply pressure reset. INTRODUCTION The authors were part of a publicly funded energy e%- ciency rese
4、arch team developing design guidelines for built- up fan systems in commercial buildings. According to previ- ous research, fan energy in new construction for commercial buildings in California accounts for 1 terawatt-hour of electric energy usage per year, representing approximately half of all HVA
5、C energy usage (CALMAC 2003). The authors research demonstrates that up to half of that fan energy is avoidable through cost-effective design practices, including fan selec- tion (size and type), fan sizing, fan staging, and static pressure control (Hydeman and Stein 2003). Five monitoring sites pro
6、vided field data on which to test the alternative fan system designs and design techniques. These sites were selected to represent a range of climates, occupancies, and fan system configurations (Kolderup et al. 2002). As part of this work, a simulation model of a fan system was sought that had all
7、of the following characteristics: Mark M. Hydeman, P.E. Member ASHRAE Accurate at predicting fan system energy over the full range of actual or anticipated operating conditions. Applicable for the full range of fan types and sizes. Easy to calibrate from manufacturers or field-moni- tored data. Abil
8、ity to identify operation in the “surge” region. Relatively simple to integrate into existing simulation tools. Ability to independently model the performance of the fan system components, including the motor, the mechanical drive components, the unloading mecha- nism (e.g., VSD), and the fan. The p
9、urpose of this model is to evaluate design alterna- tives for fan selection and control through simulation. Opti- mally, simulation tools would directly utilize the manufacturers fan curves to evaluate fan system operation at each discrete step of evaluation. Since this is not currently available, t
10、he authors sought models that simulation tools could easily incorporate that replicated fan performance. MAIN BODY Literature on component models for fans was reviewed, including the models used in the DOE-2 simulation program (DOE 1980) and in the ASHRAE Secondary Toolkit (Bran- demuehl et al. 1993
11、; Clark, 1985). We also looked briefly at the models embedded in commercial simulation software, such as Trace and HAP, but found these suffered from the same problems as the model in DOE-2. DOE-2 uses a black-box regression model that produces the fan system power draw as a function of percent desi
12、gn airflow using a second-order equation as follows: Jeff Stein is a senior engineer and Mark Hydeman is a principal at Taylor Engineering, LLC, Alameda, Calif. 02004 ASHRAE. 347 This model is implicitly built on several assumptions: Each fan operates on a single system curve that uniquely maps airf
13、low to static pressure. Fan system efficiency is directly a function of airflow. A second-order equation sufficiently models both of these effects. The DOE fan model implicitly combines the operating system curve with the models for each of the fan system components. Power is directly produced as a
14、function of airflow only, and there is no opportunity to have different conditions of fan static pressure at a given airflow. Real VAV systems do not remain on a fixed system curve. System pres- sure as a function of airflow behaves differently depending on the location of the boxes that are modulat
15、ing, the location of the static pressure sensor(s), and the static pressure control algorithm. Although this model is simple to use, it does not allow the user to independently model and evaluate each of the fan- system components. Thus, if designers wanted to evaluate the impact of motor oversizing
16、, they would have to independently assemble fan and motor models to develop the DOE-2 perfor- mance curve that represented the combination of the two together. This model also does not directly account for the variation in fan system component efficiencies as the fan unloads, nor does it allow for e
17、valuation of a multiple fan system, where fan staging will change both the operating effi- ciency and potentially the individual fan static as they are staged on and off. The model in the ASHRAE Secondary Toolkit is a gray- box fan component model that uses the perfect fan laws through application o
18、f dimensionless flow (4) and pressure (v) coefficients. This model uses a fourth-order equation to predict fan efficiency from the dimensionless flow parameter. 1. 2. 3. CFM NxD3 cp = c,x- (3) qfan = a + b x Q +C x Q2+dx Q3 + e x cp4 (4) where CFM = airflow N = fanspeed D = fandiameter P = average a
19、ir density AP = fan static pressure and C, and C2 = constants that make the coefficients dimensionless This model allows the user to calibrate an entire family of fan curves with data from a single model. Unfortunately, this model does not permit the direct calculation of fan efficiency from airflow
20、 and pressure; rather, it correlates efficiency to the dimensionless flow term (+I), which requires both airflow and fan speed as inputs. As elaborated below, a designer (and most simulation tools) will use airflow and fan pressure as inputs to the fan system model in order to calculate fan speed an
21、d effi- ciency. A second problem is that this model assumes a fixed peak efficiency for fans of all sizes. This simplification reduces the applicability of the fan model for comparative analysis of fan options as peak efficiency tends to increase with fan diameter. As a result of these shortcomings,
22、 the authors set out to develop a new component model that could directly be driven by airflow and pressure. Based on the “fan laws” (ASHRAE 2000), the core assumption of this new “characteristic curve” fan model is that the efficiency of a fan is constant as the fan rides up and down on a particula
23、r characteristic system curve. Extensive testing with manufacturers fan selection software demonstrates that the manufacturers also use this simplifying assumption for developing fan performance data in both the surge and non-surge regions. ANSUASHRAE Standard 5 1 - 1999 (ANSUAMCA Standard 2 10-99)
24、(ASHRAE 1999) explicitly permits this. For this model, a “characteristic system curve” is defined as a second-order equation, equating fan static pressure to airflow (cfm) with a zero constant and no first-order coefficient. For example, a VAV supply fan with a fixed duct static pressure setpoint of
25、 1.5 in. W.C. will ride up and down on a system curve that runs through the design point and through 1.5 in. at O CFM. A “characteristic system curve” is a particular type of system curve in that it must run through the origin (O in. at O CFM). A characteristic system curve is characterized by a sin
26、gle coefficient, SCC (system curve coef- ficient). The equation for any characteristic system curve is (5) AP CFh? SCC = - Using this assumption, it is only necessary to find fan performance at a single point on a characteristic system curve to define its performance along that curve at all speeds.
27、As depicted in Figure 1, there are three characteristic system curves of particular importance: the curves at the minimum and maximum ends of the tuning data set and the curve that represents the highest efficiency for the fan. As described below and depicted in Figure 3, fans behave very differentl
28、y on either side of this peak efficiency. The minimum and maxi- mum curves represent the boundaries of the model tuning data. The triangles in Figure 1 depict points of data that were sampled from the manufacturers fan selection software. Each point represents the fan efficiency for all points on a
29、charac- teristic system curve. The fan efficiency is calculated from the fan brake horsepower (BHP), airflow (CFM), and fan static pressure (AP) reported by the software through the following equation: 348 ASHRAE Transactions: Symposia CFMx AP fan = 6350 x BHP The model can be used to predict the fa
30、n power for any point whose system curve is between the two extreme system curves. Figure 1 is overlaid on top of an output screen from a manufacturers selection program. Notice that the peak effi- ciency line is also the boundary of the manufacturers “Do Not Select” or surge region. This is typical
31、 for plenum, backward inclined, and vane-axial fans. For airfoil, mixed flow, and propeller fans, the peak efficiency is well to the right (ie., outside) of the surge region (see Figure 5). I click on link for Large WAC Integration. The Department of Energys Motor Challenge market transformation pro
32、gram () and MotorMastert Program (. Clark, D.R. 1985. HVACSIM+ building systems and equip- ment simulation program: Reference Manual. NBSIR ASHRAE Transactions: Symposia 355 84-2996, US. Department of Commerce, Washington D.C. DOE (Department of Energy). 1980. DOE 2 Reference Man- ual, Part 1, Versi
33、on 2.1. Lawrence Berkeley National Laboratories, Berkeley Calif., May. Gao, X., S.A. McInemy, and S.P. Kavanaugh. 2001. Efficien- cies of an 11.2 kW variable speed motor and drive. ASHRAE Transactions 107(2). Atlanta: American Soci- ety of Heating, Refrigerating and Air-conditioning Engineers, Inc.
34、Hydeman, M., J. Stein. 2003. A fresh look at fan selection and control. HPAC Magazine, May. Kolderup, E., M. Hydeman, M. Baker, and R.L. Qualmann. 2002. Measured performance and design guidelines for large commercial HVAC systems. ACEEE Conference on Energy Efficiency, August. DISCUSSION David Yuill
35、, Principal, Building Solutions Inc., Omaha, Neb.: We have done some similar work in which we devel- oped a model to predict airflow through a fan using the design fan curve, fan head, and fan speed as inputs. We set up an experiment to test this model, but we found that the manufac- turers fan curv
36、es were not accurate. Have you found this, and are you aware of any data on fan curve accuracy? Jeff Stein: As noted in the paper we did not find a good corre- lation between measured fan energy and predicted energy (based on manufacturers data). There are many possible reasons including: (1) The pr
37、essure and air flow sensors used in the field tests may be inaccurate. (2) Even if the sensors are accurate, field tests of fan operating static cannot match the AMCA test conditions of “fan static“ in the lab: it is impossi- ble to measure “fan static“ (the Y axis on a fan curve) in the field since
38、 the installation conditions are completely different. System effects play havoc with fan performance in the field. (3) Manufacturers do not test all sizes. The AMCA rating standard for fans allows them to test a fan and extrapolate the results to all larger fans of the same type. (4) Accuracy of th
39、e manufacturers tests. We noticed that performance data for some fan types got worse and then better as you move to larger sizes. This suggests that there may be variability in the manu- facturing or testing processes. We are not aware ofany data on fan curve accuracy. 356 ASHRAE Transactions: Symposia
