Home | Business News | Browse by Publication | H | HVAC & R Research

Enhancing the reliability of chiller control using fused measurement of building cooling load.

Article Excerpt
INTRODUCTION

Centralized chilling systems with multiple centrifugal chillers are commonly used in large commercial buildings to cool occupied spaces. More than one-third of total energy consumption in most air-conditioned commercial buildings is used by these chillers in a climate like Hong Kong's (Lam 2000). When multiple chillers are employed, it becomes important to properly sequence their operation to ensure that the operating chillers achieve an overall coefficient of performance that is as high as possible, while also fulfilling the demanded cooling load (Hackner et al. 1984; Honeywell 1997; Chang et al. 2005a). In principle, the chiller sequencing control aims to determine how many and which chillers are to be put into operation. It plays a significant role in the overall performance of the whole air-conditioning system.

There are various methods of chiller sequencing control being used in various buildings. The differences in these methods lie mainly in how the instantaneous building cooling load is measured (Honeywell 1997). Nowadays, the total cooling load sequencing control is often adopted, in which a direct way of measuring building cooling load is used. The direct way determines the building cooling load by measuring the total flow rate of chilled water and the difference between the chilled water supply and the return temperature. The total cooling load sequencing control is, in principle, the best approach for optimal chiller sequence control (Honeywell 1997). However, according to the surveys in Hong Kong and elsewhere, the direct measurement of the building cooling load often does not provide reliable measurement of the building cooling load in practice due to the noises, outliers, and systematic errors in measuring the water temperature and the water flow rate (Kwan 2001). A site study was conducted in a number of "normally" maintained chilling systems in Hong Kong by using one redundant set of sensors to measure the differential temperature (Kwan 2001). The results showed that the difference between the two sets of sensors was more than 30% in a large proportion of the chilling systems investigated.

The building cooling load computed by the direct measurement is actually the cooling load of the chillers. Because power consumption is a direct indicator of the chiller cooling load, the application of the power measurement to estimate the building cooling load is also developed, which measures the building cooling load in an indirect way (Wang et al. 2000). There are several benefits to measuring the building cooling load based on the power consumption of chillers for chiller sequencing control. First, the power measurement is now more affordable because of its regular instrumentation in building automation systems (BAS). Second, electrical variables can be measured accurately and reliably when compared with measuring thermo-physical variables. However, correlating the power consumption to building cooling load is complicated due to the influence of the operating condition, such as part-load chiller efficiency and the capacity change due to the change of condenser and evaporator water temperature (Wang et al. 2000). A chiller model is therefore required to be developed for precisely describing the relationship of the building cooling load with power consumption. The model-based building cooling load measurement is called indirect measurement in this paper. An obvious disadvantage of the indirect measurement is that the quality of the building cooling load measurement depends on the accuracy of the chiller model. A precise chiller model should take all of the possible operating conditions into account. Besides, the chiller performance degradation will affect the chiller model accuracy as well. Therefore, measurement derivation, due to model errors, should exist universally in the indirect measurement.

Accurate and reliable measurements are essential for the accuracy and reliability of the chiller sequencing control as well as for building air-conditioning system performance monitoring and optimization (Wang and Cui 2005; Chang et al 2005b; Xu et al. 2008). Improving the building cooling load estimation will improve the performance of the chiller sequencing control. Data fusion is one of the methods that can be used to improve the accuracy and reliability of the measurements (Yager 2004; Esteban et al. 2005; Ruhm 2007; Urbanski and Wasowski 2003). Current fusion methods are primarily based on statistical theory and always lead to a weighted average of the observations from different sources (Grewal 2001; Soderstrom et al. 2001; Ozyrut and Pike 2004; Tan 2006).

In this paper, the data fusion concept will be used to improve the building cooling load measurement by removing the noises, outliers, and systematic errors in the direct measurement using the information provided by a related and simplified physical model, the indirect measurement. Since the indirect measurement is not so statistical (suffering mainly from model errors), a simple data fusion algorithm is developed to capitalize the benefits of the direct and indirect measurement. The accuracy and reliability of the fused measurement is then increased by combining the complementary advantages of the two different measurement methods (Huang et al. 2007). The fused measurement is used instead of the direct or indirect measurement in the chiller sequencing control. The reliability of the chiller sequencing control will be improved, because the fused measurement suffers little from measurement noises, outliers, systematic errors, and model errors. The degree of confidence, which indicates the quality of the fused measurement, is evaluated systematically. Furthermore, the fusion parameters are updated periodically in order to improve the performance of the fusion algorithm, as well as the reliability of the chiller sequencing control.

The paper starts by illustrating the...