If a variable speed pump control is applied in Case

If a variable-speed pump control is applied in Case 2, then the electricity consumption of the water pumps can be reduced. However, handling a situation where the cooling consumption is extremely low and unevenly distributed among different households is relatively difficult (as Fig. 8 shows). Other pumping systems, such as primary-secondary and primary-tertiary systems, may mitigate the problem, which requires the introduction of simulation methodology and further analysis. Therefore, from the above analysis, the main reason for the low system efficiency is that the cooling load in residential buildings exhibits desynchrony and low load ratio, and these features considerably differ from those in public buildings. During the design phase, the engineering load calculation method usually focuses on the peak load for equipment sizing, whereas load sphingosine 1-phosphate receptor modulator is rarely considered. In addition, even when the hourly cooling load is calculated, the internal load and schedule of each household are the same; thus, the disparity due to desynchrony and low load ratio would not be considered. In Case 2, the AC terminals are adjustable; however, the water system has a constant flow rate, which is in conflict with the users’ decentralized cooling load. Particularly, in Case 2, the heat distribution process of chilled water between the refrigerating machines and AC terminals becomes the contradiction boundary. This contradiction makes the distribution and consumption of chilled-water pumps the main reason for high energy consumption. From Table 5, the transport coefficient of chilled water in Case 2 is only 5, which is significantly below the standard limit value of 30. The non-adjustability of the chilled-water system influences the adjustability and operating condition of chillers and cooling water systems, thereby resulting in low system efficiency.
Conclusion The data from the three cases are analyzed and discussed in this study, and the following conclusions are presented:
Acknowledgments This work was supported by the Key Projects in the National Science and Technology Pillar Program during the 13th Five-year Plan Period (No. 2016YFC0700102) and Jiangsu Natural Science Foundation (BK20160683).
Introduction The automated layout design problem is referred to as a space allocation problem in the field of computer-aided architectural design. Various approaches represent architectural layouts as two-dimensional (2-D) geometries and multi-story buildings as the combination of multiple layers. Galle (1981) adopted a grid system for layout representation and proposed an approach to enumerate all arrangements of rooms within the grid. The exponential growth in the number of possible arrangements makes electron approach infeasible. A similar grid-based strategy can be found in Rosenman (1996), Lopes et al. (2010), and Peng et al. (2014). The methods adopted for layout generation are different in these approaches, which range from genetic algorithms (Rosenman, 1996) and procedural modeling (Lopes et al., 2010) to integer programming (Peng et al., 2014). The latter two have addressed the generation of multi-story building layouts, where vertical spaces, such as stair cases, were also involved. However, in Peng׳s approach, these spaces were manually positioned instead of optimized by computer. Another model widely used for layout representation is 2-D polygons. Michalek et al. (2002) represented rooms as 2-D rectangles and adopted mathematical optimization to decide the positions and the aspect ratios. Martin (2006) represented rooms as rectangles, and the positioning process in his approach was carried out by procedural modeling. Neither of these approaches addressed the multi-floor layout. Similar strategies can also be found in Goetschalckx and Irohara (2007), Afrazeh et al. (2010) and Ramtin et al. (2010). Doulgerakis (2007) generated internal layouts from the polygon that represents the boundary of the building. Rooms were generated by dividing the original polygon into small parts, and the division operations were encoded using genetic algorithms. Multi-floor layouts were addressed in this approach, but vertical spaces, such as a staircase and a two-floor-high living room, were not mentioned. Rodrigues et al. (2013a, 2013b, 2013c) and Merrell et al. (2010) also achieved the generation of multi-floor layouts. Staircases were addressed in the former, while other vertical spaces, such as two-floor-high living rooms, were addressed in the latter.