An integrated energy performance-driven generative design methodology to foster modular lightweight steel framed dwellings in hot climates

This paper presents a study on the application of lightweight steel framed (LSF) construction systems in hot climate. A generative design method created 6010 houses, with random geometry and random roof and exterior wall types with different insulation levels, and EnergyPlus was used to evaluate the energy consumption for air-conditioning of each building. The main goals were to determine which geometric variables correlate with the energy performance, and to provide some guidelines to foster efficient LSF buildings in hot climates. By correlating six geometry-based indexes with the energy consumption for each construction element type group, it was verified that roofs do not show significant correlation, while exterior walls presented weak to moderate positive correlation with the building volume, very weak to weak negative correlation with the relative compactness, no correlation with the shape coefficient, moderate to strong negative correlation with the window-to-floor, window-to-wall, and window-to-exterior surface ratios. The results also show that buildings with larger windows and greater level of insulation have better energy performance. No significant difference of energy performance was found between different LSF construction systems with equivalent thermal resistance.


Introduction
1 Lightweight steel framed (LSF) buildings have a widespread use in the USA, Australia and 2 Japan and they are gaining market in Europe (Veljkovic & Johansson, 2006). Indeed, the popularity systems, typically employed to guarantee indoor thermal comfort. Therefore, PCMs will be out gram of the Ministry of Electricity and Water (MEW, 2010) establishes several requirements to 126 improve the energy performance of buildings (including insulation, glazing, lighting and ventilation 127 requirements) and to reduce power ratings of air-conditioning systems. 128 Based on a TRNSYS-IISIBAT environment DSEB parametric study, Al-ajmi & Hanby (2008) 129 proposed several features that should be adopted in hot climate conditions to achieve more energy 130 efficient residential buildings, such as: the control of the window area and the "north-south di-131 rection" placement of the main windowed facades, the use of treated glazing to reduce solar heat cooling demand and peak-loads, and they have concluded that a 4 cm thick PCM-wallboard with 142 a melting-peak temperature of 24°C yielded the lowest annual cooling demand (annual cooling 143 energy savings of 4-5%) across a variety of room orientation and window-to-wall ratio (W W R), as-144 suming a cooling-setpoint of 24°C. Moreover, they concluded that cooling demand and peak-loads 145 can be reduced by 5-7% during summer months. 146 In all the references listed above, only heavyweight constructions were evaluated in the studies, 147 and no information about the behavior of lightweight residential buildings in hot arid climate con-148 ditions was found in the literature. Therefore, to complement the previous works, this manuscript 149 explores the thermal performance of LSF low-rise air-conditioned residential buildings in Kuwait. 150 As far as the authors know, this paper is the first study devoted to such analysis. 151 2.2. Construction system 152 In this paper, the "LSF System B(A) a " will be used. It is available on the market (urb, 2017) 153 and it was developed by Balthazar Aroso Arquitectos Lda. (bal, 2017). The main particularity of 154 this LSF system is that a single cold-formed shape profile (C100 x 45 x 1.2 mm) is used for all the 155 steel framing elements, which makes the construction more rational. 156 Regarding thermal behavior, LSF construction elements are typically classified according to the 157 location of the thermal insulation layers as cold-framed, hybrid, and warm-framed construction 158 (Fig. 2). In cold-framed construction, the thermal insulation is placed inside the wall between 159 the steel studs; in hybrid construction, the thermal insulation is distributed between the external 160 surface and the wall gap between steel studs; and finally, in warm-framed construction, all thermal 161 insulation is placed outside the steel framing on the external surface.

162
In order to evaluate the thermal performance of these different LSF construction systems in 163 hot arid conditions, and to assess the best level of insulation, several exterior wall design solutions 164 are considered in the DSEB runs. This is done by varying the thicknesses of both the thermal insulation within the steel framing, th ins1 , and the thermal insulation placed from the exterior, 166 th ins2 (Fig. 2). th ins1 can be assigned one of the 11 predefined values th ins1 = {0, 1, 2, · · · , 10} cm 167 and th ins2 can be equal to any of {0, 1, 2, · · · , 5} cm. Regarding the roof system, the thickness 168 of the XPS layer can vary within the range th ins3 = {0, 1, 2, · · · , 10} cm ( Fig. 2). Therefore, a 169 set of 66 predefined discrete exterior walls (11 cold-framed, 5 warm-framed and 50 hybrid walls) 170 and 11 roof solutions can be considered in the simulations, which means that 726 combinations of 171 different exterior walls and roofs are possible. Fig. 3 shows a sketch of the main components of an 172 LSF building. Fig. 4 also shows the cross-section of some construction elements considered in the 173 model. Table 1 lists the thermophysical properties of the materials considered in this study.

174
The non-homogenous layers and the effect of thermal bridges (originated by the steel framing) 175 are considered in the DSEB according to the methodology described in Soares et al. (2014). Fol-176 lowing this methodology, a fictitious equivalent material is defined to replace the heterogeneous 177 layers; for instance, the space between steel frames filled with insulation. As a result, the thermal 178 conductivity of the equivalent material is adjusted so that the effective thermal resistance of the 179 equivalent layer is equal to that of the heterogeneous layer. The density and the specific heat of 180 the equivalent material are also adjusted to match the thermal capacity of the heterogeneous layer 181 as proposed by Soares et al. (2014).

182
In addition, U -values are obtained by varying the thickness of the thermal insulation layers as 183 explained above. The simplified method proposed by Gorgolewski (2007) and Doran & Gorgolewski  and hybrid walls as explained by Gorgolewski (2007). Generally speaking, the method involves 187 the calculation of the upper and lower limits of the thermal resistance of the LSF elements, R max 188 and R min respectively. The conductances associated to R max and R min are then calculated on 189 an area-weighted basis. For the walls, the stud and nogging spacing is equal to 625 mm. The (1) The value of p is equal to 0.5 for warm-framed construction. In cold-framed and hybrid con-  ignored, assuming that they together amount to less than 3% of 1/R T , as prescribed by Doran & 213 Gorgolewski (2002). Table 2 summarizes (as an example) the U -value of some LSF elements.
214 Table 2. U -value of some LSF construction elements (materials listed along the cross-section area away from the steel element).

Material
Thickness  The specified spaces/rooms requirements are summarized in Table 3. For each space, there are 239 exterior openings, which are listed and detailed in Table 4. These rooms were grouped into clusters 240 according to Table 5. The interior openings and rooms relations are presented in Table 6. The 241 listed functioning architectural program will be used in the generative design study.

242
To evaluate the energy performance of the villa model in an urban context, the Al-Qadisiyah 243 residential area in Kuwait City was selected as case study, as shown in  Table 3. Rooms geometry and topologic specifications.
Living room 2.0 C sn -name, C sf -function, C ri -relative importance, C sl and C su -served lower and upper stories, C ss -minimum space side, C ssr and C slr -space small side and large side ratios shading objects, thus influencing the energy performance of the building under investigation. This

257
The envelope of the building shall be made to prevent air infiltration. Positive pressure must 258 be maintained inside the building by the air-handling system to minimize air and dust infiltration.  Table 4. Exterior openings geometry and topologic specifications. Oe 5 Window -1.00 1.00 1.00 Oe 24 Window -0.50 0.50 1.50 C os -space, C oet -opening type, C oeo -orientation, C oew -minimum width, C oeh -minimum height, C oev -vertical position Table 6. Interior openings geometry and topologic specifications.

Interior Openings
Opening C oit C oia C oib C oiw (m) C oih (m) C oiv (m) Door S 27 S 30 1.00 2.00 0 C oit -type, C oia -opening's space, C oib -destination space, C oiw -minimum width, C oih -minimum height, C oiv -vertical position Exhaust 25 a a -intermittent matter, half of the air leakage maximum legal limit for swinging doors is assumed (1.3 L·s −1 ·m −2 ), 276 as the doors are not permanently opened and these zones are also pressurized.

277
The characterization of the occupancy patterns, the operation schedules of appliances, light-  Table 8. 285 The requirements from the Kuwaiti energy conservation code (MEW, 2010) are also considered     (Fig. 8a) and high outdoor temperature (Fig. 8b)  for all windowed zones, the window shadings (exterior PVC roller shutters) are considered to 296 permanently cover the windows during the high outdoor temperature period, and to only cover 297 them at night-time during the low outdoor temperature period. For the remaining zones, single 298 yearly schedules are considered (Fig. 8c), independently of the dual window shading profile, as 299 their lighting profiles can be considered constant throughout the year, due to these zones typology 300 and occupancy.

301
The internal heat gains due to electric equipment are defined by the maximum design wattage and occupancy, and are depicted in Fig. 9 for the different zones. Schedules for bathrooms and 306 servant bedrooms are not presented since they correspond to short usage periods. Additionally, 307 a 2230 W gas oven is also considered to contribute to the kitchen's internal heat gains (radiant 308 fraction of 0.07, convection fraction of 0.93). The oven usage schedule corresponds to the kitchen's 309 exhaust ventilation schedule (see Fig. 6).

310
The villa is air-conditioned considering an ideal loads air system model in the EnergyPlus runs, 311 which allows to assess the performance of the building without modelling a full HVAC system,    in the cooler months. A 50% dehumidification setpoint is also considered during the cooling season.

315
The heating season -when heating is available -was defined for the period between 1 November and 316 31 March, when the average daily temperature is permanently, or, at least, for long periods of time, 317 below the heating setpoint. On the other hand, the cooling season-when cooling is available-was 318 defined for the period between 1 March and 30 November, when the average daily temperature is 319 permanently, or, at least, for long periods, above the cooling setpoint (Kuwait air temperatures  Fig. 6). Moreover, due to the high electric equipment heat gains in the kitchen 325 and laundry, there is only cooling available in these zones. 326 Fig. 9. Electric equipment schedules in each zone type.  ume, C f -shape coefficient, RC -relative compactness, W F R -window-to-floor ratio, W W R -376 window-to-wall ratio, and W SR -window-to-exterior surface ratio).   In graphs a) and b), the grey background corresponds to roof elements, white to hybrid construction, yellow to warm-framed construction, and blue to cold-framed construction. In graph a), the arithmetic mean of all buildings performance is marked as a vertical red line. In graph b), maximum U -value for roofs and walls defined by the Kuwaiti building code for light construction with medium light external color are marked as vertical red lines (MEW, 2010). In graph c) blue color corresponds to positive and red to negative correlation with E. In graph d), only the results with p-value above or equal to 0.01 are illustrated. In graphs c) and d) the geometry-based indexes are V -volume, C f -shape coefficient, RC -relative compactness, W F R -window-to-floor ratio, W W R -window-to-wall ratio, and W SR -window-to-exterior surface ratio.
It is observable that the energy consumption mean average of each subgroup (black dot) follows 378 the corresponding element U -value (black diamond). It is also noticeable, especially in the cold and 379 warm-framed wall types, that the range of each subgroup diminishes as the U -value also decreases, 380 thus indicating a decreasing influence of the geometry variables. When comparing different exterior 381 wall types with equivalent U -values, such as EW41 (hybrid wall) and EW55 (warm-framed wall), 382 or EW53 (warm-framed wall) and EW60 (cold-framed wall), the performance range is similar thus  Considering in Fig.13c