Performance Optimization of a Simplified Passive Design Space in Diversified Climatic Contexts

1.1. Research background

Building performance evaluation is a critical step to evaluate various design options for obtaining satisfying design outcome fulfilling multiple requirements. Building performance could be predicted during the design phase and verified in actual operation. Especially in the early design stage, the design could be optimized through various simulation packages such as TRNSYS, EnergyPlus and ESP-r. Design optimization tools are also available such as GenOpt, JEPlus, BeOpt and MultiOpt. Some of those tools are embedded in building performance simulators.

Several precedent studies attempted building design optimizations for best performance. Fig. 1. illustrates some studies that dealt with building design optimization classified into their number of objective function variables adopted for optimization. Two-variables optimization (mostly energy and cost) marks the highest frequency among 59 examined studies[1][2]. These are for either retrofitting existing buildings or early-design of new buildings.

Fig. 1.

Building design optimization researches with different number of objective function variables[1][2]

Hamed S.G., et al. (2018) investigated a multi-objective optimization model for four different climate regions in Iran to improve building design specifications. This study adopted two objective function variables for optimization, including total life-cycle cost (investment, NPV based maintenance and operation cost) and thermal energy loads (heating and cooling loads). Envelope assembly and HVAC system options are significant design variables in this research[3]. Raphael Wu, et al. (2017) also developed a multi-objective optimization scheme for building retrofit in Switzerland through which annual maintenance cost and life cycle GHG (Green House Gas) emission were evaluated[4]. Guo, et al. (2018), attempted residential building passive design optimization and suggested a framework for optimizing a building design considering both energy demand and thermal comfort at the same time[5]. Cho, et al. (2018) investigated apartment-building’s energy performance in relation with different locations and orientations in Korea. It turned out to be that the vertically middle-part apartment units consume the least heating energy[6]. Park, et. al. (2020) focused on window passive design in office buildings for three different climate conditions in Korea. Regarding this work, external shade and daylight control could save up to 6% of energy consumption[7]. As a particular case, Schito, et al. (2018) explored a museum design optimization based on energy demand, thermal comfort and artwork lifetime influenced by varying HVAC system options[8]. Kang, et al. (2020) analyzed passive design elements in a hospital by considering daylight factor and energy saving[9].

There are also several works targeting algorithm improvement to support multi-objective optimization such as OBEM, NSDA-II algorithm and ant colony optimization algorithm for mixed variables (ACOMV)[10–12].

Several key points were identified in relation to building design optimization including building types, design objectives variables and alternative evaluation algorithms. After reviewing precedent studies, the necessity of a more generalized multi-criteria design optimization framework for supporting various building types in different global locations caught our attention.

2.1. Adiabatic surface modeling

A typical office unit has at least single wall exposure to outside environment, therefore, no notable heat is transferred through other surfaces such as sidewalls, roof and floor. Since consecutive neighboring office units do not trigger temperature differences associated with internal walls, ceilings, and floors, heat transfer could be nullified through adopting adiabatic surfaces for modeling this condition. For the comparison purpose, both adiabatic and non-adiabatic surface configurations of the test space were evaluated under the same simulation conditions.

2.2. Performance assessment simulation and optimization

A series of performance evaluation simulations and passive design optimizations are performed using DesignBuilder v4.7.0.027. Optimization is carried out for minimization of occupant’s discomfort and life-cycle cost.

DesignBuilder only allows two objective variables, ten design input variables as well as user-input-based constraints that can be assigned to different variables for optimization. Building design optimization is carried out based on two stages. Initial optimization considers only passive design elements without HVAC and lighting system installations while the extended optimization includes HVAC and lighting systems. Table 1. and Table 2. show the design variables that are used in the initial and extended optimizations. Further details of walls, glazing and HVAC templates used for optimizations are described in Appendix Table A.1., Table A.2. and Table A.3.

Table 1.

Passive design variables for initial optimization

Table 2.

Selected design variables for extended optimization

This work also includes different climate zones based on Köppen climate classification[13]. Five different climatic regions being tested are shown in Table 3. Those regions represent diversified climate zones covering North-South Hemisphere and West-East Hemisphere.

Table 3.

Alternative test space locations with their climatic conditions[17]

2.3. Objective functions

Two objective variables for design optimization are selected for minimizing both discomfort hours (based on ASHRAE 55 Standard) and cost. As for the cost variable, in initial optimization, capital cost (Capex) is calculated, whereas, in extended optimization, life-cycle cost (LCC) is used. Explanations on each optimization objective variable is presented below based on DesignBuilder’s calculation condition and limitation.

3) Building design optimization algorithm

Genetic algorithm (GA) is an AI technique utilized in the DesignBuilder v4.7.0.027 building design optimization process. This GA is based on non-dominated sorting genetic algorithm II (NSGA-II) method that provides a fair tradeoff between a well-converged and a well-distributed solution set[18]. NSGA-II was developed by Kalyanmoy Deb, et al. (2002) and is believed to have better-spread solution and converges better for obtaining a non-dominated front as is graphically described in Fig. 3.[19].

Fig. 3.

NSGA-II Algorithm for optimization[19]

3.1. Adiabatic vs. non-adiabatic wall model

Comparison between adiabatic and non-adiabatic wall models is performed to verify adiabatic wall’s fitness to modeling heat-transfer neutralization effect for internals surfaces of an office unit. Table 4. illustrates the heat-transfer difference between adiabatic and non-adiabatic wall models under the same simulation settings (activity, construction, glazing, HVAC, and lighting system).

Table 4.

Comparison between adiabatic and non-adiabatic wall models

Notable differences are observed in the simulation outcomes. The roof area in adiabatic wall model is virtually zero since no heat transfer occurs through the roof surface in contrast with the non-adiabatic one which has 20.25m² effective roof area. Effective wall area which is subject to heat transfer also varies in non-adiabatic and adiabatic wall models. The adiabatic wall model’s total site energy turns out to be lower than the non-adiabatic wall model because solar gain for the adiabatic wall model is made only through one wall whereas the non-adiabatic wall model has all four walls which are subject to heat transfer. This is reflected as heat gain profiles with the wall and glazing materials in Table 4. Therefore, a single-room type space with adiabatic surfaces could represent an apartment or office unit in the middle of a multistory building in terms of heat transfer characteristics of the constituting surfaces.

3.2. Thermal comfort model

Fig. 4. to Fig. 8. illustrate the thermal comfort level for each simulated region which is represented by the Psychometric Chart (based on ASHRAE 55) drawn with the Climate Consultant 6.0 weather data processing method[20]. Red dots show uncomfortable weather conditions, whereas green dots indicate comfortable weather instances. Those dots are marked based on the hourly weather dataset for each region. Blue-parallelogram shape in Fig. 4. to Fig. 8. indicate comfort-zone criteria based on ASHRAE 55 calculation framework. Two assumptions are being made for each figure; the right side blue box represent summer clothing (0.5 clo) case while winter clothing (1.0 clo) case is illustrated by the left side one.

Fig. 4.

Psychrometric chart for Jakarta weather

Fig. 5.

Psychrometric chart for Cairo weather

Fig. 6.

Psychrometric chart for Buenos Aires weather

Fig. 7.

Psychrometric chart for Seoul weather

Fig. 8.

Psychrometric chart for Resolute weather

As are shown in the figures, weather conditions in Jakarta and Resolute do not meet the ASHRAE 55 comfort classification. Both Cairo and Buenos Aires show 17.6% and 12.9% of weather instances meet the comfort criteria, respectively. Less than that, Seoul shows only 5.9% of the weather instances meet the criteria.

Cooling Degree Days (CDD) and Heating Degree Days (HDD) are calculated in Table 3. As illustrated in the table, tested regions could be classified into cooling-dominated, mixed and heating-dominated areas. Jakarta, Cairo, and Buenos Aires would be classified as Cooling-dominated regions. Seoul is a mixed-region, whereas Resolute is a heating-dominated region.

3.3. Initial (passive) optimization outcomes

Two-objective variable based optimizations are performed for all tested regions targeting the minimization of both occupancy discomfort and capital cost. The outcomes are presented in Fig. 9. As for Jakarta, all tested passive strategies tested through optimization show the same result of discomfort hours (8,760 hours). This result indicates that any passive design strategy simulated by DesignBuilder optimization could not meet the ASHRAE 55 comfort criteria for Jakarta's weather condition. Further explanation would be discussed in the following section.

Fig. 9.

Initial optimization with discomfort and capital cost objective variables

3.4. Extended optimization outcomes

Another two-objective-variables optimization is performed for extended (passive + active) optimization including HVAC system installation for the single-room space unit design. The HVAC systems considered for the optimizations are listed in Appendix A, Table A.3. Lighting system is also included in this optimization phase with 3.3W/m²/100 lux power density regulated by a linear control system. In this extended optimization, the objective function is to minimize both occupancy-discomfort and cost (based on the LCC calculation). Extended optimization outcomes are presented in Fig. 10.

Fig. 10.

Extended optimization with discomfort and LCC objective variables

4.1. Construction material selection

Since the least capital cost strategy shows only a slight difference, the least discomfort strategy in the initial optimization process catches our attention (Table 5.). There are different optimum material characteristics for the cooling-dominated regions (Cairo and Buenos Aires). Cairo case has an advantage over higher R-value compared to Buenos Aires case. With higher R-value, heat transfer between outdoor and indoor space gets slower. Buenos Aires prefers lower R-value and higher U-value, indicating non-insulated walls in Buenos Aires could perform better than insulated walls. For Seoul and Resolute, both regions call for the same material characteristics with 2.8 R-value. Corresponding materials use the same type of insulation, namely XPS Extruded Polystyrene with approximately 79 mm - 117.5 mm thickness. In the heating-dominated regions, insulation is critical to maintaining comfortable indoor temperature. Detailed construction based thermo-physical characteristics are made available in Appendix A, Table A.2.

Extended optimization results are shown in Table 6. Here we find that the recommended materials for each city are diversified in the initial (passive) optimizations. For most tested cities, ‘Best Practice (Lightweight or Heavyweight)’ material is recommended for comfort and LCC related strategies. This materials type is quite different from the recommendation suggested in initial optimization. ‘Best Practice’ material includes XPS Extruded Polystyrene as the insulation with 117.5 mm (for Heavyweight) and 121.9 mm (for Lightweight) thickness. R-Value for this material is quite high (3.979). Resolute case gets the same type of material recommendation such as ‘NCM 2010 Notional Building (Metal Cladding) with 150.0 mm insulation having 3.846 R-Value. This insulation material would fit in with the region showing higher HDD (Heating Degree Days).

Nevertheless, in this optimization we introduce HVAC system to maintain comfortable indoor temperature. As for HVAC system installation, walls need to maintain stable indoor thermal condition without losing much heat from indoor to outdoor space. For this reason, insulated wall is better choice than non-insulated wall especially for higher CDD (Cooling Degree Days) characterized regions.

4.2. Glazing material selection

As for glazing material, Table 5. shows the preferred material selection for each region. Evaluated thermo-physical characteristics are SHGC (Solar Heat Gain Coefficient) and U-value of the windows. SHGC represents the fraction of incident solar radiation admitted through a window including directly transmitted, absorbed and subsequently released-inward solar components. Detailed glazing thermo-physical characteristics are available in Appendix A, Table A.1.

As are shown in Table 5. and Table A.1., SHGC values gradually increase as HDD (heating degree-days) increases (or CDD decreases) for each region. Cairo has the lowest SHGC and the lowest HDD. On the other hand, Resolute has highest SHGC and highest HDD. Considering these findings, the SHGC value of a material is positively correlated to HDD and yet shows negative correlation to CDD.

Extended optimization results are presented in Table 6. From here, we also figure out that the recommended glazing for each region is different from the initial optimizations. For example, ‘Project Glazing Template’ recommended through extended optimization has 2-layer glass with 1.978 U-value and 0.687 solar transmission value.

4.3. HVAC system selection

The optimum design alternatives identified through extended optimizations are presented in Table 6. As clearly shown from the result, the optimum HVAC template is different for each region and each strategy. However, it turns out that the VAV (Variable Air Volume) template is the most favorable choice for several tested strategies in the target regions. For the ‘least discomfort strategy’, the cooling set-point is around 24-25℃ and the heating set-point is around 21℃. Seoul and Resolute cases call for a lower heating set-point between 19℃ and 17℃ while minimizing the cost. The heating setback temperature is still relatively high in Resolute case and that is 16℃. These optimization results also consider the operational cost while maintaining minimized discomfort hours.

4.4. Special case: Jakarta vs. Resolute

Jakarta and Resolute are the two most distinctive regions that show extreme climate variations. Jakarta is characterized by whole day cooling, whereas Resolute calls for whole day heating. In this study, discomfort level differences could not be identified across several attempted strategies. Climate Consultant 6.0 tool [20] is used to visualize the different weather conditions in both cities.

Psychrometric chart and passive design strategies for Jakarta are generated by Climate Consultant 6.0 in Fig. 4. Even with the adopted passive design strategies, any of them still could not satisfy the comfort criteria suggested by ASHARE 55. In further observation, passive design strategies give significant indoor comfort improvement in lower humidity ratio climate conditions. On the other hand, Jakarta climate condition has a high humidity ratio (refer to Fig. 4.).

If HVAC system is introduced to the test space, the green dots portion gets 100% indicating that 100% comfort level is acquired (in Fig. 11.). Therefore, for tropical climatic regions, passive design strategies could help but they can not significantly shift the condition to meet comfort criteria. For this reason, HVAC system is indispensable in such region like Jakarta where the weather is characterized by higher temperature and higher humidity ratio.

Fig. 11.

Psychrometric chart and HVAC design strategies in Jakarta

As for Resolute case, the general weather condition shows very low temperature (Fig. 12.). However, if passive design strategies are properly implemented, there could be at least 5% of annual hours shifting into the comfort zone. Such passive design option is ‘passive solar direct gain’and that is why glazing selection becomes essential. For instance, Resolute case prefers the highest SHGC (Table 5. and Table A.1.). Several passive strategies could be implemented and tested based on the actual weather conditions (temperature and humidity ratio) considering these strategic differences.

Fig. 12.

Psychrometric chart and passive design strategies for Resolute