被动式设计
- 格式:pdf
- 大小:1.17 MB
- 文档页数:12
Optimization of passive design measures for residential buildings in different Chinese areasXinzhi Gong *,Yasunori Akashi,Daisuke SumiyoshiDepartment of Architecture,Graduate School of Human-Environment Studies,Kyushu University,Hakozaki 6-10-1,Higashi-ku,Fukuoka 812-8581,Japana r t i c l e i n f oArticle history:Received 11April 2012Received in revised form 20June 2012Accepted 20June 2012Keywords:Building thermal load performance Optimization Passive designOrthogonal method Listing methoda b s t r a c tPassive design,which has strong dependency of climate and areas,is the most economical effective strategy for reducing energy consumption inside residential buildings.A total of 25representative cities of China were selected for optimum analysis.This paper presents an approach in which the orthogonal method and the listing method are integrated to explore how energy consumption is minimized in residential buildings by optimizing 7passive design measures for each city:wall thickness (WT),roof insulation thickness (RIT),external wall insulation thickness (EWIT),window orientation (WO),window-wall ratio (WWR),glazing type (GT),and sunroom depth/overhang depth (SD/OD).With the optimiza-tion,the passive design could reduce annual thermal load of building considerably,even could replace air-condition systems in winter for the areas with high solar radiation such as Lhasa.The results reveal that neighbor cities have the same optimal combinations and the similar effects of energy-saving measures.A total of 7passive design zones were summarized in a map.The architect could be guided in his choice of appropriate passive measures directly from the map or handling of exact passive design by the optimal approach proposed in this paper for different areas in China.Ó2012Elsevier Ltd.All rights reserved.1.IntroductionWith the rapid development of economy and the increase of people ’s living standard,a signi ficant portion of the energy is consumed by today ’s residential buildings.In China,residential building sector consumed 192.7million tons of standard coal in 2003,which accounted for 11.3%of the total primary energy used in the nation [1]and continuously rise to 17.2%in 2006.Every year,approximately 34.1%of total energy in building sector is used for household space heating [2].These facts emphasize that residential building sector has become a key energy end user.In order to alleviate the growing building energy demand,it is widely accepted that passive design is the most economical effec-tive strategy to reduce thermal load of residential buildings [3e 6].It was reported by measurements that passive design buildings can save more than 50%of total primary energy consumption [7].A good passive design with the objective of minimizing the energy demand for comfort involves all the various aspects of building design [8]like orientation of the main façades and windows,wall thickness,insulation levels,window details,sunroom for passive solar heating,shading,etc.[9].In existing studies on passive design measures for residential thermal load conservation in China,Liu and Yang [10]presented a new-style passive solar house appropriate to western Chinese climate.Passive design measures of sunroom and suf ficient external wall thickness ensure that this style of houses have zero energy consumption and high indoor comfort [11].Liu et al.[4]researched further passive design strategies such as solar heating,natural ventilation,thermal mass with/without night ventilation and evaporative cooling for 18representative cities of China.According to their study,a total of 9passive design zones were identi fied by bioclimatic approach,which pay more attention to indoor comfort rather than thermal load ter,Zhang had analyzed the regional characteristics of heating load of detached house by simulating four different types of passive design integrated with insulated envelopes for 46of Chinese cities [12].He suggested that the larger the latitude is,the thicker the insulation thickness should be utilized in Eastern China.Yoshino et al.[13,14]had investigated the residential energy consumption in 9of Chinese cities and revealed its in fluence factors.His team also claimed the importance of passive measures including insulation thickness,window types,and airtightness for residential buildings in Beijing and Shanghai [15].However,these researches have not optimized passive design measures for residential buildings,which may neglect the potentially important interactive effects between parameters.*Corresponding author.Tel.:þ81926423341;fax:þ81926423368.E-mail address:gongxinzhi@ (X.Gong).Contents lists available at SciVerse ScienceDirectBuilding and Environmentjournal homepage:w /locate/buildenv0360-1323/$e see front matter Ó2012Elsevier Ltd.All rights reserved./10.1016/j.buildenv.2012.06.014Building and Environment 58(2012)46e 57Therefore,in order to identify the exactly optimal combination of passive design measures andfinally determine the different passive design zones in China,this paper proposes an approach in which the orthogonal method[16]and the listing method[17]are integrated to select the optimal combination based on thermal load simulation.This research summarizes7passive design measures for each of25cities for investigation,outlines the thermal load simulation system for a simple box model,and interprets the methodology of optimization through an example of a city.Then it analyzes the results of the optimal combination of passive design measures,andfinally,determines the passive design zones where the cities have the same optimal combination result.This paper will provide the architects with an overall selection of appropriate strategies for thermal load conservation of passive design zones during the initial design stage and further guidance for optimum analysis for various Chinese areas.2.Overview of simulation case study2.1.Climates and selection of targeted citiesIn China,a major classification of climatic regions is related to architecture and thermal design of buildings,which is defined as the thermal climatic regions by the Chinese government[18].In this classification,China is divided intofive thermal climatic regions:the severe cold region,the cold region,the mild region,the hot summer&cold winter region,and the hot summer&warm winter region.The thermal climatic regions are determined based on the criteria of average temperatures in the coldest and the hottest months of the year,and complementarily classified by the numbers of days that daily average temperature is below5 C or above25 C[19].Apart from the criteria of the thermal climatic regions,the passive design zones proposed in this investigation are classified on the criteria of the optimal combination of passive design measures based on thermal load simulation.It is strongly depending on the local climatic conditions.Thus,the selected cities should cover the five thermal climatic regions with uniformly geographical distri-butions,and should have local weather data for simulation.A total of25cities;ten are in the severe cold region,six are in the cold region,two are in the mild region,five are in the hot summer&cold winter region,and two are in the hot summer&warm winter region;were selected for this study.Fig.1shows an overall layout of the target cities within thefive thermal climatic regions in China. Table1shows a summary of relevant information on the25cities.2.2.Control parameters and levels of passive designBased on the previous studies,this research has chosen the next passive design measures which have a great influence on building thermal load:wall thickness(WT),roof insulation thickness(RIT), external wall insulation thickness(EWIT),window orientation (WO),window-wall ratio(WWR),glazing type(GT),and sunroom depth/overhang depth(SD/OD).Other than the previous parame-ters,the parameter OD is applied only for Guangzhou and Haikou, while the parameter SD is discussed for the other23cities.Hence, a total of7parameters are optimized for each city.Four levels are selected for each parameter because of the nonlinearity interest:the highest,the middle1,the middle2and the lowest,respectively,as shown in Table2.The region of interest can be sufficiently covered using these four levels.Several Chinese cities are colder than Japan;however,the envelope design of Japanese energy-saving standard is much better than China.In order to propose optimum passive design measures for China,the Fig.1.Geographical layout of the25selected cities and thermal climatic regions across China.X.Gong et al./Building and Environment58(2012)46e5747highest values for the parameters WT,RIT,and EWIT refer to Japanese energy-saving houses [20].The middle 1and 2values of the three parameters refer to energy codes of China [21,22],while their lowest values are those without any energy-saving methods.For the parameters WO,WWR,GT,and SD/OD,the four different conditions are discussed.It is relatively simple to add new parameters and levels,although the optimization complexity increases with the number of simulations.2.3.Simulation model and system descriptionA simple box model with a window,which has floor area of 16m 2(4m Â4m)and floor-to-floor height of 3m,is used to simulate the indoor thermal conditions.The basic envelope mate-rial of the box model is concrete.The annual thermal load of the model is calculated by the dynamic simulation program called THERB [23].It can estimate heating/cool-ing loads,indoor temperatures,and humidity for the whole building,taking into consideration the complete heat,air,and moisture (HAM)features.THERB is one of the of ficial software approved by the Japa-nese government and is applied nationwide in Japan.The 8760hourly records of the typical year weather data for the 25selected cities developed by Zhang [24,25]is used for simula-tion.The total internal heat sources (human body,electric appli-ances,etc.)are assumed equal to zero.Air conditioning system (COP ¼3.1)is the only energy consumer in the model,which controls indoor temperature above 18 C in winter and below 26 C in summer for a whole year on a 24h basis.Indoor relative humidity is controlled at the high peak 60%and the low peak 40%.Ventilation change rate is 0.5h À1.3.Methodology of optimizing passive design through an example of Hailar cityCurrently,genetic algorithm (GA)[26]and orthogonal method are two major optimal methods.Analysis of the GA can be very close to the best result;however,it needs advanced knowledge of mathematics and computer programming because the program of GA has to be combined with simulation software for optimal design,which is dif ficult for architects.Notably,orthogonal method,which can determine the preferable level and analyze signi ficance of parameters and their interactions by a straightfor-ward process without the need of computer programming,is suitable for this work.We only need to know the mathematic knowledge of orthogonal array for data analysis.In order to optimize the passive design by manageable number of simulation cases,this paper utilized the orthogonal method first to evaluate the signi ficance of 7parameters and their interactions,and then identify the preferable level (standard level)between the highest and the lowest levels listed in Table 2,of each parameter.Identi fication of the standard level aims to reduce the possible interactive effects on thermal load for calculation in listing method.At the second step,the listing method [17],which studies one parameter at a time while keeps other parameters fixed as the standard level,is used to further analyze the optimal level for the each parameter with a total of 4levels,and then,determine the optimal combination of passive design measures for every city.TheTable 1Summary of general information on the 25target cities.CityLatitude Longitude Altitude HDD18CDD26Annual mean temperature Thermal climatic region( N)( E)(m)( C $d)( C $d)( C)Hailar 49.22119.75611671331Severe cold Harbin 45.75126.77143503214 4.2Severe cold Altay 47.7388.0873******* 4.6Severe cold Zhangye 38.93100.431483400167.2Severe cold Gangca 37.33100.133********À0.3Severe cold Shenyang 41.77123.43453929258.5Severe cold Karamay 45.684.8542842341968.6Severe cold Urumqi 43.887.659184329367.1Severe cold Xining 36.62101.77229644780 6.1Severe cold Golmud 36.4294.9280944360 5.3Severe cold Hotan 37.1279.93137525957112.5Cold Beijing 39.93116.283126999412.3Cold Xi ’an 34.3108.93398217815313.7Cold Lanzhou 36.05103.8815183094109.8Cold Lhasa 29.6791.133649342508.3Cold Barkam 31.9102.232666339008.6Cold Xichang 27.9102.271599983616.9Mild Kunming 25.02102.6818921103014.9MildShanghai 31.4121.473154019916.2Hot summer &cold winter Wuhan 30.62114.1323150128316.7Hot summer &cold winter Changsha 28.23112.8768146623017Hot summer &cold winter Pingwu 32.42104.5289417101214.7Hot summer &cold winter Chengdu 30.67104.0250613445616.2Hot summer &cold winter Guangzhou 23.17113.334237331322.1Hot summer &warm winter Haikou20.03110.35147542724.1Hot summer&warm winterTable 2Control parameters and their values for optimization of passive design.Parameter aValues of each level for each parameter The highestMiddle 1Middle 2The lowest A:WT 250200150100B:RIT 3002001000C:EWIT 200140700D:WO South North East West E:WWR 0.150.30.450.6F:GTDouble low-E Triple Double Single G:SD/OD1200/1800600/1200300/6000/0a7parameters are identi fied as follows.WT:Wall thickness,(mm);RIT:Roof insulation thickness,(mm);EWIT:External wall insulation thickness,(mm);WO:Window orientation;WWR:Window-wall ratio;GT:Glazing type;and SD/OD:Sunroom depth/Overhang depth,(mm).X.Gong et al./Building and Environment 58(2012)46e 5748process of using the orthogonal method and the listing method for optimization has been presented in the flowchart in Fig.2.In the following Sections 3.1and 3.2,the analysis of a single city Hailar is assisted to show how the optimization of passive design measures is achieved.3.1.Identi fication of the signi ficance and standard level of each parameter by orthogonal methodTaguchi and Konishi [27]tabulated many standard orthogonal arrays for using orthogonal method.One of these arrays can be used directly as planning matrix simulation cases.An appropriate stan-dard orthogonal array for a case study is decided by the degree of freedom required.Number of rows of a candidate array must be at least equal to the required degrees of freedom and has the smallest number of simulations.In this step,a total of 22variables,which are 7parameters with the highest and the lowest level (listed in Table 2)and 15potential interactions among WT and EWIT (A ÂC),RIT and EWIT (B ÂC),EWIT and WO (C ÂD),EWIT and WWR (C ÂE),EWIT and GT (C ÂF),EWIT and SD/OD (C ÂG),WT and WWR (A ÂE),RIT and WWR (B ÂE),WO and WWR (D ÂE),WWR and GT (E ÂF),WWR and SD/OD (E ÂG),WT and SD/OD (A ÂG),RIT and SD/OD (B ÂG),WO and SD/OD (D ÂG),GT and SD/OD (F ÂG),are to be estimated.The degree of freedom for the 2-level param-eter and the interaction between two 2-level parameters is 1.Hence,the required freedom is 1Â7þ1Â15¼22,and the standard orthogonal array L 32(231),which has 32rows corresponding to 32simulation cases and 312-level columns cor-responding to arrangement of variables,becomes the candidatearray.Fig.2.Flowchart of the methodology integrated with the orthogonal method and the listing method for optimizing passive design.X.Gong et al./Building and Environment 58(2012)46e 5749Table3presents the selected orthogonal array L32(231),with the assignment of7control parameters and15interactions.Number1 or2of7parameters of the matrix in Table3represents the highest level or the lowest(listed in Table2)used in a simulation case.The annual thermal load per unit offloor square of the model is simulated by complying with this orthogonal array and is placed in the last column of Table3.For instance,referring to the columns1, 2,4,8,15,16,and19,case1uses each level of the7parameters at the highest for simulation,which consumed57kWh/(m2yr)for space heating and cooling for Hailar.Similarly,case4is conducted at the highest level for WT,RIT,and EWIT,and the lowest level for SD,WWR,GT,and WO,which consumed445kWh/(m2yr).The columns5,6,7,9,10,11,12,13,14,20,23,24,27,28,and31present the15interactions.For instance,the column5indicates the inter-actions between parameters WT and EWIT(AÂC).The other9 empty columns are identified by errors.Number1or2of interac-tions and errors of the matrix is just for calculation of their significance[16].Notably,all the two levels of each parameter are equally rep-resented in the32simulations.Hence,the mean square value, which is corresponding to the variance of two levels,can reflect the significance of parameters and interactions.Furthermore,the preferable level(standard level)can be chosen between the highest and the lowest levels,which achieves the lower value of average annual thermal load.The significance of each parameter and interaction should be analyzed before indicating the standard level,because the signifi-cant parameters should have the higher priority to be considered. Significance and contribution of each parameter and interaction are calculated by the Analysis Of Variance(ANOVA)based on the orthogonal method.The significance of variables is assessed by the variance ratio and the degree of freedom for the error.The higher the variance ratio increases,the higher the significance of param-eter is.For this orthogonal array in which the degree of freedom for the error is31e1Â7e1Â15¼9,the variance ratio above10.6 indicates99%confidence level of the parameter significance and that above5.1indicates95%.Table4presents that the total7 parameters and5interactions of AÂC,BÂG,CÂE,CÂD and EÂF are very significant for annual thermal load for Hailar.The EWIT and SD,which make the top two contributions of the total sum of square on thermal load at75.4%and10.2%respectively,are controlling parameters.The interactions of BÂG and AÂC cause large contributions at3.0%and1.9%,which should be highly valued in design.For an independent2-level parameter,the standard level is indicated to the lower average thermal load.However,if the interaction between two parameters is very significant,the combination of their levels corresponding to the lowest average thermal load may be changed.For the above reason,the average thermal load for thefive significant interactions(AÂC,BÂG,CÂE, CÂD and EÂF)should be considered to confirm the standard level for Hailar city.Average thermal load for a level of one parameter is calculated by the level’s arrangement of the matrix in the orthog-onal array(shown in Table3).It is equal to the mean thermal load of the cases in which the parameter is selected as the level. For instance,the average thermal load for the highest level of EWIT is(57þ170þ130þ445þ82þ240þ512þ566þ84þ242þ167þ230þ60þ234þ468þ765)/16¼278,as shown in Table5(a).Similarly for calculation on the interaction between WTTable4The ANOVA of annual thermal load data for Hailar.Variables Degree offreedom Sum ofsquareMeansquareVarianceratioContributionf S S/f F r(%)A11208401208402272 B1193418193418363 3.3 C144558274455827836875.4 D134******** E173667366140.1 F11790191790193363 G1604809604809113610.2 AÂC1109934109934206 1.9 BÂC131******** GÂE115315300 AÂG117817800 BÂG11783011783013353 CÂE159505950110.1 CÂG193393320 BÂE137637610 AÂE182182120 CÂF12319231940 CÂD11131311313210.2 GÂF154254210 DÂG125925900 EÂD144344310 EÂF12865128651540.5 Errors94791532e0.3Table5Summary of average thermal load for parameters and significant interactions cor-responding to each level for Hailar.(a).Average thermal load for7parametersParameter a The highest level The lowest levelA:WT590713B:RIT574729C:EWIT2781025D:WO641662E:WWR636667F:GT577726G:SD/OD514789(b)-1.Average thermal load for the interaction between WT and EWITLevel of A:WTThe highest level The lowest levelLevel of C:EWIT The highest level275281The lowest level9051145(b)-2.Average thermal load for the interaction between RIT and SDLevel of B:RITThe highest level The lowest levelLevel of G:SD The highest level511517The lowest level637941(b)-3.Average thermal load for the interaction between EWIT and WOLevel of C:EWITThe highest level The lowest levelLevel of D:WO The highest level2491033The lowest level3071016(b)-4.Average thermal load for the interaction between EWIT and WWRLevel of C:EWITThe highest level The lowest levelLevel of E:WWR The highest level2501023The lowest level3071026(b)-5.Average thermal load for the interaction between WWR and GTLevel of E:WWRThe highest level The lowest levelLevel of F:GT The highest level562591The lowest level681771The bold italic identifies the standard level for each parameter.a7parameters are identified as follows.WT:Wall thickness,(mm);RIT:Roof insulation thickness,(mm);EWIT:External wall insulation thickness,(mm);WO: Window orientation;WWR:Window-wall ratio;GT:Glazing type;and SD/OD: Sunroom depth/Overhang depth,(mm).X.Gong et al./Building and Environment58(2012)46e5751and EWIT,the average thermal load for the combination of their highest levels is (57þ170þ130þ445þ82þ240þ512þ566)/8¼275,as illustrated in Table 5((b)-1).Based on the data analysis,the average thermal load for the highest level is lower than that for the lowest level,which is further demonstrated in the five signi fi-cant interactions,as highlighted in Table 5.Hence,we can conclude that the highest level of each parameter is the standard for Hailar.3.2.Identi fication of the optimal level of passive measures and their optimal combination by listing methodOnce the standard level of each parameter is decided,the listing method is utilized for further optimization for the total 4levels(listed in Table 2).The additional 22simulations are planned by the listing method,which changes the values of 4levels of one parameter while the value of other parameters does not change as the standard level,as shown in Table 6.Table 6also displays the level assigned for each parameter.Number 1represents the standard level which is determined by the orthogonal array in the Section 3.1;numbers 2and 3represent the values of the middle 2and 3listed in Table 2respectively,number 4represents the remaining level for each parameter at Table 2.The optimal level of each parameter is representing the lowest result in those of the four levels,which can comprise the optimal combination of minimum annual thermal load.According to the results of Table 6,values of the optimal levels for the parameters WT,RIT,EWIT,WO,WWR,GT and SD are determined respectively as 250mm,300mm,200mm,South,0.15,Double low-E glazing and 1200mm.The optimal combination consumes 50.5kWh/(m 2yr)by simulation for Hailar.Based on the explanation how to minimize building thermal load mentioned above,the following Section 4shows the opti-mized results by the orthogonal method and the listing method for total of 25selected cities.4.Results and discussion of optimization and passive design zones for 25selected cities4.1.Signi ficance and standard level of each parameter by orthogonal methodThe annual thermal load for 32of seven 2-level parameters combinations designed by orthogonal array was simulated using THERB for each targeted city.Signi ficance and contribution of 7parameters and their 15interactions were calculated.Table 7shows the variance of the annual thermal load,which is mainly in fluenced by roof insulation thickness,external wall insulation thickness,and window-wall ratio for Guangzhou and Haikou,with a large contribution of more than 15.7%,50.2%and 15.0%,respectively.For the other cities,the most signi ficant parameter of external wall insulation thickness contributes around 70%of the total square sum,and the second signi ficant parameter of sunroom depthTable 6Parameters assignment of simulation cases by listing method and annual thermal load for Hailar.Case no.Level assigned for each parameter Thermal loadkWh/(m 2yr)A:WT B:RIT C:EWIT D:WO E:WWR F:GT G:SD/OD1111111157.32211111157.93311111158.54411111159.151********.46131111157.67141111157.88112111178.891131111140.5101141111685.511111211179.112111311169.913111411176.114111121158.115111131159.116111141160.317111112177.818111113195.9191111141153.120111111252.821111111350.5221111114157.2Table 7Contributions of total parameters and their interactions for the target cities based on the ANOVA calculative method.VariableContribution rate of each variable for each city (%)HailarHarbin Shenyang Urumqi Xining Hotan Beijing Xi ’an Lhasa Kunming Shanghai Changsha Chengdu Guangzhou Haikou A 2 1.9 2.1 2.1 2.1 2.4 2.22 1.9 3.42 2.2 2.1 1.9 1.5B 3.3 3.1 3.1 3.13 3.3 3.3 3.14 2.6 3.1 3.232415.7C 75.474.974.375.373.673.87475.57068.975.474.875.452.350.2D 000.10.100.10.10.30.90.30.20.20.30.40.3E 0.10.10.10.400.60.40.500.20.5 1.20.61526.3F 33 2.8 2.5 3.12 2.3 2.3 2.7 2.6 2.2 1.9 2.42 1.9G 10.21111.310.112.110.8119.81313.810.19.39.70.10.1A ÂC 1.9 1.822 1.9 2.2 2.12 1.7 2.9 1.9 2.1 2.1 1.9 1.5B ÂC 00000000000000.20.2G ÂE 00000.10.100.100.30.10.20.10.50.3A ÂG 000000000000000B ÂG 3 3.1 3.13 2.9 3.2 3.2 3.14 2.5 3.2 3.3 2.900C ÂE 0.10.10.10.20.10.30.20.2000.20.30.20.7 1.1C ÂG 0000000000.300000B ÂE 000000000000000.1A ÂE 000000000000000C ÂF 0000000000000.10.10.1C ÂD 0.20.20.20.10.20.10.20.20.30.30.20.10.200G ÂF 0000.100.20.10.1000.10.20.100D ÂG 000000000000000E ÂD 000000.10.10.10.20.20.10.10.100E ÂF 0.50.50.50.60.50.60.60.60.50.50.60.60.50.70.6Errors0.30.20.20.20.30.30.20.20.71.40.20.10.10.10.1X.Gong et al./Building and Environment 58(2012)46e 5752contributes around10%,while window orientation and window-wall ratio are less involved,with a contribution below0.7%.According to the analysis of the annual thermal load by orthogonal array,the standard level of5parameters(wall thick-ness,external wall insulation thickness,window orientation, glazing type,and sunroom depth/overhang depth)is determined as their highest levels for the total25cities,as shown in Table8. Notably,the standard level of roof insulation thickness is at0mm for the cities:Shanghai,Wuhan,Changsha and Pingwu;while it is at300mm for the other21cities.The standard level of window-wall ratio is at0.6for the cities:Xining,Golmud,Lanzhou,Lhasa, Barkam,Xichang and Kunming,while it is at0.15for the other18 cities.4.2.Optimal level of passive measures by listing methodTo optimize the7passive design measures,they were varied on 4levels for a total of25cities after deciding the standard levels by orthogonal array,based on the listing method.In order to obviously illustrate the variation of the annual thermal load of each param-eter for different areas(from Figs.3to9),7representative cities are chosen:Harbin(severe cold),Urumqi(severe cold),Xining(severe cold),Xi’an(cold),Kunming(mild),Changsha(hot summer&cold winter),and Guangzhou(hot summer&warm winter).Wall thickness(WT)causes a small effect on annual thermal load for the total selected cities under the passive design condi-tions.Fig.3shows that the change of WT setting from100mm to 250mm leads to a few decreases in annual thermal load as other parameters were set at the standard levels.Roof insulation thickness(RIT)changes have a distinctive effect on the areas of different latitudes.As shown in Fig.4,only under the condition of setting the RIT from0mm to100mm,thermal load descends strongly by63.3%from56.1kWh/(m2yr)to20.6kWh/ (m2yr)for Guangzhou,while the increasing ratio of the RIT are small for the other6cities.By external wall insulation thickness(EWIT)0e70mm,the thermal load reduces greatly,then gradually approached to 0kWh/(m2yr)when the EWIT ascends continually,as shown in Fig.5.Increasing the EWIT from0to200mm can lead to a sharp decrease in the thermal load reduction rate at least up to72%for the total cities.In China,a window is usually installed in the south facing façade to catch solar radiation and reduce heating load,which is demon-strated in Fig.6.Notably,the thermal load is the lowest when windows are positioned at the north façade,and the highest if positioned at east only for Guangzhou.This is because the cooling load is merely considered for Guangzhou within the hot summer& warm winter region,where the heating load is not required for indoor comfort in winter.Large effects of window-wall ratio(WWR)on thermal load in hot summer areas demonstrated by this investigation are often ignored by architects.For instance,in a room with a glass curtain wall or large window surfaces,the thermal load can be adversely affected by direct solar gain and re-radiated heat of absorbed energy from those window surfaces.Fig.7shows that when WWR is increased from0.15to0.6,the thermal load ascends from 17.2kWh/(m2yr)to39.3kWh/(m2yr)for Guangzhou,whereas it fluctuates steadily in other cities.As shown in Fig.8,double-glazing with a low-emissivity coating is very appropriate in cold and mild areas in China.Double low-E glazing can lower annual thermal load compared with the single glazing with a rate of76.3%in Harbin,99.2%in Xining,43.6%inTable8Results of the standard level for the25target cities based on the orthogonal method.Value of the standard level for each parameter City A:WT B:RIT C:EWIT D:WO E:WWR F:GT G:SD/OD250300200South0.15Doublelow-E 1200/1800Hailar;Harbin;Altay;Zhangye;Gangca;Shenyang;Karamay;Urumqi;Hotan;Beijing;Xi’an;Chengdu;Guangzhou;Haikou2500200South0.15Doublelow-E1200/e Shanghai;Wuhan;Changsha;Pingwu;250300200South0.6Doublelow-E 1200/e Xining;Golmud;Lanzhou;Lhasa;Barkam;Xichang;KunmingFig.3.Variation of thermal load with different wallthickness.Fig.4.Variation of thermal load with different roof insulation thickness.X.Gong et al./Building and Environment58(2012)46e5753。