biomimetic-computationaldesignfordoublefacadesin...

Biomimetic-Computational Design for Double Facades inHot ClimatesA Porous Folded Façade for Office Buildings

Salma El Ahmar1, Antonio Fioravanti21,2Sapienza University of Rome1,2{salma.elahmar|antonio.fioravanti}@uniroma1.it

Biomimetic design is an approach that is gaining momentum among architectsand designers. Computational design and performance simulation softwarerepresent powerful tools that help in applying biomimetic ideas in architecturaldesign and in understanding how such proposals would behave. This paperaddresses the challenge of reducing cooling loads while trying to maintaindaylight needs of office buildings in hot climatic regions. Specifically, it focuseson double skin facades whose application in hot climates is somewhatcontroversial. Ideas from nature serve as inspiration in designing a porous,folded double façade for an existing building, aiming at increasing heat lost byconvection in the façade cavity as well as reducing heat gained by radiation. Thecooling loads and daylight autonomy of an office room are compared before andafter the proposed design to evaluate its performance.

Keywords: Biomimicry, Parametric design, Double facades, Performancesimulation

INTRODUCTIONCooling loads represent around 20% of building en-ergy consumption in Egypt (Attia, et al., 2012) and isbecoming one of major reasons of power shortagesespecially during the summer months. The buildingskin plays a critical role in determining the amountof heat gain in buildings and is therefore the focus ofthis study.

Most investigations about double facades weredone in temperate climates, aiming at increasingbuoyancy-driven not wind-driven natural ventilationand the difference between cavity and ambient tem-peratures could reach 20°C (Barbosa & Ip, 2014). In

fact the bigger the temperature differences the bet-ter the buoyancy effect and natural ventilation. How-ever, having such a great increase in cavity tempera-ture will not be favourable in hot climates where am-bient temperatures are relatively higher. Thereforetheir application in hot climatic areas is controversialand still needs investigation.

An overview of current techniques for naturallyventilatingdouble skin façadeswasneeded tounder-stand how airflow would generally behave and whatthe factors that affect it are. An interesting reviewpa-per by Barbosa and Ip (2014) identifies the main pa-rameters influencing the thermal and energy perfor-

Smart and Responsive Design - Applied - Volume 2 - eCAADe 33 | 687

mance of such facades. They provide a set of guide-lines for each design parameter which were impor-tant to take into consideration. Some of these guide-lines are suitable for hot climatic areas. For examplepreferred cavity depth shouldbebetween0.7 and1.2m, the multi-storey and shaft type structures have astronger stack effect increasing ventilation rates, im-portance of using double glazing, bigger air open-ings that aid in extracting warm air out of the cav-ity, and that decrease in cavity temperature does notvary in a linear way with opening sizes.

Hamza (2008) compared a single façade and adouble façade with three possible glazing options;transparent, tinted and reflective. The single façadeacts as a benchmark base case representing a typicaloffice building in Cairo. Results indicate that carefulmaterial choice is critical to the thermal performanceof the double façade, as reflective glazing providesthe most reduction in cooling loads. The façade hadair openings at the top and bottom. The use of shad-ing devices was not addressed.

Another study was conducted by Radhi et al.(2013) in the United Arab Emirates. They comparedan east-facing double façade system that had open-ings at top, bottom and at each floor, with a classicsingle façade system. The upper cavity opening in-duced the stack effect causing upper floor to expe-rience more heat gain, but at the same time it wasresponsible for removing the heat out of the cavity.They estimated a 17% reduction in cooling energy ona typical summer day due to the performance of thedouble façade.

An interesting study done in central Italy byBaldinelli (2009) explored the use of an unconven-tional double façade made of L-shaped movableglass and aluminium shading panels that could ro-tate using hydraulic jacks to take one of two posi-tions; an open state for summer and a close state forwinter. Cooling loads were 10.3 KWh/m2 for the pro-posed design, 151 KWh/m2 for a fully glazed singlefaçade and 77 KWh/m2 for 50% glazed single façade,which shows that double facades could be used inwarm climates if certain design aspects were taken

into consideration.Studies regarding double facades usually con-

sider the outer façade layer a as a flat vertical sur-face with not much geometrical complexity, and thedaylight performance resulting from the presence ofthese double facades has been rarely addressed incorrelation with their thermal performance. In thesearch for new ideas for buildings skins in hot cli-mates, the researchers turn to biomimetic design toexplore possible solutions.

AIM ANDMETHODOLOGYIn a previous paper (El Ahmar and Fioravanti, 2014)we defined a number of possible biomimetic inspira-tions and categorised them based on the heat trans-fer methods; radiation, conduction, convection andphase change. One inspiration related to decreasinggain through radiation was chosen (which was fold-ing strategies) as a starting point and was appliedin the design of a shading screen for a typical officeroom.

This paper builds upon previous work andwidens the scope to study the reduction of heat gainby convection as well as radiation. The biological in-spiration addressed in this case is termitemound be-haviour in natural ventilation. The main design chal-lenge is decreasing heat gained by radiation and in-creasing heat lost by convection in the cavity of aproposed double facade, with the aim of decreasingcooling loads and improving thermal comfort. An-other important aim is to study visual comfort simul-taneously, so that decreasing cooling loads wouldnot be at the expense of reducing daylight perfor-mance.

The design is developed using computationalparametric software, together with real-time envi-ronmental analyses which provide instant feedbackas the design progresses. These tools proved to beparticularly useful especially when there are conflict-ing design needs at hand. The performance of theproposed skin is evaluated by comparing the simu-lated cooling loads and daylight performance of anoffice room in anexistingbuilding inCairobefore and

688 | eCAADe 33 - Smart and Responsive Design - Applied - Volume 2

after the skin is placed. Figure 1 describes the overallapplied biomimetic-computational design method-ology applied in this research.

Figure 1Generalresearch/designmethodology.Scope of this paperis in bold.

BIOLOGICAL INSPIRATIONTermite mounds are one of the famous biomimeticexamples known for natural ventilation strategies.Macrotermes michaelseni termite mounds are foundin sub-Saharan Africa, they are closed cone-shapedmounds that could extend several meters in height,with a porous skin and an underground nest with adiameter of 1.5 to 2meters (Turner, 2001). Turner andSoar (2008) highlight somemisconceptions often as-sociatedwith these ventilationmechanismsandoffer

a more accurate explanation of how the ventilationsystem actually works.

They explain that airflow in termite mounds ismore complex than the two previously assumedairflow models; the thermosiphon model in closedmounds and the induced flow model in openmounds. The famous Eastgate building althoughsuccussful, is based upon these models that werefound not be entirely accurate. Rather, the systemis closer to a 'tidal' ventilation model in which airflow depends on temporal variations of the wind.Air in the the middle of the chimney, is under theeffect of natural/free convection (buoyancy) push-ing upwards, and forced convection pushing eitherup or down depending on the wind. In the end,air is in constant movement up and down in roughsynchrony with wind conditions outside the mound.Wind is usually the dominant force.

It is beyond the scope of this paper to fully ex-plain how the ventilation systemworks, however oneof its important features is addressed here serving asthe main inspiration for the architectural proposal.The porousmound surface plays an important role asair enters through it to a network of surface conduitsthat extend all around the mound. So regardless ofwind direction, air could enter from one side of themound and reach the other through the surface con-duits. Air movement in this network is always due towind. Then it passes through a reticulum of tunnelstill it reaches the chimney causing the air to move ei-ther up or down depending on the resultant force asexplained. The porous surface and the surface con-duits act as a buffering layer protecting the interiorof the mound from turbulent winds and enable theuse of wind regardless of its direction

Analogy in architectureAs mentioned earlier, ventilation in termite moundsis a complex process. Therefore just a couple of im-portant features serve as the main inspiration here.These features are simplified and abstracted to beable to apply them in an architectural context. Figure2 explains the addressed features of the mound ven-


tilation system and the corresponding architecturalanalogy.

Figure 2Inspiration fromtermite moundventilation systemand thecorrespondinganalogy in anarchitecturalcontext fordesigning a doublefaçade.

Design objectives of the analogy

• To design a double façade with a porous sur-face that mimics an aspect of the ventilationbehaviour in termite mounds

• Air flow in the cavity would be in constantmovement and is a result of both wind andbuoyancy forces either reinforcing or oppos-ing each other, thus attempting to increaseconvective heat loss

• To design a folded surface that reduces heatgained by solar radiation due to self-shading

• Maintain visual comfort in the office space atleast as it was before the folded façade, if notimproved

Other buildings inspired by termite mound ventila-tion such as the Eastgate building in Harare, Zim-babwe, and the Davis Alpine House at Kew Gardens,usually apply the analogy on a whole building scale,by designing a central atrium within the buildingalongwith other features (Pawlyn, 2011). However inthis research the scope is focused only façade designfor an existing building and therefore the biomimeticinspiration is applied only to the façade to see its con-

tribution alone in decreasing cooling loads.Regarding the desired self-shading effect, many

folding patterns can achieve this requirement. Aes-thetic aspects are also important when decidingwhichpattern to choose. For this designproposal thetriangular pinwheel pattern is chosen (Figure 3), it isseen as aesthetically pleasing and also all folded sur-faces would be triangular and therefore flat, makingit relatively easy to construct as opposed to double-curved surfaces.

COMPUTATION AND ANALYSISThis section describes how the biomimetic conceptwas translated to a double-skin façade system. Pre-liminary sketches were transformed into a digitalparametric model. Software used include Grasshop-per visual programming language for Rhino 3Dmod-eller, Octopus pluginwhich performsmulti-objectiveoptimization of design requirements using evolu-tionary algorithms, ArchSim Plugin which performssimulations in EnergyPlus v.8.2, and finally DIVA fordaylight simulations.

Digital model of existing roomOnly one office room from an existing building is as-sessed in this optimization process, representing atypical mid-floor space in the South Eastern façadewith the dimensions of 5 m in width, 8 m deep,and 4.1 m high. This orientation was chosen as astart as it is one of the most challenging for ther-moregulation of this building. The construction ma-terials assigned are the same as those in the exist-ing building; most importantly the façade has non-operable double-glazed tinted curtain wall panels(with light transmittance of 37%) and aluminiumcladding. The technical specification of thesemateri-als have been requested and obtained by interviewsand correspondence with the building owner/archi-tect. The simulated annual cooling loads of this roomare 139KWhm−2 . This is in close accordance withactual readings of the building which indicated thataverage annual consumption is 113 KWhm−2.

The Daylight Autonomy (300 lux, for half of oc-


cupied time) is achieved for 51% of the room area,however 37.5% of this area is over-lit, receiving morethan 3000 lux inmore than half of the occupied time.Daylight Factor minimum (which is a value of 2) wasachieved in 36% of the room area. This fact is onlyimportant for later comparison of optimization re-sults, as Daylight Factor calculations are much fasterso they are used in the optimization process as an in-dication. Daylight Autonomy will be calculated onlyfor the selected result for accurate comparison withthe existing room.

Digital model of double façadeThe triangular pinwheel is an iterative pattern thattakes an input triangle and divides it in a certain way,then applies the same division logic again to the re-sulting triangles and so on. To have a folded surfacemade from this pattern, a certain point in input tri-angle is moved perpendicularly to its surface (Figure3). Themoved distance of this point controls the folddepth of each iteration. When applied to the doublefaçade, the first iteration was extracted to act as themain structural elements that would bear the load ofthe façade. They also serve another function as theycontain a network of small 2x2 cm perforations. Thisnetwork extends throughout the façade to create theintended porous effect. Figure 4 demonstrates thecomposition of the proposed double façade.

Figure 3Left: Hierarchy of 3iterations of thetriangular pattern.Right: adding athird dimension tothe pattern tocreate a foldingeffect. Eachiteration could havea different folddepth that could bepositive (outwards)or negative(inwards).

The main input in the grasshopper script is a 'base'vertical rectangular surface placed in front of the of-fice room, exceeding the width of the room fromboth sides by 2 m. It starts at the first floor level till 1m above the third (and last) floor. The folding logicis applied to this base surface. The double façadealso has bigger openings at the very top at the heightabove the existing building to promote either windcatching and/or getting rid of hot air rising up.

Performance criteria.According to the required de-sign goals, a number of performance criteria werechosen to represent the behaviour of the proposedidea. They include cavity operative temperature, cav-ity air flow measured in air changes per hour (ach)and Daylight Factor (DF) in the office space. Theyrepresent the 'fitness' in the evolutionary algorithmicsolver Octopus which attempts to optimize designvariables to reach the solution that achieves the bestbalance or trade-off between these criteria.

EnergyPlus simulations of the double façadetake place on just one day that represents a typicalhot summer day in Cairo (2nd of July) in which aver-age ambient temperature is 32°C, average site windspeed is 4.9m/swith aNorthWest directionmeaningthat the façade at hand is in the leeward side of thebuilding. This was due to the large computation timeneeded to runannual energy simulations for this rela-tively complexmodelwith ahugenumberof small airopenings. When a solution is selected, cooling loadsfor themonthof Julywill be compared to those of theexisting room.

Design variables.Anumber of design variables con-trol the double façade morphology and therefore itsbehaviour. Table 1 describes each variable, the crite-ria that it influences and its search range in the evo-lutionary solver.

Preliminary testing of variables. The variables arenumerous and simultaneously affect performancecriteria. At the beginning all variables were givena default fixed value (mentioned in Table 1), andthen each one is tested individually to see its ef-fect on performance criteria and assign a suitablesearch rangeaccordingly. Default valueshave the fol-


Figure 4Diagram illustratingdigital model of thestudied room andthe proposedfolded porousfaçade.

lowing performance: cavity temperature=35.2°C, airflow=7.7 ach (due to infiltration only as no air open- Table 1Design variables,their descriptionand affectedperformancecriteria. The searchrange of eachvariable is set afterpreliminary testingof a wider range foreach variable aloneto understand itsinfluence.


ings are assigned in default state), DF=17% of theroom area.

The effect of some of them is clear. For exampleincreasing glazing area increases cavity temperature(up to 35.6°C) and daylight value (up to 30%of space)as expected with slight increase in airflow. Air open-ings increase air changes per hour (up to 106 ach) asthey increase, and decrease temperature (to 34.0°C).

As fold depths increase inwards (negative value)they slightly decrease temperature as insolation oneach face decreases, and also daylight decreases, butairflow increases. For the folds of the second andthird iteration, temperature slightly decrease even ifthey are folded outwards (positive value) due to theshading they create. Cavity depth had different ef-fects on each criterion. As it increases, it slightly de-creases temperature, clearly decreases airflow, andincreases daylight as it approached 1.5 m then af-ter that value daylight decreases. For fold and cav-ity depths specifically, their effect changes a lot withdifferent default values of other variables, and withdifferent combinations of each other.

An important observation is that an increase inairflow does not necessarily mean a decrease in tem-perature. This was true only to certain limit of airchanges per hour (around 60 ach), after that limit theincrease in airflow had no effect in decreasing tem-perature. This was important to understand; as it im-plies that airflow should not be assigned as an objec-tive in the evolutionary solver because it would justtry to increase it asmuch as possible. It would do thatnot understanding that it shouldn't try to increase itafter this limit especially if this increase is at the ex-pense of another objective being optimized in thesame time. Therefore, optimizing the air-openingswill take place in a separate process aiming to opti-mize temperature onlywithout daylight aswell, sincedaylight isn't affected by them.

Optimizing temperature and daylightThe search range for each variable has beenmodifiedbased on these preliminary results, to avoid wastingtime in trying solutions thatwill not producegood re-

sults. The assigned objectives are minimizing cavitytemperature and maximizing daylight, and the vari-ables are the first five of those listed in Table 1.

Best results (shown in Figure 5) for cavity temper-ature were obtained inmany cases, but usually whenfirst and second fold depths were at opposite ex-tremes, and the third fold depth was always at max-imum positive (folded outwards). Values near zero(complete flatness) for all three folds depths were al-ways avoided. There was no clear preferred range forcavity depth. Best results for daylight did not show aclear preferred variable range as well. It was noticedthat only thefirst folddepth tended tohave apositivevalue (folded outwards) rather than a negative one.Many different combinations produced the same re-sults.

A solution achieving a trade-off between cavitytemperature and room daylight was selected (Figure6) in which cavity temperature decreased to 34.7°C,and DF=23% of space. The DF is much better thanthe default case but still less than the performance ofthe existing room.

Optimizing temperature and airflowThe selected solution went through a final single-optimization process in which the variables assignedwere the distance between porous openings andcavity topopening scale factor to see the effect of theporous surface and consequently convective heatloss on the cavity temperature and airflow. The fit-ness is set to minimize cavity temperature.

Results show that the presence of this network ofperforations in addition to openings at the top had abig effect in increasing the airflow rate to an averageof 60 ach and decreasing the cavity temperature toan average of 33.7°C after an average of 34.7°C with-out them. The difference between cavity and ambi-ent temperatures is within the range of 1.7°C whichimplies that overheating is prevented at least in thespecified typical day inwhich simulationswere calcu-lated. This shows the importance of increasing heatloss by convection in the thermoregulation of thedouble façade. Airflow is due to both natural and


Figure 5Visualising designvariables thatproduced atrade-off betweentemperature anddaylight. Imageproduced by'Pollination'web-basedapplication [1].

forced convection (wind), but forced convection isdominating. All solutions produced very similar re-sults with a temperature range of 33.7°C. This is prob-ably due to the narrowed search range that was as-signedbased on the results of the preliminary testingin the beginning. The chosen solution had openingsat the top with total area of 7.2m2, and total area ofperforations=1.48m2 (each is 2x2 cm2, 0.19mapart)and the cavity volume=237.6m3.

Figure 6Selected solution;max. scale factor forglazing=0.99(values inverselyproportional toinsolation on eachface), cavity depth=1.8m, 1st folddepth=0.5m,2nd=0.24m, 3rd=0.15m , scale factorof upper airopenings= 0.9,distance betweenthe smallperforations=0.19m.


Figure 7Chart illustratingthe performance ofthe cavity (itstemperature and airchanges per hour)with changingambient conditionsthroughout themonth of July. Winddirection is usuallyNW except for the1st day and lastweek of the monthwhen it is S/SE.

Figure 8Daily cooling loadsin July of the officeroom before andafter placing thedouble facade

EVALUATING SELECTED SOLUTIONSimulationshavebeen repeated for thewholemonthof July. Figure 7 shows that cavity temperature is al-ways around 1.7 to 2.0°C above ambient temperatureevenwith changingwind speed and direction. This isquite important so that the proposed design wouldnot be always dependent on a specific wind direc-tion, increasing its efficiency. Cavity airflow through-out the month closely follows wind speed and had aminimum of 24 ach when wind speed was 2.7 m/s.

Table 2 summarizes theperformanceof the roombefore and after the double façade. The presenceof this façade had a positive effect on cooling loadsas they decreased by 15%. However, it had a neg-ative result on Daylight Autonomy which decreasedto 45.2% of the space. This means that further mod-

ifications need to be done to improve daylight, suchas using light shelves and more importantly using aglass type with higher visual transmittance than theone already used in the existing room, and then cool-ing loads should be re-calculated for evaluation anddesignmodification. Figure 8 shows a comparison ofthe daily cooling loads of the roomwith and withoutthe double façade.

Table 2Comparison ofdaylight andcooling loads of theoffice room beforeand after theplacement ofdouble façade.

Software limitationsIt is important to point out the limitation of Ener-gyPlus in modelling double facades, as the AirflowNetwork model it uses assumes that each thermalzone has a uniform temperature distribution, and itdoes not take into consideration the cavity airflowpattern (EnergyPlus, 2014). Several studies (Zhang,et al., 2013; Sabooni, et al., 2012; Kim & Park, 2011)


attempted to test the appropriateness of the AirflowNetwork model for simulating double facades, andconcludedwith the recommendation of coupling En-ergyplus with a Computational Fluid Dynamics (CFD)tool to complement each other's limitations.

CONCLUSIONThe study demonstrates the potential of couplingbiomimetic design and computational tools in de-signing a double façade for hot climates. It aimed atreducing heat gain by radiation by using a folded sur-face for self-shading, and increasing heat loss by con-vection by using small perforations inspired by theporosity of termite mound surface. Computationalsoftware played an important role in the develop-ment and analysis of the proposed design and un-derstanding its performance to make modificationsin early design stages when decisions are most im-portant. All design variables affected performancecriteria in different ways, the evolutionary optimiza-tion process proved useful in finding a trade-off be-tween different conflicting objectives.

The proposed design was capable of reducingcooling loads, however its performance regardingdaylight still needs improvement. It gives a positiveindication that double facades can be used in hot cli-mates if well shaded and ventilated. This paper is apart of an on-going research in which the followingstep is to repeat the same experiment in other façadeorientations. This would be followed by CFD simula-tions of the double façade to get more accurate re-sults regarding cavity temperature and airflow. Thisis important in order to know whether the results ofEnergyPlus could be relied on to give an indication ofthe performance of complex double façades in earlydesign phases in which the use of CFD tools is notpractical.

REFERENCESEl Ahmar, S. and Fioravanti, A. 2014 'Botanics and Para-

metric Design Fusions for Performative BuildingSkins - An application in hot climates', Proceedingsof eCAADe 2014, Newcastle upon Tyne, pp. 595-604

Attia, S., Evrard, A. and Gratia, E. 2012, 'Developmentof benchmark models for the Egyptian residentialbuildings sector.', Applied Energy, 94, pp. 279-284

Baldinelli, G. 2009, 'Double skin façades forwarmclimateregions: Analysis of a solution with an integratedmovable shading system', Building and Environment,44, p. 1107–1118

Barbosa, S. and Ip, k 2014, 'Perspectives of double skinfaçades for naturally ventilated buildings: A Review',Renewable and Sustainable Energy Reviews, 40, p.1019–1029

EnergyPlus, initials missing 2014, InputOutput Reference:The Encyclopedic Reference to EnergyPlus Input andOutput, US Department of Energy

Hamza, N. 2008, 'Double versus single skin facades in hotarid areas', Energy and Buildings, 40, p. 240–248

Kim, D. and Park, C. 2011, 'Difficulties and limitations inperformance simulation of a double skin facade', En-ergy and Buildings, 43, p. 3635–3645

Pawlyn, M. 2011, Biomimicry in Architecture, RIBA Pub-lishing, London

Radhi, H., Sharples, S. and Fikiry, F. 2013, 'Will multi-facade systems reduce coolingenergy in fully glazedbuildings? A scoping study of UAE buildings', Energyand Buildings, 56, pp. 179-188

Sabooni, M.A., Vaseti, H.M., Maerefat, M. and Azimi, A.2012 'Development of the capability of Energyplussoftware to simulation of building double-skin fa-cade', International Symposium on Sustainable En-ergy in Buildings and Urban Areas (ICHMT), Kusadasi

Turner, J.S. 2001, 'On the Mound of Macrotermesmichaelseni as an Organ of Respiratory Gas Ex-change', Physiological and Biochemical Zoology,74(6), pp. 798-822

Turner, J.S. and Soar, R.C. 2008 'Beyond biomimicry:What termites can tell us about realizing the liv-ing building', Proceedings of the First InternationalConference on Industrialized, Intelligent Construction(I3CON), Loughborough

Zhang, R.L., Yao, S.C. and Zhang, Y. 2013, 'Coupled En-ergyplus and computational fluid dynamics natu-ral ventilation simulation', Building and Environment,68, pp. 100-113

[1] http://mostapharoudsari.github.io/Honeybee/Pollination


biomimetic-computationaldesignfordoublefacadesin...

Documents