Protection by Generative Design

Designing for full-culm bamboo durability using sunlight-hours modelling in Ladybug

John Osmond Naylor1 1University of Newcastle upon Tyne [email protected]

High yield cultivated construction materials such as bamboo could reduce our overconsumption of concrete and sand. Full-culm bamboo has low natural durability which in construction makes it imperative that the design affords protection from rain and sunlight. This paper presents and advocates a generative design workflow for full-culm bamboo using widely applicable architectural design software. A series of trials were carried out to modify the geometry of a planar truss and gablet roof with input parameters tested to determine the optimal roof surface area which could provide full solar protection at three different sites. This algorithmic process tested both straight and curved poles. Depending on the site, when compared to a symmetrical uniform 45 degree overhang, less or greater roof surface area is required in order to provide full solar protection. The use of curved poles and an asymmetrical truss could maintain full protection yet reduce the roof surface area further.

Keywords: Full-culm bamboo, Generative design approach, Ladybug, Architectural design, Digital materiality

INTRODUCTION ston, 2018). Replacing conventional materials with By 2050, some 50% of the world’s population will bio-based materials that store carbon can be one so- live in the tropics (State of the Tropics, 2020). Trop- lution to help us reduce our overconsumption of con- ical Low-to Middle-Income Countries (LMIC’s) have crete and sand (Pomponi, Hart, Arehart, & D’Amico, a shortage of housing which lacks structural qual- 2020; UNEP, 2019; van der Lugt, 2017). In many re- ity and durability (UN-Habitat, 2016) yet have a la- gions of the world, timber will most likely never be tent opportunity to utilise locally sourced bamboo able to provide the sustainable alternative we need. (Lobovikov, Paudel, Piazza, Ren, & Wu, 2007). Our Given the speed of growth and quantity required, ex- global construction industry has a decisive role to traction could have a deleterious effect on ecosys- play in climate-change mitigation with annual ce- tems (Pomponi et al., 2020). Therefore we need to ment production accounting for around 8% of an- develop the use of high yield cultivated construction thropogenic carbon dioxide emissions (Lehne & Pre- materials such as bamboo (van der Lugt, 2017). Bam-

Digital design for sustainable buildings - Volume 1 - eCAADe 39 | 315 boo can absorb carbon dioxide and stabilise slopes to and though there are minor amounts of resins, waxes tackle the effects of deforestation (Tardio, Mickovski, and tannins, none of these have enough toxicity to Stokes, & Devkota, 2017). provide any natural durability (Kumar, Shukla, Dev, & Dobriyal, 1994). Bamboo is hollow (Figure 1) and with typically thin walls, this means that a small amount Figure 1 of decay can have a significant effect on the bamboo Sketch section and (Janssen, 2000). There are abiotic factors which can terminology of a affect bamboo in a structure (Liese & Tang, 2015a). If bamboo culm bamboo is exposed to the sun and rain and in con- through the node. tact with soil the lifespan of the bamboo in the struc- Sketch by author. ture can be only 1-3 years (Janssen, 2000). Changes in temperature and humidity may produce steep mois- ture gradients between surface and inner layers, and direct exposure to sun causes unbalanced and re- peated swelling and shrinkage (Liese & Tang, 2015a) (Figure 2).

Figure 2 Full-culm bamboo It can be a challenge however to incorporate bamboo poles cracked and into contemporary design and construction prac- showing signs of tices. This is due to the natural variability of bamboo mould due to between the over 1200 species of bamboo, individ- exposure of excess ual plants, and individual culms. Within culms there moisture and is a non-regular distribution of node locations, a ta- bleaching due to pered diameter of the bamboo culm, and a reduced sunlight exposure. width of the culm wall over the length of the culm. Photograph taken Even within the same species, mechanical proper- by author in 2020. ties can differ since bamboo grown in drier areas and on slopes may have a higher fibre density and in- creased strength properties (Liese & Tang, 2015b). One response to this has been to standardise the ma- UV and visible light radiation also causes pho- terial through the manufacture of Engineered Bam- todegradation which breaks down bonds of the lig- boo Products (EPBs) (Sharma, Gatoo, Bock, Mulligan, nocellulosic polymer causing the bamboo surface to & Ramage, 2015). However, the greater challenge turn grey and coarse (Liese & Tang, 2015a). Such pro- is to use the raw form of bamboo, full-culm bam- cesses have contributed to a societal attitude of bam- boo. In doing so, we can ensure that the most af- boo as a temporary “poor man’s timber”. When bam- fordable and sustainable form of bamboo is enfran- boo is exposed or inappropriately applied in a struc- chised (Harries, Sharma, & Richard, 2012). Another ture and degrades, bamboo will often be blamed for challenge for architects when designing with bam- the action of the architect. As Janssen writes, “No boo, and the focus of this paper, is the low natural chemical treatment will be good enough to solve the durability of bamboo (Kaminski, Lawrence, Trujillo, & problems caused by incorrect design”. (Janssen, 2000). King, 2016). Bamboo is more prone to decay than This is a rallying call for architects to make sure that timber. Bamboo does not develop reaction wood, fundamental to any design process for bamboo, the

316 | eCAADe 39 - Digital design for sustainable buildings - Volume 1 bamboo is protected from rain and UV light. This is Computational design processes and full- a concept known as protection by design. The title culm bamboo of this paper refers to this phrase. Large roof over- Generative Design refers to a design approach that hangs provide protection to bamboo members from uses algorithms to generate designs (Caetano, San- wind driven rain (Figure 3), with a rule of thumb of 45 tos, & Leitão, 2020). Such processes save time and al- degrees (Kaminski, Lawrence, & Trujillo, 2016a). The low the testing of various iterations in order to find a angle of wind driven rain is largely consistent world- more optimal design solution (Jabi, 2013). The pro- wide, however the sun angle is not. cess of form-finding emerges from analysis with the output exclusively determined by function (Laiserin, Figure 3 2008). In the case of this paper, generative design A large roof tools applied to the architectural design process are overhang provides designed to deliver an optimal design output which protection to the provides full protection from sun and rain with min- structural full-culm imal material usage and minimal cost on a site by bamboo members. site basis. A wider challenge of computational de- ZERI Pavilion, Simon In some locations and orientations the overhang of sign tools with their great accuracies and full-culm Velez, 1999. Sketch 45 degrees will always be required to protect against bamboo is the difficulty in accurately modelling an by author based on wind driven rain but this may not provide enough anisotropic material with natural variability (Crolla, the ZERI Pavilion solar protection particularly if a site is further away 2017; Qi et al., 2021). In 2005, Willis and Woodward image from the from the equator. A design decision can be to suggested design parameters such as material flaws, book, Grow your cover the bamboo behind adobe or cement mortar grain directions and inconsistent densities will make own house: Simón such as bahareque construction techniques (Kamin- it difficult to achieve a direct correlation between dig- Vélez and bamboo ski, Lawrence, et al., 2016a). However for cost, aes- ital data and a constructed building, but note this gap architecture (Vélez thetic or biophilic reasons the designer may decide to between the building and the model will continue et al., 2000). make the full-culm bamboo visible. There are impres- to narrow (Willis & Woodward, 2005). Far from try- sive examples of buildings with structural full-culm ing to bridge this gap, mainstream architectural de- bamboo visible externally such as the Amairis Fac- sign practice operates on “reduced matter-models de- tory in Puerto Caldas, Colombia, by Ruta 4 [4], or the signed to behave like pristine, controlled numerical mi- ZERI Pavilion, by Simon Velez in Manizales, Colom- lieu” (Kwinter, 1996, p. 70). Using a generative de- bia (Figure 3). Conversely, where rainfall is low and sign approach to design with full-culm bamboo is sit- characteristic of a dry steppe climate or subtropical uated in a wider discourse within architectural design ridges, this overhang can be surplus to needs. Large crystalised by two terms. Digital materiality as coined overhangs can also reduce the occupiable footprint by Fabio Gramazio and Matthias Kohler (Gramazio of a building if the necessary extension of the roof & Kohler, 2008), and digital materiallurgy coined by requires an encroachment into a neighbouring site. Adam Fure in 2011 (Fure, 2011). Kohler and Gramazio Large overhangs can also be aesthetically cumber- use the term digitalmateriality to suggest a departure some and also prone to wind uplift. Whenever bam- from the design of purely form, but a design that is in- boo is exposed, the design should afford the bamboo formed by the constructive organisations and meth- protection to maintain its structural and aesthetic ods of implementation (Willmann, Gramazio, Kohler, properties. & Langenberg, 2013). Instead of realising a design or an image, a comprehensive design and building pro- cess is conceived. Constructive principles can be de- termined that define the production of architectural

Digital design for sustainable buildings - Volume 1 - eCAADe 39 | 317 components as interrelated production steps (Will- sented which use digital modelling and robotic fab- mann et al., 2013). Digital materiallurgy (a play on the rication to enfranchise bamboo poles by mapping ancient craft of metallurgy) builds on this and sug- them locally (each individual pole) and designing for gests to intentionally cede limited design control to each distinct characteristic, as opposed to negating the material’s innate ability to produce and to take the natural variability (Lorenzo, Lee, Oliva-Salinas, & advantage of unexpected formal and material com- Ontiveros-Hernandez, 2017). An example of Type plexity (Fure, 2011). Such an attitude can be the pivot 1B is the paper Working with Uncertainties: An Adap- point around which we use computational tools with tive Fabrication Workflow for Bamboo Structures which materials with natural variability, where contrary to uses vision augmentation research to respond to the materials waiting to be formed a “productive slack deviations between bamboo as built- and designed between materials and digital form” emerges (Fure, form in real-time so as to compensate for cumula- 2011). Examples of these two terms in use can be tive deviations caused by material uncertainties (Qi seen in the ongoing movement to incorporate bam- et al., 2021). An example of looking to implement boo into a design process which utilises computa- global variabilities of bamboo into a design process, tional design tools. This happens on a spectrum of Type 2, is the ZCB Pavilion by the Chinese Univer- scales from mapping the pole, to looking to model sity of Hong Kong built in October 2015 in Kowloon the variable nature of bamboo as part of the build- Bay, Hong Kong (Crolla, 2017). This was an event ing system. Before I highlight some examples and space for the Zero Carbon Building (ZCB) with a forty- in order to better situate the design process I detail yard span (Crolla, 2018). A physical scale model built in this paper, I have listed what I believe to be two from bamboo was brought into Rhinoceros 3D and ways in which bamboo is used in a computational de- Grasshopper using the Kangaroo plug in. Bending sign process (Figure 4). The first is where an inven- forces present in the physical model were abstracted tory of pre-selected bamboo is mapped in order to into vector forces and applied on a discretised curve establish parameters for the design, and the second network. The digital setup would find its force equi- is where an algorithm determines the specification of librium in comparable emerging geometries (Crolla, bamboo required. In other words a specification as 2017). Unlike those processes I classify as Type 1, the an input (Type 1), or a specification as an output (Type ZCB Pavilion workflow produced a digital emulation 2). I have split Type 1 into A and B, where B looks to model which embodied variability, not an accurate compensate for the variability of bamboo by design- model of individually digitised culms with their local ing additive components. variances that would be replicated. Using the gablet roof type as the basis of this study, the goal of this pa- per is to advocate both awareness for protection by Figure 4 design for bamboo structures and demonstrate the Conceptual applicability and efficiencies of computational de- representation of a sign tools when working with bamboo. This paper is categorisation of not about mapping the eccentricities of bamboo in how computational order to organise their role in a building system. The design processes design process in this paper is situated between Type address the natural 1 and 2. The algorithm builds immaterial geometry variability of as a representative roof form in order to test the shad- full-culm bamboo ing performance against cylindrical geometry which in the design process. An example of Type 1A is the paper BIM Bamboo. represents the bamboo poles which would be visible Here the conceptual details of a framework are pre- externally. This model can be used as both a quantifi-

318 | eCAADe 39 - Digital design for sustainable buildings - Volume 1 cation tool to suggest the curvature of the bamboo be no more than a span of 8m. The choice of a pla- that should be used, but the model also allows the nar truss for roof construction has many advantages input of material parameters. These inputs attempt for buildability. A planar truss can be built flat on the to standardise pole parameters through allowing the ground then raised and pivoted into place. This pro- ability to adapt to curvatures and pole diameters in a vides a safer work environment due to less working global manner. The question of building future toler- at height. ances into the construction information would occur Figure 5 after the steps identified in Figure 4. Planar truss spanning up to 8m Tools (Minke, 2016, p. 51). The software used in this study are widely used in the Sketch by author architecture profession. Firstly Rhinoceros 3D (Ver- based on truss sion 7) which is a three-dimensional computer graph- design referenced ics and computer-aided design software [2]. Sec- from Building with ondly, Grasshopper [1] which is a graphical algorithm Bamboo by Gernot editor integrated with Rhinoceros 3D. Grasshopper is Minke, 2016. used to build generative algorithms. Ladybug [3] is an open source environmental plugin for Grasshop- per. Ladybug allows the designer to explore the di- The truss (Figure 5) is modelled, within the algorithm rect relationship between site specific environmental generating secondary members. Producing the ge- data through the importation of standard EnergyPlus ometry for the individual poles in the truss will allow Weather (EPW) files, and the generation of graphical for drawings to be produced which show bamboo el- data outputs which can also act as inputs into the ements to cut with node and bolt locations. This truss geometry building process (Sadeghipour Roudsari & defines the transverse section of the roof geometry, Pak, 2013). Galapagos is an evolutionary solver em- where this truss will be replicated and arrayed along bedded within Grasshopper. It is a heuristic solver, the length of the building. The roof type used as the which are used when there are a large number of vari- basis of this study (Figure 6), is known in the UK as a ables to consider and an exact solution cannot be Dutch gable roof or gablet roof. For straightforward- found so a best fitness to the problem is sought. ness this is the terminology I use in this paper. There are examples of this roof design present in Kerala (He- METHODOLOGY ston, 1996), and Indonesia, known as a Javanese kam- Constructing the roof geometry in the algo- pung roof (Samodra, 2009). A conceptual representa- rithm tion for this process is shown in (Figure 8). This process starts by drawing a closed quadrilat- Input parameters eral shape in Rhinoceros which should have sides no The algorithm which generated the gablet roof ge- larger than 6m and this ‘closed curve’ is assigned in ometry was built with a series of variable input pa- Grasshopper. This allows the roof geometry to be rameters (Figure 6): (1) The peak height of the roof built over any quadrilateral site. The first edge of truss in meters; (2) The position of the ridge line as this quadrilateral will define the location of the planar a percentage of the width of the building; (3) The truss and the truss will be constructed perpendicular height of the gable as a percentage of the height of to the base plane. The design of the planar truss is the roof truss (Input parameter 1); (4) The width of the based on a design in Building with Bamboo, by Ger- roof in meters; (5) The depth of the roof overhang in not Minke (Minke, 2016). This truss is suggested to meters which is mirrored on the opposite edge; (6)

Digital design for sustainable buildings - Volume 1 - eCAADe 39 | 319 The height of the arc segment as a percentage out perform an aesthetic and performative role. It could of straightness which defines the curved rafters; and be the case that we want to use a species of bam- the additional length of the overhang (each opposite boo which is generally more curved than another. In edge can be independently input). (7A) is the exten- Trial 1, the algorithm negates any curvature of the sion in meters of the straight rafters in Trial 1, and (7B) bamboo, assuming that the bamboo that would be is the extension of the arc of the curved rafters used used in construction would be less than 1% out of in Trial 2 and 3. Within the input parameter 7A and straightness (Kaminski, Lawrence, & Trujillo, 2016b). 7B, minimum values can be set to ensure that the 45 Using curved poles as rafters (Figure 7) is later imple- degree guideline roof overhang is maintained, if re- mented in the algorithm and in Trial 2 and 3, this is quired. Input parameter 2 allows the truss to become compared to the results of the Trial 1 environmental asymmetrical to assist in finding the optimal roof ge- analysis. Trial 2, generates curved rafters with a 5% ometry and surface area. out of straightness and Trial 3, connects the input pa- rameters for the rafter curvature to the Evolutionary Solver in order to let the algorithm suggest the op- Figure 6 timal curvature of rafters we should use in order to Parameters of the reduce the surface area of the roof. gablet roof design which are input into Selected climates and optimisation the algorithm in methodology order to define the In order to run the solar hours study in Grasshopper gablet roof we need a series of input geometry and components geometry. which are provided in Grasshopper through the La- dybug plug-in (Ladybug Tools LLC, 2020). The ge- ometry to be exposed to the solar study (the bam- Figure 7 boo columns), the geometry which will provide the Planar truss design shade, known as the Context (the roof), and the lo- from Figure 5 cations of the Sun we wish to test which are avail- spanning up to 8m able through the Ladybug component ‘Ladybug_- (Minke, 2016, p. 51), SunPath’. This generates a 3D sun path in the with rafters Rhinoceros 3D window and with the sun path ori- replaced with ented east to west as default. In all tests in this study curved members. the building is positioned north/south, however this Sketch by author, can be easily changed by rotating the sun path as referencing truss necessary. A file path container and the ‘ImportEPW’ design from component is used to link this algorithm to an EPW Building with Bamboo by Gernot Consideringthecurvatureofbambooculms file. Three sites were chosen to reflect a variety of ge- ographies in which bamboo could be locally sourced Minke, 2016. Bamboo has a natural arch in growth yet straight and utilised. These were Kunming, China, which is poles are often used. This can render the majority of north of the tropics, Bengaluru, India, situated be- the available bamboo stock an unused latent asset, tween the Tropic of Cancer and the Equator, and La pushing up the price of available poles and creating Libertad, Ecuador, which will be exposed to equato- an architecture which seeks to find bamboo which rial sunlight but is characterised by a local steppe cli- fits the design, rather than designing for the natural mate with little rainfall throughout the year and prox- state of bamboo. The use of curved poles can both

320 | eCAADe 39 - Digital design for sustainable buildings - Volume 1 imity to local bamboo resources. For Kunming and put as part of the ‘genome’ to find this objective func- Bengaluru, the 45 degree guideline roof overhang is tion differed in each of the 3 trials. These are refer- set as a minimum value in input 7 given the protec- enced individually for each of the three trials in this tion required for rainfall and the script looks to val- study. idate or add to this in order to provide full protec- Baseline geometry: Roof with 45 degree over- tion from UV light. In La Libertad, given the steppe hang. As a baseline roof geometry to compare the climate and lack of rainfall, the goal here is to opti- studies to, a gablet roof is modelled which maintains mise and potentially reduce the roof overhang there- a uniform overhang on all sides of the building of 45 fore there is no minimum value set for input 7. EPW degrees. weather files were used for each of these three lo- Trial 1: Optimised roof using straight poles. For cations. Hourly sun locations for each of these sites Trial 1, all members in the truss and roof rafters are were taken from 10:00 to 16:00 for each solstice (21st straight. In this trial, input parameter 6 and 7B (Fig- June and 21st December), which is when the sun ure 6) are not used. would be at extreme angles. This information as well as the input geometries were input into the ‘Sun- Trial 2: Optimised roof with percentage straight- lightHoursAnalysis’ component which ran the calcu- ness as an input. The goal of Trial 2 and Trial 3 is to lation and output the combined hours of sunlight for compare a roof design with curved rafters to the roof both periods and dates. design with straight poles in Trial 1. In Trial 2 the user can input the % out of straightness value that is re- Figure 8 quired. This can be determined from the available Conceptual material. In this case 5% has been used. representation of Trial 3: Optimised roof with percentage straight- the generative ness as an output. It could be the case that we want design process in to find the optimal roof area prior to selecting the this paper, based on bamboo poles we will use in construction. Therefore Laiserin (2008) and in Trial 3, the inputs which define the straightness of (Tedeschi, 2014). the bamboo poles used as rafters were input into the Evolutionary Solver to find the most optimal curva- ture of pole to use.

RESULTS AND ANALYSIS The resultant geometry is shown in Figure 9 and the resultant values are shown in Table 1. It is interest- ing to note that in Kunming where the uniform 45 The optimisation methodology involved the Galapa- degree overhang was tested, there were still parts of gos evolutionary solver running for 100 steps, with 50 the columns that received hours of direct solar expo- iterations per step, with an additional 50 iterations in sure. In order to provide full shade in Kunming, the the first step. Therefore 5,050 total iterations. The ob- roof was required to be larger than the uniform base- jective function is the minimum value for: one hun- line roof geometry. In Bengaluru the algorithm val- dredth the value for the roof area plus the hours of idated the required 45 degree overhang as provid- direct sunlight on the bamboo poles. The set of vari- ing enough solar protection. In La Libertad, where ables which define the gablet roof geometry and the the dry climate meant we could overlook the 45 de- objective function are each defined by a range, a start gree guideline roof overhang, the roof was found to and an end value. The variable sliders which were in-

Digital design for sustainable buildings - Volume 1 - eCAADe 39 | 321 Table 1 Baseline roof information and results of all trials. For clarity, the values of input parameters 2, 3, 4, 5 and 7, as referenced in Figure 6, are not shown. (*the 45 degree guideline roof overhang was not applied and there is no minimum value set for input 7).

be significantly less area yet provide full solar protec- CONCLUSION tion to the columns. By implementing curved poles Architects should be aware of the low natural dura- into the algorithm this showed in some cases there bility of bamboo and align their design tools more could be marginal reduction in the roof surface area closely for bamboo to ensure durability. This paper whilst still ensuring full solar protection. In all cases demonstrates one such process. Results also showed an asymmetric truss was found to be optimal with the that using curved poles instead of straight poles can height at the minimum value of 1m. In Trial 3 which reduce the surface area of the roof and the material looked to find the most optimal curvature of bam- quantity required. Contrary to negating the curva- boo pole, the optimal curvature was between 1 and 4 ture of bamboo, if a design process following a gener- out of 100. Given the many inputs, the evolutionary ative design approach can use the curvature of bam- solver may need more than 100 steps for an improved boo as an opportunity this can potentially provide fitness to be found. A next step could be to run the cost savings. If this design was more than a bespoke solver by fixing certain values such as the 1m truss dwelling, and was a mass building programme, scal- height and the curvature of the bamboo poles at 1% ing this material reduction over many projects could to 4% to reduce the possible combinations of inputs see substantial cost savings. A next step could be to in future tests. The roof structure and joinery is also embed rainfall data to decide whether the minimum something which needs to be considered and I have 45 degree overhang is required or not as part of the not covered it here given the scope of this paper. In algorithm. The performance of bamboo structures this paper a centre line model was used as the basis of and the societal attitude to bamboo can be greatly the truss, however this can be taken further to ensure improved through the design decisions of architects. the node, bolt and cut locations are all taken into ac- At least, this is an awareness of the need to protect ex- count when placing secondary members. These can posed bamboo from the sun and rain. At best, archi- be easily added to the algorithm. tects should follow a design process which balances the formal and functional requirements of a design with the material considerations of using high yield

322 | eCAADe 39 - Digital design for sustainable buildings - Volume 1 Figure 9 Screenshot front elevations and axonometric views of output geometry visualisation from Rhinoceros interface, following the Sunlight Hours Analysis in Grasshopper and Ladybug.

cultivated materials such as bamboo. We need to Crolla, K 2018, ’Bending Bamboo Rules: Beyond Century- align our tools and processes to consider materials Old Typologies’, Journal of Architectural Education, which will provide ecological restoration, particularly 72(1), pp. 135-145 Fure, A 2011 ’Digital Materiallurgy: On the productive in tropical LMIC’s in which bamboo is available. If this force of deep codes and vital matter’, ACADIA 2011, happens we can see a construction industry as one University of Calgary, pp. 90-97 of the biggest catalysts to improved ecosystems and Gramazio, F and Kohler, M 2008, Digital materiality in ar- climate change, instead of the present reverse condi- chitecture, Lars Muller / Springer distributor, Baden / tion. London Harries, K, Sharma, S and Richard, M 2012, ’Structural Use of Full Culm Bamboo: The Path to Standardization’, REFERENCES International Journal of Architecture, Engineering and Caetano, I, Santos, L and Leitão, A 2020, ’Computational Construction, 1(2), pp. 66-75 design in architecture: Defining parametric, gener- Heston, M B 1996, ’The Nexus of Divine and Earthly Rule: ative, and algorithmic design’, Frontiers of Architec- Padmanābhapuram Palace and Traditions of Archi- tural Research, 9(2), pp. 287-300 tecture and Kingship in South Asia’, Ars Orientalis, 26, Crolla, K 2017, ’Building indeterminacy modelling – the pp. 81-106 ‘ZCB Bamboo Pavilion’ as a case study on nonstan- Jabi, W 2013, Parametric design for architecture, Laurence dard construction from natural materials’, Visualiza- King Publishing, London tion in Engineering, 5(1), pp. 1-12

Digital design for sustainable buildings - Volume 1 - eCAADe 39 | 323 Janssen, JJA 2000, Designing and Building with Bamboo, Qi, Y, Zhong, R, Kaiser, B, Nguyen, L, Wagner, HJ, Verl, INBAR A and Menges, A 2021, ’Working with Uncertain- Kaminski, S, Lawrence, A and Trujillo, D 2016a, Engi- ties: An Adaptive Fabrication Workflow for Bamboo neered Bahareque Technical Report, INBAR Structures’, in Yuan, PF, Yao, J, Yan, C, Wang, X and Kaminski, S, Lawrence, A and Trujillo, D 2016b, ’Structural Leach, N (eds) 2021, Proceedings of the 2020 Digital- use of bamboo: Part 1: Introduction to bamboo’, The FUTURES, Springer Singapore, pp. 265-279 Structural Engineer, 94(8), pp. 40-43 Sadeghipour Roudsari, M and Pak, M 2013 ’Ladybug: A Kaminski, S, Lawrence, A, Trujillo, D and King, C 2016, parametric environmental plugin for grasshopper to ’Structural use of bamboo: Part 2: Durability and help designers create an environmentally-conscious preservation’, The Structural Engineer, 94(10), pp. 38- design’, Proceedings of BS 2013: 13th Conference of 43 the International Building Performance Simulation As- Kumar, S, Shukla, KS, Dev, T and Dobriyal, PB 1994, Bam- sociation, pp. 3128-3135 boo preservation techniques: A review, INBAR Samodra, FXT 2009, ’Analysis Of Solar Geometry Influ- Kwinter, S 1996, ’Flying the bullet, or when did the fu- ences To The Roof Of The Roof Architecture In The ture begin?’, in Kwinter, S (eds) 1996, Rem Koolhaas: Tropical Region’, REGOL - Journal of Architecture and Conversations with students, Rice University, Prince- Environment, 8, pp. 35-48 ton Architectural Press, pp. 67-94 Sharma, B, Gatoo, A, Bock, M, Mulligan, H and Ramage, M Laiserin, J 2008 ’Digital Environments for Early De- 2015, ’Engineered bamboo: state of the art’, Proceed- sign: Form-Making versus Form-Finding’, First Inter- ings of the Institution of Civil Engineers - Construction national Conference on Critical Digital: What Mat- Materials, 168(2), pp. 57-67 ters(s)?, Cambridge, USA, pp. 235-242 Tardio, G, Mickovski, SB, Stokes, A and Devkota, S 2017, Lehne, J and Preston, F 2018, Making Concrete Change: ’Bamboo structures as a resilient erosion control Innovation in Low-carbon Cement and Concrete, measure’, Proceedings of the Institution of Civil Engi- Chatham House Report: The Royal Institute of In- neers - Forensic Engineering, 170(2), pp. 72-83 ternational Affairs, London Tedeschi, A 2014, AADAlgorithms-AidedDesign: Paramet- Liese, W and Tang, TKH 2015a, ’Preservation and Dry- ric Strategies using Grasshopper, Le Penseur ing of Bamboo’, in Akinlabi, ET, Anane-Fenin, K and State of the Tropics, Report 2020, State of the Tropics 2020 Akwada, DR (eds) 2015a, Bamboo: The Multipurpose Report, James Cook University, Townsville, Australia Plant, Springer, pp. 257-297 UN-Habitat, Report 2016, Slum Almanac 2015/2016: Liese, W and Tang, TKH 2015b, ’Properties of the Bam- Tracking Improvement in the Lives of Slum Dwellers, boo Culm’, in Akinlabi, ET, Anane-Fenin, K and Ak- UN-Habitat, Nairobi wada, DR (eds) 2015b, Bamboo: The Multipurpose UNEP, Report 2019, Sand and Sustainability: Finding Plant, Springer, pp. 227-256 new solutions for environmental governance of global Lobovikov, M, Paudel, S, Piazza, M, Ren, H and Wu, J 2007, sand resources, UNEP, Geneva, Switzerland World bamboo resources - A thematic study prepared Vélez, S, Vegesack, Av and Kries, M 2000, Grow your own in the framework of the Global Forest Resources As- house: Simón Vélez and bamboo architecture, Vitra sessment 2005, FAO, Rome Design Museum, Weil am Rhein, Germany Lorenzo, R, Lee, C, Oliva-Salinas, JG and Ontiveros- Willis, D and Woodward, T 2005, ’Diminishing Dificulty: Hernandez, MJ 2017, ’BIM Bamboo: a digital design Mass Customization and the Digital Production of framework for bamboo culms’, Proceedings of the In- Architecture’, Harvard Design Magazine, 23, pp. 71- stitution of Civil Engineers - Structures and Buildings, 83 170(4), pp. 295-302 Willmann, J, Gramazio, F, Kohler, M and Langenberg, S van der Lugt, P 2017, ’Engineered Bamboo Products - 2013 ’Digital by Material’, Rob | Arch 2012, Vienna, A Sustainable Choice?’, in Hebel, DE and Heisel, F Austria, pp. 12-27 (eds) 2017, Cultivated Building Materials: Industrial- [1] https://www.grasshopper3d.com/ ized Natural Resources for Architecture and Construc- [2] https://www.rhino3d.com/ tion, Birkhäuser, Basel, pp. 86-95 [3] https://www.ladybug.tools/ladybug.html Minke, G 2016, Building with Bamboo, Birkhäuser [4] https://www.dezeen.com/2020/07/28/amairis-cloth Pomponi, F, Hart, J, Arehart, JH and D’Amico, B 2020, ing-factory-ruta-4-colombia/ ’Buildings as a Global Carbon Sink? A Reality Check on Feasibility Limits’, One Earth, 3(2), pp. 157-161

324 | eCAADe 39 - Digital design for sustainable buildings - Volume 1