Vladimir ShukhovA Critical Review onDigital Architecture

Elizaveta Edemskaya1, Asterios Agkathidis21,2University of Liverpool1elizaveta.edem@gmail.com 2asterios.agkathidis@liv.ac.uk

This paper is a critical review on advantages and disadvantages of contemporarydigital architecture, in retrospect to Vladimir Shukhov's design techniques,applied in the early 20th century. After investigating Shukhov's structuralsystems, this paper explores the relationship between performance and form,questioning the necessity of high-complexity structures. It will presentunpublished archive material of his early work and stimulate a valuablediscussion by comparing it with contemporary projects designed by renownedarchitects. The study on Shukhov focuses on his tessellation method ofdouble-curved surfaces using simple standardized elements. The study of presentdigital approaches revolves around leading architects using computational tools(e.g. Foster and Partners, Buro Happold and Arup), who have materialized highcomplexity structures composed by irregular units. Our findings highlightadvantages and disadvantages of contemporary computational approaches.

Keywords: Fabrication, Material Studies, Architectural History, DesignConcept, Performative Design

INTRODUCTIONTechnology, materials and construction are three in-terdependent components that change architecture.Contemporary architecture started with the Indus-trial Revolution, which brought new technologiesand new constructionmaterials: iron and steel. In re-cent decades, computationhas becomeanext signif-icant impetus for architectural development ( Castle2013) . Computation redefined the practice of archi-tecture. Instead of working on compositions, design-ers construct a parametric computational system,where the form can be generated simply by varyingparameter values. On one hand, computational tech-nologies make it possible to design a building of any

level of geometric complexity and optimize any de-sign proposal. 'If you find a nice curve of surface some-where with interesting properties you can incorporateit in your design' (Peters, Peters 2013). On the otherhand, computational technologies create a gap be-tween architects and the final product. Because ofinsufficient knowledge of algorithmic concepts (Pe-ters, de Kestelier 2013) , most practices come upwitha building shape and a concept; then invite compu-tational designers to optimize the project. Compu-tational designers generate and explore architecturalconcepts by writing and modifying algorithms, buthow efficient is this approach?

This paper is aiming to rethink and evaluate ad-

Shape, Form and Geometry - Concepts - Volume 1 - eCAADe 33 | 395

vantages and disadvantages of computational de-sign techniques, by comparing contemporary perfor-mative design methods with Shukhov's pre- compu-tational approach, themost efficient approach of thepre- computational era. In order to narrow down theresearch area, this investigation will focus on grid-shell structures, based on following main researchquestions:

• What were Vladimir Shukhov's design tech-niques?

• HowdoShukhov'smanufacturing and assem-blingprocesses differ fromcontemporary dig-ital design techniques? Howdoes their designlogic differ?

• How can today's advanced performative de-sign techniques benefit from Shukhov's de-sign heritage?

ABOUT VLADIMIR SHUKHOVVladimir Shukhovwas one of themost innovative en-gineers of the late 19th and early 20th century in Rus-sia. Contemporaries used to call him the "Russian Edi-son" due to his extensive inventions in numerous sci-ence and engineering subfields. For Shukhov, designprocesses had always been associatedwith extensiveanalysis and research. In the design process Shukhovused to encourage his fellow workers to think 'sym-phonically,' (Kovelman 1958) which means 'complexthinking'. He perceived a construction as one organ-ismconsistingof hidden interconnections (Shukhova2003) that he aimed to reveal during deep investiga-tions.

Vladimir Shukhov realized the commercial na-ture of the construction industry and the powerof optimization as aiding competitive advantagein the growing building industry. Thus, his de-sign approach was based on two aspects, which, inShukhov's opinion, fundamentally affected the totalcost of a construction and its quality: material con-sumption and labor expenditure.

Due to the crucial interdependence betweena structure and material consumption, Shukhov al-

ways started withmathematical models that he usedto find the most efficient construction system. Hismathematical approach was based on geometryas an analytical tool. In addition to calculations,Shukhov relied on physical models. He believed thateven the smallest papermodelwas able to reveal hid-den forces that could be missed during the analysisstage (Kovleman).

Shukhov successfully optimised structures aswell as each step of the construction process. Try-ing to avoid different-type elements and keep clearof anydetails andelements that could complicate thestructure, he aimed to standardize their size withinthe structure. Another important part of Shukhov'sapproach was preparing an elaborate plan for theassembling process. In an attempt to optimize theon-site building process and reduce faults, Shukhovlimited the number of working drawings and triedto combine all the necessary information on a fewsheets, developing detailed assembly process in-structions for each specific task.

SHUKHOV'S LATTICE STRUCTURESOne of the most significant of Shukhov's inventionsin the field of architecture was the thin metal latticeor grid shell structure. Itwasdevelopedafter detailedinvestigations searching for themost rational type ofrafters that weighed and cost the least and could bequickly assembled. Shukhov suggestedaproportion,which at first sight seemed senseless:

a = e = c

wherea- lengthof panels, e -minimal distancebetweenframes, and с - distance between two purlins, depen-dent on the actual situation (Shukhov 1897)

According to the formula, the minimal coveringweight could be achieved only if the constructionhadnopurlins, and the distance between trusseswasequal to the distance between the missing purlins.The answer to this riddle was the spatial latticestructure, where trusses and purlins were the same,and the distances between trusses and purlins wereequal. The new structures were first presented to the

396 | eCAADe 33 - Shape, Form and Geometry - Concepts - Volume 1

general public at the All-Russian Industrial Art Exhibi-tion inNizhniyNovgorod in 1896,where Shukhovde-signed a number of objects using three types of lat-tice structures: suspension, vaulted and rigid spatialshell. (Figure 1)

Figure 1The first Shukhovhyperboloid watertower erected atthe 1896 NizhniyNovgorodexhibition, Archiveof the RussianAcademy ofScience,F.1508/Op.2/81(16)

Since engineers and architects have started usingcomputer technologies, they rediscoveredgrid struc-tures that do not dictate the shape of the build-ing and allow forming more complex shapes. How-ever, the shift to relying on algorithms for captur-ing and communicating designs in architecture hasbeen slow, as many architects still do not have suf-ficient computer modelling skills. Additionally, sinceeach grid structure is unique and no guides or design

and construction recommendations exist regardingthese buildings, only large practices with the knowl-edge and research back-up available, for example,BuroHappold andOver Arup& Partners, have agreedto take part in such projects (Paoli 2007). This gapbe-tween the conceptual idea and the skill set needed tooperate computational technologies contradicts thegist of the parametric approach and leads to a trans-formation fromamethod toa style. Striving tounder-stand the current situation inmore depth, in the nextchapter we will compare the computer-generatedgrid structure design process with the most effec-tive design and construction methods from the pre-computational era.

To assess the effectiveness of computational de-sign process we have chosen four grid-shell struc-tures with comparable parameters: 30St Mary Axeand the Great Court of the British Museum de-signed by Foster+Partners, and the Radio Tower onShabolovskaya St (1921) and Viksa Works (1897-98)by Shukhov. The comparison is focused on fourmainaspects: design process, grid-shell structure, fabrica-tion and assembling.

SHABOLOVSKAYA TOWER AND 30 STMARY AXE TOWERDesign processThe Radio Tower in Moscow (Figure 2) and The SwissRe building, later rebranded 30 St Mary Axe, are twotowers with the similar height (150 and 180 m) withthe biggest diameter of 40.3 m and 56.15 m, re-spectively. The Radio Tower in Moscow is a grid-shell structure comprises of six hyperboloid blocksformed with angle rods and horizontal hoops. Itsminimal surfaceandopen lattice structure reduce thewind load, the main challenge for high-rise build-ings. The dense intersections between elementsand wide cross-sections granted the tower stability,while themulti-level construction system creates ad-ditional intersections in the tower's trunk, which rein-force the structure with minimal material consump-tion (Rainer, Perchi, Shukhov 1995). The logic ofShukhov's calculations that led him, to hyperboloid

Shape, Form and Geometry - Concepts - Volume 1 - eCAADe 33 | 397

Figure 2Radio Tower’s latticestructure, by author

398 | eCAADe 33 - Shape, Form and Geometry - Concepts - Volume 1

spatial system, were based on exploring structural in-terdependences and embracing the most importantparameters in generative formulas.

The elliptic shape of the Swiss Re building wasdesigned also for reducing the wind load. By thetime, Foster+Partners architects invited Mark Burry,a specialist from the Arup Group, to optimize the fi-nal shape according to aerodynamic requirements;they developed the initial tower concept using hun-dreds of card and plastic scale models (Burry 2011).The structure of Swiss Re's building consists of a cen-tral core and a perimeter steel grid-shell tightened toeach other with rolled-steel radial beams. The grid'sinterlocking horizontal hoops turn the structure intoa stiff triangulated shell with a lateral working loadthat resistswind force andmakes thewhole construc-tion stable (Hart 2001).

Grid-shell structureThe most significant difference between Shukhov'sand computational approaches reveals in the logicof the grid shell structure. The main idiosyncrasyof Shukhov's lattice towers was that he never usedequal rings and regular intersections: the intersec-tion points between straight lines in the upper andlower parts are not symmetrical (Khan-Magomedov2010) . Rod connections were also shifted from oneto another, trying to create as much small-scale in-tercrossing as possible, like in a knitted garment;whereas, The Swiss Re's diagrid system representsa polyhedron compiled with diagonal columns andnodes.

The lattice mesh of the hyperboloid pylons inthe Shabolovskaya Tower is made of two layers of di-agonal double 140 mm U-section rods aligned be-tween two rings. These rings have a truss structurecomprisingof twoL-section rods,which simplifiespy-lons' connections andmakes it possible to fix rods se-curely. Intermediate U-section rolled metal holdingrings fix the rods between the main structural rings.In the process of connecting the diagonal rods tothe rings, they were slightly twisted along the wholelength, which was done quite easily due to high ma-

terial flexibility and because the rod section was rel-atively small. That granted additional structural stiff-ness to the construction. In order to stabilize the con-struction, the number of the rods reduced from thelower pylon to the top.

The Swiss Re's diagrid system comprises of a se-ries of steel two-story A-frames. Each frame consistsof two tubular diagonal columns bolted with a two-meter-height node. Nodes connect the diagrid shellto the radial beamsof the central core andgovern thecurvature of the building that makes them crucial inthe overall structure of the construction. Due to thebuilding's elliptical shape the geometry of each nodeis different, so the Arup team designed each node indetail during the computer modelling stage (Powell2006). Additionally, in order to stabilize the tower,Arupdesigned twocolumn types: bigger andheavierfor bottom levels and lighter for upper floors.

Fabrication and assemblyThe fabrication and assembly of the Radio Tower wasquite simple due to the identical elements through-out the building and highly original 'telescopic' as-sembly method implemented by Shukhov. All sec-tions were assembled and lifted in large blocks in-side the structure with five basic wooden cranes andpulleys (Khan-Magomedov 2010). The only issuewasbending horizontal U-section rings according to thestructure's diameter because itwas anexpensivepro-cess at the time (Shukhova 2003). The simplicity ofthe design and the assembling method made it pos-sible to build a complex structure using primitiveequipment and relying on low-skilled workers. Be-sides, the 'telescope' assembly method was highlyaccurate. In another famous hyperboloid tower withsimilar structural features to the Radio Tower the lee-way from thedead centerwas only 24mm(Kovelman1961).

Themain issue in assembling thegrid-shell struc-ture for 30 StMary Axewas that it depended on accu-rate fabrication (Powell 2006). 'With a triangular grid,there's nothing you can do if all goes wrong' (Peterson2002). Building such a structure was possible only

Shape, Form and Geometry - Concepts - Volume 1 - eCAADe 33 | 399

with the aid of advanced 3D-modelling and moderncomputational fabrication technologies that enableaccurate calculation and construction element pro-ductionwithminimumdefects and errors. There wasthe most difficult step of construction, when all 18nodes were in place around the circumference form-ing a horizontal hoop, and tie-sections were addedto link the nodes. To close the bolt holes, all thetie-sections would have to line up; the process de-manded high precision.

THE GREAT COURT OF THE BRITISH MU-SEUMAND VIKSAWORKSDesign process and grid-shell structureThe covering for the Viksa Works (Figure 03) repre-sents the first time in the world's building practicewhendouble-curved spatial vaultswere createdwithsingle-type rod elements (Khan-Magomedov 2010).Overall, the roof is divided by three-pinned archesinto five segments coveredwith symmetrical double-curvature shells. The single-curvature system wastransformed to a double-curved surface simply bybending the longitudinal beams. From the cross tothe long sections, the curcular-cut dome edges havea bend size equal to 1/6 of the span. The symmetricshape of the double-curved domes made it possibleto form themwith identically bent rods (Rainer, G, Ot-mar, P and Shukhov, 1995). 68° was considered themost optimal angle of rods' intersections. One pro-fessor, Shukhov's contemporary, describing his struc-tures, proposed the angle of 90° instead, which, as aresult, led to a 31% increase in the structure's weightduring the calculations (Shukhova 2003).

The character of the Great Court covering struc-ture is more complicated. The covering works as alattice-glazed canopy stretched between the dome'sdrum of the Reading Room and four sides of theMuseum's quadrangle. Its design was generated intwo stages. At first, engineers from Buro Happoldcalculated its geometry using standard static, or lin-ear, computer programming. Then, Chris Williams,a pioneer of design computation, was invited tostudy the deformation of the roof structure. Us-

ing computer software, he designed a 3D simula-tionmodel and optimized themesh of the grid-shell.The grid size was determined by the maximum glasspanel size available, so the structure consists of 3,312uniquedouble-glazedpanels. The totalweight of thecanopy without glazing is 478 tons.

Figure 3The Viksa Workscovering, Archive ofthe RussianAcademy ofScience,F.1508/Op.2/37(91)

Fabrication and assemblingChris Williams designed the highly complicatedshape consisting of thousands unique elements rel-atively easily. Yet although the grid-shell detailswere successfully prefabricated following 3Dmodels,the process of assembling such a non-standardizedstructure was expensive and difficult. The final grid

400 | eCAADe 33 - Shape, Form and Geometry - Concepts - Volume 1

construction was formed of radial hollow steel sec-tions (box beams), whichwerewelded to 1,826 struc-tural nodes, each node having a unique design (An-derson 2000). The engineers were highly concernedabout the reliability of the structure; therefore, Hap-pold, instead of using lower grade steel that mightcontain impurities, chose Grade D steel material,which is usually used in marine and offshore appli-cations (Hart 2001).

In contrast to the canopy of the Great Court, theconstruction system of the roof for the Viksa Works,while highly original, was relatively simple and usedno extended scaffolds, which are necessary in theassembling of complicated spatial shell structures.This construction system was 40% lighter than otherroof structures. Kovelman (1961) writes that initiallybuilders even refused to climb on the roof becausethey could not believe that such a light lattice struc-ture could sustain their weight.

DISCUSSION AND CONCLUSIONBy looking at the two grid-shell structures presented,it becomes clear that theywerebuilt following funda-mentally different principles. Shukhov's design logic,based on crossed rods, minimal asymmetry and dis-placed detail interconnections is similar to the logicof a wicker basket structure. Thin fragile elementsjoint together form a strong elastic spatial construc-tion. In contrast, Foster's computational structureis composed by polyhedrons made of beams. It iseasy to trace this back to computermodelling, wherethe most straight-forward way to design and calcu-late a smooth form is by using a polygon mesh. Thisdiscrepancy poses a question: do we use computertechnology as a supporting tool, or has it started dic-tating the design process? Computational technolo-gies make it possible to generate shapes of any levelof complexity, but Lynn stresses another issue:

'The computer is not a brain. Machine intelligencemight best be described as that of mindless connec-tions. When connecting multiple variables, the com-puter simply connects them, it does not think criticallyabout how it connects... Even in the most scientific ap-

plications of computer simulations it is argued that firstan intuitionmust bedeveloped inorder to recognize thenonlinear behaviour of computer simulations.' (Lynn1997).

Nowadays, architects and engineers designforms using the most convenient ways of compu-tation that lead to structures that are difficult tobuild. The presented comparison demonstrates that,despite new opportunities given by new technolo-gies, the design logic did not go much further thanpost and lintel system. If the architecture gener-ated by computational logic is called 'smartgeome-try,' perhaps engineers and architects should strivefor 'wisegeometry,' geometry based not as much oncalculations as on optimal construction logic in thephysical world, such as that of Shukhov's structures?Perhaps 'wisegeometry' would start addressing ChrisWilliams's contemplation in Smartgeometry: 'Com-puters are no longer a new technology, but their impli-cations for the ways in which people will work are stillunclear' (Peters and Peters 2013).

REFERENCESAnderson, R 2000, The Great Court and The British Mu-

seum, The British Museum Press, LondonBurry, M 2011, Scripting cultures: architectural design and

programming, Wiley, LondonCastle, H 2013, 'Editorial', Architectural Design, 1(222), p.

5Galankin, AN 1958, 'My briefmemories', in Archive of the

RussianAcademyof Science, NA (eds) 1958, VladimirGrigorevich Shukhov – the prominent Russian inven-tor, engineer 1853-1939, Archive of the RussianAcademy of Science, Moscow, p. F1508/Op.2/37

Hart, S 2001, 'A Brilliant Shell Game at the British Mu-seum', Architectural Record, 1(189), pp. 149-154

Khan-Magomedov, S O 2010, Vladimir Shukhov,Gordeev, Moscow

Kovelman, GM 1961,Works of Honorary Academician En-gineer Vladimir Grogorevich Shukhov, State Press ofliterature about construction, architecture and ma-terials, Moscow

Kovelman, G M 1958, 'Chapter 2. Metal structures of in-dustrial buildings and civil buildings', in Archive ofthe Russian Academy of Science, NA (eds) 1958, Thegreat Russian engineer Vladimir Grigorevich Shukhov

Shape, Form and Geometry - Concepts - Volume 1 - eCAADe 33 | 401

(1853-1939), Archive of the Russian Academy of Sci-ence, Moscow, p. F.1508/Op.2/37

Lynn, G 1997, Animate Form, Princeton ArchitecturalPress, Princeton

Paoli, C 2007, Past and Future of Grid Shell Structures,Master's Thesis, Diploma, Massachusetts Institute ofTechnology

Peters, B and de Kestelier, X 2013, 'The building of algo-rithmic', Architectural Design, 1(222), pp. 9-15

Peters, B and Peters, T (eds) 2013, Inside Smartgeometry:Expanding the Architectural Possibilities of Computa-tional Design, John Wiley & Sons, Chichester

Peterson, A 2002, 'London will never look the sameagain', Buildng, 1(46), pp. 42-49

Pottmann, H 2007, Architectural Geometry, Bentley Insti-tute, Exton

Powell, K 2006, 30 St Mary Axe: A Tower for London, Mer-rell Publishers Ltd, London

Rainer, G, Otmar, P and Shukhov, F 1995, Shukhov V.G.(1853-1939): Art of construction, Mir, Moscow

Russel, J S 2004, 'In a city averse to towers, 30 St. MaryAxe, the “towering innuendo” by Foster and Part-ners, is a big eco-friendly hit', Architectural Record,1(192), pp. 218-226

Shukhov, V G 1897, Rafters, Russian Press, MoscowShukhova, E M 2003, Vladimir Grigorevich Shukhov: the

first engineer of Russia, Bauman MGTY, Moscow

402 | eCAADe 33 - Shape, Form and Geometry - Concepts - Volume 1