Retro-digitising non-Euclidean physical models in construction history: Challenges, results and potentials

Baris Wenzel, Christian Vöhringer, Benjamin Schmid, Christiane Weber

doi:10.37190/arc240404

Summary

The article examines the process of digitizing historical non-Euclidean physical models used in construction. It discusses the challenges faced during digitization, presents results from case studies, and explores the potential applications of these digital models. Special attention is paid to 3D scanning methodology and data processing in the context of architectural heritage preservation.

✨ Summary generated by Artificial Intelligence. The actual Abstract can be found in the article below.

Full article view is only available on bigger screens.

2024

4(80)

Baris Wenzel*, Christian Vöhringer**, Benjamin Schmid***,

Christiane Weber****

Retro-digitising non-Euclidean physical models in construction history:

Challenges, results and potentials

DOI: 10.37190/arc240404

Published in open access. CC BY NC ND license

Abstract

A range of established methods are currently available for the production of digital models to document, survey, or further develop buildings.

Equipment and other technologies have been adopted from other disciplines, such as geodesy, and further developed specically for the construction

industry. These methods, therefore, have a very specic purpose and quickly reach their limits when they leave this primary area of application. As

part of a research project “Last Witnesses – Physical Models in Civil Engineering” funded by the German Research Foundation (DFG SPP 2255), an

attempt was made to transfer known methods of retro-digitisation from architecture to smaller-scale objects to create digital representations. These

small-scaled objects are so-called physical models (Messmodelle), a particular type of model mainly used in civil engineering in the past to test

load-bearing structures. The models were mainly used in the 20

century to calculate the behaviour of complex structures; however, both the model

building and the mathematical calculations were very time consuming and prone to errors. From the 1960s and 1970s onwards they were replaced by

the rst powerful computers in the construction industry.

This contribution will discuss the special characteristics of these physical models and why they were replaced by computers. We describe the appli-

cability of the dierent non-destructive methods of retrodigitisation and analyse the advantages and disadvantages of the structured light 3D scan, the

3D laser scan, and photogrammetry. In a further step, the potential of the produced digital twins in dierent contexts, such as restoration, engineering,

or architectural history, will be demonstrated.

Key words: architecture, physical models, reverse engineering, digital twin, data storage

Research Project “Last Witnesses

– Physical Models in Civil Engineering”

Physical testing models form a special class of models

that were mainly used in the eld of civil engineering, un-

til the advent of powerful computers in the construction

industry from the 1960s onwards. Such models served

to analyse, understand, and nally, even dimension and

test load-bearing structures (Bühler, Weber 2022). While

form-nding models were used to establish the geome-

try, measurement models were used to calculate forces,

but also allow for experimentation with dierent forms

and materials. Such physical models – in German called

Messmodelle – were mainly in use during the 20

centu-

ry, when drawing, dimensioning, and calculating the shape

and construction of cable bridges, dams and wide-span

lightweight structures or shells was time consuming, or not

even possible with the known calculation methods

The research project “Last Witnesses – Physical Models in Civil

Engineering” was part of the priority program “Cultural Heritage Con-

struction” (SPP 2255) funded by the German Research Foundation DFG

from 2020 to 2024. Scientists from the elds of construction history,

engineering and restoration sciences have been cooperating in the pro-

ject (DFG 2024).

ORCID: 0009-0000-5738-2722. Institute of Architectural Hi -

story, University of Stuttgart, Germany, e-mail: baris.wenzel@ifag.uni-stutt-

gart.de

** ORCID: 0000-0002-0596-0989. Institute of Architectural

History, University of Stuttgart, Germany.

*** ORCID: 0000-0003-0593-8340. Institute of Architectural

History, University of Stuttgart, Germany.

**** ORCID: 0000-0001-5446-8707. Institute of Architectural

History, University of Stuttgart, Germany.

32 Baris Wenzel, Christian Vöhringer, Benjamin Schmid, Christiane Weber

One early example is the scale model for a suspension

bridge over the Rhine at Rodenkirchen near Cologne, built

in the 1940s by the Stuttgart Material Testing Institute (Ma -

terialprüfungsanstalt Stuttgart – MPA). As it was very di-

cult to calculate the deformation of this statically overdeter-

mined structure, the Stuttgart civil engineer Fritz Leonhardt

(1909–1999), who led the project, had commissioned this

physical model on a scale of 1:100, which is documented

in photographs, today preserved in the Stuttgart University

archives (Fig. 1).

Like architectural models, physical models at the same

time could serve as a communication medium and planning

tool, but unlike these, they often were damaged or even de-

stroyed in the measuring process. For example, models of

concrete shells have been tested for cracking until failure,

resulting in the destruction of the model. Thus, the model

itself was not preserved, but data from the impact under

load was documented as shown in photographs – or even

more abstract data deduced from these. The ndings of the

research project are therefore only a very small number of

surviving objects, mostly only fragments of experiments.

The experimental set-up, including the measuring equip-

ment, has not survived at all, as it was reused for further

experiments. The appendix lists all the models researched

(based on Addis 2021), including those objects that can

only be reconstructed using images or archival material

(appendix). The Table of Surviving and Lost Models al-

lows for a preliminary evaluation of the ratio between pho -

tographic and material remains of testing models and it can

be stated that only a few physical models, or more precisely

fragments of these, have survived or found their way into

collections or archives. This may generally have occurred

due to the engineers’ lack of historical awareness and their

general underestimation of the importance of construction

as cultural heritage (Schmid 2023).

Fig. 1. Physical model (scale 1:100) for the bridge over the Rhine

at Rodenkirchen proposed by Fritz Leonhardt

(photo: saai Karlsruhe, collection Fritz Leonhardt)

Il. 1. Model fizyczny (skala 1:100) mostu nad Renem w Rodenkirchen

zaproponowany przez Fritza Leonhardta

(źródło: saai Karlsruhe, kolekcja Fritza Leonhardta)

Fortunately, certain physical models like the model for

the Olympic roof construction for Munich in 1972, de-

signed by the famous architect Frei Otto (1925–2015),

have ever been an object of broad public interest. The

ligree wireframe model is still on public display in the

visitor centre of the Olympic Park in Munich today and of-

fers detailed insights into a very courageous phase of civil

engineering (Fig. 2b). It fascinates as an in-between object:

At the same time, it is very specic concerning forces but

also abstract concerning form.

From analogue to automated

– limits of physical models

The very few still existing models are precious objects

and knowledge sources of engineering practice between

1920s and 1970s. All the more so, as they took place at

the thresholds from analogue experimentation to early

data processing with computers, developing novel in-

terfaces. For the Munich Olympic Roofs, Frei Otto’s in-

stitute for lightweight construction (Institut für Leichte

Flächentragwerke – IL) at Stuttgart University built sev-

eral form nding models of polyester mesh on a scale of

1:200. For the statical calculation of the execution of the

cable net construction, physical models at a scale of 1:125

were developed (Fig. 2). The cable net in the model is

made of stainless steel wires with a diameter of 0.2 mm

and the edge cables are made of copper wires. In order

to determine the deformations under load, the model was

loaded with steel weights that could be moved by means

of a pneumatic substructure. In addition to strain gauges

and protractors, photogrammetry with a large number of

cameras was also used in these tests, which is why they are

also known as multimedia tests. After these tests, the mod-

els were also to be used to determine the cut pattern, used

for the production and installation of the rope net. Due to

the size of the structures and the maximum working range

of the IL measuring table, the chosen scale was 1:125. The

small scale, as well as the high preload and its tolerance

values, meant that a very high level of precision in mod-

el building was required. The cutting pattern information

was obtained through photogrammetry at the Institute for

the Application of Geodesy in Civil Engineering (Institut

für angewandte Geodäsie im Bauwesen – IAGB) at the

Stuttgart University under the direction of Klaus Linkwitz

(1927–2017). As it would not have been possible to build

all the necessary detailed models in time, two new, com-

puterized methods were used to determine the dimensions

of the roof cutting patterns of the Olympic Park, in addi-

tion to the classic manual measurement at the IAGB. Klaus

Linkwitz and his Stuttgart colleague John Hadji Argyris

(1913–2004) set about calculating the geometry and,

derived from this, the cuts of the cable nets, taking into

account material and temperature inuences under pre-

tension. A CDC 6600 super computer from Control Data

Corporation at the Stuttgart University was used for this

purpose. Argyris worked on the sports hall based on the

nite element method (FEM), which he co-developed. In

FORTRAN IV, a program was created in which the topolo-

gy (Gade et al. 2022) of a network is described, an iterative

Retro-digitising non-Euclidean physical models in construction history: Challenges, results and potentials 33

determination of equilibrium is carried out, various load

cases can be simulated, and drawings can be created.

Linkwitz and his team developed a dierent calculation

program based on the method of least square deviations,

which was used immediately after its completion in May

1970 for the calculation of the stadium, and subsequently

for the cable nets of the two minor cable net constructions

at the Munich Olympic site. The very eective and at the

same time extremely robust and computationally ecient

numerical method developed by Linkwitz under great time

pressure is known today as the force density method for

determining the shape of prestressed cable meshes and is

also used for textile membrane structures.

The justied concerns of engineers as well as their curi-

osity about innovative computer-aided methods have con-

tributed to the fact that, at the height of the development

of measurement modeling statics, its end was also in sight.

Two essential computer calculation tools were developed

in this way, which are still used in daily calculations today

(Schmid et al. 2024; Schlaich et al. 1972).

Materiality in models

– appearance and behaviour

Models are always a reduction to the important aspects

of an object. Everything that seems unimportant is left out.

The simpler the model, the better, as the German physicist

Heinrich Hertz (1857–1897) said (Ortlieb 2008). When

a physical model was created for a building in the high

modern era, the reduction to the essentials was an import-

ant goal. Creating the models was usually a time-consum-

ing process. In the model for Munich’s Olympic Stadium,

for example, all the nodes of the cable mesh had to be man-

ually knotted, which led to only representing every fourth

mesh in the model.

The question of materiality played a major role in phys-

ical models. According to similarity mechanics, a material

with similar properties was usually chosen. But there were

limits to scalability, as the model of the Alster indoor swim-

ming pool that has survived at Stuttgart University shows.

This huge model realised by the Stuttgart Institute of

Model Statics (Institut für Modellstatik – IMS) is made of

acrylic glass, while the roof of the realized building (com-

pleted in Hamburg 1973) was made of reinforced concrete

(Fig. 3a) (Schmid, Weber 2021). The problem with scaling

the material was the grain size of the sand added to the

concrete. Preliminary investigations at the IMS resulted in

a minimum thickness of 1 cm for the planned model made

of micro-concrete. The resulting scale would have made

the model very large. The ratio of the internal and external

surfaces changes drastically when scaling, because while

the surface area only increases quadratically, the volume’s

growth is cubic. Furthermore, the theory of elasticity re-

quires materials with a homogeneous continuum for the

calculation of load-bearing structures, which is dicult to

achieve with natural building materials. Therefore, a mate-

rial with similar physical characteristics, such as a similar

modulus of elasticity, was chosen to build the model.

Acrylic resin proved to be a suitable modelling ma-

terial for simulating concrete elements, not only for the

Hamburg shell model but also for the many models created

by the Swiss engineer Heinz Hossdorf (1925–2006). In ad-

dition to its homogeneous nature, it was easier to process

and prepare for the experiments. As it is transparent, cracks

and air inclusions are easily recognizable and the strain

gauges could be placed exactly on top of each other, which

Fig. 2. Multimedia experiments on the model of the stadium roof in Munich at a scale of 1:125:

a) in 1964–1972 (source: ILEK Stuttgart), b) in 2023, at Visitor Centre of the Olympic Park in Munich (photo by B. Schmid)

Il. 2. Eksperymenty multimedialne na modelu dachu stadionu w Monachium w skali 1:125:

a) w 1964–1972 (źródło: ILEK Stuttgart), b) w 2023 r. w Centrum dla zwiedzających Parku Olimpijskiego w Monachium (fot. B. Schmid)

a b

34 Baris Wenzel, Christian Vöhringer, Benjamin Schmid, Christiane Weber

led to more accurate measured. Strain measurements were

made with strain gauges, an automatic measuring device

and edited via pinhole printers (Fig. 3b) (Hossdorf 1971).

Only a few of these models and of the complex mea-

suring equipment, which illustrate the transition from an-

alogue structural calculations to computer-based methods

in the 1970s, have survived. These few surviving models

store knowledge for the engineering and technical practic-

es of their time and are only handed down in fragments.

Retrodigitisation/retro-engineering

of the surviving physical models – methods

Digital representations of these physical models were

created as part of our project “Last Witnesses” as a basis

for further scientic research into these objects of transi-

tion. We have carried out a retro-digitisation of these ob-

jects like the model for the IL pavilion (Schmid, Weber

2021; Wenzel et al. 2021) (Fig. 4). The mast and cable-roof

construction were in early 1967 a 1:1 testing model for the

German Pavilion at the EXPO67 in Montreal, designed by

Rolf Gutbrod (1910–1999), Fritz Leonhardt, and Frei Otto.

The construction of this mock-up was later translocat-

ed and became Frei Otto’s institute, Institut für Leichte

Flächen tragwerke (IL) (Weber 2011, 68–83; Kleinmanns

et al. 2020). We refer to these retro-digitised models as dig-

ital twins, although we are aware that this term has a some-

what broader meaning in the mechanical engineering: the

concept of digital twins in general drove the digitisation

of industrial product development because it consequently

avoids costly and time-consuming material testing, mock-

ups and prototypes (like the IL Pavilion at its time). Digital

twins help to simulate and predict the present and future

behaviour of a physical object, which in turn allows for

Fig. 3. A surviving model and its measuring device: a) the acrylic model of the Alster swimming pool in Hamburg (photo by B. Schmid),

b) automatic measuring system and pinhole printer at the IMS (source: Stuttgart University Archive, collection IMS)

Il. 3. Zachowany model i jego urządzenie pomiarowe: a) akrylowy model basenu pływackiego Alster w Hamburgu (fot. B. Schmid),

b) automatyczny system pomiarowy i drukarka otworkowa w IMS (źródło: Archiwum Uniwersytetu w Stuttgarcie, zbiory IMS)

a b

Fig. 4. Digital representation of the physical model of IL Pavilion.

It consists of very thin spring steel wire, which is difficult to capture

using conventional recording methods (drawing by B. Wenzel, J. Nett)

Il. 4. Cyfrowa reprezentacja fizycznego modelu Pawilonu IL.

Składa się z bardzo cienkiego stalowego drutu sprężynowego,

który trudno uchwycić za pomocą konwencjonalnych metod rejestrowania

(rys. B. Wenzel, J. Nett)

Retro-digitising non-Euclidean physical models in construction history: Challenges, results and potentials 35

object optimisation and to oer variants easily, altogether

improving business eciency and hopefully sustainabili-

ty. While the term and procedure are standard in industrial

product development, with aerospace and automotive in-

dustries being forerunners, there are up to now only few

known applications for the architecture and engineering

sector. Industry 4.0 oers important orientation to the

construction industry in dealing with digital planning and

management. Comprehensive networking of data sources

and the coupling of simulation models are increasingly im-

portant for building processes such as BIM and integrative

computational design and construction

The digital representations of the surviving physical mod

els in the “Last Witnesses” project were created using state-

of-the-art digital tools. Algorithms as well as dierent mod

elling programmes were tested for their applicability. There

are two dierent methods of reversed engineering available

and both have been tested for the generation of digital twins:

A. By using the structured light 3D scan, a 3D mesh-

face is computed by projecting a simple stripe pattern on

the object. The degree of distortion is measured by com-

paring the projected stripes with detected stripes, which

allows the calculation of the relevant information about

depth. Evaluation of the depth applies for all correspon-

dences with respect to the focal points of the camera. The

result is a cloud of 3D points.

B. By using 3D laser scan, the scanner emits a laser

beam, which results in reections from the surroundings,

and these reections are received by the optics. In this

case, a rotating mirror deects the beam and the laser light

received by the scanner is evaluated accordingly and again

results in a cloud of 3D points.

Both methods were tested on a representative replica of

a cable net model, but neither provided satisfactory results.

The diculties in collecting data for reverse engineering

For the Stuttgart based cluster of excellence on Integrative Com-

putational Design and Construction for Architecture see https://www.

intcdc.uni-stuttgart.de/

are the often very thin wires or transparent or translucent

components like acrylic plastics or glass bre-reinforced

plastics, which formed the physical models of lightweight

constructions and shells.

These objects are hard to recognise by laser technolo-

gies, so it was necessary to turn to the most time-consum-

ing method, photogrammetry, which is manual, but most

of the time leads to the best results (Fig. 6). For the photo-

grammetric method, an object is rst recorded with a cam-

era from various angles. Every externally visible point

must be clearly visible in at least two photos. By using at

least two corresponding image points from two dierent

recording positions, if the mutual position is known, the

two rays can be brought to intersection and each object

point can be calculated three-dimensionally.

Therefore, the process of retro-digitization usually be-

gins by taking photographs of the model with a digital sin-

gle-lens reex camera. In the next step, the results serve

as the basis for the extraction of relevant 3D points with

PhotoModeler. In order to make this possible, the photos

have to be referenced using target points and an overlap of

the images is calculated. Then, the program Rhinoceros3D

is used to scale the geometry and align the extracted points

in a global coordinate system. Grasshopper3D allowed the

authors to automate the creation of dierent parts of the

models. Kangaroo2 (a particle-based live physics engine)

enabled the simulation of elastic models’ conditions. Some

eort was spent to connect Grasshopper with a database

for the models, via a Plug-In written in C#.

For the retro-digitisation of more solid types of models,

like a large 1:5 model of a reactor container made of steel,

which belongs to Leipzig University, the laser technique

(B) was combined with photogrammetry (Fig. 7). The

5-meter-tall object was photographed by a drone.

Use and benefits of retro-digitised data

Today’s structural analysis programs allow for much

more freedom when planning digital models. You can assign

Fig. 5. Diagrams of function from retrodigitization tools, which were tested for their suitability for the thin wires (up to 0.8 mm):

a) structured light scanner, b) LIDAR scanner, c) photogrammetry (drawing by B. Wenzel)

Il. 5. Schematy funkcjonowania narzędzi retrodigitalizacji, które zostały przetestowane pod kątem ich przydatności do cienkich drutów (do 0,8 mm):

a) skaner światła strukturalnego, b) skaner LIDAR, c) fotogrametria (rys. B. Wenzel)

a b c

36 Baris Wenzel, Christian Vöhringer, Benjamin Schmid, Christiane Weber

ny existing material and even create materials with prop-

erties that did not previously exist. In most cases, col-

umns, beams and other elements are only available as

lines, to which properties such as material and diameter

are assigned. Discretised models usually only visualise the

diameters of the supporting elements. The surface prop-

erties are generally not important and therefore omitted

for the sake of the simplication, as described at the be-

ginning. When it comes to such matters of surfaces and

external appearance, there are already AI-supported tools

Fig. 6. Surviving model and its cloud of digital 3D Points: a) low point T2 physical model of the Munich Olympia Park Swimming Pool,

b) points extracted by using photogrammetry (photo and drawing by F. Brauner)

Il. 6. Zachowany model i odpowiadająca mu chmura punktów: a) model fizyczny T2 niskiego punktu basenu w monachijskim Parku Olimpijskim,

b) punkty wyodrębnione za pomocą fotogrametrii (fot. i rys. F. Brauner)

a b

Fig. 7. 3D-Scan of a surviving model and its 3D-Resinprint:

a) for the reactor containment model state of the art tools like drones and terrestrial laser-scanners were used.

This model is represented by a polygon-mesh with 220355 vertices and 309975 faces and is fully textured,

b) the same model as a 3D resin print for an exhibition

(drawing and 3D prints by B. Wenzel)

Il. 7. Skan 3D zachowanego modelu i jego wydruk trójwymiarowy:

a) do modelu obudowy reaktora wykorzystano najnowocześniejsze narzędzia, takie jak drony i naziemne skanery laserowe.

Model ten jest reprezentowany przez siatkę wielokątów z 220355 wierzchołkami i 309975 ścianami i jest w pełni teksturowany,

b) ten sam model jako wydruk 3D z żywicy na wystawę (rys. B. Wenzel)

a b

Retro-digitising non-Euclidean physical models in construction history: Challenges, results and potentials 37

Fig. 8. Damage mapping for

restoration purposes

as exemplified in the lowpoint

model T2 (see: Fig. 6)

for the Munich Olympic stadium

(drawing by F. Brauner)

Il. 8. Mapowanie uszkodzeń

na potrzeby renowacji,

jak pokazano na przykładzie

modelu punktu najniższego T2

(por. rys. 6)

dla Stadionu Olimpijskiego w

Monachium (rys. F. Brauner)

that automatically assign UV maps to models. Work that

a few months ago had to be done laboriously by hand can

now be done by an AI in just a few moments, if you enter

the right prompt. The associated polygon meshes are also

partly generated by text prompts. We can therefore assume

that very soon AI will create a structural analysis model

and its calculations.

So, the thus produced digital representations Will serve

as a basis for further scientic processing in disciplines

such as restoration science, architectural history and civil

engineering. Concerning restoration and conservation, the

digital model helps to document the materiality and surface

quality, including signs of use, and oers possibilities for

damage mapping to provide a valuable basis for the resto-

ration process (Fig. 8). The digital twin of the retrodigitised

object can be developed as an annotation tool, useful for

recording archival sources as well as the documentation of

the restoration process by integrating the reports or photos.

In this regard the further development of retro-digitisation

presents us with the same challenges as HBIM methods,

where the aim is to store information on damage, changes,

and sources for historical buildings in a BIM model.

From an engineering and mechanical perspective, static

simulations, and calculations of structural parameters for

changing conditions (caused, e.g., by climate change, such

as greater amounts of precipitation and increased snow

loads or wind eects) could be simulated and evaluated.

The inuence of increasing precipitation could for exam-

ple be tested on the digital twin of the Munich Olympic

Sports Hall (Fig. 9) (Wenzel et al. 2021).

One research perspective is the digital reconstruction of

the lost test facilities. With the help of virtual animations,

the mechanical testing process of the physical model can

be visualised as a lost practice of civil engineering. For

example, Frei Otto’s suspended model for the Multihalle

Mannheim, which makes use of the reverse principle,

currently kept in the German Architecture Museum in

Frankfurt (Fig. 10).

Like their material brothers, the digital models store

knowledge, carrying information about an innovation-pro-

ducing lost engineering practice and about how this in-

formation was communicated and transferred into build-

ing practice. Within the research project, for archival and

conservation purposes, data contained in the digital twins

is the basis for cataloguing and categorising the objects.

A follow-up project intends to store the model-data in an

open database according to principles of FAIR for an in-

terested public. In the future, a constantly expanding data-

base of digital models should be made accessible in virtual

spaces, for example for academic institutions or museums,

to make the objects virtually tangible by using augmented

reality.

Conclusion

When the applicants asked for funding in 2019, the

expectations to nd more last witnesses of physical test-

ing models and measurement models from construction

and engineering were high. We assumed that there might

well be larger numbers of unknown objects. As an open

call which led to very poor responses, we had to revise

Information layer

Plumbs as aid

for projection

and verification

Projection layer

(also substructure)

Acrylic glass as aid

for tesnioning

Construction layer

Substructure

38 Baris Wenzel, Christian Vöhringer, Benjamin Schmid, Christiane Weber

Fig. 9. The digital twin of the measurement model of the Olympic Park sports hall in Munich in use for a qualitative rainwater runoff

analysis algorithm on the polygonmesh created by photogrammetry and a physics based particle spring system

(drawing by B. Wenzel)

Il. 9. Cyfrowy bliźniak modelu pomiarowego hali sportowej w Parku Olimpijskim w Monachium używany do algorytmu analizy jakościowego

odpływu wody deszczowej na siatce wielokątnej utworzonej za pomocą fotogrametrii i opartego na fizyce układu sprężyn cząsteczkowych

(rys. B. Wenzel)

Fig. 10. Digital model of the physical model of Multihalle Mannheim:

The so-called inversion principle (below) was applied to the load-bearing structures subjected to compression

(drawing by B. Wenzel)

Il. 10. Cyfrowy model fizycznego modelu hali widowiskowej Multihalle Mannheim:

Do konstrukcji nośnych poddanych ściskaniu zastosowano tzw. zasadę inwersji (poniżej)

(rys. B. Wenzel)

Retro-digitising non-Euclidean physical models in construction history: Challenges, results and potentials 39

Acknowledgements

This work was supported by the Deutsche Forschungs gemein schaft

(DFG, German Research Foundation) under Ger many’s Excellence Stra-

tegy – EXC 2120/1 – 390831618 and the SPP 2255.

our assumption: for German-speaking countries, no more

than a dozen physical models could be identied. At the

same time, the aesthetic, technical and techno-historical

estimation and our construction knowledge of the rare sur-

viving objects has grown immensely. The reasons why the

age of model statics came to an end in the 1970s could be

claried and the models that fall into this transition phase

could be examined for traces of the beginning automated

technology.

For these few “surviving witnesses”, matters of resto-

ration and conservation concepts became even more im-

portant and resulted in an ongoing campaign to save and

valorise the remaining models and model-fragments of

outstanding scientic as well as public interest. For this

purpose, the digital representations of the models had to

be created specically, which led us to develop methods

for the production of digital models, exploring and assess-

ing methods from other disciplines such as geodesy. It is

noteworthy, that techniques from the days of early light-

weight constructions like photogrammetric approaches

still proved the most reliable method for our project, albeit

now done with digital cameras and eventually semi-auto-

mated. This transfer of established methods from geode-

sy in architecture to smaller-scale objects also opens up

further research perspectives and concepts for their future

preservation considering scientic purposes in teaching

and experimentation by exploring an integrative AI envi-

ronment for example, to reconstruct destroyed or no longer

preserved models and their testing environment on the ba-

sis of photographic and text-based records.

Our aim is an evaluation that focuses on developing

concepts for their future preservation considering scientic

purposes in teaching and experimentation even more as the

last witnesses of model statics allow us to re-engage with

past ingenuity and cultures of evidence, which in turn may

foster nowadays approaches (Bühler, Weber 2021).

References

Addis, Bill, ed. Physical models: their historical and current use in civil

and building engineering design. Ernst & Sohn, 2021.

Bühler, Dirk, and Christiane Weber. “Epilogue: A future for models from

the past.” In Physical models. Their historical and current use in

civil and building engineering design, edited by Bill Addis. Ernst

& Sohn, 2021.

DGF, SPP 2255. Construction as cultural heritage – Principles for en-

gineering-based and interlinked conservation strategies for the built

heritage of the high modern era. Accessed May 24, 2024. https://

gepris.dfg.de/gepris/projekt/422730777?language=en.

Gade, Jan, Ekkehard Ramm, Karl-Eugen Kurrer, and Manfred Bischo.

“Marc Biguenets Beitrag zur Berechnung der Seilnetztragwerke für

die Olympischen Spiele 1972.” Stahlbau 91, H. 9 (2022): 612–21.

https://doi.org/10.1002/stab.202200048.

Hossdorf, Heinz. Modellstatik. Bauverlag, 1971.

Kleinmanns, Joachim. “Der deutsche Pavillon der Expo 67 in Montreal:

ein Schlüsselwerk deutscher Nachkriegsarchitektur.” DOM, 2020.

Ortlieb, Claus Peter. “Heinrich Hertz und das Konzept des mathema-

tischen Modells des Modellbegris.” In Heinrich Hertz (1857–

1894) and the Development of Communication. Proceedings of the

International Scientic Symposium in Hamburg Oct. 8–12, 2007,

Norderstedt 2008. Accessed May 17, 2024, at https://www.math.

uni-hamburg.de/personen/ortlieb/OrtliebHertzModell.pdf.

Schlaich, Jörg, H. Altmann, R. Bergermann, K. Gabriel, K. Horstkötter,

K. Kleinhanss, et al. “Das Olympiadach in München.” IABSE 9,

(1972): 365–77. https://doi.org/10.5169/SEALS-9577.

Schmid, Benjamin, Eberhard Möller, Andreas Putz, Christiane Weber,

and Baris Wenzel. “Die Seilnetzmodelle zur Olympiadachlandschaft

in München. Letzte Zeugen der Modellstatik. Ein Bericht aus dem

DFG-Schwerpunktprogramm Kulturerbe Konstruktion”. Stahlbau

93 (2024). https://doi.org/10.1002/stab.202400017.

Schmid, Benjamin, and Christiane Weber. “Das Messmodell für die Al-

ster-Schwimmhalle Hamburg – ein interdisziplinäres Zusammen-

wir ken Stuttgarter Bauingenieure auf dem Gebiet der Modellstatik.”

Schrif tenreihe der Gesellschaft für Bautechnikgeschichte 5 (2021):

207–22.

Schmid, Benjamin, Christiane Weber, Baris Wenzel, and Eberhard Möl-

ler. “Tool – Object – Fragment: The Afterlife of Physical Measure-

ment Mo dels.” In Structural Ana lysis of Historical Constructions.

SAHC 2023, edited by Yohei Endo, and Toshikazu Hanazato. Hei del-

berg, 2023. https://doi.org/10.1007/978-3-031-39450-8_75.

Weber, Christiane. “Der deutsche Pavillon auf der Weltausstellung in

Montréal 1967.” In Rolf Gutbrod. Bauen in den Boomjahren der

1960er, edited by Klaus Jan Philipp. Müry Salzmann, 2011.

Wenzel, Baris, Eberhard Möller, Benjamin Schmid, and Christiane We-

ber. “Last Witness and digital twin – physical and digital modeling

the Munich Olympic sport hall – a case study.” In X. International

Conference on Textile Composites and Inatable Structures – Struc

tural Membranes 2021, edited by Kai-Uwe Bletzinger, Roland

Wüch ner, and Carlos Lázaro, at https://www.scipedia.com/pub-

licWenzel_Moeller_2021a.

Streszczenie

Retrodigitalizacja nieeuklidesowych modeli zycznych w historii budownictwa. Wyzwania, efekty i możliwości

Obecnie dostępnych jest wiele metod tworzenia modeli cyfrowych do dokumentowania, badania lub dalszego rozwoju budynków. Sprzęt i inne tech-

nologie zostały przejęte z innych dyscyplin, takich jak geodezja, i są dalej rozwijane specjalnie dla branży budowlanej. Dlatego też metody te mają bar-

dzo konkretny cel i ich zastosowanie jest ograniczone przy zmianie głównego obszaru zastosowań. W ramach projektu badawczego „Ostatni świadkowie

40 Baris Wenzel, Christian Vöhringer, Benjamin Schmid, Christiane Weber

Appendix

Table of surviving and documented measurement models* (elaborated by authors based on Addis 2021)

Aneks

Tabela zachowanych i opracowanych modeli pomiarowych (oprac. autorzy na podstawie Addis 2021)

Object Time Page

Technische Hochschule München Tonnenechtwerk 1895 274 f.

Gebäude 23 Zeiss Süd Jena 1924 276 f.

Schott Dome 1924 275 f.

Dywidagschale Bootsbauhalle Düsseldorf 1925 278 f.

Zylindrische Dywidagschale 1/2 1925 279 f.

Zylindrische Dywidagschale 2/2 1925 280 f.

Markthalle Frankfurt 1927 282 f.

Markthalle Leipzig Kuppel 1927 285 f.

Markthalle Leipzig Kellerdecke 1927 286 f.

Markthalle Budapest Schale 1930 287 f.

Markthalle Budapest Träger 1930 288

Kuppelartige Dywidag Schale (Markthalle Dresden) 1931 289 .

Beton Hangars Orvieto, Orbetello, Torre del Lago 1/2 1935–1936 306 f.

Beton Hangars Orvieto, Orbetello, Torre del Lago 2/2 1938–1939 308

Pavillon Messe Mailand 1947 –

E42 Bogen Weltausstellung Rom 1939 310 f.

Kragarmplatte Hospital Clinico Universität Madrid 1929 324 .

Innenhofüberdachung Escuela elemental de trabajo 1934 326 f.

Hipodromo de la Zarzuela / Überdachung Tribüne Pferderennbahn Madrid 1935 347 f.

Oberlicht Operationssäle Hospital Clinico Universtität Madrid 1935 329 f.

Markthalle Algeciras Spanien 1934 330 .

Frontan Recoletos Madrid 1935 333 .

White River Brücke Lake Taneycomo 1931 402

Yadkin River Brücke Albermale und Nount Gilead – 402

Ashtabula River 1927 402

Bahn Brücke Weikersheim 402

Rheinbrücke Köln-Rodenkirchen 1939 416

Elbbrücke Hamburg – 417

Druckring Kuppel München Hauptbahnhof 1940 420

Kuppel München Hauptbahnhof 1940 418 .

Rheinbrücke Emmerich 1961 420 f.

Alsterschwimmhalle late 1960s 422 .

Deutscher Pavillon Montreal Expo 67’ late 1960s 427 .

Olympia Stadion München late 1960s 431 .

King Abdul Aziz Sporthalle Jeddah 1981 435

Pirelli Hochhaus Mailand 1955 449 .

Boden Galfa Hochhaus Mailand 1958 452

Stütze Velasca Hochhaus Mailand 1956 45

– modele zyczne w budownictwie lądowym” nansowanego przez Niemiecką Fundację Badawczą (DFG SPP 2255) podjęto próbę przeniesienia zna-

nych metod retrodigitalizacji z architektury na obiekty o mniejszej skali w celu stworzenia reprezentacji cyfrowych. Te obiekty o małej skali to tak zwane

modele zyczne (Messmodelle), szczególny typ modelu stosowany w przeszłości głównie w inżynierii lądowej do testowania konstrukcji nośnych.

Modele te były używane w XX w. do obliczania zachowania złożonych konstrukcji, jednak zarówno tworzenie modelu, jak i obliczenia matematyczne

były bardzo czasochłonne i podatne na błędy. Od lat 60. i 70. XX w. zostały one zastąpione przez pierwsze wydajne komputery w branży budowlanej.

W artykule omówiono szczególne cechy tych modeli zycznych i powody, dla których zostały zastąpione komputerami. Opisano przydatność

różnych nieniszczących metod retrodigitalizacji i zanalizowano zalety i wady skanowania 3D światłem strukturalnym, skanowania laserowego 3D

i fotogrametrii. W kolejnym kroku zademonstrowano potencjał wytworzonych cyfrowych bliźniaków w różnych kontekstach, takich jak renowacja,

inżynieria lub historia architektury.

Słowa kluczowe: architektura, modele zyczne, inżynieria odwrotna, cyfrowy bliźniak, przechowywanie danych

Retro-digitising non-Euclidean physical models in construction history: Challenges, results and potentials 41

Table of surviving and documented measurement models (based on Addis 2021)

Object Time Page

Bodenplatte Velasca Hochhaus Mailand – 452

Börsenhochhaus Place Victoria Montreal 1962 452 .

Parque Central Hochhaus Caracas Venezuela 1969 460 f.

St. Mary’s Kathedrale San Francisco 1/3 1965 461 .

St. Mary’s Kathedrale San Francisco 2/3 1965 461 .

St. Mary’s Kathedrale San Francisco 3/3 1965 461 .

St. Mary’s Kathedrale Tokyo ca. 1964 462

Australia Square Hochhaus Sydney 1964–1965 462

MLC Centre Hochhaus Sydney ca. 1972 462

SCOPE Cultural and Convention Center Norfolk 1967 462

Hyperbolische Paraboloidelemente Dach Newark International Airport New Jersey 1968–1969 462

Rupert C. Thompson Arena Dartmouth College New Hampshire 1970–1971 462 f.

Mole Antonelliana Turin 1955 465 f.

Brückendeck Lake Maracaibo Venezuela /

Brückendeck General-Rafael-Urdaneta-Brücke Venezuela

1958 465 .

Brückenpfeiler Polcevera Brücke Genua 1962 466 f.

Fahrbahn Eisenbahnbrücke Venedig 1963 466

Brücke Basento Potenza 1/4 1967–1975 466 . + 589 .

Brücke Basento Potenza 2/4 – 467 f. + 589 .

Zarate-Brazo Lago Brücke Argentinien 1971 467 f.

Kirchendach Xeralli Pyrenäen – 482 f.

Schalendach für National Automotive Company ENASA – 484 f.

Dach Kirche Sant Felix und Ragula Zürich 1/3 1949 485 .

Dach Kirche Sant Felix und Ragula Zürich 2/3 1949 485 .

Dach Kirche Sant Felix und Ragula Zürich 3/3 1949 485 .

Experimentelles Fertigteil-Schalendach 1949 487 .

Lagerhaus Fabrik Nadam Havenwerke 1956 491 .

Mensadach Universidad Laboral Tarragona Campus 1956 493 f.

Club Tachira Complex Caracas 1957 494 f.

Bürodach Barcadi Havana 1959 498 .

Kirche La Paz Barcelona 1961 500 f.

Canadrome Madrid 1/2 1961 502 .

Canadrome Madrid 2/2 1961 503 .

Palau Blaugrana Sportstadion Barcelona 1970s 505 f.

Strassenbrücke 1954 C&CA 1954 514 f.

Fleet Brücke Hampshire 1954 514 f.

Clifton Brücke Nottingham 1954 517 .

Auto Showroomdach Lincolnshire Motor Company 1958 520 .

Texas Instruments Factory Bedford 1959 522

Dach Commonwealth Institut London 1959–1960 522 f.

Dach Smitheld Poultry Markthalle London 1960–1961 523 f.

Medway Viadukts 1959 526 f.

Hammersmith Überführung Hochstrasse London 1959 526 .

Huntley’s Point Overpass Glasesdale Australia 1961 528

Dach Sydney Opera House 1962 529

Metropolitan Kathedrale Liverpool 1961–1964 529 .

Cumberland Basin Viadukt Bristol 1962 531 .

Mancunian Way 1963–1964 534 f.

Tinsley Viaduct M1 Sheeld 1964 536 f.

West Way London 1/2 1965 537 f.

West Way London 2/2 1965 537 f.

42 Baris Wenzel, Christian Vöhringer, Benjamin Schmid, Christiane Weber

Object Time Page

CEGB Cooling Tower 1965–1968 538 f.

West Gate Bridge Melbourne Australien 1967 539 f.

Gateshead Viadukt A1 1967–1970 540

Erhöhter Abschnitt M11 London Cambridge Autobahn 1970–1971 540

Vischer Haus Elsass 1960 553 f.

Holz Pavillon nahe Basel 1959 553 .

Bruder Klaus Kirche St. Gallen Schweiz 1957 554 f.

Lagerhaus Verband Schweizerische Konsumvereine Wangen bei Olten 1958–1961 555 f.

Überdachung Lesesaal der Universitätsbibliothek Basel 1964 557 f.

Dach Stadttheater Basel 1968 557 .

Kiessilos Günzgen Solothurn 1960 560 f.

Brückenübergang Köhlbrandbrücke – 560 f.

Pavillon Kern-Element Expo’64 Lausanne 1964 561 f.

Pavillon Element Expo’64 Lausanne 1964 561 .

National Westminster Tower London 1971 566

Brücke Basento Potenza 3/4 – 467 f. + 589 .

Brücke Basento Potenza 4/4 – 467 f. + 589 .

Bellinzona Schale 1964 620 f.

Museum Dübendorf bei Zürich – 623 f.

Dach Konzerthalle Hotel Kreuz Langenthal 1954–1955 625 f.

Heilig Geist Kirche Lommiswol bei Solothurn 1967 626 f.

Dach Sicli SA Fabrik Genf 1968 627 f.

Dach Gips Union Bex 1968 626 .

Essener Gitterschale 1/2 1962 639 f.

Essener Gitterschale 2/2 – 648 .

Multihalle Mannheim 2/3 – 652 f.

Multihalle Mannheim 3/3 1976 656 f.

Olympia Halle München late 1960s 1036 f.

Neue Kleiner-Belt-Brücke 1964 1036 f.

IL Pavillon 1966 426 .

Multihalle Mannheim 1/3 – 641 .

Müther Schale Sport- und Kongresshalle Rostock 1970 –

Stadtzentrum Tucuman 1949 314 .

Tiefpunkt T2 late 1960s –

Reaktor Containment Leipzig early 1980s –

Table of surviving and documented measurement models (based on Addis 2021)

* All models mentioned in this article are underlined.