Engineering Failure Analysis 43 (2014) 107–119

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/engfailanal

Partial collapse of the Berlin Congress Hall on May 21st, 1980

R. Helmerich ⇑, A. Zunkel
Federal Institute for Materials Research and Testing, Unter den Eichen 87, 12205 Berlin, Germany

a r t i c l e i n f o a b s t r a c t

Article history: The Berlin Congress Hall was a gift of the United States of America to Germany for the
Available online 11 December 2013 Berlin World Exhibition in 1957. The elegantly double-curved roof was made from an
advanced prestressed concrete construction. The construction followed the design by the
Keywords: American architect Hugh Stubbins with special structural modifications to comply with
Partial collapse German construction rules. No early indications for failure initiation were detected before
Berlin Congress Hall a sudden partial collapse. On May 21st, 1980, the Southern external roof overhang col-
Analysis
lapsed. In December 1980, Jörg Schlaich and his co-authors published a summarizing
Stress corrosion cracking
Reconstruction
expert opinion about structural causes of failure. The present paper mainly introduces
results of cause analysis made at the Federal Institute for Materials Research and Testing
(BAM), mandated by the Public Prosecutor at the Berlin District Court to specify the causes
of the sudden failure. This paper refers to BAM-publications about analyses that were per-
formed under this mandate and published – most of them in German language – within the
first years after failure. The expert’s opinions comprise structural considerations materials
investigations, metallographic analyses and corrosion. Nowadays experts have learnt from
the failure and built a slightly modified roof in the original shape at the 750th birthday of
the city of Berlin and re-opened the former Berlin Congress Hall to the public on May 9th,
1987. The Hall is now serving as the House of the Cultures of the World, following the
initial intention of the hall.
Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.

1. Introduction

The Berlin Congress Hall is a building that is of importance not only from the architectural, but also from the historical
point of view. The gift from the United States of America was of high politic and symbolic relevance after WWII in the period
of the cold war between the two international political blocks. The architect of the Berlin Congress Hall, the American archi-
tect Hugh Stubbins (1912–2006), started his carrier as an assistant of Walter Gropius (1883–1969). He was working together
with other famous architects such as Marcel Breuer (1902–1981) and Alvar Aalto (1898–1976). At the time of erection, only
few shell structures existed in the world such as the Raleigh arena (now Dorton Hall) in North Carolina, designed by the
architect Maciej Nowicki (1910–1950), realized by his colleagues in 1953, and the Kresge-Auditorium of the Massachusetts
Institute of Technology in Cambridge built by Eero Saarinen between 1953 and 1956 [1]. Early shell structures in Germany
are the Schwarzwaldhalle in Karlsruhe built by Erich Schilling, the master transmitting station Felsberg in Saarland by Jean
Francois Guédy – both built in 1954 – and the retirement home Knapsack of the Hoechst AG in Hürth by Karl Hell, which
followed in 1956. Most of the previous structures were supported along their roof arches.
The Berlin Congress Hall roof with a new type of very slim prestressed concrete structure and cantilevers forming the
external roof, belongs to the most famous international structures of the postwar modernity. The structurally demanding

⇑ Corresponding author.
E-mail address: rosemarie.helmerich@bam.de (R. Helmerich).

1350-6307/$ - see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engfailanal.2013.11.013
108 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119

and venturing prestressed roof structure provoked special awareness. Great public attention was paid to the construction
and finally to the big opening of the Berlin Congress Hall which represents the recent architecture from the fifties until today
[1]. A hyperbolic paraboloid (Hypar shell) as a roof erected on only two bearings with a wide brim – as designed by the
American architect Hugh Stubbins – has never been built in this manner before. The initial design of a roof with only two
supports was sensitive to fulfil stability requirements under extreme load cases such as one-sided wind or snow. To ensure
a safe performance of the structure, the structural design of the double-curved roof had a modification of the roof details.
23 Years after the big opening, in May 1980, a considerable external part of the Southern roof suddenly collapsed. One
journalist lost his life; several persons were injured [3]. In the end of 1980, Professor Schlaich, Professor Kordina and Profes-
sor Engell published the summarizing expert opinion, prepared under the mandate of the owner and operating company
Ausstellungs-Messe-Kongress-GmbH (AMK) [2]. The Office of the Public Prosecutor at the Berlin District Court mandated
the Federal Institute for Materials Research and Testing (BAM) to carry out an expert opinion with special attention to mate-
rials behavior [3] to ascertain the circumstances of the partial collapse in more detail. Under this mandate BAM experts coop-
erated with the building operating company AMK during inspection of the failure and the dismantling of the remaining roof
overhang. The root cause analysis finally resulted in the conclusion that nobody was declared to be exclusively responsible
for the partial collapse and the Office of the Public Prosecutor at the Berlin District Court stopped all further investigations.
The coincidence of all subsequent insufficiencies during construction, execution and survey of the structure led to the fail-
ure. A single deficit would not have caused this catastrophe [3]. The Office of the Public Prosecutor could not accuse any of
the responsible persons for full misconduct. No single acting person was able to foresee that his doing together with the
activities of others could result in the partial collapse of the roof. The event was discussed in expert groups; the findings
and the lessons learnt from the collapse were published, e.g. in [2–4], and resulted finally in generating new knowledge
for the design of prestressed concrete structures and optimization of technical approvals, consequences in standards few
years later, for prestressing steel.
Prestressed concrete structures have been a quite young technology based on the ideas and patents of Freysinet between
1928 and 1936. German practical experience with prestressed concrete structures reached back to the first prestressed
bridges in Aue (1937–1993) with external prestressing cables, designed by Franz Dischinger. The first prestressed bridge
with prestressed tendons embedded in the concrete according to Freysinet was erected in Germany in 1938. As Leonhardt
mentioned in his Prestressed concrete for the practice (1955) [10], there was a need for developing higher strength steel for
reaching a sufficient remaining prestressing force in the structure, high enough to prestress the concrete. That was, at the
same time, a demanding requirement for the early application of the high strength steel wires for the erection of the Berlin
Congress Hall.

1.1. Historical and political importance

The history of the Berlin Congress Hall began during the preparation for the Internationale Bauausstellung IBA (Interna-
tional Exhibition for Construction), which was planned to take place in Berlin in 1956. At the same time, the construction
was part of the American initiated economic support in the West German federal states after WWII. The engagement was
affected by the relation to the Soviet Union and the confrontation between the two political blocks. Thus, besides the inten-
tion to build an exhibition hall and conference venue, the hall gained a prominent symbolic role in the political context.

Fig. 1. Original structure of the congress hall before sudden collapse, photograph in 1960 [3].
 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119 109

The architecture and design and not least the location in the divided city of Berlin intended to be a symbol for the love for
freedom in the Western hemisphere of the world. The venturesome drive of the shell roof symbolized this intention and was
consistent with the prevailing taste in this period (Fig. 1).
The United States of America were making the Berlin Congress Hall a gift for the destroyed Germany after WWII at the
occasion of the Berlin International Construction Exhibition (IBA) 1956, later postponed to 1957. They were also sponsoring
the construction to a large extend. The Berlin Senate sent just a formal invitation. The Berlin authorities had discussed the
design concept in detail before construction and introduced unique details in the static system to transfer the loads safely
into the ground. For horizontal stabilization a horizontal foundation was added consisting of two plates on both sides of the
regular foundation plate resting on piles. A tensile member connects the East and West abutment (Fig. 2) [3]. While config-
uring the structural details, the engineers were following the intended system by the architect Hugh Stubbins as far as pos-
sible [2]. The German contribution according to the contract was delayed because of the dispute about the German
construction law. Under enormous time pressure, the construction was finalized according to German construction law
and opened to the public during the IBA in 1957.

1.2. Architectural concept and realized construction

The architectural concept followed the general idea designed by the American architect Hugh Stubbins [1]. The Berlin
Congress Hall roof arch was – in contrast to the other mentioned existing shell structures – a cantilever that was not
directly supported by external walls. The roof was stabilized along the cantilever edges by a hollow section but it was
not vertically supported along its roof arch (Fig. 1). Architect Stubbins’ design of a roof with only two main supports
(Fig. 2) and a wide cantilever was hard to realize at the former-days knowledge about shell structures and because
of the expected instability under one-sided wind or snow. The example for the design of the Berlin Congress Hall
was the double curved roof concept of the Raleigh arena according to the architectural idea of Maciej Nowicki. The
construction of the Raleigh arena was realized after his death with a shell-roof by his colleague, engineer Fred Severud,
in the city of Raleigh, North-Carolina USA in 1953 [1]. This hall roof shell without overhang was stabilized along the
edges of the roof by an outer wall structure and by additional stiffening elements inside the wall. An additional tendon
was supporting the roof ring carrying the shell. In difference to the Raleigh arena, the Berlin Congress Hall roof was
designed on two supports, only, symbolically bridging the structure between its west and the east bearing with cantile-
vering arches.
Berlin authorities responsible for structural safety required structural measures for a safe load transfer from the roof un-
der extreme loading conditions into the abutments. The roof overhang with the limiting arches caused significant transverse
loading on the 7 cm slim shell at the connection to the massive auditorium wall support. Engineers had additional concerns
about the stability under asymmetrical load, because the overhang of the arch was exceeding the walls of the auditorium by
up to 8 m.
The structural concept was finally realized in accordance with the German construction law. Some details are described
below (Figs. 2–4):

– The whole structure was stabilized using an invisible circular reinforced concrete beam (ring beam) above the walls of the
auditorium on the upper side of the roof guaranteeing a sleek appearance.
– The external roof tensioning belts were separated from the roof above the inner auditorium and anchored into the ring
beam.
– A tensile (tie) member in the foundation level connected and tightened the east with the west abutments (Fig. 2).

Fig. 2. East–west section of the original Berlin Congress Hall (1957–1980) [3].
110 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119

Fig. 3. Connection of the inner and the external roof in a ring beam with detailing [3].

Fig. 4. Ground view with the location of corroded and broken tendons [3].

– Arches served as tensioning belts to prestress parallel plates with two or more tendons forming the external roof canti-
lever against the ring beam (see Fig. 4). The tendons were designed to be in the axis of the plate with the same concrete
cover of only 2.25 cm on both sides according to former construction rules.
– The ring beam above the auditorium walls anchored the prestressing tendons of both, the inner roof and the external
roof; see Fig. 4, [3].
– Additional horizontal foundation plates and vertical walls contributed to the stability of the structure against tilting.

The anchorage of the tendons against the ring beam was carried out stepwise. The last action was the filling of the groove
with concrete after prestressing, shown in Fig. 3, detail A.

1.3. Description of the failure event

On May, 21st 1980, the external cantilever with the arch of the Southern roof structure of the Berlin Congress Hall
collapsed suddenly (Fig. 5).
 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119 111

It was visible from the debris that numerous prestressing wires for fixing the external roof had failed. The concrete cover
depth of these tendon ducts was only 2 cm and probably too thin to prevent corrosion [2].The inner roof structure remained
almost undamaged. All failed tendons were located in the south-east quarter of the building. Some heavily corroded
wires seemed to be broken already long time ago. Three main conditions of the failed wires were distinguished and described
in [3]:

– ductile fracture,
– corroded fracture surface,
– partly corroded surface.

The location and condition classification of broken tendons is shown in Fig. 4.

1.4. Objectives

Investigations, carried out at BAM focused on visual inspection, analysis of the corrosion damage, analysis of two mate-
rials, concrete in the final groove of the ring beam (Fig. 3) and steel wires in the tendon ducts. The fracture surfaces were
analysed by means of light and scanning electron microscopy.

2. Investigations

2.1. Visual inspection and observations after failure

The visual inspection of the roof elements after failure [3] revealed that eight tendons had failed completely and two ten-
dons failed partly. An additional awkward fact was the location of seven of these failed tendons next to each. The majority of
tendons were corroded to some extent in the failed location next to the final groove of the ring beam. Many fracture surfaces
were covered with corrosion products indicating that the wires were broken already some time ago. It was observed that all
failed tendon ducts were located in the south-east quarter of the Congress Hall, just opposing the so called mirror lakes in
front of the Congress Hall with permanent fountains causing a continuous foggy atmosphere in their vicinity, i.e. adverse
microclimatic conditions. In the same quarter of the roof, the isolation of the roof was humid and heavily deteriorated.
The roofing sheets were broken in this section [3].
Directly on-site and later in the laboratory colleagues from BAM [4] have inspected the single wires fracture surfaces to
capture the information of possible different types of fracture.
All investigated broken wires belonged to the outer roof only and were concentrated in the field of the groove in the cir-
cular ring beam (see Fig. 3, detail A) in the south-east end of the failed roof cantilever. Insufficient embedded tendons sus-
pected low passivation by the concrete in this extreme case.
According to the structural design it was intended to locate the tendons in the axes of the external roof plates. In reality
was observed that there was no concrete between the bitumen sheet in the groove and the tendon duct (Fig. 6) at their con-
nection to the ring beam (Fig. 3). Remaining bituminized roofing from the groove ground and colour protection layer below
the tendon duct illustrate the described situation (Fig. 6). The tendon duct was corroded and the broken wires showed heavy
corrosion. From the same figure it is obvious that the grouting of the tendon duct for passivation of the steel was insufficient

Fig. 5. View from the South on the collapsed roof overhang in 1980 [3].
112 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119

Fig. 6. Remaining bituminized roofing on a completely failed, non-grouted and heavily corroded tendon [3].

or not even existent (Figs. 6–8). Unfortunately some surfaces were completely covered with cauliflower-like corrosion prod-
ucts (Fig. 9), [3,4].
The extreme situation was completed by bending of the tendon ducts at the connection of the prestressed outer roof
plates to the ring beam, probably caused by the self-weight already during construction. The influence of the low cover
on the durability was probably not considered sufficiently in former standards. During stepwise prestressing, the bending
of the external roof plates including their tendons led to an unexpected disorder of the strand́s location inside the tendon
duct leaning to the lower surface of the inner wall of the tendon duct (Figs. 7 and 11). At the ends of the broken wires heavy
corrosion was visible, probably starting from the lower surfaces [3].

2.2. Structural analysis

The roof of the Berlin Congress Hall consisted originally of a:

– prestressed inner roof above the auditorium with anchorage in the ring beams on top of the auditorium wall in Fig. 3 and
a,
– prestressed outer roof prestressing 24 concrete plates (width 2 m, length between 2 and 8 m with 2–4 tendons in each
plate) against the ring beam, Fig. 3.

Each tendon in the outer roof plate contained between 7 and 10 prestressed wires [4]. The arches are the boundary of the
outer roof and ended at the foot points above the foundation in the east and in the west, Fig. 2. The foundation consisted of
plates anchored by piles in the ground. A spanning tensile member in the ground connected both anchorage points. Trans-
verse walls to the main axis of the roof and stiffening stripes stabilized the roof structure against tilting [2,7].
For the prestressing of the tendons in the outer roof plates, creep and shrinkage were considered for the unloaded case by
intending to have the summary force in upward direction. As earlier described, the critical point was the load transfer

~ 130 mm

Fig. 7. Prestressing wires are almost not embedded in the protective grout (left) or insufficient grouted (right). Only one broken wire (left) shows a non-
corroded brittle broken fracture surface.
 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119 113

~ 46 mm

~ 23 mm

Fig. 8. Insufficient grouted tendon duct Sigma oval 140/165 with ten wires in carbonated concrete [4].

~ 4.2 mm
~ 9 mm

Fig. 9. Heavily pre-existing corroded fracture surface event.

Fig. 10. Intended (upper) and real location (lower) of tendons (left) and wires in the tendon duct (right) [4].

between the outer roof plates to the ring beam (Fig. 10). The real distribution of the wires in the tendons and their location at
the lower inner wall of the duct (Fig. 11) differed from the intended axial location. This fact may have occurred already dur-
ing construction and during the stepwise prestressing. The outer roof was completed after the inner roof was finalized.
114 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119

Fig. 11. Grouting conditions of tendon XI in a distance of 50–80 cm from the fracture. The wires lean to the lower surface [3].

2.3. Analysis of the concrete

The last cast concrete was for the closure of the groove on the ring beam. The concrete quality in the groove of the ring
beam was insufficient (see Fig. 12). The oven-dry density and the porosity related to open pores of the concrete differed
along the length of the ring beam groove, ranging from 2.12 to 2.18 kg/m3 and from 13.7% to 19.7%, respectively. The water
uptake under vacuum conditions is here 20%, resulting in carbon content of 7–8%. These are good conditions for the initiation
of corrosion. For comparison, the oven-dry density for the roof plates was 2.2 kg/dm3 and the porosity related to open pores
was 13–14%. The resulting possible carbon hydroxide content corresponds to 2–3%, only. Locally, in failed cross sections, the
sand and gravel were observed almost without cement matrix.
At the time of construction (1957), the requirements for reduced corrosion-inducing and advancing components, as chlo-
rides, in the concrete mass did not exist. As early as in December 1958, the declaration of chlorides in the cement was imple-
mented in the German standard for cements, DIN 1164 Portland cement [16] (today substituted by DIN EN 197-1). A remark
was added about the natural content of chlorides up to 0.1% Cl. The analysis of the chloride content after failure (1980) in the
ring beam groove (0.15–0.27%) and in the tendon duct grouting (0.11–0.16%) exceeded this limit. According to the limit of
0.2% compared to the cement mass, valid at the time of the expert opinions (1980), the chloride content in the groove was
exceeding the limit for one specimen, only.

2.4. Analysis of the prestressing steel

Close to both surfaces of the 7 cm thin roof was mild steel reinforcement, Fig. 3, Ø 6 mm as a near surface grid in both
directions with grid size 15  15 cm made of the former steel category BStIIIb. Between the two layers in the middle of
the roof plate was the oval reinforcing metal tendon duct with 7–10 wires made of ribbed high strength steel Sigma
oval St 145/160, (30). St 145/160 was tempered steel with a yield strength of 1450 N/mm2 and ultimate strength of
1600 N/mm2 (Table 1). The number 30 in brackets in the second column of Table 1 describes the cross section area of an
oval wire in mm2. Tempering reduces the grain size, but increases the ductility, yield and ultimate strength. The resulting
high strength wires are more sensitive to heat and to corrosion.

Fig. 12. Poor concrete of the ring beam groove at the tendon duct no. VIII 3 with a piece of reinforcement and corroded tendon duct [3].
 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119 115

Table 1
Mechanical properties of the prestressing steel Sigma oval as obtained in the tensile tests [3], reference values acc. to Leonhardt [10] and the technical
approvals as far as cited in [3].

St 145/160 Cross section Elastic limit or Ultimate strength Elongation (%) Technical creep
F mm (mm2) r0,2 (fy or f0,2) (MPa) rBr (fu) (MPa) limit (kg/mm2)
Sigma-oval rolled and tempered [10] 9  4.2 (30) 1450 1600 5 120
Technical approvals [3] – 1420 1570 6 –
Test results [3] – 1389–1475 1528–1615 5.5–9.5 –

Nowadays we know, that the old high strength Sigma steel 145/160, oval, produced before 1965, is very sensitive to stress
corrosion cracking (SCC) [4]. If this material is continuously under tensile stress in the structure, the exposure to rain during
transport of the wires or before construction would be sufficient to initiate corrosion followed by stress corrosion cracking. In
former ductile materials under tension, the hydrogen ions penetrate into the metal grid and the exposure to low corrosive
environment may cause a sudden brittle fracture. Therefore, the production of the high strength Sigma steel 145/160, with
oval cross section (‘‘old type’’), susceptible to stress corrosion cracking was stopped in 1965.

2.4.1. Tensile tests

The mechanical properties of the presstressing wires such as elongation and ultimate strength were tested in tensile tests.
Objective was to exclude or confirm that a deficient material property would be the cause for the roof collapse. The yield
point of 15 prestressing wires extracted from two tendon ducts was not distinct. Thus, the 0.2%-offset yield strength was
determined from the stress–strain curve with values between 1389 MPa and 1475 MPa. The proportional elongation (A10)
was between 5.5% and 9.5%. The corresponding values for proportional elongation yield and ultimate strength in the approv-
als were 6%, 1420 MPa and 1570 MPa. It was assumed that the small deviation from standard values cannot be the cause for
the failure of the prestressing tendons of the roof brim [3]. Table 1 gives an overview on values resulting from tests compared
with limits given by Leonhardt in 1955 [10] and in the technical approvals of the Deutsche Institut für Bautechnik, DIBt, as
far as described in [3].

2.4.2. Fractography
When investigating the fracture surfaces of the prestressing wires using a scanning electron microscope it was possible to
distinguish between failure due to corrosion (brittle fracture) and failure due to static overload during the partial collapse
(ductile fracture). Heavily corroded surfaces showed corrosion products even after cleaning [3].
Fractographic evaluations showed a significant difference between the two mentioned failure modes. While previously
undamaged wires showed a ductile fracture due to overload after necking (local deformation) and rupture because of its duc-
tility, the fracture of brittle material shows a low–grad (brittle) deformation without necking. Brittle fracture is a result of an
unexpectedly sudden failure due to fracture of wires under high tension under aggressive environment, so called stress in-
duced corrosion (stress corrosion cracking). The ductile fracture was identified macroscopically by local necking and micro-
scopically by the dimpled structure of the fracture surface. Contrary, there is almost no plastic deformation of the fracture
surface and no necking detectable after brittle fracture. Fig. 13 shows a surface of a specimen after ductile failure with the
characteristic plastic deformation. Brittle fracture is characterized by no apparent deformation and trans crystalline cracking.
The scan in Fig. 14, here shown with deteriorated grain boundaries, shows a trans crystalline fracture. The investigations in a
scanning electron microscope (SEM) also proved the existence of corrosion on the cracked prestressing wire surfaces. The
cracks of the corroded wires develop along the grain boundaries of the steel microstructure (intergranular). Corrosion caused
deterioration of the grain surfaces, see Fig. 14.

2.4.3. Corrosion protection

Alkaline atmosphere of the pore solution in the cement matrix with a pH-value of approximately 12.6 regularly gives the
steel a passivation and protects the steel surface against corrosion. The anodic reaction to resolve the iron almost completely
stops. Under certain conditions, this protection can get lost. This is the case if carbonation propagates into the concrete. The
pH-value is reduced below 9 and the steel begins to corrode under the presence of moisture and oxygen from the air. As
found in the visual inspection, see Section 2.1, the corrosion of the eight failed tendon ducts was at an advanced stage. This
was very likely caused by missing or insufficient concrete cover layer above the steel in the related areas. Typically, 2 cm
concrete cover was required according to the standards at design, in reality it was not reached in several cases. In these cases,
the corrosion attack may have started already shortly after completing the construction, since the required protection and
passivation of the steel surface was simply not available [3].

2.4.4. Wire protection during construction

For corrosion protection of the wires during transportation, the construction process and prior to grouting, the wires were
painted with zinc oxide. Although theoretically, the zinc oxide could chemically react with the hydrated lime of the grout,
but this factor is assumed to be negligible [3].
116 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119

Fig. 13. Ductile fracture [3].

Fig. 14. Corroded and cleaned surface with deteriorated grain boundaries; inter-granular fracture assumed [3].

3. Discussion

3.1. Root cause analysis

The partial collapse of the Congress Hall roof is a spectacular example for a series of corrosion induced failures of high
strength wires in prestressed concrete structures with complicated detailing. The technology of prestressing concrete struc-
tures by means of wires in tendons was a quite modern construction principle in its early years of application. At the time of
erection, no practical experience about failure mechanisms or long-time durability existed.
Complicated construction detailing and prestressing technology prevented entire by grouting the wires on their lower sur-
face for corrosion protection. Due to the wire disorder in the tendons resulting from the complicated load path, the injected
grouting mass could not reach the full surface coverage of the wire surfaces for anticorrosive protection (see example in Fig. 7).
Because of the disastrous partial roof collapse and the lack of knowledge about stress corrosion cracking, a BAM research
project was initiated to study the increase the understanding and knowledge about hydrogen induced stress corrosion crack-
ing [5]. Finally, this type of prestressing high strength steel (145/160, Sigma-oval, rolled and tempered [11]) has not been
applied after 1965 because of its sensitivity to stress corrosion cracking.
 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119 117

Mandated by the Prosecutor at the Berlin District Court, BAM has carried out detailed investigation of materials taken
from the Berlin Congress Hall after collapse. Materials as the steel and concrete were analysed regarding their possible con-
tribution to the cause of the partial collapse of the roof [3].
Tensile tests of wires without cracks taken from two tendon ducts were carried out at room temperature. The yield and
ultimate strength were not significantly below the required values.
Broken prestressing wires were analysed under the scanning electron microscope. Surfaces from ductile fracture of wires
after tensile test differed from broken wires with corroded surfaces without the typical necking. The fracture behavior has
changed from ductile to brittle. Hydrogen induced stress corrosion cracking was analysed [4] and further investigated [5].
Concrete related issues were published in [3,13].
The concrete quality of the grooves was insufficient. The unique construction of the arched outer roof was analysed and
modelled [6].
For better understanding of the failure mechanisms, the expert opinions [2] were cited. Since then, testing methods for
the investigation of stress corrosion cracking have advanced [5] and the standard for prestressed concrete structures [15]
was developed further. Finally, a new roof structure with clear load path was developed [12].
The cause for the partial collapse was the local concentrated reduction of the load carrying capacity of the arched external
roof. As many prestressed wires as possible failed due to hydrogen induced stress corrosion cracking until the load carrying
capacity of the remaining wires was exceeded. Suddenly they failed with ductile fracture. The final collapse of the Southern
roof followed quite fast. The microscopic analysis of the single prestressing wires allows distinguishing between

– corroded = initially damaged (brittle) and

– non-corroded = overload induced (ductile) fracture.

The corrosion damage was identified to be stress corrosion cracking. Stress corrosion cracking may occur to prestressing
wires under high permanent tensile stresses while being exposed to a crack initiating medium. Additional local tendon bend-
ing at the joint enhanced this process.
The structure was built within a quite short period because of the high pressure to finalize the congress hall before the IBA
1957, already one year postponed. The coincidence of unclear detailing and insufficient execution quality especially in the
ring beam and in the ring beam groove resulted in the stress corrosion cracking, which has finally caused the partial collapse.
The material-characteristic sensitivity of the Sigma steel oval against stress corrosion cracking in the vicinity of humid
microclimate is the third component that may cause the failure of a structure if all three conditions exist at the same time.
Cracks in and leakage of the waterproofing allowed moisture to penetrate the roof structure and reached finally the crit-
ical groove on the top of the ring beam. At the connection of the roof plates between ring beams and the outer roof death load
and deformation of the cantilever caused unexpected high bending forces resulting in early cracks in the concrete. The cracks
finally allowed infiltrating moisture to reach the tendons.
Unexpected deformation in the ring beam due bending and locally bent tendons – more than allowed – loaded the wires
with extremely high stresses. Mortar and concrete in the ring beam groove were porous, sandy and permeable in locations
with broken wires. The corrosion protection in these locations was insufficient and thus the corrosion process must have
been initiated just after finalizing the construction in the 1950 ties. In difference to the original plans, some tendons were
not in the middle axis of the outer roof plates, but they were pressed against the bitumen roofing on the bottom of the ring
beam groove. This reduced execution quality resulted in the fact, that some tendons had insufficient or no concrete cover.
The expected passivation of the tendons by the concrete was thus not given.

3.2. Consequences

Both summarizing expert’s opinions [2,3], came to the same result that the partial collapse was the result of the coinci-
dence of the factors of which none of them alone would have resulted in the same disastrous result.
The partial collapse resulted in discussions about regular maintenance and inspection of public buildings [14]. A further
research project analysed the influence of environmental influences between production and grouting on the susceptibility
of prestressing wires to hydrogen induced stress corrosion cracking [5].
After the analysis of the detailed investigations, the new roof structure was planned as a post-card-identical solution. Con-
dition for the bid was the consequent preservation of all possible structural elements such as supports, tie member in the
ground, pile foundation and the ring beam. The decision was made to keep the inner roof structure above the auditorium
in place and to use it as formwork for the new roof to be built without framework. The architect Professor Bomhardt offered
new simple reinforced concrete shell with a thickness of 11 cm instead of only 7 cm in the former complicated prestressed
roof. The new roof spans over the whole ground as a membrane about 90 cm above the old inner roof above the auditorium.
The new roof is made of light weight elements with reinforcement in the joints having an architectural function, only. A main
problem to solve was the differential deformation of the architectonic (new shell structure) and the functional roof: the old
inner roof still fulfils all functional requirements, e.g. on thermal and noise insolation. This task was solved with an isolating
gap ring between the new roof and the wall of the auditorium (Fig. 15) preventing all weaknesses related to fire engineering,
thermal and acoustic insulation [12].
118 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119

Fig. 15. New roof structure of the former Berlin Congress Hall, reopened to the public in 1987 at the 750th birthday of Berlin as the ‘‘House of the Cultures
of the World’’. The new roof shell spans between the outer ring 90 cm above the preexisting inner roof.

3.3. Initiated research

Analysed failures indicated an early influence of environmental factors affecting the wires between production, during
transport to the construction site and until grouting of the tendon ducts. Numerous possible influences can cause pre-exist-
ing defects that would possibly lead to stress corrosion induced cracking. Basis for the investigation was the former German
Standard DIN 4227 Prestressed concrete, issue 1967 with the related commentaries. Only those construction sites were con-
sidered in the investigations that had fulfilled all requirements of the mentioned standard. At the same time, validity and
practicability of these instructions were screened for the revision of the DIN 4227. Artificially induced corrosion processes
with initial corrosion as observed in practice accompanied the site tests in laboratory durability tests for more information
about the typical corrosion processes.
As the result the prestressing steel types were classified according to their resistance to salty environment. Prestressing
steel A (St 835/1030, d = 26 mm) and E (St 1375/1579, d = 12.2 mm) showed the highest resistance against hydrogen-in-
duced embrittlement. The prestressing steel type B (St 1080/1230, d = 26 mm) and D2 (St 1420/1560, d = 12.2 mm) experi-
enced considerable reduction of the ultimate tensile strength and crack initiation during durability testing. Except nitrate the
other applied ions advanced corrosion. Sulphates, sulphites and rhodanide promote the hydrogen activity [3].

3.4. Impacts on the shell roof design

New types of constructions need discussion and time for the consideration of all eventualities possibly to happen during
the service life of structures. The roof of the Berlin Congress Hall with the wide cantilevering roof built with prestressing
tendons was such a new type of construction. The structure of the inner roof was not critical. The complicated detailing
and estimation of the load path was related to the outer roof. The experiences with the new technology, of the new shell
structure and the knowledge about the durability behavior of the prestressing Sigma steel oval were limited. Last but not
least, the results of the corrosion behavior investigation influenced the actualization of rules as the technical approvals
for prestressing steel by the Centre of Competence in Civil Engineering (DIBT), as the admission office for approving all build-
ing materials and construction procedures. New procedures and new standard test setups were developed for standard test-
ing procedures prestressing wires.

4. Conclusions

Finally, the expert opinions concluded that several causes arose during design, construction and in service and a sum of all
led to the partial collapse of the Southern roof arch. Today we also know that the prestressing steel used at the Berlin Con-
gress Hall roof belongs to the susceptible type of Sigma steel 145/160 oval produced before 1965 that is especially sensitive
to stress corrosion cracking (SCC) [4,5]. The expert opinions summarize the causes for collapse as follows: The collapse of the
Southern external roof was caused by insufficiencies during structural planning, unsatisfactorily executed construction of
the outer roofs and, consequently, to corrosion-induced fractures in the tendons bearing the roof arch [2], cited in [8].
In detail, the following factors contributed to the partial collapse of the outer roof arch:
 R. Helmerich, A. Zunkel / Engineering Failure Analysis 43 (2014) 107–119 119

(1) The new type of the apparent Hypar shell structure was planned and erected in a very short period of about one
year [3].
(2) The structural design followed the architect́s design and resulted in a locally non-transparent load path at the outer
roof connection to the ring beam on the top of the auditorium walls [7].
(3) The execution quality of the ring beam mainly of the groove did not reach the required precision according to the
drawings. The concrete quality was insufficient and caused a reduced strength [3].
(4) The applied Sigma steel was susceptible to stress corrosion cracking [4,5,10].
(5) Environmental conditions and especially the fountain on the Southern side of the congress hall caused a continuously
wet spray [7].
(6) The wet conditions led to wet thermal isolation below the roofing paper [3].
(7) Low concrete quality of the ring wall beam together with tendon bending near the anchorage caused a dense concen-
tration of the wires in the lower part of the tendon, leading to reduced grouting between the wires and thus to too low
passivation and reduced durability [2,3,10].
(8) Wet conditions together with the detailing and the insufficient execution quality finally caused the hydrogen induced
stress corrosion cracking of the tendons [3].

One of the insufficiencies alone would probably never have caused this catastrophic failure, but the coincidence of all was
causing the partial collapse [7]. The new construction was opened to public for the 750th birthday of the city of Berlin.
Many years later, in 2002, international cooperation has started in COST action 543 to increase the durability and to
extend the service life of prestressed structures. RILEM recommendations summarized the outcome of European research.

Acknowledgements

We kindly acknowledge the authors of the publications presenting the results of tests carried out in the laboratories of
BAM between 1980 and 1982, mainly Dr Arno Plank, Dr Werner Struck and Manfred Zschätzsch for summarizing the expert’s
opinion for the Berlin district court in the BAM report 80 [3], Dr Bernd Isecke and Wolfgang Stichel for corrosion related tests
and research [4,5], Dr Jürgen Hundt and Ekkehardt Porzig for materials testing [13], Dr Frank Buchardt, Dr G. Magiera, and Dr
Wolfgang Matthies for modelling and discussion about the load carrying capacity [6]. We acknowledge Professor Jörg Schla-
ich and his co-authors Professor Karl Kordina and Dr Hans-Jürgen Engel, for the detailed expert opinion published in 1980 [2]
as well as for the informative presentation for the Berlin VDI group Construction history together with Professor Bomhardt
and the short informative talk after the presentation in 2013 [7].

References

[1] Rudder S. American modern architecture of the fifties in Berlin, the architect Hugh Stubbins (Amerikanische Moderne der Fünfziger Jahre in Berlin, Der
Architekt Hugh Stubbins). jovis Verlag GmbH; 1997. ISBN 978-3-939633-23-5 [in German].
[2] Schlaich J, Kordina K, Engell H-J. Partial collapse of the Berlin Congress Hall – Causes of the failure, sumarizing expert opinion, (Teileinsturz der Berliner
Kongresshalle – Schadensursachen Zusammenfassendes Gutachten), Beton- und Stahlbetonbau 1980(12):281–94. ISSN 0005-9900 [in German].
[3] Plank A, Struck W, Tzschätzsch M.: Causes of the partial failure of the Congress Hall in Berlin-Tiergarten, (Ursachen des Teileinsturzes der
Kongresshalle in Berlin-Tiergarten). BAM-Research report 80, ISSN 0172–7613, 1981 [in German].
[4] Isecke B. Collapse of the Berlin Congress Hall prestressed concrete roof. Mater Perform 1982;21(12):36–9.
[5] Isecke B, Stichel W. Influence of environmental conditions on the construction site on the corrosion behavior before injection (Einfluss baupraktischer
Umgebungsbedingungen auf das Korrosionsverhalten vor dem Injizieren). BAM-Research report 87. ISBN3-88314-240-9; 1982 [in German].
[6] Buchardt F, Magiera G, Matthees W.: Assessment of the load carrying capacity of the Berlin Congress Hall, (Berechnungen zum Traglastvermögen der
Berliner Kongresshalle). Die Bautechnik; 1982 [in German].
[7] Schlaich J, Bomhardt H. Safety = transparent load path + clear detailling, The Berlin Congress Hall (1957) – concept, realization, partial collapse,
reconstruction, (Sicherheit = transparenter Kraftfluss + saubere Details, Die Berliner Kongreßhalle (1957) – Konzept, Realisierung, Teileinsturz,
Wiederaufbau). Presentation at the Society of German Engineers (VDI) – meeting, Berlin, June 06th; 2013 [in German].
[8] Official website of the House of the Cultures of the World: <http://www.hkw.eu/de/hkw/gebauede/architektur/index.php>, [accessed 18.06.2013].
[10] Leonhardt F. Prestressed concrete for the practice (in German: Spannbeton für die Praxis). Berlin: Wilhelm Ernst & Sohn; 1955.
[11] Structurae.de , J.S. Dorten Arena (Raleigh 1953) (Ernst & Sohn), <http://de.structurae.de/structures/data/index.cfm?ID=s0000395 [accessed
18.06.2013].
[12] Kuehl G, Bomhardt H.: Reconstruction of the Berlin Congress Hall – Structural task and its solution (Wiederaufbau der Kongreßhalle Berlin – Die
Bauaufgabe und ihre Lösung; 1986 http://dx.doi.org/10.1002/best.198600040 [in German].
[13] Hundt J, Porzig E.: Material investigationsof the roof of the Berlin Congress Hall), (Materialtechnische Untersuchungen des Daches der Berliner
Kongresshalle), Die Bautechnik, vol. 59, 8/1082. ISSN 0341-1052; 1982 [in German].
[14] Annual report of the Federal Institute for Materials Research and Testing; 1981. p. 31–41.
[15] DIN 4227:1979. Prestressed concrete (Spannbeton. Ausgabe 1979), ed.; 1979 [in German].
[16] DIN 1164-1958. Portland cement (Portlandzement), now replaced by DIN EN 197 [in German].

Further reading

[9] Ruske W.: 100 Years materials testing in Berlin, A contribution to the history of technology, (100 Jahre Materialprüfung in Berlin, Ein Beitrag zur
Technikgeschichte), issued by BAM, DK 620.1 (430-2.1 091), 1971 [in German].