Model analysis for structural safety and

optimization

Autor(en): Oberti, Guido

Objekttyp: Article

Zeitschrift: IABSE congress report = Rapport du congrès AIPC = IVBH

Kongressbericht

Band (Jahr): 8 (1968)

PDF erstellt am: 18.08.2018

Persistenter Link: http://doi.org/10.5169/seals-8731

Nutzungsbedingungen
Die ETH-Bibliothek ist Anbieterin der digitalisierten Zeitschriften. Sie besitzt keine Urheberrechte an
den Inhalten der Zeitschriften. Die Rechte liegen in der Regel bei den Herausgebern.
Die auf der Plattform e-periodica veröffentlichten Dokumente stehen für nicht-kommerzielle Zwecke in
Lehre und Forschung sowie für die private Nutzung frei zur Verfügung. Einzelne Dateien oder
Ausdrucke aus diesem Angebot können zusammen mit diesen Nutzungsbedingungen und den
korrekten Herkunftsbezeichnungen weitergegeben werden.
Das Veröffentlichen von Bildern in Print- und Online-Publikationen ist nur mit vorheriger Genehmigung
der Rechteinhaber erlaubt. Die systematische Speicherung von Teilen des elektronischen Angebots
auf anderen Servern bedarf ebenfalls des schriftlichen Einverständnisses der Rechteinhaber.

Haftungsausschluss
Alle Angaben erfolgen ohne Gewähr für Vollständigkeit oder Richtigkeit. Es wird keine Haftung
übernommen für Schäden durch die Verwendung von Informationen aus diesem Online-Angebot oder
durch das Fehlen von Informationen. Dies gilt auch für Inhalte Dritter, die über dieses Angebot
zugänglich sind.

Ein Dienst der ETH-Bibliothek

ETH Zürich, Rämistrasse 101, 8092 Zürich, Schweiz, www.library.ethz.ch

http://www.e-periodica.ch
la

DISCUSSION PREPAREE / VORBEREITETE DISKUSSION / PREPARED DISCUSSION

Model Analysis for Structural Safety and Optimization

Analyse sur modeles de la securite et de l'optimisation des structures

Modelluntersuchung der Bausicherheit und -Optimierung

GUIDO OBERTI
Prof.
Italy

i. - Foreword-
a) In a short note presented at the Rio de Janeiro
Congress of the IABSE in 1964 I stated that the possibility
of analyzing on modeis, even V3 failure,of large structures,
particularly piain or reinforced concrete structures, has
long been proved by nie in a great number of cases.
In fact, a model study under elastic conditions
furnishes the values of the prototype stresses under
working load, which is important for several reasons«
Firstly, the results obtained, unaffected by the assumptions
and limitations which impair the classical methods of
calculation, can profitably be compared with those supplied
by these methods. Secondly, it is not hard to solve on
modeis unusual three-dimensional problems, contrary to what
is the case with the conventional analytical procedures
both because of extreme complexity (only partly reduced by
the finite element method) and difficult mathematical
schematization of accurate boundary conditions.
Extension beyond the elastic ränge is still always
invaluable to the structural engineer as may enable himit
to locate possible weak points in the design and thus ausist in
securing greater safety and optimization.
Models may be classified in elastic (tested within the
elastic ränge only), structural (carried to failure) arid
geomechanical (when the foundation influences the structural
Performance).
Precent trends, bascd on experience at ISMES, are:
increasing emphasis on structural modeis;
constant improvement of model materials to better suit
the aims pursued;
4 la
- MODEL ANALYSIS FOR STRUCTURAL SAFETY
growing interest in thermal stress investigation,
especially for concrete dams and reinforced
(prestressed or not) concrete vessels of nuclear
reactors;
dynamic testing on large shake tables and marked
concern for earthguake effects,
b) Theme I
has been treated by the general reporters
prof. A. Freudenthal and prof. J.Courbon.
M.
In a first theoretical and critical paper regarding
topic la, prof. Freudenthal deals with the evaluation of
overall structural safety based on probabilistic criteria
related to the operating loads, which seems fit for
statically determinate structures only. In a second paper
concerning topic lb and also of a theoretical probabilistic
nature, the same author discusses the possibility of
predicting ultimate safety based on the physical properties
of the materials and their influence at failure. None of
the papers mentions structural model analysis.
Prof. Ccurbon's paper treats with topic lc of Theme I.
It concludes by mentioning, all too briefly, the great
Services rendered by model studies in the design of dams,
thin shells and shields of nuclear reactors.
I, therefore, believe it of use tc outline, the
present-day possibilities of model analysis in evaluating
the safety degree of large statically indeterminate
structures.
d) Model investigation primarily concems statically
highly indeterminate structures and may be regarded as:
I) a modern method of stress analysis;
II)a tool for failure load evaluation.
In any case, it is possible to consider or predict the
Statistical disPersi°n °f the operating loads and of the
structural resistance of the prototype material.
In case II), when several modeis are tested, it is
possible to evaluate the ultimate carrying capacity R of
the structure for each type of load S, so that the model
functions as a tool for determining the overall safety
factor V •
This factor may vary for each type of structure,
depending on the probabilistic possibility assumed for the
operating loads and the structural resistance, associated
with a definite risk of failure,
Thus,for concrete dams, the loads are practically known
(excepting those for earthguake-resistant^ design), and the
uncertainties about the concrete resistance are quite small,
The highly redundant type of these structures generally
reduces the importance of the concrete strength dispersion,
The safety factor in this case, therefore, serves rather as
a coefficient of security against the insecurity of the
analytical results, especially in relation to the real
properties of the rock foundation.
GUIDO OBERTI 5

2.- Actual possibilities of model analysis.

a) Elastic modeis are based on linear elasticity
(Hocke's law) and, hence, a superposition of effects is
allowed, They also permit to proportionately modify the
loadings so as to obtain the most suitable testing
conditions. In parti,cular, it is possible to operate
at strains that are amplified with respect to these
required by similitude (which demands that the strains
in the prototype and in the model be the same).
Elastic modeis, widely used in "stress-analysis",-
may be divided in two groups.
The first group concerns plane elastic structures,and
for them the deformeter, photoelastic and Moire methods
are predominant,
Deformeters are based on the well-known reeiprocal
theorems (Maxwell, Betti, Müller-Breslau). Photoelasticity
is a first-rate research method, most used in structural
engineering laboratories. The Moire method is primarily
used in flat slab investigations.
However, it
should be observed that the importance
of these methods has lately decreased due to the use of
Computers in solving problems relating to plane elastic
structures.
The se:ccndgroup deals with three-dimensional modeis.
In statical tests the loading equipment is usually made
up of calibrated weights or hydraulic jacks, the pistons
of which react against an external rigid frame; the loads
are applied to the model through wooden cork-soled pads.
Strain gages, ordinarlly applied to the surface of the
model, are used for measuring the direction and magnitude
of the principal strains.

Young's modulus and Poisson's ratio of the model

material are determined as usual, the former by tensile
and flexural tests and the latter by torsional tests, The
material may quite differ from that of the prototype,
provided it obeys Hooke's law and its Poisson's ratio is
similar to that of the structure. The model then functions as
a "stress Computer",and its results may be compared with
the theoretical ones.
For elastic mcdels, ISMES has recently sueeeeded in
using epoxy resins mixed with various aggregates.They permit
obtaining a wide ränge of elastic moduli in accordance with
the requirements of each case, and stress-strain
relationships that are similar even when the stresses are
high.
b) Structural modeis are best made of the same material
as the prototype. This is generally possible for steel or
prestressed concrete structures when suitable scale
(1:4-1:?0) modeis are used. But for very large structures,
such as concrete dams, we are forced, also for economic
6 la - MODEL ANALYSIS FOR STRUCTURAL SAFETY
reasons, to adopt greatly reduced scales (1:30-1:100) and
hence to use model materials whose mechanical
characteristics are reduced compared with those of
concrete in accordance with similitude requirements,
For structural modeis, I have long since used special
materials simulating the mechanical properties of concrete,
by introducing the technique cf 'wet mcdels" (with a
waterproof coating) practically free of internal stresses.
The tests are then divided into two successive stages.
In the firststcge,;called "normal load tests", the
deformations are investigated for values close to
similitude conditions (that is, e e') under loads
corresponding to those of the structure in operation»0\,
The second stage concerns ultimate load tests and the
transition to them is gradual. The ratio cf the highest
actually supported load to that of the design load is
generally assumed as overall "factor of safety".
This ratio can easily be referred to all the operating
loads equally or differently increased following a
probabilistic coefficient applied tc each independent load.
In the case of statically high indeterminate structures it
differs frcm the classical ratio cf ultimate Lo working
stress, and its meaning is greater since it takes into
account the bi- and triaxial strength of the material under
stresöes in different directions and the plastic-adjustment.
One can by expedients increase on the model solely the
loads which in the prototype may rise through extraordinary
action. Such are wind load for skyscrapers and water
pressure for dams. The horizontal loads alone may undergo
increases of consequence for the stability of these
structures. In setting up a model study it is, therefore,
cf basic importance that the factor of safety shall be
evaluated as simultaneously affected by:
- loads having a fixed value (dead load);
- loads which may increase with respect tc their ordinary
value (wind effect);
- actions the occurrence of which is only probable
(earthqaake).
In practice, when the so-called "weight" cf each of
the above phenomena has beon established, one can obtain
(°) It is advisable to secure, through repeated loading
cycles, non-elastic displacements (settlement of the
foundation, adjustment and opening of joints,
localized plasticity) which are likely to occur since
the first loading in order to obtain an elastic and
uniform model Performance fit
for repeated measurements
.,...¦! -cntrols. Thir remits cbtai1-.*ng the stresses,
displacements and structural behavior of the prototype
under workir-g conditions.
GUIDO OBERTI

the factor of safety by experimentally increasing all the

loads up to the failure of the model, considering the
"weight" corresponding to each type of load
Therefore, not only cn© but a number cf faotors cf
safety can be secured, each of which corresponds to a
given set of phenomena the influence of which is to be
analyzed.
c, Geomechanical modeis investigate structures
on foundations whose equilibrium conditions
may affect the sarety of the structures, as is the case of
dams, large bridges and power or highway tunnels,
The stability of block foundations has lately beer
simulated and studied on geomechanical modeis the
characteristics of which had conveniently been schematized
on the basis of gecgnostic tests.
It may also be pointed cut that in-situ and laboratory
investigations cf the geomechanical features cf the reck
and soil mass are increasingly used and recemmended as an
aid tc model studies,
The modeis, therefore, must faithfully simulate the
rock and soil conditions and its mechanical preperties,
The tests are usually carried to failure.
These investigations are tc be considered as basic
when extensive discontinuities (faults, cavities) or a
prenouneed anisotrepy (stratificatiens and diaclases) are
present in the rock mass, especially when sliding er
least-resistance planes may develop or, more generally,
when large low-strength block formations are involved.
In these modeis, cohesion and angle of friction must
also be faithfully reproduced. The difficulty encountered
in establishing the true values of the angles of friction
makes it in the rnodeling conservative to assume reduced
values wnich are still within the approximation allcwed by
field tests.
strains have to be the same
The prototype and model
and, therefore, the scale ratio must be reduced. The model
materials then shall have high densities and low mechanical
properties (i.e.,very low moduli of elasticity, yield-point
and ultimate loads) in order to ccmply with similitude.

3.- Ässessment of (structural) safety at the desiem stage.

The adeption of model techniques is firstly cf
considerable importance at the design stage of structures,
especially if these are statically highly indeterminate.
Structural safety can then be evaluated by modern
probabilistic criteria as suggested in the Freudenthal
report when:
the expected Statistical dispersion of the loads, i.e.
of the external or operating forces S, is taken into
consideration by determining the dimensions of the
8 la - MODEL ANALYSIS FOR STRUCTURAL SAFETY
prototype on the basis of a force ys«S (where Vs,
the load safety factor, is >1) and adopting equivalent
working forces in the model;
- the Statistical dispersion of the strength dr of the
prototype material is taken into account by assuming an
ultimate load, or a yield point, equal to Or/Vr (with
Yj>, the rupture safety factor, > 1) and comparing the
hignest internal stresses furnished by the model at that
value.
The model then becomes a very efficient tool for a
"structural analysis".
Äs typical examples I shall mention:
- the statical and dynamic investigations carried out on
an elastic model of the Polcevera viaduct, of the
Maracaibo bridge type, designed by prcf. R. Morandi
(fig. l);
- the far more elaborate analysis, made particularly on a
structural model carried to failure, of the new San
Francisco Cathedral designed by prof. P.L.NervK figs.2,3);
- the study cf the safety degree of the Kurobe IV Dam and
its foundation (figs. 4, 5, 6).

1 j
<Z* Jü
%
l-^THH:
¦

«Uli

Fig. 1 Polcevera viaduct, Italy. Elastic model:

Scale 1:50
GUIDO OBERTI

All n *L<
Fig. 2 SanFrancisco Cathedral, U.S.A.
General view of model. Scale 1:15,

f *

•
fe v ¦«-

I
4*1 !-*"
¦::

fl
Fig. 3 San Francisco Cathedral.
Structural model under failure tests,
10 la - MODEL ANALYSIS FOR STRUCTURAL SAFETY

*iV

3^

!£
» * »»

CHARÄCTEPt5riC_ DATA FCTT ACTUAL FAULTS

rMt THOMV.
RIGHT B. *
IKIO KIUJ1 er
¦"U
lEFT
Tffl
EANK
2B5.m&?
ZOO ICO oE no ¦goir'
Fl fWO 200 25= wo _T? Oü) 0.LO TD°
FW Oü) 25° i OU) CiO w
F5 040 200 25= 0 0«3 043 10°
F 1 OK 200 25= rao ¦

200 25°
fQ
FB 075 200 25° TO ¦

Pig.4- Kurobe IY Dam, Japan.

General Layout of geomechanical model.
iTOOSWIC WO
u>c j. oiMT qinmt y. *0"»»lt _LO*p GH..T-*ftj1 Ct»IW3

• awooon owa Ktxrt

nuj
r-
a&\jcn<nn
LENGTHS
nun aw«

AHCH *T EL U» r

Pig.5 Kurobe IV Dam. Displacements recorded at

a horizontal section (el. 1320 m).

Pig.6 Kurobe IV Dal. Structural model at failure,

showing the opening of joints.
11
GUIDO OBERTI

Evaluation of the safety degree cf an alreadv existing

structure.
Mcdels may be of great assistanr-o not only when
structures are being designed but also when the stability
and safety degree of er^"(-ed wcrks are being checked. This
is particularly f^-~ when verifying large structures which
have undergor- statical conditions unpredicted or
unprer11--' ---^-Le at the design stage.
I shall briefly illustrate some examples in
Here, toc,
which testing cn mcdels has yielded highly significant and
conclusive results, with particular regard to the safety
degree of the structure«
After a few years of Operation, extensive subhorizontal
microcracks were found at the upstream face cf a large
arch-gravity dam completed in 1958. The influence of these
cracks on the structural Performance and safety of the dam
at füll reservoir has been investigated cn a large structural
mpdel in which the number and pattern of the microcracks
had faithfully been reproduced (figs. 7, 8).
Interesting tests were also conducted on a 1:4 scale
model tc verify the compression safety degree of the main
columns of the Cathedral in Milan (fig. 9). The two materials
(Candcglio and Serizzc mnrbles), of highly different moduli
of elasticity, and the geometry of the individual blocks
were identical with those of the prototype (fig, 10). The
pattern of the stresses in the masonry dorne carrying the
main spire of the Cathedral has then been anaJyzed on a
large elastic model (fig, 11),
The effect of the horizontally stratified bedrock
anisotropy cn the stability of a recently constructed dam
was investigated by means of geomechanical modeis. The
various expedients devised tc raise the safety degree of the
dam-foundation unit were also examined (fig, 12).
Finally, the model tests carried out for the
double-curvature arch Vajont Dam should be mentioned. As is
known, this dam has brilliantly withstood the extraordinary
sliding of Mount Toc into the partly filled reservoir and is
new sustaining the enormous aSymmetrie mass of slide
material (fig, "13). After the di säst er, model studies were
conducted to determine the safety degree of this imposing
structure under the present exceptional live load (fig. 14).
12 la
- MODEL ANALYSIS FOR STRUCTURAL SAFETY

~:--,-

I
\ />

l camti. j)ij»iroB «sncc er r>C OWOtS

Fig. 7 Flumendosa Dam, Italy. Microcracks on upstream

face of model.

:u
a j

Fig. 8 Flumendosa Dam. Downstream view of model

under test.
GUIDO OBERTI 13

— —T -.
I STRATO i

ti.
n STRATO i

fryy^ IV 3TM.TO •
m STRATO
m STWTO V

«& I
D STftATO

I STtftTO
«

ti
IV STtpO i
Hl 3TUTO B
STMTO 1

I
i snucro >
V
iv vnuTo t
m VTWLTO V
IV STRATO
n tSTVftTO •
I kTHTO j c

\—^—. r

Fig. 10 Milan Cathedral.Cross-sections

of columns.

-^t;^!

I
ffl

Fig. 9 Milan Cathedral. Model of column

under test. Scale 1:4.
14 la - MODEL ANALYSIS FOR STRUCTURAL SAFETY

nl
i il

+*
E

Fig. 11 Milan Cathedral. Elastic model of masonry dorne.

¦
,;.

»"?»»»-.

^m
F ig. 12 Mequinenza Dam, Spain. Plane model on
geomechanical foundation.
GUIDO OBERTI 15

Fig. 13 Vajont Dam. Asymmetrie slide material

acting on upstream face.

^ i
U
I

Fig. 14 Vajont Dam. Model under

asymmetric load test.
16 la - MODEL ANALYSIS FOR STRUCTURAL SAFETY
5. - C'onclusions.
It is believed that the above brief outline has
clearly shcwn the contribution given, and which can be
given, by testing on modeis when evaluating the safety
degree of a structure at its design stage and after
ccnstruction.
When it is assumed, in accordance with modern trends,
that a rational determination cf safety involves the
adoption cf an acceptable risk cf failure, a design
procedure for uniform safety, and hence optimization, can
be based on structural model investigation.
Finally, the arduous problem of structural reliability
of statically indeterminate structures, related to the
failure mechanisms depending on the ccnsecutive loads
mentioned also by prof. Freudenthal at the end cf his
paper, can satisfactorily be solved through a judicious
adopticn of the present model test technique.
SUMMARY
After a short introduction the paper outlines the
actual possibilities of evaluating the safety degree of
a structure by testing elastic, structural and geomechanical
models.
The evaluation may concern: 1)structural safety at the
design stage; 2) safety degree of an existent structure and
cf one operating under extraordinary conditions.
The importance of model investigation particularly for
the optimization of statically highly indeterminate
structures is then emphasized.
RESUME
Apres quelques mots d'intreduction le rapport sousligne
les possibilites actuelles donnees par les differents types
de modeles(elastigues,structureaux,gecmechanigues)pour
l'analvse de la securite dea grandes structures.
On considere apres:1) l'examen du coefficient de
securite dans la phase du projet de l'ouvrage; 2)l'evaluatior
du degre de securite d'un ouvrage dejä acheve ou soumis ä
des actions exceptionelles.
Le rapport termine en souslignont les possibilites des
modeles surtout pour l'etude et l'optimization des structures
hautement hyperstatiques.
ZUSAMMENFASSUNG
Nach einer kurzen Einleitung werden die wirklichen
Möglichkeiten einer Untersuchung des Sicherheitskoeffizienten
eines Bauwerkes an elastischen, strukturellen und
geomechanisehen Modellen beschrieben.
Der untersuchte Sicherheitskoeffizient kann sich auf den
Entwurf, \ein bestehendes oder ein unter ausserordentlichen
Verhältnissen befindliches Werk beziehen.
Die Wichtigkeit der Modelluntersuchungen für die
Optimisierung statisch hochunbestimmter Werke wird nachdem
besonders unterstri^chen.