STRUCTURAL MODEL ANALYSIS
Introduction:
In the field of structural engineering, the use of models has steadily become more and
more prevalent in the recent past. Today model analysis of structures not only is extremely
important as a tool for research and development but also forms an important supplement to the
mathematical methods used in the actual design of structures. A structural model is required to
reproduce an actual structure (the prototype) to some convenient scale and in such a manner that
it is possible to predict the behavior of the prototype subjected to its applied loads from the
behavior of the model subjected to a system of proportional loads. Model analysis may be
employed in the design of a complex structure for which a mathematical analysis is not possible
or may be used to check the validity of a new theoretical method of analysis. In addition, simple
model techniques may be utilized to develop an appreciation of structural behavior in young
engineers. No simplifying approximations are necessary for a model analysis, and all secondary
effects are automatically included in the analysis.
Objectives:
Stress analysis of the model
Determination of stress distribution,
Determination of critical or buckling loads,
Analysis of the characteristics of the normal modes of vibration.
Necessity of Model Analysis:
Model analysis of structural problems encountered in either research or actual design may be
used for one or more of three reasons:
Because mathematical analysis of the problem concerned is virtually impossible;
Because the analysis, though possible, is so complex and tedious that the model analysis
offers an advantageous short-cut; and
Because the importance of the problem is such that verification of the mathematical
solution by model test is warranted.
The stress distribution in an irregularly shaped member may be investigated by use of a
model for the first reason; a model test may serve as the basis for the analysis of a complex
building frame for the second reason; and a model study of the proposed design of a suspension
bridge may come under the third classification.
Design of Models
Whenever a reduced-scale model is used to study an actual structure, it is necessary, of
course, that the model be designed so that full-scale behavior of the prototype may be deduced
from the observations of the behavior of the model. In order that this may be accomplished, it is
necessary that the dimensions of the model and the characteristics of the material used in its
construction bear certain definite relations to the dimensions and material of the prototype. The
principles which govern the relationship between a model and its prototype are called the
principles of similitude. Certain of these principles govern the design of the model, and others
establish the means of extrapolating the results of the model tests to predict the performance of
the prototype.
The choice of the proper material for the construction of models is of great importance.
Not only must the material be such that its structural action is suitable to its use, but also the ease
with which it can be fabricated for a small model must be borne in mind. For many models, the
materials of the prototype may be used. Steel is often used, and reinforced concrete may be
used if the model is sufficiently large. It is often desirable to use a material which has a lower
modulus of elasticity than the material of the prototype so that distortions which are large enough
to be measured accurately may be obtained without the application of forces which are too great.
The use of Duralumin or brass in place of steel is sometimes convenient for this reason. Brass
has the additional advantage that it may be soldered easily, thereby facilitating the construction
of the model.
Celluloid is one of the most widely used materials in the construction of the models used in
conjunction with the more common model methods of stress analysis. The selection of the scale
of a model depends on many factors such as:
the properties of the materials available for its construction;
the capacity of the equipment to be used in loading the model,
the dimensions of the instruments to be used in testing of the model,
the limitations of machinery to be used in fabricating the model, and
the funds and time available for the experimental program.
As the scale of a model is reduced, it becomes increasingly difficult to maintain exact
geometric similarity, and the duplication of all the details of the prototype is physically
impossible. Some details of the design are obviously unimportant and may be omitted from the
model. In other cases, the details of the structural connections have a great influence on the
result, and a large enough scale must be used so that the structural action of the model is
adequate.
Properties of Model Materials
The properties of model materials such as steel, brass, Duralumin, wood, and concrete
are well known and need not be reviewed here. Celluloid is widely used for structural models
and has, in common with certain other plastic materials, certain properties which are not well
known and require further discussion. Celluloid (or cellulose nitrate) has some very desirable
properties as a model material, but it also has some that are very undesirable. It is very readily
machined, has a low modulus of elasticity, is homogeneous, and may be readily welded by using
acetone. On the other hand, its elastic properties change decidedly with age, temperature, and
humidity. More serious still, celluloid creeps under a constant load; that is, if a load is applied,
some 85 per cent of the deformation occurs within a few seconds, but the remaining 15 per cent
takes place more slowly. It would be necessary to wait an appreciable periodabout 15 min
before motion would be essentially complete. Even then small movements would still be
occurring.
METHODS OF MODEL ANALYSIS:
Model analysis may be classified under the headings direct methods and indirect methods.
Direct Method
In this method, the actual structure (prototype) is represented in a reduced form and
loading is applied directly. The elastic deformation of the model is then similar to that of the
prototype, and strain measurements then lead to the required results. Such a method is said to be
a direct method of model stress analysis. Suspension-bridge model studies are usually conducted
in this manner.
Indirect Method
Contrasted to direct methods of model analysis are those methods in which the model is
loaded in a manner which bears no direct relation to the actual loading on the prototype. Such
methods are called indirect methods. Usually such methods involve first finding influence lines
for the model. These results may be extrapolated to the prototype, and the stresses in the
prototype due to the given condition of loading may then be computed from these extrapolated
influence lines. Hence the indirect method is applicable only to linear structures whereas the
direct method is applicable to structures in both the elastic and inelastic states.
Principles of Similitude
The principles of similitude governing the relationships between a model and its
prototype may be determined by either of two approaches. The conditions of similarity may be
expressed in mathematical form, using established laws of structural mechanics, and the
principles of similitude rigorously deduced from them; or the principles may be deduced by
using the methods of dimensional analysis. The first method is generally used for structural
models since the mathematical laws which structures follow are usually well known. However,
in cases where the mathematical laws are not known but the factors affecting the phenomena are
known, the principles of similitude may be determined by the dimensional-analysis approach.
Derivation Using Established Laws of Structural Mechanics
To illustrate the derivation of the principle of similitude by this approach, consider the
problem of extrapolating influence-line data from a model to its prototype. For this purpose,
consider a beam which is statically indeterminate to the first degree, such as shown in Fig.
below. Suppose we wished to obtain the influence lines for the vertical reaction at a, Ra , and the
moment at b, Mb. In either case, we could apply Muller Breslau's principle and thereby obtain the
desired influence lines as indicated in Fig. below. Of course, this procedure could be applied
either to the prototype or to its model. If this procedure were applied to the prototype, the
expressions for the ordinates of these two influence lines would be
where the index P indicates that these quantities refer to the prototype.
On the other hand, if Muller-Breslau's procedure were applied to the model, the ordinates for the
two influence lines for the model would be
where the index M refers to the model.
Of course, if the principles of similarity between model and prototype were known, the
ordinates of these influence lines could be measured experimentally on a small-scale model, and
then these results could be extrapolated to the prototype. In other words, to do this we need to
know the relation between Ra M and Ra p and between Mb M and Mb p. From above equations, it is
apparent that these relationships between the reactions or moments on the model and those on
the prototype depend on the relationship between deflections.
The relationship between deflections of the model and the prototype may be investigated by
considering the computations of such deflections by the method of virtual work. To compute
beam deflections, the law of virtual work may be expressed as
If a force Fp acted at point a of the primary structure of the prototype, as shown in Fig.b, the
vertical deflection of point n could be computed from
In the same manner, to compute a similar deflection on the model due to a force FM at a, or
Suppose that the model had been constructed so that the following relations existed between the
model and prototype:
In view of these relations, it is then apparent that the following additional relations are true for
the statically determinate primary structure shown in Fig. b,
By comparing the above equations, we have
These relationships are true for any of the vertical deflections in Fig. b. Therefore
We will find that
It can be also concluded that
Of course, this discussion has been limited to a beam which is indeterminate to the first
degree, but it could now be extended successively to include beams or frames which were
indeterminate to the second, third, etc., or to any degree. Such considerations would lead to the
following general conclusion for any indeterminate beam or frame, the stress analysis of which
can be carried out satisfactorily by considering only the effect of bending distortion. A model
should be dimensioned so that the axial length of its members is k times those of the prototype;
the moment of inertia of its cross sections is times those of the prototype; and the modulus of
elasticity of the model is times that of the prototype. If this is done, then the ordinates of the
influence line for any reactive force, shear, or axial stress on the prototype are equal to the
corresponding ordinates on the model; but the ordinates of the influence line for any moment on
the prototype are equal to 1/k times the corresponding ordinates on the model. In this manner, the
principles of similitude may be developed for models of trusses and other types of structures,
providing the solution of the problem can be formulated in mathematical form.
Structural similitude using Dimensional Analysis
The relationship between the n fundamental variables affecting structural behavior can be
expressed as a function of (n-i) dimensionless products, where i is the number of dimensions
involved. The dimensionless products are referred to as the pi terms. The variables affecting
structural behavior are given in the table below together with the exponents of their dimensions
of mass, length, and time. Only the first eight of these variables are fundamental; the remainder
has been included in order to derive their interdependence. The fifteen variables contain three
dimensions, and thus the structural behavior can be expressed as the relationship between twelve
pi terms.
The pi terms are obtained by considering the Model analysis exponents of the dimensions of the
product formed by g, W, l, and one other variable at a time. Thus
Any number of the dimensionless products may be combined to give a different expression for
the pi term, as shown by the five different expressions given for 6. For complete similitude,
each variable affecting the behavior of the model and the prototype must be in a fixed ratio,
which is dimensionless and is referred to as the scale factor. In addition, the pi terms of the
model and the prototype must be in a fixed ratio and establish the relationships between the scale
factors. Thus, for geometrical similarity, lengths on the model must be in a fixed ratio to the
corresponding lengths on the prototype and to, the linear scale factor. Considering each pi term
in turn:
When it is known that some variables have a negligible effect on the behavior of a particular
structure, it is unnecessary to fulfill all the above conditions of similitude, and a distorted model
may prove satisfactory and be easier to construct.
The method of dimensional analysis is now illustrated by applying it to the problem of
determining stresses, strains, displacements, or redundant forces or reactions, in any elastic
structure of material obeying Hooke's law. The structure may be specified by the elastic
constants E (Young's modulus) and (Poisson's ratio), by one length a, and the ratios r1, r2 etc.,
of all other required lengths to a. The loads may be specified by one of them P, and the ratios- r1',
r2', etc., of the others to P. The directions may be specified by ratios r1'', r2", etc. A particular
stress component is to be found at some point x, y, z. A formula for this would be a relation
between the quantities:
Only two fundamental units are required for the measurement of all these quantities, and these
can be force and length. From Buckinghams theorem, dimensionless groups can be formed
from the list of symbols to a number equal to the number of symbols minus the number of
fundamental units. In the present problem this means two less and the groups are
Any other dimensionless groups will be merely products of powers of these and so are not
independent. The sense of dimensionless is simply that the numerical value remains the same
when the units are changed. The stress formula must take the form of a relation between these
groups, not merely a relation between the original symbols 1. Thus,
where the right-hand side denotes a function of unknown form, definite in a given problem
although we may not be able to find it. For instance, the beam formula = My/I for a simple
beam of rectangular cross section (width b, depth a, span c) simply supported and loaded by a
central force P becomes, written in the form
where r1= b/a and a is the fiber stress evaluated at a section x from one end, at a depth y below
the neutral axis.
The formula 3 is now applied to both model and prototype. Although the functional form
is unknown, it is the same for both. If, therefore, the groups on the right of formula 3 are made
the same for both, it will follow that the group on the left is also the same for both.Making r1, r2,
, the same for both(prototype and model) means geometrical similarity; making r1', r2', , and
r1'', r2", etc the same for both means an obvious similarity of load distribution. Making x/a, y/a,
z/a, the same means that the stress is to be taken at the point in the model corresponding to the
point in the prototype at which the stress is required. These conditions, of course, are usually
taken for granted, but the method of dimensional analysis, being free from such assumptions,
insures that there are no unnecessary restrictions. Next must be the same for both, if it matters.
Otherwise, it is not required in the list of symbols. Finally P/Ea2 must be the same for both. If the
subscript m is used for the model and p for the prototype, this means
The loads must, therefore, be scaled down according to this rule. We are free to make the model
of a different material. The change is allowed for in the factor Em /Ep, but Poisson's ratio must
be kept the same if it is significant. All dimensionless groups on the right of formula 3 being now
the same for model and prototype, it follows that the group on the left is also the same, that is
Substituting from above equation for Pm/Pp the equation becomes
from which p may be calculated when m has been measured.
The analysis is carried out in a similar way for quantities other than stress. For a displacement u,
the dimensionless group u/a is used instead of a2/P. For a strain e, e itself may be used, and for
a force R, the ratio R/P. Then instead of formula 3 we should find
Where f 2, f3 denote other functional forms.
The similarity rules up to and including equation 4 must again be observed, and it then follows
that
