Decision Support to Engineering Analysis Process
Bojan Dolšak and Marina Novak
Faculty of Mechanical Engineering, University of Maribor, Smetanova ul. 17, SI-2000 Maribor, Slovenia
Keywords: Computer-aided Design, Analysis-based Design Optimisation, Finite Element Analysis, Intelligent Decision
Abstract: Analytical aids represent a group of the most widely applied intelligent computer systems for supporting
design process. Such systems capture the expertise of a specialist in the application of a design technique
for instance in the development of an analytical model, in the forming of assumptions or in the
interpretation of results. An example discussed here is an "expert" aid to be applied within structural
analysis using finite element method. The system is constituted from three modules, one for finite element
selection, the other for finite element mesh design and the third one for results‟ interpretation.
Analysis-based design optimisation is an integrated
part of the design process for many components.
Moreover, computer aided structural analyses are so
extensively applied within design process that
analysts are no longer the only specialists dealing
with this issue. Designers have to carry out different
types of analyses very frequently themselves.
However, the existing conventional computer
aided design (CAD) tools are not adequate as a
proper aid to be used by designer in the process of
analysing a new product. Instead of being oriented
particularly in mathematical aspects of the analysis,
they should provide a continual stream of advice and
information to assist in decision-making. For
example, CAD system can be used as a powerful
computer graphic tool for developing an idealised
model for the analysis, but it would give no advice
on what type or density of idealisation is appropriate
for the particular design case. For this reason, the
quality, effectiveness and reliability of the structural
analysis-based design optimisation still depends
mostly on the level of designers‟ knowledge and
The way in which it is hoped to get more
intelligent computer support to structural analyses-
based design optimisation is to increase the
intelligence of the existing CAD systems. In order to
do that, some intelligent modules need to be
developed and integrated into the analysis process.
Analytical aids represent a group of the most widely
applied intelligent computer systems for supporting
design process.
Such systems capture the expertise of a specialist
in the application of a design technique for
instance in the development of an analytical model,
in the forming of assumptions or in the interpretation
of results.
An example discussed here is an "expert" aid to
be applied within structural analysis using finite
element method. The system is constituted from
three modules, one for finite element selection, the
other for finite element mesh design and the third
one for results‟ interpretation.
The purpose of structural design analysis is to
simulate and verify the conditions in the structure, as
they will appear during its operational life. Physical
and mathematical modelling simulations are
computationally intensive but offer immense insight
into developing product. The results of structural
engineering analysis are often basic parameters for
design optimisation process.
The analysis‟ results can confirm the design
candidate, but this is very rare at the first attempt.
Mostly, the results show that the structure is under-
sak B. and Novak M. (2009).
INTELLIGENT ANALYTICAL AIDS IN DESIGN - Decision Support to Engineering Analysis Process.
In Proceedings of the International Conference on Knowledge Engineering and Ontology Development, pages 129-134
or over-dimensioned. An under-dimensioned design
needs to be improved and changed, because it will
most certainly break. On the other hand, design
changes are unnecessary for over-dimensioned
structures, although they should be carried out
unless the improvement / optimisation process is
financially unsound, costing more money than the
customers are prepared to pay. In that case, the
calculated displacements and stresses in the structure
simply need to be within the allowable limits. Where
design improvement is needed, certain design
changes should be applied, such as the use of
another material or geometry modification.
Nowadays, the results of the structural analyses
are usually very-well presented. The analysing
software is very helpful at this point, as it offers
adequate computer-graphic support in terms of
reasonably clear pictures showing the distribution of
unknown parameters (for example: stresses,
deformations, temperatures) inside the body of the
The definition of the design material also defines
the allowable values for stresses / deformations /
temperatures, which should be specified by the
materials supplier. When the computed values
exceed the allowable limits in some critical areas,
design improvement is necessary. Thus, when the
analysis is concluded many questions need to be
answered in order to come to the correct decisions.
However, a lot of knowledge and experience is
needed, first to prepare the correct idealised model
for numerical analysis, and second to be able to
understand the results of the analysis and to choose
necessary design optimisation steps (Ong and
Keane, 2002).
Young inexperienced engineers often do not
understand basic principles of the structural design
analysis and make wrong conclusions quite
frequently. They even have problem to see whether
the results are within expected limits considering the
original problem. In most of such cases, the existing
software cannot help them, as the support provided
by geometry-based CAD systems is limited, mainly
because of the wide semantic gap between
geometry‟s expressive power and the abstract
features of a product (Mili, Shen, Martinez, Noel,
Ram and Zouras, 2001).
Thus, very extensive, time consuming and
expensive analysis often become meaningless.
Moreover, as a consequence of wrong problem
definition or miss-interpretation of the results of the
analysis, the structure may even break down during
its exploitation. The experiences gained by many
design iterations are of crucial importance. When
considering this, an intelligent support is needed in
this phase of the design process. In order to provide
this support, the knowledge and experiences need to
be encoded in an intelligent advisory system to help
the designer to perform analysis-based design
optimisation process.
Finite element method is the most frequently used
numerical method to analyse stresses and
deformations in physical structures (Zienkiewicz,
Taylor and Shu, 2005). Finite element analysis
(FEA) is divided into three phases:
pre-processing phase of the analysis
matrix calculation
post-processing phase of the analysis
Usually, FEA software is divided into three
modules, each dedicated to one phase of the
analysis. The reliability and accuracy of the results
are strongly related to the overall quality of the
analysis process. Thus, every phase of the analysis is
important and has to be performed carefully and
3.1 Pre-processing Phase
of the Analysis
In the pre-processing phase of the analysis, the real
structure has to be idealized with the appropriate
mesh model that ensures low approximation errors
and avoids unnecessary computational overheads.
For that purpose, the correct finite elements need to
be selected for the analysis and the appropriate
density of the mesh needs to be defined.
Whilst many FEA pre-processors will
automatically create the finite element mesh, such
automatic creation still requires data, such as the
type of the elements, the mesh density and the
position and type of loads and boundary conditions
to be applied.
The selection of finite elements is strongly
related with the meshing task. The quality of the
results derived by using inadequate type of element
for a certain problem is usually very poor. The main
differences between the elements are related to the
basic polynomial approach and to the geometry of
an element (Rieg and Koch, 2001)
3.2 Matrix Computation
After setting up the loads and boundary conditions,
the matrix computation is performed to solve a
KEOD 2009 - International Conference on Knowledge Engineering and Ontology Development
system of linear algebraic equations determining
displacements (and stresses) in the nodal points of
the computational model. This phase of the analysis
is the most optimised and the most user independent.
3.3 Post-processing Phase
of the Analysis
In the post-processing phase of the analysis,
numerical results have to be examined and correctly
interpreted. Post-processing phase represents a
synthesis of the whole analysis and is therefore of
special importance. It concludes with the final report
of the analysis, where the results are quantified and
evaluated with respect to the next design steps that
have to follow FEA in order to find an optimal
design solution.
In spite the fact that the results are pretty well
ordered, the numerical figures are hard to be
followed in case of complex real-life problem, when
the data file is usually both, complex and extensive.
FEA software offers an adequate computer graphics
support in terms of reasonably clear pictures
showing a distribution of unknown parameters
inside the body of the structure. However, the user
still has to answer many questions and solve many
dilemmas in order to conclude the analysis
The application of artificial intelligence (AI) to
design is generally concerned with studying how
designers apply human intelligence to design, and
with trying to make computer aids to design more
knowledgeable. The part of AI that is particularly
concerned with the development of such
representations is known as decision support
intelligent systems (Turban, Aronson and Liang,
Although the AI technology is still a subject of
extensive research, many successful AI applications
in reallife domains already proved the usefulness of
these technologies when dealing with problems that
are nondeterministic and as such cannot be treated
adequately by using conventional approaches, unless
the user is possessed of special skills and experience.
It is becoming increasingly evident that adding the
intelligence to the existing computer aids, such as
CAD systems, leads to significant improvements of
the effectiveness and reliability in performing
various engineering tasks, including design. In this
context, structural analyses are of special
importance, representing a crucial part of the
modern design optimisation cycle.
Actually, AI applications to design improvement
process are reality and subject of intensive
development and implementations. Proceedings of
the international scientific conferences “AI in
Design”, edited by J.S. Gero, constitute a good
collection of papers related to this area (Gero, 2002),
and some more recent developments can also be
found in (Clarkson and Eckert, 2005).
Intelligent computer support to design may be
classified into four broad groups, as follows:
guiding inexperienced users;
automated design of particular products;
intelligent analytical aids;
intelligent “design for X”.
In this paper, our interest is oriented in the
intelligent analytical aids for supporting FEA
It order to support the designer in overcoming three
major FEA bottlenecks (finite elements selection,
finite element mesh design, and results
interpretation) three separate stand-alone intelligent
analytical aids (one for each task) are being
developed in our laboratory.
First two KB modules are meant to be applied in
the pre-processing phase of the analysis, when mesh
parameters need to be set. The third KB module
should support the post-processing phase of the
In continuation, basic architecture of the KB
modules that should provide intelligent aid to FEA-
based design optimisation process is presented.
5.1 Intelligent FE Type Selection
Nowadays FEA software tools offer to the user a
wide range of different, but often also very similar
elements. Even the elements that are meant to be
used for the same generic type of analysis may have
different geometric shape and polynomial function.
Thus, selection of the most appropriate type of the
elements to be used for certain analysis is a complex
task that requires a lot of knowledge and experience.
INTELLIGENT ANALYTICAL AIDS IN DESIGN - Decision Support to Engineering Analysis Process
Figure 1: Knowledge-based finite elements selection.
Figure 2: Knowledge-based finite element mesh design.
Most novice designers need advice to select the
correct type of elements that ensures quality results
at reasonable consumption of computing resources.
Figure 1 presents basic idea for the KB support
to the pre-processing phase of the analysis. The
proposed scheme has been realised by development
of the KB system named Z88FESES (Novak, Rieg,
Dolšak and Hackenschmidt, 2006), which is
adjusted to the freeware finite element analysis
program Z88 (Rieg, 2006).
The knowledge base comprises 24 data-driven
production rules that are applied to select the
appropriate finite element type out of the list of 20
different types that are available in the current
version of Z88 program. The most appropriate type
of finite elements is proposed by the system
considering problem description given by the user,
who needs to answer some questions interactively.
In current version of the system the selection of
the most appropriate finite elements type to be used
for the analysis is based on the following selection
criteria: space dimension, dimension of the structure,
cross-section (only for beams), the expected quality
of the results, geometry, and loading case
complexity. The way of adding the relation between
the element type and the mesh density into KB finite
element type selection process is still a subject of
The knowledge base was constructed manually
after some interviews of the human experts. The Z88
user manual was also used as a source to construct
the rules, as it contains a detail description of all
available elements. A thorough presentation of the
knowledge base can be found in (Novak, Rieg,
Dolšak and Hackenschmidt, 2006).
5.2 Intelligent FE Mesh Design
Within finite element analysis usually a few
different mesh models need to be created until the
right one is found. The trouble is that each mesh has
to be analysed, since the next mesh is generated with
respect to the results derived from the previous one.
There is no clear and satisfactory formalisation
of the mesh design know-how. Finite element design
is still a mixture of art and experience, which is hard
to describe explicitly. Defining the appropriate
geometric mesh model that ensures low
approximation errors and avoids unnecessary
computational overheads is still very difficult and
time-consuming task.
As alternative to the conventional “trial-and-fail”
approach to this problem, we have developed the
intelligent computer system named FEMDES
(Dolšak, 2002). The system was designed to help the
user to define the most appropriate density and
pattern for the finite element mesh model. The
KEOD 2009 - International Conference on Knowledge Engineering and Ontology Development
Figure 3: Knowledge-based analysis results‟ interpretation.
system application enables the designer to conduct
the finite element mesh model easier, faster, and
more experience independent.
For this system the knowledge base was
constructed by using inductive logic programming
algorithm Golem (Dolšak, Bratko and Jezernik,
1998). Machine learning techniques were used on
numerous examples to develop more than 1700
classification rules.
Figure 2 shows the idea of this KB module
application within the pre-processing phase of the
analysis. In any case, the user has to define the
problem (geometry, loads, and supports). The data
about the problem need to be converted from the
FEA pre-processor format into symbolic qualitative
description to be used by the KB module.
The task of the intelligent system is to determine
the appropriate mesh resolution values. A command
file for the mesh generator can be constructed
according to the results obtained by the intelligent
5.3 Intelligent Aid for Analysis Results’
When numerical part of the engineering analysis is
finished, designer has to be able to judge, whether
the results of the analysis are correct and reliable,
and decide what kind of design changes are needed,
if any.
Most of the users need “intelligent” advice to
perform the results interpretation adequately
(Pinfold and Chapman, 2004). Unfortunately, this
kind of help cannot be expected from the present
software. The traditional systems are rather
concentrated on numerical aspects of the analysis
and are not successful in integrating the numerical
parts with human expertise.
In order to support this crucial phase of the
analysis-based design optimisation, a prototype of
the intelligent consultative system PROPOSE has
been developed (Novak and Dolšak, 2006).
PROPOSE provides a list of redesign
recommendations that should be considered to
optimise a certain critical area within the structure,
considering the results of a prior stress/strain or
thermal analysis.
As a rule, there are several redesign steps
possible for design improvement. The selection of
one or more redesign steps that should be performed
in a certain case depends on the requirements,
possibilities and on requests.
Figure 3 presents a basic idea for the KB
analysis results‟ interpretation. The user has to
define design problem and present the results of the
engineering analysis. In addition, critical areas
within the structure need to be qualitatively
described to the system. These input data are then
compared with the rules in the knowledge base and
the most appropriate redesign changes are
determined and recommended to the user.
The qualitative description of the problem area
should be as common as possible to cover the
majority of the problem areas, instead of addressing
only very specific products. In cases when the
problem area can be described to the system in
different ways, it is advisable to run the system
several times, every time with different description.
Thus, the system will be able to propose more
design actions, at the expense of only a few more
minutes at the console.
At the end, the user can get the explanation how
the proposed redesign changes were selected as well
as some further guidelines how to implement a
certain redesign proposal.
The knowledge base of the PROPOSE system is
quite complex and was constructed by using
different approaches for knowledge acquisition,
including experts‟ interviews, study of literature and
some project elaborations, etc. Development of the
knowledge base is described in detail in (Novak and
Dolšak, 2006).
INTELLIGENT ANALYTICAL AIDS IN DESIGN - Decision Support to Engineering Analysis Process
Structural analysis-based design optimisation is a
part of development process for almost every new
product. Thus, it has very important role in
nowadays high-tech world, where only optimal
solutions can win the game on the market. However,
development of the optimal analysis proven design
solutions is very complex domain, which cannot be
treated adequately by using the conventional CAD
tools, unless the user is possessed of special skills
and experience. The main reason for that lies in the
fact that the present CAD tools are still mainly
mathematically oriented and are not able to provide
an adequate expert advice when some crucial
decisions in product development process need to be
On the other hand, advanced computing
applications are changing the way in which
designers interact with computers. Knowledge
representation formalisms and advanced reasoning
techniques are no longer the sole territory of AI
community. New approaches have earned
acceptance in design sphere and have started to
emerge in commercial software.
For this reason, many research activities are
oriented in making analysis-based design
optimisation process more intelligent and less
experience-dependent (Chapman and Pinfold, 2001).
Many experts share the opinion that it can be done
by supplementing the existing CAD systems with
some intelligent modules that will provide advice
when needed.
The intelligent modules discussed here represent
some crucial parts of the overall design optimisation
cycle, where in addition to the structural analysis,
some other design aspects, such as for example the
ergonomics and aesthetics of the product (Kaljun
and Dolšak, 2006) also play an important role.
The intelligent analytical aids in design that are
presented in this paper have already proved to be
very useful in the university education as well as in
engineering practice. Some practical examples
demonstrating the use of intelligent decision support
in FEA-based design optimisation process are
presented in references that are listed for each
intelligent module discussed in this paper.
Chapman, C., Pinfold, M., 2001. The Application of a
Knowledge Based Engineering Approach to the Rapid
Design and Analysis of an Automotive Structure. In
Advances in Engineering Software, 32, 903-912.
Clarkson, J., Eckert, C. (eds.), 2005. Design Process
Improvement - a Review of Current Practice. Springer.
Dolšak, B., 2002. Finite Element Mesh Design Expert
System. In Knowledge-based Systems, 15, 315-322.
Dolšak, B., Bratko, I., Jezernik, A., 1998. Knowledge
Base for Finite Element Mesh Design learned by
Inductive Logic Programming. In Artificial
Intelligence in Engineering Design, Analysis and
Manufacturing, 12, 95-106.
Gero, J.S. (ed.), 2002. Artificial Intelligence in Design '02.
Kaljun, J., Dolšak, B., 2006. Computer Aided Intelligent
Support to Aesthetic and Ergonomic Design. In
WSEAS Transactions on Information Science and
Applications, 3 (2), 315-321.
Mili, F., Shen, W., Martinez, I., Noel, P., Ram, M.,
Zouras, E., 2001. Knowledge Modeling for Design
Decisions. In Artificial Intelligence in Engineering,
15, 153-164.
Novak, M., Dolšak, B., 2008. Intelligent FEA-based
Design Improvement. Engineering Applications of
Artificial Intelligence, 21 (8), 1239- 1254.
Novak, M., Dolšak, B., 2006. Intelligent Computer-aided
Structural Analysis-based Design Optimisation. In
WSEAS Transactions on Information Science and
Applications, 3 (2), 307-314.
Novak, M., Rieg, F., Dolšak, B., Hackenschmidt, R.,
2006. Intelligent Support to Finite Element Type
Selection. In WSEAS Transactions on Information
Science and Applications, 3 (9), 1617-1624.
Ong, Y.S, Keane, A.J., 2002. A Domain Knowledge
Based Search Advisor for Design Problem solving
Environments. In Engineering Applications of AI, 15,
Pinfold, M., Chapman, C., 2004. Using Knowledge Based
Engineering to Automate the Post-processing of FEA
Results. In International Journal Computer
Applications in Technology, 21 (3), 99-106.
Rieg, F., 2006. Z88 The Compact Finite Element
System, version 12.0. In Chair for Engineering Design
and CAD, University of Bayreuth, Germany.
Rieg, F., Koch, F., 2001. Selection of Finite Elements
Considering Loadcases and Geometry. Design
Methods for Performance and Sustainability. In
Proceedings of the int. conference on Engineering
Design, Glasgow, 107-114.
Turban, E., Aronson, J.E., Liang, T.P., 2004. Decision
Support Systems and Intelligent Systems. Prentice
Hall. 7th edition.
Zienkiewicz, O.C., Taylor, R.L., Shu, J.Z., 2005. The
Finite Element Method: Its Basis and Fundamentals.
Butterworth-Heinemann. 6th edition.
KEOD 2009 - International Conference on Knowledge Engineering and Ontology Development