See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/341140607
Structural Optimization Computing Platform (SOCP) and SCIA Software
Article in Procedia Manufacturing · January 2020
DOI: 10.1016/j.promfg.2020.02.276
CITATION READS
1 212
5 authors, including:
Chara Ch. Mitropoulou Georgios Kazakis
National Technical University of Athens National Technical University of Athens
29 PUBLICATIONS 447 CITATIONS 13 PUBLICATIONS 194 CITATIONS
SEE PROFILE SEE PROFILE
Stefanos Sotiropoulos Nikos Lagaros
10 PUBLICATIONS 116 CITATIONS
National Technical University of Athens
260 PUBLICATIONS 5,466 CITATIONS
SEE PROFILE
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
OptArch project: “Optimization Driven Architectural Design of Structures” (No: 689983) belonging to the Marie Skłodowska-Curie Actions (MSCA) Research and
Innovation Staff Exchange (RISE) H2020-MSCA-RISE-2015. View project
Webinars on Optimization - Driven Architectural Design of Structures View project
All content following this page was uploaded by Nikos Lagaros on 05 May 2020.
The user has requested enhancement of the downloaded file.
Available online at www.sciencedirect.com
ScienceDirect
Available online at www.sciencedirect.com
ScienceDirect
Procedia Manufacturing 00 (2019) 000–000
www.elsevier.com/locate/procedia
Procedia Manufacturing 44 (2020) 410–416
1st International Conference on Optimization-Driven Architectural Design (OPTARCH 2019)
Structural Optimization Computing Platform (SOCP) and SCIA
Software
Chara Ch. Mitropouloua, Giorgos Kazakisb, Stefanos Sotiropoulosb, Nikos D. Lagarosb*,
Nikos Ath. Kalliorasb
ACE-Hellas, 6 Aigaiou Pelagous St., Agia Paraskevi Athens 15341, Greece
a
b
National Technical University of Athens, School of Civil Engineering, 9 Iroon Polytechniou St., Zografou Athens 15780, Greece
Abstract
Structural analysis programs usually compute a static model with cross sections given by the engineer or set to reasonable
defaults. Any optimization of the structure with respect to a variable, for example weight/cost, is left to the experience of the
engineer, usually because adding optimization capabilities to a structural analysis program is not trivial. Many optimization
algorithms are available, each one with their own peculiarities, requirements, and performance. The variable to optimize is
different for different projects such as material cost, construction cost, first eigen frequency and others. Adding to the complexity
is the fact a static or dynamic analysis is computational expensive (time consuming) and thus the algorithms needs to be tuned to
perform as few static analyses as possible. Optimization Computing Platform (OCP) is software developed at ISAAR-NTUA,
which in combination with widely known structural analysis programs such as SAP2000 and ETABS, can optimize any structure
with various algorithms, and with respect to various variables or a combination of them. Recently, OCP is actively developed to
be program agnostic, so that it can be linked with any structural analysis program. The program gains painlessly mature and
sophisticated optimization capabilities. In this case study OCP platform was linked with SCIA engineering, structural analysis
and design software to provide addition optimization capabilities. To validate the software connection one structural examples is
tested with very good results.
©
© 2020
2019 The
The Authors.
Authors. Published
Published by
by Elsevier
Elsevier B.V.
B.V.
This
This is an open access article under the
is an open access article under the CC
CC BY-NC-ND
BY-NC-ND license
license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the scientific committee of the 1st International Conference on Optimization-Driven
Peer-review under responsibility of the scientific committee of the 1st International Conference on Optimization-Driven
Architectural Design
Architectural Design
* Corresponding author. Tel.: +30 210 7722625; fax: +30 210 7721655.
E-mail address: [email protected]
2351-9789 © 2019 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the scientific committee of the 1st International Conference on Optimization-Driven Architectural Design
2351-9789 © 2020 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the scientific committee of the 1st International Conference on Optimization-Driven Architectural Design
10.1016/j.promfg.2020.02.276
Chara Ch. Mitropoulou et al. / Procedia Manufacturing 44 (2020) 410–416 411
2 Chara Ch. Mitropoulou et al./ Procedia Manufacturing 00 (2019) 000–000
Keywords: Optimizaiton; SCIA Engineer; PQN.
1. Introduction
In the field of structural optimization it is very important to be able to combine the power of strong mathematical
optimization algorithms for sizing, shape and topology optimization with the modelling, analysis and design tools of
a professional structural analysis software [1]. Although many finite element analysis software’s have implemented
optimization tools in their products like Tosca optimization software for the Simulia of Dassault Systems, structural
analysis and design software’s have been left behind in their optimization capabilities. This has mainly to do with
the complexity and demand for applicability of real word structural applications. The objective of HP-OCP is to
facilitate this problem by a set of procedure which formulate the mathematical taking into account many different
aspects of the structural model, analysis results and most importantly design. In this paper the main focus will lean
to the component enabling HP-OCP to connect to different structural analysis software’s in order to formulate and
run the optimization procedure and more precisely to the component build for the connection of HP-OCP to the
SCIA Engineer structural analysis software.
Nomenclature
Haunch position of minimum cross section height
H1 minimum cross section height
H2 maximum cross section height
Bh top flange width
Bs bottom flange width
th top flange thickness
ts bottom flange thickness
2. HP-OCP Optimization Platform
HP-OCP [2] is an optimization library able to perform structural optimization on simple and complex structures
using a number of advanced optimization algorithms. The types of optimization supported by HP-OCP are sizing,
shape and topology optimization with or without constraints. To perform the optimization procedure HP-OCP is
equipped with many different optimization searching algorithms both deterministic as well as probabilistic. The
most notable probabilistic algorithms available in HP-OCP are a typical Genetic Algorithms, Ant Colony
algorithms, the Differential Evolution algorithm and Particle Swarm Optimization algorithms [2]. For the
deterministic algorithms the most notables available in HP-OCP are the PQN (Projected Quasi-Newton) algorithm
[3,4] and COLYBA (Constrained Optimization by Linear Approximation) algorithm [5-7]. A typical optimization
flowchart can be seen in Fig. 1. All algorithms mentioned are used in the Compute New Design Variables step. In
most of the optimization problems, HP-OCP spends a varying number of the first optimization iterations
implementing a random sampling search of the feasible solutions domain and after that the searching algorithm is
called and initialized with the best feasible design variable combination found during the sampling procedure.
A small description for the PQN algorithm will be given here because it will be used for the test case presented at
the end of the paper. PQN is a steepest descent algorithm and a hybrid of the projected quasi-Newton or Gauss-
Newton algorithm and a deterministic grid-based search algorithm. The gradients for the quasi-Newton and the
Jacobians for the Gauss-Newton are approximated using a finite differences scheme with the difference increment
changing during the optimization procedure. HP-OCP is able to implement the whole optimization procedure but
cannot perform the tasks of structural analysis and design. For this reason, the connection with a professional
structural analysis software is required. To enable this connection an interface between HP-OCP and the structural
analysis software must be implemented. This interface translates the tasks that HP-OCP requires from the structural
analysis software and applies them. A sketch of a very basic configuration of the three components can be seen in
412 Chara Ch. Mitropoulou et al. / Procedia Manufacturing 44 (2020) 410–416
Chara Ch. Mitropoulou et al./ Procedia Manufacturing 00 (2019) 000–000 3
Fig. 1. The interface varies for each structural analysis software and it is a component, which must be constructed in
order for HP-OCP to be able to connect to each different structural analysis software.
Fig. 1. Optimization flowchart
By using the analysis and design results obtained from the structural analysis software HP-OCP is able to
compute and evaluate a number of objective and constraint functions. This is achieved by the forming of a database
which holds all necessary information for the computations of this functions as well as what parts of it should be
updated during the optimization iterations. Due to the limitations of each software, each interface would have some
but not all functionalities available for HP-OCP to use. This is taken into account automatically by HP-OCP.
Depending on the functionalities available in each interface, HP-OCP is able to compute:
▪ Structures volume
▪ Structures total material cost
▪ Structures total construction cost
▪ Structures compliance
▪ Structures Drift for each floor
▪ Structures life Cycle emission & energy
The prices for the materials as well as the labour costs for the computation of the total material cost and construction
cost can be set by the user. In the case of no data for this values HP-OCP will assume a number of default values
taken from the typical prices of Greek structural materials and labour. In addition HP-OCP can perform optimization
using other objective functions provided that the structural analysis software can perform their calculation.
Regarding the constraints available in HP-OCP, all the aforementioned objective functions can be used as
constraints as well as the collection of structures design violations and other arbitrary constraints computed by the
structural analysis software.
3. HP-OCP interfaces
For each interface to facilitate the communication between HP-OCP and the structural analysis software it
requires to implement a number of functionalities. In Fig. 2 a very brief representation of the functionalities needed
to be implemented by an interface is presented. The most important of these functionalities is the running of the
structural analysis and the setting of the new design variables (in the case of sizing optimization the new section
dimensions). For the case of the running of the structural analysis, the interface is responsible for running the
structural analysis and return any values of an objective or constraint function set to be computed by the structural
analysis software. For the case of setting the new design variables the interface is responsible for getting the new set
of design variables form HP-OCP and setting them to the model inside the structural analysis software. Additionally,
Chara Ch. Mitropoulou et al. / Procedia Manufacturing 44 (2020) 410–416 413
4 Chara Ch. Mitropoulou et al./ Procedia Manufacturing 00 (2019) 000–000
if extra components of the interface are implemented then other functionalities of HP-OCP would be available such
as the setting of default values to the design variables by taking into consideration extra geometrical constraints and
the computation of the objective and constraint functions mentioned in the previous page. The degree of which each
of this functionalities are available depends on the capabilities of the interface.
Fig. 2. Brief representation of the functions implemented by each interface for a sizing optimization formulation
As mentioned before, connecting each structural analysis software to HP-OCP requires a different interface to be
constructed and implemented (but no changes need to be made to the HP-OCP Optimization library). Thus, the
construction of each interface depends on the connection capabilities of the structural analysis software. Some of the
tools used for the construction of these interfaces are the software’s API (Application Programming Interface), if the
software provides any, commands that go directly into the software source code (for open source programs) to set or
get the necessary data and xml files of input data loaded automatically into the program and output written
automatically after the structural analysis. Below there will be presented some example of interfaces with different
structural analysis software’s. The third example which is the interface created to connect HP-OCP with SCIA
Engineer is the one that will be explained in more detail. The first interface example is the interface connection with
the CSI products (SAP2000 and Etabs). CSI provides an API for each of its products which enables a programmer to
perform almost all available GUI (Graphical User Interface) actions using scripts. Thus taking advantage of this
functionality the interface is built on top of the provided CSI API. A brief sketch of this integration can be seen in
Fig. 3.
Fig. 3. Basic diagram for the connection of HP-OCP with SAP2000 using as interface the CSI API
For the second interface example we used methods that performed the necessary tasks by accessing the source
code of the structural software. The software used was Truss2D, an educational software able to solve 2D truss
structures only written is C++. In this case only the functions for setting new dimensions and running the analysis
were implemented inside the interface. Lastly, the interface for connecting HP-OCP to SCIA Engineer software was
created using xml files for setting and getting functionalities. This Xml component interface is able to perform only
the two main functionalities of running the analysis and setting the new design variables.
To set the new section dimensions into the SCIA Engineer software the Parameters component imbedded into the
SCIA Engineer software was used. This component enables a user to create parameter values that can then be
414 Chara Ch. Mitropoulou et al. / Procedia Manufacturing 44 (2020) 410–416
Chara Ch. Mitropoulou et al./ Procedia Manufacturing 00 (2019) 000–000 5
assigned to parts of the structure like cross section dimensions or member lengths. By changing the value of each
parameter, the values of the parts that it is assigned to update and take the new value automatically. Thus by creating
a parameter value for each design variable of the optimization problem HP-OCP is able to change cross section
dimensions and other parts of the structure by only accessing the Parameters component of the SCIA Engineer
software. Thus the functionality of setting the new design variables is completed by building the ability inside the
Xml component interface of creating an xml file as input and then using some system commands to automatically
import this xml file into the SCIA Engineer software, target the parameters component and changing the proper
values. This functionality would be better understood in the example presented at the end. To run the analysis and
get the values for the objective function (only structure volume available for this configuration) and the maximum
design violation the Run analysis functionality is build inside the Xml component interface. This functionality
basically uses system commands to run the analysis in SCIA Engineer and then reads the Xml output file written
automatically by the software at the end of the analysis. The data contained in the xml file are predefined before the
start of the optimization procedure by the user using the Xml output component of the SCIA Engineer software.
Once the interface reads the necessary data it passes them to HP-OCP to continue with the evaluation and
computation of new design variables.
4. Optimization Test Case
4.1. Optimization Example
The optimization problem presented in this section was set by the SCIA Engineer Company. The provided
structure can be seen in Fig. 4. The part of the structure to be optimized are the beams on the roof of the structure
noted with pink dotted lines. All the beams have the same cross section which is a I cross section with varying
height from H1 (minimum height) in the middle to H2 (maximum height). The loading combination applied was a
combination extracted from the LRFD code containing self-weight, wind and snow loads and the design capacity
check is a LRFD verification check as well.
Fig. 4. SCIA Engineer example
4.2. Optimization Formulation
The formulation of the optimization problem was also provided by the SCIA Engineer Company. The volume
minimization was chosen as the objective function of the optimization problem and no design verification check
violation as the constraint. The design variables of the optimization problem can be seen in Table 1. The Haunch
variable represents the position of the minimum height as a percentage of the beam length. The other design
variables represent all the beam cross section dimensions except the web thickness. The algorithm chosen to
Chara Ch. Mitropoulou et al. / Procedia Manufacturing 44 (2020) 410–416 415
6 Chara Ch. Mitropoulou et al./ Procedia Manufacturing 00 (2019) 000–000
perform the optimization was the PQN algorithm and the convergence criterion used was to have change in the
objective function less than 5%.
Table 1. Design Variables of the Optimization Problem.
Description Name Starting Value
Position of the minimum height Haunch 0.5 (50%)
Minimum Height H1 610mm
Maximum Height H2 915mm
Top flange width Bh 305mm
Bottom flange width Bs 305mm
Top flange thickness th 16mm
Bottom flange thickness ts 16mm
Both the objective function as well as the constraint are computed by the SCIA Engineer software and are
written at the end of each analysis at the output xml file automatically using the xml output component descripted
in the section 3. Regarding the design variables, all seven of them are formulated in the Parameters component and
then set to each dimension inside the program so the values can be updated automatically as well. The final values
of the design variables after the optimization procedure can be seen on Table 2. The reduction of the objective
function achieved was 6.59% and the iterations perform by the algorithm were 94. The maximum capacity
violation of the proposed optimal design in near its limit of 1 at 0.98.
Table 2. Optimization results
Design Variable Original Value Final Value
Volume 5.61m 3
5.24m3
Haunch 50% 50%
H1 610mm 666mm
H2 915mm 789mm
B1 305mm 263mm
B2 305mm 263mm
th 16mm 9.6mm
ts 16mm 9.6mm
Design Violation 0.77 0.98
5. Further Development
To enrich the interface between HP-OCP and SCIA Engineer the next step would be to build a hybrid interface
component which takes advantage of the xml capabilities already used but implements the API provided by the
latest version of SCIA Engineer as well to enable to component to collect more information for the model which
will help in setting the optimization problem to produce realistic results with tackle more complex structures and
many different design variables automatically.
Acknowledgements
This research has been co-financed by the European Union and Greek national funds through the Operational
Program Competitiveness, Entrepreneurship and Innovation, under the call RESEARCH–CREATE–INNOVATE
(project code: T1EDK-05603, “HP-OCP: Holistic, High Performance Optimization Computing Platform”).
416 Chara Ch. Mitropoulou et al. / Procedia Manufacturing 44 (2020) 410–416
Chara Ch. Mitropoulou et al./ Procedia Manufacturing 00 (2019) 000–000 7
References
[1] N.D. Lagaros, The environmental and economic impact of structural optimization, Structural and Multidisciplinary Optimization, 58(4) (2018)
1751–1768.
[2] N.D. Lagaros, A general purpose real-world structural design optimization computing platform, Structural and Multidisciplinary Optimization,
49 (2014) 1047–1066.
[3] C. Audet, J.E. Dennis, Mesh adaptive direct search algorithms for constrained optimization, SIAM Journal on Optimization, 17 (2006) 188–
217.
[4] R.M. Lewis, V. Torczon, Pattern search algorithms for linearly constrained minimization, SIAM Journal on Optimization, 10 (2000) 917–941.
[5] M.J.D. Powell, An efficient method for finding the minimum of a function of several variables without calculating derivatives, Computer
Journal, 17 (1964) 175-162.
[6] M.J.D. Powell, A method for minimizing a sum of squares of nonlinear functions without calculating derivatives, Computer Journal, 7 (1965)
303-307.
[7] M.J.D. Powell, A tolerant algorithm for linearly constrained optimization calculations, Mathematical Programming, 45(1-3) (1989) 547–566.
View publication stats
