Content uploaded by Sebastian Charles Dietrich
Author content
All content in this area was uploaded by Sebastian Charles Dietrich on Sep 14, 2017
Content may be subject to copyright.
Gülen Çağdaş, Mine Özkar, Leman F. Gül and Ethem Gürer (Eds.)
Future Trajectories of Computation in Design
17th International Conference, CAAD Futures 2017
Istanbul, July 12-14, 2017
Proceedings
th
Future Trajectories of Computation in Design – Proceedings of the 17 International
Conference on Computer Aided Architectural Design Futures, Istanbul Technical
University, Istanbul, Turkey, 12-14 July 2017. Edited by Gülen Çağdaş, Mine Özkar,
Leman F. Gül and Ethem Gürer.
ISBN
978-975-561-482-3
Copyright © 2017
Cover Design: S. Zeynep Bacınoğlu and Begüm Hamzaoğlu
Print: Cenkler Matbaa
I.Karaoglanoglu Cad. Civan Sok. No:7Seyrantepe / 4 Levent – Istanbul
T: 0212 283 02 77 – 264 18 21 – 269 04 99
F: 0212 264 05 31 www.cenkler.com
All rights reserved. Nothing from this publication may be produced, stored in
computerised system or published in any form or in any manner, including electronic,
mechanical, reprographic or photographic, without prior written permission from the
publisher.
128 - CAADFutures 17
RhinoRstab
Introducing and Testing a New Structural Analysis Plugin for
Grasshopper3D
Sebastian Dietrich 1, Sven SCHNEIDER 1, Dimitry Demin 2
1 Bauhaus University Weimar, Germany 2
ETH Zurich, Switzerland
sebastian.charles.dietrich@uni-weimar.de, sven.schneider@uni-weimar.de
Abstract. This paper presents a new open-source structural analysis plugin for
Grasshopper – RhinoRstab. The plugin bridges data between the worldwide
established software: Rhinoceros3d and Dlubal RSTAB. The basic idea behind
the approach is to create an interactive workflow between the architectural design
on the one hand and a structural analysis tool on the other hand. In contrast to
RhinoRstab, other analysis tools for Grasshopper predict the structural behaviour
independent of its structural capacity. Thus, additional standalone software is
necessary to verify the analysis of these plugins subsequently. To test the validity
of this new tool, it is compared to a similar application, namely Karamba (a widely
used structural analysis plugin for Rhinoceros/Grasshopper). Both tools are tested
in different scenarios. The study shows that for some elements in a structural
system and some calculation methods RhinoRstab and Karamba results differ
strongly. However, regarding the runtime, Karamba operates faster than
RhinoRstab.
Keywords: Automation, Structural Analysis, Structural Design, Optimization
1 Introduction
With parametric design gaining currency among structural engineers, information
exchange between architectural designs and structural analysis software receives major
interest as a design tool [1]. This paper presents a plugin for the workflow of a dynamic
inter-process communication between the architectural design and structural analysis
software. It combines already existing software and uses their advantages as reliable
design and structural calculation tools. The plugin RhinoRstab connects the 3D
graphical software Rhinoceros/Grasshopper [2] and the structural analysis program
RSTAB by Dlubal [3].
With regards to similar and already existing applications, the Rhino [4] and
Rhinoceros/Grasshopper communities offer different solutions for structural analysis.
These plugins are able to be executed during the design process, providing a deeper
understanding of the structural performance influenced by alterations of the
see
Erratum
CAADFutures 17 - 129
parametric design. This structural observation has relevance, especially for complex
structures such as the ones created by parametric modelling approaches. During the
conceptual design phase this observation is useful for a behavioural estimation and
optimisation of structural elements. Besides RhinoRstab one of the most popular
parametric plugins is Karamba, which strength lies within the usage in an early design
stage [5]. Since this tool creates only stress resultants, accurate structural calculations
are required to verify the results subsequently. In this further step the structure is
determined according to Building Codes, such as DIN, Eurocode, International Building
Code, etc. Different researches show that additional, standalone structural analysis
software is used to verify the results of Karamba [6]. Furthermore in the later design
stages of projects with higher degree of complexity the usage of certificated program in
structural performing is required [1].
A direct connection between Grasshopper and a verified structural analysis software
such as RSTAB would therefore eliminate the needs to verify the results in another
standalone program and thus allow a more flexible usage throughout all design stages.
As the structural design and the calculation happen within the same software
environment, the possibility of a straightforward structural optimising is no longer
limited to the early design stage, but could influence more advanced design phases as
well. Linking the two software components would therefore result in more reliable
design concepts already in early phases, more flexibility throughout all design stages, as
well as a great reduction of time and workload for all project participants. In this paper
such a connection is established by a new plugin RhinoRstab. It provides a quantitative
benchmark based on a structural analysis of spatial structures and discusses the
advantages and disadvantages of the software plugins, RhinoRstab and Karamba.
2 Structural Analysis Tools for Grasshopper
The majority of structural analysis tools for Grasshopper is based on Finite Elements as
RhinoRstab, Karamba [5] and Millipede [7]. As well as RhinoRstab, all other calculation
tools consist of common Grasshopper utility elements and are fully embedded in its
friendly interface. The embedding into Grasshopper allows a direct link between the
parametric model, the finite element calculator and also optimisation algorithm.
Derived from the notion to create a benchmark for RhinoRstab the plugin is compared
with Karamba, following the basic function: specification of material and crosssection
properties, support definition, first and second order analysis, and result visualization
methodology. Paragraphs 3 and 4 describe the benchmarking on spatial structures,
which showing the detail results.
see
Erratum
130 - CAADFutures 17
2.1 RhinoRstab: an Open-Source Plugin
RhinoRstab [8] is an open-source developed plugin connecting RSTAB and
Grasshopper3D through the RS-COM interface provided by Dlubal [9]. To create an
interactive workflow between these programs, the plugin is designed in such a way that
the workspace is embedded in the Grasshopper environment. Using its convenience of
a visual programming language, the whole structural analysis process such as support
definition, force application and the actual calculation can be controlled in
Grashopper3D without switching between programs.
Forming the RhinoRstab-plugin, it is separated in different components. One of the three
major members is the export tool, which transfers the parametric model from Rhinoceros
to RSTAB including specifications of the structural members regarding support
conditions, material and cross section. Another main component is the analysis tool.
Exporting the load definition to RSTAB, the tool starts the structural analysis and imports
the outputs back to Rhino. The result-component provides the choice of visualising
different analysis results such as the deformation, internal forces and support reactions.
Fig. 1 presents all different plugin components.
Fig. 1. Illustration of RhinoRstab Plugin Elements
2.2 Comparison to Karamba
Comparing the application of Karamba and RhinoRstab, both plugins are similar to use.
The plugins provide element libraries, including material and cross sections. Where
RhinoRstab provides a full selection of international standardized materials and cross-
sections, the selection of Karamba is limited to a smaller range. However, the main
difference from RSTAB to Karamba lies within the calculation, or more precisely in the
verification of the results. The significant difference is that the
Support
Element
Export
Model
Loads
Results
Structural
Analyses
Structural
Members
see
Erratum
CAADFutures 17 - 131
analysis issued by Karamba is not proven by any construction standards. In contrast
RSTAB enables a user to choose from different international building codes and thus
provides a verified proof of stability. RSTAB allows a more elaborated and detailed
calculation in terms of buckling, 2nd order analyses, dynamic calculation, etc.
Depending on the outcomes of the structural analysis, it may lead to an alteration of the
initial design. Thus it is important to provide reliable analysing results in an early design
phase. The presented plugin aims at the realization of that goal. Pointing out the
difference between the analysis of Karamba and RhinoRstab, the comparative study
shall present an evaluation of the results concerning accuracy and runtime. In order to
provide a detailed comparison of both programs, in the following the plugins are tested
in different scenarios such as the structural analysis of structural systems. The
parametric models are defined with variable form and cross section parameters
following the aim to create an optimized structure.
3 Analysis and Benchmarking of Spatial Structures
In the following two scenarios the analysis of treelike, spatial structures is presented.
The design of the first structure (Fig. 2) is kept rather simple in order to prove the
authenticity of the results by comparing the software analysis with manual calculations.
The second structure demonstrates the usage of both tools analysing a more complex
construction. Structure 1 is analysed following theory first order. The theory considers
stresses in a simplified manner, analysing the structure as an unformed system. Structure
2 is analysed according to theory first and second order, where theory second order also
considers the deformation of the system.
3.1 Analysis of Structure 1
Structure 1 consist of 62 steel rods of different length, from 2.96 to 4.50 meters. Its
structural elements are S275 steel profiles, type RO 57x4 | DIN EN 10220_1. Allocating
the most convenient position of the trunk, the fixed support is placed in such a way that
all moments equal zero. The geometry of structure 1 is shown in Fig.
2.
16 kN
Fig. 2.
Tree Structure 1
10
kN
see
Erratum
132 - CAADFutures 17
The first part of the calculation does not take the self-weight into account, in the next
step the analysis is repeated also considering the self-weight of the structure. In
Appendix1 the results of the structural analysis are displayed. It shows the support forces
of the structure. In addition Table 1 demonstrates the runtime of both plugins and the
coefficient of determination R², which point out how strongly the results of both
calculations correlate. Whereby a R² of 1.0 represents identical results and 0.0 indicates
no result correlation at all. Furthermore the results of each structural member was
analysed and compared. The maximal deviation of each result field was determined
regarding the mean value of both programs. In Table 1 the maximal deviation is shown
proportional and by its real value.
Table 1. Runtime & Statistical Analysis of Structural Calculations
Software
Assemble + analysis + results
[ms]
Total runtime [ms]
RhinoRstab
320ms + 675ms + 77ms
1072ms
Karamba
2ms + 6ms + 56ms
64ms
Theory
R²
max result deviation
1st order
support-forces
1.000
0.00
0.00
1st order
max. normal-forces in members
1.000
13.2%
0.09kN
1st order
max. shear-forces in members
0.999
58.6%
0.04kN
1st order
max. moments in members
0.986
71.8%
0.03kNm
1st order
max. deformation
0.909
4.8%
78.21mm
3.2 Analysis of Structure 2
The second example shows the structural analysis of a pavilion supported by several
treelike pillars (Fig. 3), which consist of 68 rods and 14 beams. The load of the roof gets
transferred into the beams, which result in point-loads on the end of each pole (Fig. 3,
right). As the whole structure is made of steel S275, the pillars consist of RO 82.5x7.1 |
DIN EN 10220_1 profiles.
(right)
Fig. 3.
Perspective of the tree structure 2 (left),
exploded view (middle) and structural model
SP2
SP3
SP4
SP5
SP6
SP1
see
Erratum
CAADFutures 17 - 133
The structure is analysed concerning its internal strain, stresses and support forces,
following theory 1st and 2nd order. In Appendix2 the results of RhinoRstab and
Karamba are presented , including the self-weight, sum of vertical loads, the total load
and its support forces in X, Y, and Z direction. In Table 2 the runtime performance of
both programs is demonstrated followed by an estimation of results similarities.
Additionally to the coefficient of determination R², the single results of Karamba and
RhinoRstab were analysed and compared. The max deviation of each result field (i.e.
maximal normal forces) is shown proportional and by its actual value. The value was
calculated, regarding the mean value of both programs.
Table 2. Runtime & Statistical Analysis of Structural Calculations
Software
Assemble + analysis + results
[ms]
Total runtime [ms]
RhinoRstab
329ms + 775ms + 112ms
1216ms
Karamba
83ms + 153ms + 156ms
392ms
Analysis
Statistical analysis of
R²
max. deviation
Th.1st order
support-forces
0.994
52.0%
0.07kN
Th.1st order
max. normal-forces in members
0.992
61.7%
0.032kN
Th.1st order
max. shear-forces in members
0.876
65.8%
-0.70kN
Th.1st order
max. moments in members
0.819
74.0%
0.26kNm
Th.1st order
max. deformation
0.867
6.2%
10.15mm
Th.2nd order
support-forces
0.994
58.5%
0.08kN
Th.2nd order
max. normal-forces in members
0.993
74.1%
0.24kN
Th.2nd order
max. shear-forces in members
0.935
54.3%
0.627kN
Th.2nd order
max. moments in members
0.782
70.0%
2.09kNm
Th.2nd order
max. deformation
0.877
6.5%
14.05mm
3.3 Results
Analysing structure 1 and 2, both programs show in general similar results, as the
coefficient values R² are mostly around 0.9. The biggest difference for structure 1 lies
in the estimation of the maximum deformation with a value of 0.9. Analysing structure
2, the largest difference lies in the results of the second order analysis of moment forces,
with 0.782. As the R² value evaluates the result in a general way, the individual result
values were examined more precisely. This observation shows that despite a high
coefficient of determination great deviations exist. The maximum deviation of structure
1 lies in the structural analysis of moment forces with 71.8%. This value is regarding
the mean value of both software results. Examining the results of structure 2, the
maximum deviation lies in the first order analysis of moment forces
see
Erratum
134 - CAADFutures 17
with a value of 74%. In the second order analysis the results of the normal forces show
a maximum deviations of 74.0%.
Concerning the runtime of both plugins, Karamba performs the structural analysis
quicker than RhinoRstab, 16 times faster in structure 1 and 4 times faster in structure
2.
4 Conclusion and Further Works
In this paper a new structural analysis plugin for Grasshopper is introduced, in which a
dynamic inter-process communication between the software Grasshopper and RSTAB
is created. Subsequently the RhinoRstab plugin is compared with Karamba, creating a
benchmark for the performance of both plugins.
4.1 Conclusion
The functions of RSTAB provide, among other things, dimensioning tools for each
building materials and thus is usable for the verification of the results. Karamba, in turn,
only shows structural behaviour but does not demonstrate its actual structural capability.
Karamba also lacks of options to superposition results, it only offers the choice of either
a single or all load cases. This load treatment is rather unfortunate in proper structural
analyses, as it presents an important part of the general analysis. Concerning the
numerical comparison of the software, it shows that both programs create different
results. Though the structural analysis of the plugins provide in general similar results,
analysing the results more precisely, it can be observed that the result of the single
elements show great differences (up to 71% in structure 1 and 74% in structure 2). A
reason for the different results can be that the result output of Karamba is imprecise. As
it only shows one result per element, it is not clear where the result force is acting on the
element and if it represents the maximum value. It also does not distinguish between
strong and weak axis of a cross section, and therefore only provides the resultant for
both axis. Comparing both programs the results of RhinoRstab were customised to the
result output of Karamba. Another great differences lies in the runtime of both plugins.
Karamba performs faster than RhinoRstab and is therefore very suitable for quick
alterations, such as performed in optimisation processes. Whereas the strength of
Karamba lies within the runtime, the advantage of RhinoRstab is the quality of the
analysis. RhinoRstab provides large object libraries and allows very detailed settings
concerning structural analysis and result visualizations.
4.2 Further Works
Further works target the improvement of the RhinoRstab plugin for optimisation
purposes, finding the most suitable form and cross section for a structure. As
optimisation processes usually require a rapid alteration of model properties, the aim
see
Erratum
CAADFutures 17 - 135
is it to optimize the operation time of RhinoRstab, by simplifying extensive calculation
processes.
In the current state of the plugin, the type of parametric models is limited to spaceframe
structures. In order to analyse plates, walls, shells, etc. it is favourable to base the
analysing tool on 3 dimensional finite elements. Additional extension of the plugin‘s
abilities target on linking it to the finite element calculator Dlubal RFEM [10], following
the same strategies as introduced in the interactive workflow between Rhino and
RSTAB.
References
1. Demin, D. Algorithms Behind Building Envelopes, (2016),
<https://www.researchgate.net/publication/308647065_Algorithms_Behind_Building_Enve
lopes>, accessed on 11 September 2016.
2. Rutten, D. (2016) Grasshopper. Available at: http://www.grasshopper3d.com/ (Accessed: 30
August 2016).
3. Dlubal Software, RSTAB, (2014), <https://www.dlubal.com/en/products/rstab-
beamstructures/what-is-rstab>, accessed on 21 August 2016.
4. Robert McNeel & Associates Rhinoceros (2016), http://www.rhino3d.com/>, accessed on 30
August 2016.
5. Preisinger C.: Linking Structure and Parametric Geometry. Archit Design, 83, 110-113 (2013).
6. Rumpf, M., Grohmann, M., Eisenbach, P. and Hauser, S.: Structural Surface - multi parameter
structural optimization of a thin high performance concrete object, IASS, 2015 Amsterdam
(2015).
7. Sawako, K. and Panagiotis, M.: Millipede March2014 release, (2014),
<www.sawapan.eu/sections/section88_Millipede/files/MillipedeMarch2014.pdf> accessed on
13 November 2016.
8. Demin, D. and Dietrich, S.: RhinoDlubalRstab.
(2016),<http://www.food4rhino.com/project/rhinorstab>, accessed on 23 August 2016.
9. Dlubal Software (2011) RS-COM 2.3. Germany
10. Dlubal Software RFEM (2014), <https://www.dlubal.com/en/products/rfem-
feasoftware/what-is-rfem>, accessed on 21 August 2016.
Appendix 1
Table 1. Analysis of Tree Structure 1
FV
[kN]
FH
[kN]
M
[kNm]
Comment
1.1
Manual calculation
26.00
0.00
0.00
self-weight not considered
1.1
Karamba
26.00
0.00
0.00
self-weight not considered
1.1
RhinoRstab
26.00
0.00
0.00
self-weight not considered
1.2
Karamba
38.46
0.00
0.00
self-weight considered
1.2
RhinoRstab
38.46
0.00
0.00
self-weight considered
136 - CAADFutures 17
Appendix 2
Table 1. Theory 1st Order Analysis of Tree Structure 2
Self-weight of
structure
[kN]
Sum of vertical point
loads
[kN]
Total load [kN]
RhinoRstab
Karamba
RhinoRstab
Karamba
RhinoRstab
Karamba
23.228
23.229
48.593
48.593
71.821
71.822
Fx
[kN]
Fy
[kN]
Fz
[kN]
SP1
-0.100
-0.090
0.244
-0.270
22.018
22.670
SP2
0.838
0.930
-0.381
-0.410
8.292
7.860
SP3
0.190
-0.060
-0.200
0.430
5.596
4.260
SP4
-0.272
-0.280
-0.034
0.060
5.604
6.950
SP5
-0.887
-0.620
0.247
0.460
8.2841
8.090
SP6
0.231
0.130
0.124
-0.280
22.026
22.000
6
0.000
0.010
0.000
-0.010
71.821
71.830
Table 2. Theory 2nd Order Analysis of Tree Structure 2
Self-weight of
structure
[kN]
Sum of vertical point
loads
[kN]
Total load [kN]
RhinoRstab
Karamba
RhinoRstab
Karamba
RhinoRstab
Karamba
23.229
23.229
48.593
48.593
71.821
71.822
Fx
[kN]
Fy
[kN]
Fz
[kN]
SP1
-0.632
-0.460
0.229
0.060
20.711
21.540
SP2
1.083
1.070
-0.369
-0.580
11.307
10.390
SP3
0.507
0.170
-0.207
0.250
3.892
2.900
SP4
-0.520
-0.620
0.164
0.070
3.890
5.390
SP5
-1.068
-0.810
0.414
0.480
11.303
10.980
SP6
0.629
0.650
-0.230
-0.280
20.719
20.620
6
0.000
0.000
0.000
0.000
71.821
71.820
see
Erratum
CAADFutures 17 - 703
Authors
Abel; Groenewolt Edgar; Rodriguez
Adeline; Stals Elif; Ensari
Agata; Migalska Elif; Erdine
Alan; Wang Elif Sezen; Yağmur Kilimci
Alexander; Hollberg Eonyong; Kim
Alexandra; Paio Ergun; Akleman
Alexandre; Dubor Evangelos; Pantazis
Alexandros; Kallegias Felipe; Veliz
Alexandros; Tsamis Fernando; García Amen
Ali; Motamedi Filipa; Osorio
Alican; Sungur Francesco; De Luca
Alireza; Borhani Francesco; Livio Rossini
Andréa; Quadrado Mussi Francis; Miguet
Andy; Song Frank; Petzold
Anetta; Kepczynska-Walczak Gabriela; Barber Sarasola
Angel Fernando; Lara Moreira Gabriele; Novembri
Angelos; Chronis George; Stiny
Anna; Pietrzak Guillaume; Moreau
Antonio; Fioravanti Gülen; Çağdaş Areti;
Markopoulou Guzden; Varinlioglu
Aslı; Ağırbaş Halil; Erhan
Athanassios; Economou Hao; Wu
Ben; Yuqiang Haruo; Adachi
Benay; Gürsoy Hayri; Dortdivanlioglu
Benjamin; Spaeth Hendrik; Voll
Benson M.; Kariuki Hideki; Nada
Bilge; Kobas Hülya; Oral
Can; Sucuoglu Ikhwan; Kim
Can; Uzun Iestyn; Jowers
Canhui; Chen Iman; Sayah
Catherine; Elsen Injung; Lee
Cemal Koray; Bingöl Ivy; Wang
Chang; Mei-Chih Jae; Seung So
Chantelle; Niblock Jane; Burry
Chao; Yan Jeong A.; Choo
Chenke; Zhang Ji-Hyun; Lee
Chikako; Takei Ji Won; Jun
Chris; Earl Joao; Rocha
Dagmar; Reinhardt José Nuno; Beirão
Daniel; Siret José P.; Duarte
David; Gerber Josep; Alcover Llubia
Desantila; Hysa Judyta M.; Cichocka Diego;
Pinochet Julia; Tschetwertak
Dina; El-Zanfaly Juliano; Lima Da Silva
Dirk; Donath Julie; Milovanovic
Duygu; Tüntaş Karaman Jürgen; Ruth
Katja; Knecht
Kenneth D.M.; Harris
704 - CAADFutures 17
Kent; Larson
Kian Wee; Chen
Kibum; Kim
Krishnendra; Shekhawat
Lale; Başarır
Lauro; André Ribeiro
Lawrence; Sass
Leman Figen; Gül
Leslie; Norford
Lukasz; Bonenberg
Mahmood; Ettehad
Mallika; Arora
Marc Aurel; Schnabel
Marcos; Lafluf
Manuel; Muehlbauer
Marilena; Sorrou
Mark; Meagher
Matteo; Silverio
Mine; Ozkar
Naghmi; Shireen Nai
Chun; Chen
Negar; Kalantar Nobuyoshi;
Yabuki
Ömer Halil; Çavuşoğlu Onur
Yüce; Gün
Orkan Zeynel; Güzelci
Özgüç Bertuğ; Çapunaman
Ozgur; Genca
P. Andrew; Williams Pablo;
Accuosto
Paloma; Gonzalez Rojas
Pantea; Alambeigi
Patrick; Janssen Peter;
Buš
Philip F.; Yuan
Pradeep; Devadass
Rachel;Dickey
Reinhard; Koenig
Robert; Woodbury
Rizal; Muslimin Rui;
de Klerk
Saeid; Zarrinmehr
Salih; Ofluoglu
Sancho; Oliveira
Sarah; Jenney
Sarah; Mokhtar
Sebastian; Dietrich
Sema; Alaçam
Serdar; Aydin
Sergio; Araya
Sergio; Pineda
Serkan; Kocabay
Seungyeon; Choo
Sherif; Abdelmohsen
Shunta; Shimizu
Siniša; Kolarić
Süheyla Müge; Halıcı
Sven; Schneider
Sylvie; Jancart
Takehiko; Nagakura
Theodoros; Dounas
Tomohiro; Fukuda
Will N.; Browne
Vernelle; Noel
Yie; He
Yufan; Miao
Yusuke; Sato
Zeynep; Vaizoglu
Erratum
The scientific paper ”RhinoRstab - Introducing and Testing a New Structural Analysis Plugin
for Grasshopper3D” contains the assertion that ”. . . for some elements in a structural system
and some calculation methods RhinoRstab and Karamba results differ strongly.”. In order to
prove their case, the authors compare the results of two structural calculations performed with
both programs.
The reported differences in the calculation results are however caused by
modelling errors, methodical errors and inaccuracies on the side of the authors of the pa-
per. The corresponding assertions and conclusions in the paper are therefore wrong.
The paper fails to mention that the pro-student version of Karamba 1.2.2 was compared with
RStab version 8.05.0030 (henceforth RStab8 for short).
Also the paper makes wrong asser-
tions regarding the features of Karamba 1.2.2.
Here a detailed account of the errors contained in the paper:
•
In the paper the results of the calculations of structure 1 are wrong because of these
modelling errors:
–
In the RStab8 calculation under ”Calculation Parameters” the option ”Activate stiff-
ness factors of materials” was enabled. This makes RStab8 divide the materials
Young’s Modulus by the partial safety factor of the material. Since RStab8 uses a
partial safety factor of
1.1
for steel, enabling this option increases the calculated
displacements by 10% relative to the unfactored case.
–
The structure consists of elements with circular hollow cross sections (CHS) of
diameter
57[mm]
and wall thickness
4[mm]
. In Karamba 1.2.2 the shear areas
Ay
and
Az
are calculated according to the standard textbook formula for a CHS:
Ay=Az= 2 ·A/π
which results in a value of
4.24[cm2]
. RStab8 uses a value of
3.31[cm]
. This causes deviations in the cross section forces for very short members
like those along the upper boundary of structure 1 which have a length of roughly
0.5[m].
Setting the shear areas in Karamba 1.2.2 to those of RStab8 and disabling the stiffness
reduction in RStab8, the following differences in the calculation results are observed. The
percentages refer the maximum deviation found in all elements to the corresponding mean
value of the Karamba and RStab results:
–Support forces: 0.0001%
–Maximum normal forces in members: 0.006%
–Maximum shear forces Vzin members: 0.8%
–Maximum moments Myin members: 0.15%
–Maximum deformation: 0.0003%
The results deviations presented in table 1 on page 132 of the paper are therefore
wrong.
•
In the paper the results of the calculations of structure 2 are wrong because of these
modelling errors:
–
Like for structure 1 a stiffness reduction of
10%
was imposed on all materials in the
RStab8 calculation.
–
The geometry of the structures calculated with RStab8 and Karamba 1.2.2 was not
identical: due to geometric inaccuracies there were two gaps in the Karamba model.
In the RStab-model these were not present because the authors of the paper used
different tolerance settings in Karamba and RStab for joining neighboring nodes.
–
The comparison of the results from second order theory between RStab8 and Karamba
1.2.2 neglects a difference in the way both programs take account of the normal force
NII
which causes the second order effects: according to the RStab8 manual RStab8
uses the mean value of the normal forces in a beam as
NII
whereas Karamba 1.2.2
uses the minimum normal force. This is documented in the Karamba 1.2.2 manual.
The latter procedure gives results which lie on the safe side.
In order to make a valid comparison between second order results of Karamba 1.2.2
and RStab8 one has to divide the beam elements of the structure into small segments
so that the difference between
NII
as calculated in RStab8 and Karamba 1.2.2
becomes negligible. In case of elements where the cross section forces change sign,
and the gradient of the cross section force is small, the method of comparing the
difference of the results to the corresponding mean value becomes meaningless for
judging the correctness of the structural analysis: Tiny variations in the results then
cause large relative deviations. Therefore the method for comparing the results of
RStab8 and Karamba 1.2.2 by calculating relative deviations is invalid and has a
pronounced influence in case of second order structural calculations when small beam
segments are used.
Setting the shear areas in Karamba 1.2.2 to those used in RStab8, disabling the stiffness
reduction in RStab8 and using the geometry with the gaps for RStab8 and Karamba 1.2.2
the following differences in the calculation results are observed. The percentages refer the
maximum deviation found in all elements to the corresponding mean value of the Karamba
and RStab results:
–Th.I, support forces: 0.0002%
–Th.I, maximum normal forces in members: 2.7%
–Th.I, maximum shear force Vzin members: 0.02%
–Th.I, maximum moments Myin members: 0.04%
–Th.I, maximum deformation: 0.3%
–Th.II, support forces: 3.9%
–Th.II, maximum deformation: 2.4%
If each beam is divided into 20 segments the following relative deviations result for the
calculation according to second order theory:
–Th.II, support forces: 0.19%
–Th.II, maximum deformation: 0.14%
The results deviations presented in table 2 on page 133 of the paper are therefore
wrong.
•
On page 129, first paragraph, it is stated that Karamba 1.2.2 creates only stress resultants.
This is not the case. Besides stresses and other properties resultant cross section forces,
local cross section forces, displacements can be retrieved from beam elements.
•
On page 130, section 2.2, first paragraph, it is stated that in Karamba 1.2.2 the selection
of predefined cross sections is limited to a smaller range than in RStab8. This is not
the case. In version 1.2.2 the cross section library of Karamba comprises roughly 6600
different cross sections. The cross section library can be easily extended by the users and
has therefore no limit on the potential number of predefined cross sections.
•
On page 131, section 2.2, first paragraph, it is stated that ”. . . Karamba is not proven by any
construction standards”.
–
In case the authors mean that Karamba 1.2.2 does not contain procedures for designing
structural elements according to building codes then this is not correct: Karamba 1.2.2
contains assessment and optimization tools based on Eurocode 3 for steel structures.
–
In case the authors mean that the results of Karamba 1.2.2 are not verified then this
is not correct: Karamba 1.2.2 comes with a selection of widely used benchmark
examples with comparisons to results known from literature.
•
In section 3.3 of the paper the authors draw several comparisons between the results of
Rstab8 and Karamba 1.2.2. These comparisons are wrong.
•
In section 4.1, first paragraph, it is stated that ”. . . Karamba, in turn, only shows structural
behavior but does not demonstrate its actual structural capability”. This is wrong: Karamba
features assessment and optimization tools based on Eurocode 3 for steel structures. It also
lets the user retrieve (besides other result properties) cross section forces and moments for
beams and principal stresses and Van Mises stresses for shells.
•
In the same paragraph it is stated that ”. . . Karamba also lacks of options to superimposition
results.”. It is not mentioned in the paper that Karamba 1.2.2 offers the option of load
superimposition.
•
In the same paragraph it is stated – with respect to the results of Rstab8 and Karamba 1.2.2
– that ”. . . it can be observed that the result of the single elements show great differences
(up to 71% in structure 1 and 74% in structure 2)”. This is wrong.
•
In the same paragraph it is stated – with reference to Karamba 1.2.2 – that ”. . . as it only
shows one result per element, it is not clear where the result force is acting on the element
and if it represents the maximum value.”. This is wrong. In case of beams Karamba
1.2.2 lets the user retrieve a user defined number of results (displacements, cross section
forces,. . . ) on equidistant points of the beam axis.
•
In the same paragraph it is stated – with reference to Karamba 1.2.2 – that ”. . . It also
does not distinguish between strong and weak axis of a cross section, and therefore only
provides the resultant for both axis.”. This is wrong. Karamba 1.2.2 distinguishes between
the strong and weak axis of a cross section and provides not only resultants of the cross
section forces but also their components in the local element coordinate system.
•For the reasons described above, the result comparisons in table 1 and table 2 for
tree structure 2 in appendix 2 on page 136 for the components of the support forces
are wrong.
