See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/276709883

MULTI-MATERIAL DESIGN OPTIMIZATION OF A BUS BODY STRUCTURE

Article  in  Journal of KONES · January 2013

DOI: 10.5604/12314005.1135327

CITATIONS READS

6 2,580

2 authors, including:

T. Uhl
AGH University of Science and Technology in Kraków
434 PUBLICATIONS   2,637 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Subsurface evaluation using georadar View project

SHM in composite structures View project

All content following this page was uploaded by Jakub Korta on 26 December 2015.

The user has requested enhancement of the downloaded file.

 Journal of KONES Powertrain and Transport, Vol. 20, No. 1 2013

MULTI-MATERIAL DESIGN OPTIMIZATION

OF A BUS BODY STRUCTURE

Jakub Korta, Tadeusz Uhl

AGH University of Science and Technology

Faculty of Mechanical Engineering and Robotics
Department of Robotics and Mechatronics
A. Mickiewicza Av. 30, 30-059 Krakow, Poland
e-mail: [email protected], [email protected]

Abstract
In the recent years the safety and eco-friendliness have gained much of attention of the automotive stakeholders.
These two characteristics are especially important in the case of mass transportation vehicles, such as buses or
coaches, which are in continues use for long periods of time, covering significant distances. In such situations, the
economical aspects play major role for the transportation companies which try to minimize operational costs of their
fleet, by choosing vehicles with reduced fuel consumption. In order to obtain improvements in all the mentioned areas
and hence to strengthen their position on the market, bus manufacturers have recently turned their attention to multi-
material design strategies. Structures built in that manner consist not only of regular steel parts, but contain also
a mix of components made from various lightweight materials like aluminum alloys or composites, which allow for
significant reduction in vehicle curb weight. However, due to the differences in mechanical characteristics which are
especially evident in the case of laminates, the material substitution is not a straightforward task. In order to find the
material distribution pattern that meets all the requirements, a great number of prototypic numerical models must be
prepared and tested. To ease the search for the final solution, optimization techniques can be applied into the design
process, allowing for automatic design modifications and assessment of the obtained results.
The paper presents an attempt of enhancing the operational characteristics of a bus body structure with
simultaneous reduction in the structural weight. In order to find the optimal component configuration, a multi-
material optimization was employed and supplemented by sensitivity and robustness analyses. Such a technique helps
to discriminate the over-optimized solutions that are often pointed out as the most desirable by the optimization
algorithms which neglect the uncertainties of the analysed system.
Keywords: weight reduction, bus structure, structural optimization, multi-material design

1. Introduction

The stakeholders on the nowadays automotive market have started to tackle with a design
process that takes advantage of various non-ferrous materials available on the market, exploiting
their specific properties. The main outcome of a multi-material design is a significant mass
reduction and enhancements in structural mechanical properties, e.g. increased stiffness of
crashworthiness.
The motivation for weight minimization comes from the fact that the mass of new vehicles has
been growing steadily for the last four decades, exhibiting 1.1% of an annual increase [1]. Despite
the fact that load bearing components are the subject of structural optimization approaches, the
mass increase takes place due to the intensive use of auxiliary systems like air-conditioning,
electronics or additional gas tanks [2]. Moreover, from the economical point of view, lower curb
weight result in decrease of operational costs. This is particularly important in the case of the mass
transport (e.g. buses or coaches), for which the annual mileage is very high.
Modern structural materials, very often adopted from the aeronautical industry, are
characterized by very good specific strength and stiffness ratios. These quantities describe the
chosen parameters of materials related to their density, comparatively indicating how much of the
selected material type must be used, in order to withstand the applied loads (Tab. 1).
 J. Korta, T. Uhl

Aluminum has been used widely in the aerospace industry for decades, but has been also
incorporated into the automotive design, replacing components traditionally made from steel. Its
lower stiffness and strength is compensated by considerable lower density. Castings and extrusions
are the most commonly encountered forms of aluminum parts preparation [3-5]. The material
drawbacks are mainly connected with relatively high primary production costs and technical
difficulties with steel to aluminum transition.
Magnesium is 75% and 30% lighter than steel and aluminum respectively. Its density
is comparable with the density of polymers. It has many advantages like very good machinability,
excellent damping capacity or the best strength-to-weight ratio among structural metals, but suffers
from the general corrosion resistance problems and difficulties with joining with other metals. The
material itself has been used in structural components for a relatively short time, thus recyclability
has not been well developed yet – only about 20-30% of the produced material is reused [5-7].

Tab. 1. Mechanical parameters of the chosen structural materials

Specific Specific
Density Young Modulus Tensile Strength
Material type Modulus strength
[g/ccm] [GPa] [MPa]
[GPa/g/ccm] [MPa/g/ccm]
Wrought Magnesium
1.8 45 380 25 211.1
AZ80 – T5
Wrought Aluminum
2.7 69 310 25.6 114.8
6061-T6
Unidirectional Carbon
Fiber Reinforced
1.6 135 1500 84.4 937.5
Plastic (standard
modulus)
E – Glass Fiber
1.9 40 1000 21.1 526.3
Reinforced Plastic

Composites consist of two or more separate phases, among which reinforcement and load
transmitting constituents can be found. A vast number of materials are commonly used in
composite design, but the most appropriate for structural applications are the fiber reinforced
plastics (FRP). Depending on the material composition, both: strength and stiffness can be higher
when compared to mild steel. The most commonly used FRPs are composed of glass, aramid or
carbon fibers in conjunction with epoxy or polymeric resins [8-10].
Modern structural materials have been also incorporated into mass transport vehicles, including
bus superstructures. Harte et al. in [11] presented a study on utilization composites in lightweight
rail cart, carrying out an optimization on walls’ thicknesses and geometrical features. Colombo and
Vergiani [12] presented a research on load carrying composite beams of a bus body, which was
pultruded from glass fiber and polyester resin. They assessed the fatigue resistance, strength and
weight of the examined parts, pointing out their superiority over the steel counterparts. Ko et al.
investigated the application of composite sandwich panels in the bus structure, showing their
usefulness in improving crashworthiness and rollover characteristics [13].
These mechanical performances along with the weight minimization were the subject of many
optimization attempts. The test procedures needed to evaluate a rollover resistance of bus
superstructures were established in 1987 by the Economic Commission for Europe in the ECE R66
regulation [14, 15]. The evaluation of the ECE R66 recommendations was done by Liang and Le
in [16], in which they investigated the structural deformations and occupants’ safety level,
comparing them with equivalent quantities obtained on the basis of similar American FMVSS 220
regulation. The same authors in [17] presented a study on enhancing the bus rollover resistance.
They identified the components that exhibited high internal energy level and applied on them
structural modifications that were obtained by means of a simple one-variable regression-based
optimization process. Similar studies were presented by Su et al. in [18] in which the authors

140
 Multi-Material Design Optimization of a Bus Body Structure

conducted a weight optimization of a bus body, considering the sidewall intrusion and static
stiffness constraints. Gauchia et al. [19] published results from a multi-objective optimization
attempt in which weight and torsional stiffness parameters were improved simultaneously. The
subject of the optimization process in both cases were only the thicknesses of sidewall beams, thus
the final structures consisted entirely of steel.

2. Optimization strategy

The target of the optimization process was to redesign the bus superstructure to improve its
mechanical characteristics. The bus body was first examined carefully in order to define the
components that could be modified in the following steps. The criterion was the easiness of
substitution and the efficiency of the possible modifications: only the elements which had a clear
impact on the structural deformations were taken into consideration. The selected group included
vertical beams that were parts of the pillars and the chosen sidewall beams – as shown in Fig. 1.
In order to save the computational time and to decrease the optimization complexity by
lowering the number of the design variables, the process was divided into two steps carried out
separately, as described in the following paragraphs.

Fig. 1. The baseline bus superstructure. The marked components were the subjects of the optimization: blue and red –
the first and the second run respectively

2.1. Design constraints

The boundary conditions described the geometry of the baseline structure and its mechanical
performance. The former limitations were related to the assembly process of a full vehicle,
preventing from changes in the components’ external dimensions. The latter assured that the
overall weight, torsional stiffness and side impact deflection of the optimized solution did not
decrease compared to the baseline design.

2.2. Design variables

The scope of the presented process was to exploit the properties of different materials used in the
nowadays automotive structural design. In the problem stated, five different material types were
considered as input variables: standard and high modulus carbon fiber reinforced plastics (SM-CFRP
and HM-CFRP respectively), glass fiber reinforced plastic (GFRP), aluminum and steel. Material
substitution was supplemented by the wall thickness optimization. Beams used in the sidewall
construction were assumed to be produced by means of pultrusion process, thus reinforcing fibers
were oriented along the components, what excluded any orientation angle changes.

141
 J. Korta, T. Uhl

2.3. Optimization objectives

Regardless of the optimization phase, the targets were defined as the (Fig. 2):
- torsional stiffness,
- rollover resistance.
The torsional stiffness is considered as the deformation of a structure resulting from the
presence of vertical loads applied in the front suspension mounting points, with the rear
attachments fixed. The torsional stiffness coefficient is expressed by Eq. 1:
F ˜d , (1)
kt
M
where:
F – force,
d – distance between the load application points,
ij – twist angle.
The rollover conditions are described in detail in ECE R66 standard and will not be provided in
this paper.

Fig. 2. Body torsional (left) and rollover (right) deflections

2.4. Robust design

The optimization process described above was carried out for input arguments without taking
into consideration their variations, which can result from the production errors. In reality, the
geometry of profiles suffers from some degree of inaccuracy, normally expressed as the dimension
tolerances. The over-optimized solution is defined as the one that exhibits good, but unstable
behavior that is easily negatively influenced by the input.
To prevent similar situations, robustness of the final design must be verified, as depicted in
Fig. 3. In the example given, design A is more attractive because of the higher target function
output, but even small changes in the input parameter can considerable deteriorate this value.
Despite lower performance, the second solution is more stable against the input variations, thus it
is more preferable.
The variations in the input parameters were obtained by monte carlo sampling method, with
normal distribution of the values. To assess the stability of the output quantities, the relative
standard deviation (%RSD) was monitored – Eq. 2:
% RSD 100 ˜ V / x , (2)
where:
ı - standard deviation of the obtained objective values,
x - their mean value.

142
 Multi-Material Design Optimization of a Bus Body Structure

Fig. 3. An idea of robust optimal design. Solution B is more preferable, because of the higher robustness against
the input function variations

However, despite the advantages, the robustness examination can be very time consuming
because of the need of additional analyses. To prevent unacceptable computational expenses, only
selected solutions that exhibited the most desirable improvements were tested.

2.5. Genetic algorithm

In order to find the optimal solution, the non – dominated sorting genetic algorithm II (NSGA-
II) was exploited. The main features of the algorithm are:
- domination estimation and fitness assignation in every generation,
- crowding distance evaluation,
- selection of parents on the basis of the above.
The subject of GAs is extensive, thus will not be covered in this article. A detailed description
of the exploited algorithm can be found in [20, 21].

3. Optimization and the final solution

As mentioned above, the optimization of the bus superstructure was divided into two stages:
modification of the vertical beams being part of the pillars (blue colour in Fig. 1) and changes of
the selected beams of the sidewall construction (red colour in Fig. 1).
After the selected number of simulation had finished, a number of designs that exhibited
improvements compared to the baseline strength and weight parameters were found. In the case of
a multi-objective optimization, the choice regarding the final design is always a matter of trade-off
between the outputs categorized as quasi-optimal in the Pareto sense. The choice is made between
solutions, for which an improvement in one quantity means a deterioration in others.
Table 2 presents the results obtained from the first optimization phase, in which the pillars
were the subject of modifications. The presented designs were selected because of having the most
profitable material distribution. In the following step, the robustness against the geometrical
variations of all the presented solutions was tested, which in conjunction with economical
premises was the decision-making criterion.
The material composition that provided the highest level of improvements, was clearly a mix of
SM-CFRP and HM-CFRP. If other materials were used, not only the improvements were less attractive,
but also the stability of the obtained solutions suffered. Aluminum was neglected by the optimization
algorithm, what was caused by the demanding boundary conditions, concerning the geometry.
Because of the satisfactory level of improvements and material composition (Fig. 4), resulting
in good material price and efficiency compromise, solution no. 7 was chosen as the input for the
following optimization phase.

143
 J. Korta, T. Uhl

Tab. 2. Results obtained from the first optimization run: pillars modification
Mean
Mean mass Mean torsional
%RSD rollover %RSD
relative to stiffness
Robust Material %RSD torsional deflection rollover
the relative to the
design id composition mass [%] stiffness relative to deflection
baseline baseline
[%] baseline [%]
[%] [%]
[%]
SM-CFRP,
P1 HM-CFRP, 97.36 0.021 106.55 0.16 96.41 0.29
GFRP
SM-CFRP,
P2 HM-CFRP, 97.36 0.015 106.59 0.15 96.33 0.19
GFRP
SM-CFRP,
P3 HM-CFRP, 97.45 0.016 106.56 0.15 95.95 0.18
GFRP, steel
SM-CFRP,
P4 97.34 0.015 106.80 0.16 95.95 0.25
HM-CFRP
SM-CFRP,
P5 97.35 0.013 106.82 0.15 95.13 0.09
HM-CFRP
P6 HM-CFRP 97.99 0.011 111.47 0.05 87.33 0.09
SM-CFRP,
P7 97.35 0.014 107.06 0.13 95.41 0.28
HM-CFRP
P8 HM-CFRP 97.98 0.011 111.68 0.07 87.42 0.10
P9 HM-CFRP 98.01 0.012 111.44 0.31 87.20 0.10

During the second optimization run, the genetic algorithm had more difficulties with finding
non-dominated designs providing an acceptable level of mechanical performance enhancements,
thus the new Pareto front was less numerous. Nonetheless, three propositions were selected for
further stability assessments, the results of which are combined in Tab. 3.
The analysis of the presented results leads to the conclusion, that the applied material
modifications are not recommended. Although some degree of improvements in the torsional
stiffness and overall weight were obtained, the rollover resistance was deteriorated. Moreover, the
small observed outcomes were found exclusively by application of HM-CFRP, what would
significantly increase the production cost.
Basing on the above, the P7 solution was chosen as the final one. As depicted in Tab. 2,
compared to the baseline all-steel bus body, the optimized structure exhibits better rollover
resistance (4.59%) and torsional stiffness (7.06%). The optimization provided also 2.65% savings
in weight.
Tab. 3. Results from the second optimization run: sidewall beams modifications
Mean Mean
Mean mass torsional rollover %RSD
%RSD
Robust Material relative to %RSD mass stiffness deflection rollover
torsional
design id composition the baseline [%] relative to relative to deflection
stiffness [%]
[%] the baseline baseline [%]
[%] [%]
B1 HM-CFRP 96.93 0.006 107.60 0.03 95.97 0.01
B2 HM-CFRP 96.91 0.006 107.55 0.02 95.98 0.02
B3 HM-CFRP 96.93 0.005 107.64 0.02 95.97 0.01

4. Conclusions

The two step optimization process was carried out in order to improve the mechanical
characteristics of the bus superstructure. The design variables were defined as the material type

144
 Multi-Material Design Optimization of a Bus Body Structure

and thickness of the selected sidewall beams, while the objectives were the mechanical
performances under torsion and rollover conditions and the structure curb weight.
The first optimization phase (pillars modification) provided significant improvements in all the
considered targets, by SM-CFRP and HM-CFRP application. Furthermore, the robustness against
the production errors of the tested Pareto-optimal solutions was confirmed.

Fig. 4. final design modifications

The following optimization of sidewalls reinforcement beams was more demanding. The
optimizer did not found simultaneous improvements in all of the targets, causing slight
deterioration in rollover resistance. Moreover, any profitable modifications were possible only by
application of the expensive HM-CFRP, thus this modification was not recommended. The final
solution has been chosen from among the designs found in the first optimization run.

References

[1] Lutsey, N., Review of Technical Literature and Trends Related to Automobile Mass-
Reduction Technology, USC-ITS-RR-10-10, Institute of Transportation Studies, University of
California, May, 2012.
[2] Fuel economy and traffic fatalities: multivariate analysis on international data” and “Light-
Duty Automotive Technology and Fuel Economy Trends: 1975 Through 2006”, U.S.
Environmental Protection Agency, Appendix D, July 2006.
[3] Salonitis, K., Pandremenos, J., Paralikas, J., Chryssolouris, G., Multifunctional Materials
Used in Automotive Industry: a Critical Review, Engineering Against Fracture: Proceedings
of the 1st Conference, pp. 59-70, doi: 10.1007/978-1-4020-9402-6_5, 2009.
[4] Fridlyander, I. N., Sister, V. G., Grushko, O. E., Berstenev, V. V., Sheveleva, L. M., Ivanova,
L.A., Alluminum alloys: promising materials in the automotive industry, Metal Science and
Heat Treatment, Vol. 44, No. 9-10, UDC 669.5:629.113/.115, 2002.
[5] Cole, G. S., Sherman, A. M., Lightweight materials for Automotive Applications, Material
Characterization 35:3-9, SSDI 1044-5803(95)00063-5, 1995.
[6] Caceres, C. H., Economical and Environmental Factors in Light Alloys Automotive
Applications, Metallurgical and Materials Transactions A, Vol. 38A, pp. 1649 – 1662,
doi: 10.1007/s11661-007-9156-z, 2007.
[7] Goede, M., Stehlin, M., Rafflenbeul, L., Kopp, G., Beeh E., Super Light Car – lightweight
construction thanks to a multi-material design and function integration, European Transport
Research Review, 1:5-10. DOI: 10.1007/s12544-008-0001-2, 2009.
[8] Cramer, D. R., Taggart, D. F., Design and Manufacture of an Affordable Advanced-
Composite Automotive Body Structure, Proceedings of the 19th International Battery, Hybrid
and Fuel Cell Electric Vehicle Symposium & Exhibition, 2002.
[9] Boeman, R. G., Johnson, N. L., Development of a Cost Competitive, Composite Intensive

145
 J. Korta, T. Uhl

Body-in-White, SAE Technical Paper 2002-01-1905, doi:10.4271/2002-01-1905, 2002.

[10] ĝleziona, J., Podstawy technologii kompozytów, Wydawnictwo Politechniki ĝląskiej, ISBN
83855718842, 1998.
[11] Harte, A. M., McNamara, J. F., Roddy, I. D., A multilevel approach to the optimisation of
a light rail vehicle bodyshell, Composite Structures 63, pp. 447-453, 2004.
[12] Colombo, C., Vergani, L., Experimental and numerical analysis of a bus component in composite
material, Composite structures 92, 1706-1715, doi:10.1016/j.compostruct.2009.12.012, 2010.
[13] Ko, H.-Y., Shin K.-B., Jeon K.-W., Cho S.-H., A study on the crashworthiness and rollover
characteristics of low-floor bus made of sandwich composites, Journal of Mechanical Science
and Technology, 23, pp. 2686-2693. Doi: 10.1007/s12206-009-0731-7, 2009.
[14] United Nations Economic Commission for Europe, ECE/R66/01, Uniform Technical
Prescriptions Concerning The Approval Of Large Passenger Vehicles With Regard To The
Strength Of Their Superstructure, Geneva 2006.
[15] RusiĔski, E., Czmochowski J., Smolnicki, T., Zaawansowana metoda elementów skoĔczonych
w konstrukcjach noĞnych, Oficyna Wydawnicza PWr, 2000.
[16] Liang, C.-C., Le, G.-N., Analysis of a bus rollover protection under legislated standards
using LS-DYNA software simulation techniques, International Journal of Automotive
Technology, Vol. 11, No. 4, pp. 495-506, doi: 10.1007/s12239-010-0061-x, 2010.
[17] Liang, C.-C., Le, G.-N., Optimization of bus rollover strength by consideration of the energy
absorption ability, International Journal of Automotive Technology, Vol. 11, No. 2,
pp. 173-185, doi: 10.1007/s12239-010-0023-3, 2010.
[18] Su, R., Gui L., Fan, Z., Multi-objective optimization for bus body with strength and rollover
safety constraints based on surrogate models, Struct Multidisc Optim, 44: 431-441, doi:
10.1007/s00158-011-0627-x, 2011.
[19] Gauchia, A., Diaz, V., Boada, M. J. L., Boada, Torsional Stiffness and weight optimization of
a real bus structure, International Journal of Automotive Technology, Vol. 11, No. 1,
pp. 41-47 (2010), doi: 10.1007/s12239-010-0006-4.
[20] Beyer, H. – G., Deb, K., On self-Adaptive Features in Real-Parameter Evolution Algorithm,
IEEE Transactions on, Vol. 5, Is. 3, pp. 250-270, 2001.
[21] Deb, K., Agarwal, R. B., Simulated Binary Crossover for Continuous Search Space, Complex
Systems, 9:115-148, April, 1995.

146

View publication stats