Structural and Multidisciplinary Optimization (2021) 63:2509–2529

https://doi.org/10.1007/s00158-020-02812-z

INDUSTRIAL APPLICATION PAPER

Designing lightweight diesel engine alternator support bracket

with topology optimization methodology
Erkan Talay 1 & Cengiz Özkan 2 & Emre Gürtaş 2

Received: 28 June 2020 / Revised: 24 October 2020 / Accepted: 4 December 2020 / Published online: 4 January 2021
# The Author(s), under exclusive licence to Springer-Verlag GmbH, DE part of Springer Nature 2021

Abstract
Decreasing fuel consumption and tailpipe emissions have been an important issue nowadays for automotive industry after new
regulations. Fuel consumption is also an important parameter that highly affects the customers’ choice. Reduction of total vehicle
weight with optimization is one of the common methods to improve fuel consumption and emissions. There are a lot of studies on
body, interior, and engine components of the vehicle to improve fuel efficiency and emissions including design modifications,
using alternative material and architectural optimizations. In this article, the optimization of an aluminum support bracket that
carry alternator and hydraulic steering pump is studied. First, design space is determined with layout analysis defining constraints
and fixing points. Then, topology optimization method is applied in order to explore the load carrying path of the component.
Final design is frozen through computational analysis verifications and industrialization phase modifications. A comparison of
the new design with the actual production showed that the mass of the current design could be reduced by 37%, while all product
expectations were met.

Keywords Powertrain . Alternator support bracket . Finite element simulations . Topology optimization . Lightweight

1 Introduction reduced by 6.9% and 7.6%, respectively, (MIT 2008) and

the emissions of CO2 will decrease by 8–11 g per kilometer
Automotive industry is very competitive today and changing (Wallentowitz 2004).
very fast. Automotive manufacturers have to reduce vehicle Weights of components on the vehicle should be decreased
cost without compromising from quality to compete and sur- to reduce total weight of the vehicle. The main approach is to
vive. Optimization methods became crucial in the last decade start by inspecting the components of the vehicle which are
for this purpose. The reduction of overall weight of the vehicle too heavy. When expected performance from parts is defined,
is also helping to decrease fuel consumption and exhaust targeted weight reduction can be realized by applying modifi-
emissions. For every 100 kg weight reduction, combined cations on design and using alternative materials. These
city/highway fuel consumption could decrease by 0.40 l/ weight reductions can be reached easily by utilizing analysis
100 km for cars, and 0.49 l/100 km for light trucks. In other programs that are widely used in automotive research and
words, for every 10% weight reduction from the average new development centers. Unforeseen problems and unnecessary
car or light truck’s weight, the vehicle’s fuel consumption cost up can be noticed and avoided beforehand with the help
of new simulation methods. Optimum designs that are con-
firmed by analyses can be produced by new prototype pro-
Responsible Editor: Ren-Jye Yang duction methods and can be verified according to serial pro-
duction conditions.
* Erkan Talay Structural optimization aim is to find optimal material dis-
[email protected] tribution according to given design domain or space. Some
common functions are to minimize mass and displacement or
1
Virtual Analysis, TOFAS, Istanbul Cad. No: 574, maximize eigenvalues. By structural optimization new holes,
16110 Bursa, Turkey voids can be added to the structure. Structural optimization
2
Powertrain Engineering, TOFAS, Istanbul Cad. No: 574, can be divided into three types: size optimization, shape opti-
16110 Bursa, Turkey mization, and topology optimization (Christensen and
2510 E. Talay et al.

Klarbring 2009). Topology optimization is the most common activities. The aim of topology optimization is to find the most
type of structural optimization and is useful for determining suitable design by not violating the constraints such as dis-
optimal material distribution and voids. Optimization is an placement, rigidity, and natural frequency. Generally, the pur-
iterative process to make a system or component as good as pose of the optimization problem is to maximize or minimize
possible based on an objective function, given design space the response function. The design space is divided into small
and constraints. Topology optimization aims to find best use elements by the finite element method. Then, some elements
of material by achieving requirements such as stiffness, eigen- which do not carry load are killed during the optimization
values, and displacement (Bendsøe 1995). Vehicle product iterations (Chang 2015a, 2015b).
development process can be divided into four stages: concep-
tual design, detailed design, prototype production, and physi-
cal testing phases. The application of structural optimization
has been generally used in the detailed design phase of auto- 2 Numerical studies
motive and aerospace engineering to reach lightweight prod-
ucts. Wei Zhong has applied a genetic algorithm–based meth- The objective of this study is to explain how to perform to-
od for multi-objective topology and sizing optimization of bus pology optimization for an alternator bracket, which is used in
body frame with manufacturing constraints where the weight automotive engine to assemble alternator and steering pump
of the bus frame is decreased 97.9 kg (3.43%) while the tor- to engine body which is shown in Fig. 1. In order to prevent
sional stiffness is increased 4.89 kNm/deg (Zhong et al. 2016). resonance, the first natural frequency of all the engine compo-
Chao Li has applied topology optimization method on engine nents must be higher than engine working frequency. In addi-
cradle design and an optimum conceptual design was deter- tion to that, the bracket should also be resistant for stresses
mined which minimizes weight while satisfying all stiffness originated from both vibrations, weight of carried parts and
constraints. Based on topology optimization results, a design belt loads during the operation conditions. Since the major
review was conducted, and a revised model was created which forces originated from belt contact during engine working,
addresses all structural and manufacturability concerns. The belt forces and first eigenfrequencies are selected for optimi-
initial design domain had the mass of 82.6 kg; topology opti- zation study. After reaching an optimum, vibration-induced
mization in conceptual design reduced the mass to 26.7 kg; stresses have been checked to guarantee the durability perfor-
and the detailed design task involving shape and size optimi- mance. In addition, stresses arose from belt forces and
zation further reduced the mass to 21.4 kg (Chao et al. 2014). bump/hole passing, and rear crash has been also checked for
When designing actual structures using topology optimization robust design to satisfy the customer’s worst case driving
methods, the manufacturability should be reasonably consid- conditions.
ered. Otherwise, it may result in an optimal design that could Element density method has been chosen for the support
not be manufactured. In order to integrate the topology opti- bracket optimization. The main goal of optimization is to
mization with the manufacturability, manufacturing con-
straints need to be included in the formulation of topology
optimization. Bo Wang has proposed to use the Helmholtz
type anisotropic filter to implement the extrusion constraint
in topology optimization. The proposed method needs less
manual setup in the preprocessing of the analysis model for
topology optimization, and thus, the background mapping
mesh and the neighborhood search are not required (Wang
et al. 2020). L. Rakotondrainibe presented a coupled method
for the optimization of diesel engine accessories bracket as-
sembly and the locations of rigid supports. Accessories are an
alternator, a belt tensioning roller, and an air conditioning
(AC) compressor. Rigid supports are modeled by Dirichlet
boundary conditions while bolt-like connections are idealized
and simplified as a non-local interaction to be representative
enough at a low computational cost. This study reveals about
1.5 kg of lightening of the bracket compared to a classical
optimization of its structure corresponding to an additional
35% improvement (Rakotondrainibe et al. 2020).
Since the shape of the part comes out of topology optimi-
zation, it is generally used at concept phase to lead the design Fig. 1 Support bracket assembly on engine
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2511

obtain both robust and producible design with minimum ma- stress and strain calculations (Bartosz and Podgórski 2017).
terial usage. Design space is created by using UG NX design Bolted connections have been created using 1-dimensional
software. On the other hand, finite element model is created by CBEAM elements as can be seen in Fig. 3.
using HyperMesh 2017 and OptiStruct is used for optimiza- When modeling the alternator and power steering pump
tion calculations. Minimum value of first natural frequency of mounted on support bracket, only the outer surfaces of these
bracket and maximum deformation value of alternator and parts are added inside the model as representative of compo-
steering pump center of gravity points at belt loading case nents. The mass and inertia moment information which can be
have been defined as optimization constraints. Draw direction seen in Tables 1 and 2 were obtained from the design group
constraint has been also added as manufacturing constraint. and added inside the model with the actual weight values as 1-
OptiStruct optimization analysis results have been reviewed dimensional CONM2 elements. While doing this, a local co-
with HyperView and in order to export conceptual design, ordinate system was created for the mass center of each part,
Ossmooth tool has been used. and the moment of inertia information was defined according
to these local coordinate systems (Figs. 4 and 5).
2.1 Modal analysis Since the engine is composed of lots of components, it will
be more realistic to create contact definitions in simulations.
All the component’s first natural frequency, which are at- The surfaces which touch each other were selected and contact
tached on the engine, must be higher than engine idle frequen- surfaces created to be used in interface definition. TIE contact
cy to prevent resonance-related problems on components. A type with friction has been selected and created for all the
very detailed finite element model of engine assembly has mating surfaces which are shown in Fig. 6.
been created for modal analysis of support bracket in Altair/ Bracket material is AlSi10Mg which is a universal alumi-
Hypermesh software as shown in Fig. 2. Weight of the bracket num alloy with very good castability suitable for pressure die
has been calculated as 1.9 kg on finite element model. Support casting. High-pressure die cast components reduce vehicle
bracket is simply mounted on the engine and carries the alter- weight, so it has positive impact on performance and fuel
nator and power steering pump. The finite element model efficiency. This trend is driven by the needs of the automotive
consists of nearly 3 million solid CTETRA elements. manufacturers to substantially reduce weight of the power
Since the support bracket geometry is very complex, it was train and chassis components (Hirsch 2004). Elastic modulus,
not possible to model it with hexagonal elements which pro- Poison’s ratio, and density of aluminum alloy have been de-
duces more accurate results. Finite elements of support brack- fined in Hypermesh with MAT1 card.
et have been created by using nearly 225,000 tetragonal ele- Alternator bracket main function is to sustain static and
ments to be able to create simulation model of such a complex impulsive loads produced by engine functioning and dynamic
geometry. However, 10-noded second-order tetragonal ele- loads produced by components fixed on alternator support
ments were chosen especially for FRF and static loading anal- bracket. Support bracket should not amplify dynamic engine
yses accuracy. Preprocessor will add extra nodes between displacements, due to resonances coming from the bracket. If
nodes of tetra elements, and this will increase accuracy for a resonance is in frequency range produced by the engine,

Fig. 2 Finite element model of support bracket

2512 E. Talay et al.

Fig. 3 Creating bolted

connections

amplification coefficient is big (from 2 to 5) and the size of 2.2 Belt forces on alternator and pump
resonance zone is high (50–100 Hz). High amplification fac-
tors in resonance can influence vehicle noise and comfort. The power steering pump and the alternator mounted on the
Support bracket first main frequency must be higher than carrier bracket are connected to each other and to the engine
maximum frequency generated from the engine from the last via a belt which delivers power to the car’s alternator, water
harmonic of significant amplitude. Minimum value suggested pump, power steering pump, and air conditioner compressor
for bracket first resonance frequency depends on velocity of (Leamy 2003). As the engine’s crankshaft pulley turns, it
engine rotation and on amplitude of superior order harmonics, drives the serpentine belt. In turn, the serpentine belt drives
in case that these do not have unimportant amplitude than the the pulleys for the other systems including the air condition-
ones with lower order. The critical frequency can be seen in ing, alternator, water pump, cooling fan, and power steering.
Fig. 7 on Campbell diagram which simultaneously represents Belt forces acting on alternator and steering pump are shown
excitations and vehicle system modes on an RPM vs in Fig. 10.
Frequency axis system which is a more convenient way of Serpentine belt drives are the automotive and truck industry
understanding the interaction among modes and excitation standard for front-end accessory drives. The steady-state ten-
(Boyce 2011). sion in each belt span is a major factor affecting belt slip and
Finite element model has been constrained on engine vibration. These tensions are determined by the accessory
block. The constrained zone can be seen in Fig. 8. loads, the accessory drive geometry, and the tensioner prop-
Modal analysis of base model performed until 800 Hz. And erties. Keeping proper belt tensions to provide adequate
first critical frequency calculated as 176 Hz for support brack- torque without belt slip is important for the durable, quiet
et which is ok according to engine rpm. First two mode shapes operation of serpentine drives and it creates forces on alterna-
can be seen in Fig. 9. tor and pump center (Beikmann et al. 1997). These belt forces

Table 1 Alternator CONM2 element Table 2 Power steering pump CONM2 element

Weight [kg] 6 Weight [kg] 1.57

Center of gravity [mm] Gx Gy Gz Center of gravity Gx Gy Gz
− 1.1 − 2.14 − 72.3 [mm] 41.35 6.85 0.034
Moment of inertia Ixx Iyy Izz Ixy Ixz Iyz Moment of inertia Ixx Iyy Izz Ixy Ixz Iyz
[kg mm2] 12,292 13,403 12,668 −345 43 151 [kg mm2] 1376.7 1739.8 2064.8 108 − 1.3 − 12.16
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2513

Fig. 4 Finite element model of

alternator

are applied on finite element model to check the strength of important to add pretension step to have more accurate stress
the support bracket which is mounted on the engine block results. Bolt pretension increases the stiffness of the structure
with 4 M8 bolts. The power steering pump and alternator and thus causes the shear stress to be transferred mostly be-
are also connected to the carrier bracket with 3 M8 bolts. tween the plates through friction (Novoselac et al. 2016).
When designing a structure which is clamped by bolts, it is Stress results under maximum belt loads can be seen in

Fig. 5 Finite element model of

steering pump
2514 E. Talay et al.

Fig. 6 Contact interface

definitions

Fig. 11. According to the simulation results, displacement at iterative procedures known as the local approximation method
alternator center is 0.22 mm and at steering pump center is to reach the global minimum or maximum by using conver-
0.14 mm which have been transformed to stiffness target at gence test and response screening. During the solution design,
optimization problem. variables are changed within their bounds to achieve a stable
convergence. In topology optimizations, variables are the el-
2.3 Topology optimization of support bracket ement densities and the algorithm tries to kill the elements
having 0–0,3 densities and leaves the elements which have
Topology optimization generates optimized material distribu- densities bigger than 0.3. Mesh quality and element types
tion for a set of loads and constraints within a given design are also important to reach meaningful results especially when
space (Lee and Lee 2003). The classical topology optimiza- first-order finite elements are used. Minimum member size
tion sets up solving the minimum compliance problem which control is used for the problem of mesh dependency of solu-
implies finding a structure with the smallest displacement (Lee tions in topology optimization. With increasing mesh density,
et al. 2003). Manufacturing constraints can also be imposed the solution of a broad class of problems tends to form an
using a draw direction constraint, extrusion constraint, and increasing number of members with decreasing size (Zhou
symmetry planes (Chao et al. 2014). Optimization codes use et al. 2001). When the optimization is converged, the

Fig. 7 Critical frequency for

support bracket
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2515

Fig. 8 Constrained areas for

modal analysis

remained elements’ shape is used to create the new solution design variable (or design voxel), E0 indicates the real
satisfying the target. Altair/OptiStruct solver has been chosen Young’s modulus of the base material, and p is the penalty
to solve the optimization problem. OptiStruct is one of the factor whose value is greater than one.
software modules of HyperWorks, which is the principal OptiStruct has been widely accepted by the automotive
product offered by Altair Engineering—a product design industry (Chang 2015a, 2015b). Since OptiStruct does not
and development, engineering software, and cloud computing have any graphical interface, problem formulation has
software company in Detroit, MI. OptiStruct uses solid isotro- been made using Hypermesh preprocessor and solved by
pic microstructure with penalization (SIMP) method. SIMP command line entries. 3D design space of support bracket
method converts a continuous optimization model into a was created by using the packaging envelope surface
discretized model in which a penalty factor is used to drive which can be seen in Fig. 19. When modal analysis was
the intermediate density to converge to either 0 (void) or 1 performed with design space, the first frequency has been
(solid). The SIMP model has been proved to be efficient in found at 257 Hz with 3.9 kg which sets the reachable upper
solving a variety of engineering problems. The penalization is limit of the frequency.
achieved by the following formulation: Topology optimization problem of support bracket has
been formulated as:minimize

E ¼ ρpi E 0 m
W ð ρ i Þ ¼ ∑ ρ i vi
i
where E represents the penalized Young’s modulus of the ith

Fig. 9 Modal analysis results of First Mode: 176 Hz Second Mode: 359 Hz
base design
2516 E. Talay et al.

Fig. 10 Belt forces on alternator and pump center

subject to
8
> K ðρÞU ¼ P
<

uj ≤Dj Fig. 11 Stress map at maximum belt loading
>
: f k ≥ Fk
0 < ρmin ≤ ρi ≤ 1
design space. Since the attachments must keep their locations
where the objective is to minimize the weight ρi indicates to be able to assemble on the engine, the elements near the
the relative density of the ith design variable and vi represents attachments are extracted from design space and defined as
the volume of ith element. P and U are the matrix of nodal frozen elements in the optimization which are shown as red in
force and displacement, respectively, and K is the global stiff- Fig. 12.
ness matrix. Uj denote the total displacements of the jth load- The optimization target has been set as minimizing the
ing point, respectively, Dj is the corresponding upper bound mass of the bracket. Base model’s weight is 1.9 kg. The ma-
for total displacements of the jth loading point, respectively. j terial of the bracket is AlSi10Mg and the bracket is produced
represents the total number of loading points, and since belt by high-pressure die casting method. With this production
loads are applied on pump and alternator centers, its value in method, by producing parts from lighter alloys, a positive
this study is 2. Fk indicates the frequency targets of kth com- effect is obtained on vehicle performance and fuel consump-
ponent and fk denotes frequency values of kth component, tion. In order to have a manufacturable output in optimization,
respectively. Since only pump and alternator are assembled the exit direction of the mold is defined as the production
on support bracket, the value of k is 2 in the present study. ρmin constraint in OptiStruct code.
is the minimum allowable bound for the density which is Stiffness targets from belt loading scenario and frequency
implemented in order to avoid numerical difficulties, and m targets for assembled components on support bracket have
is the total number of design variables (around half million). been defined as constraint in topology optimization problem.
Design space is created mainly with 3D hexagonal ele- Stiffness targets have been transformed to displacement con-
ments which is shown as green in Fig. 19. Optimization target straints and frequency constraints have been defined as the
is to decrease the volume of the defined design space without first modes of the alternator and steering pump components.
violating constraints that come from modal and static analysis. Constraints are listed in Table 3 and simulation results can be
Design variables are the element densities inside the defined seen in Fig. 13. Topology optimization was performed on
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2517

Fig. 12 Aluminum support

bracket design space

AL Support Bracket Design Space

Intel(R) Xeon(R) 16 CPU E5-2670 0 @ 2.60GHz Linux serv- Again, as was done in the first place, the mesh model of
er. After 30 iterations which took about 3.5 h, optimization this latest design is created, and modal analysis is per-
converged with all constraints satisfied. formed. As a result of the analysis, the first natural fre-
Optimization outputs of OptiStruct generally look like quency of the alternator support bracket increased to
paths instead of design. The outputs of OptiStruct need to be 194 Hz with 1.2 kg which means a 37% weight reduction.
corrected with a separate tool to better understand the results. The comparison of results with base design and design
The Ossmooth tool in OptiStruct is used to transfer the con- space can be seen in Table 4.
cept design to the CAD data. The aim is to obtain a smooth
and producible geometry. The corrected output of load paths
with Ossmooth tool can be seen in Fig. 14. When modal
analysis was repeated with topology optimization output, the
first frequency of the alternator support bracket was found as
184 Hz with 0.7 kg which means a 63% weight reduction with
respect to the base design.
Topology optimization results and load carrying paths have
been discussed with design and production departments.
Since the bracket is produced with high-pressure die casting
method, production voids inside the component were a big
problem. After several loops, new design idea was brought
to life in UG environment which can be seen at Fig. 15.
After reaching the feasible design, it is necessary to
check whether this latest design also fulfills the objectives.

Table 3 Topology optimization constraints

First frequency [Hz] 190

Second frequency [Hz] 210
Pump center displacement [mm] 0.14
Alternator center displacement [mm] 0.22
Fig. 13 Topology optimization on OptiStruct
2518 E. Talay et al.

Table 4 Topology optimization results comparison

Weight 1st 2nd

Frequency Mass
Reduction Mode Mode
Comparison (kg)
(kg) (Hz) (Hz)
Base Design 1.90 - 176 359
Design Space 3.90 - 257 496
Optimized Shape 0.70 1.20 / 63% 184 230
Producible Design 1.20 0.70 / 37% 194 305

can cause the fatigue problems. Stress levels at all load cases
have been checked and compared with respect to the base
design to validate the robustness of the solution.

2.4.1 Vibration induced durability load case

Fig. 14 Extracted load carrying paths with Ossmooth
Vehicle structures and engine-mounted components are most-
ly subjected to dynamic random loads. The engine excitations
2.4 Stress check at static and durability load cases are the main dynamic loadings for engine-mounted compo-
nents which are exposed to high vibrations produced by com-
While the vehicle is running, different types of loads are trans- bustion firing, misalignment of the pistons, and balancing is-
mitted to the support bracket in transient domain. Static anal- sues. Engine vibrations have been measured with a tri-axial
ysis scenarios could be designed to measure the strength of the engine dynamometer running from idle speed to the maxi-
bracket under the forces exposed by the support bracket dur- mum engine speed (Wang et al. 2017). These vibrations are
ing engine operation throughout the vehicle’s lifetime. These both harmonic and random in nature. Engine behavior is har-
mission loads are classified as belt load, bump passing, and monic at low frequency range up to 500 Hz and it gets random
rear crash. Apart from the static load cases, engine vibrations in the high frequency range because of variations in

Fig. 15 New support bracket

design
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2519

combustion, valve closing impact, sticking of pistons, and Engine vibrations can initiate cracks on engine-mounted
other moving components. The accelerometer sensors are components and random loads finally lead to fatigue fail-
placed on cylinder heads during the test. Measured engine ure. Fatigue life of engine-mounted components can be
vibrations at vertical, lateral, and longitudinal directions in estimated with stress approximation method according to
time domain can be seen at Fig. 16. the infinite life design approach. If calculated stress level is
For all the components mounted on vehicle engine, it is under the 60% of yield strength of the material, it can be
important to understand the stress state under dynamic load- said that components will have infinite life (Mayer et al.
ings. The frequency response function (FRF) is the measure 2005). Applied dynamic loads on cylinder heads create
of the dynamic response output of a system in response to an stress on support bracket. According to the simulation re-
input load with a range of frequencies and is used to character- sults, peak stress at 167 Hz has been calculated as 18 MPa
ize the dynamics of the system (Lima et al. 2015). Measured for actual production part and stress levels have been in-
engine vibrations in time domain are transformed to frequency creased to 19.6 MPa with optimized bracket but it stays in
domain to be able to calculate the realistic response on alterna- a very safe region for aluminum material. Stress compari-
tor and steering pump. Transformed accelerations into frequen- son can be seen in Fig. 19.
cy domain and load application points can be seen in Fig. 17.
Simulation results are viewed in Altair/Hypergraph. 2.4.2 Static load cases
According to the simulation results, accelerations on alternator
center have been increased from 5.9 to 7.2 g which can be The most important loads are generated on support bracket
evaluated as ok according to the FCA (Fiat Chrysler that comes from belt assembly. As it is described at
Automobile) core design documents. Calculated accelerations Section 2.2, when engine is working, belt follows its path
on alternator component can be seen in Fig. 18. and creates forces on alternator and steering pump centers.

5000
4000
3000
2000
1500

19.4
10.0
0
-10.0
-18.2

15.3

-10.0

-18.9

14.8

-10.0
-15.6

Fig. 16 Vertical, lateral, and longitudinal engine vibrations

2520 E. Talay et al.

Fig. 17 a Loading in frequency Acceleraon

domain. b Load application
points
(a)

Acceleraon [m/s2]
Frequency [Hz]

(b)

Fig. 18 Accelerations on
alternator
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2521

Fig. 19 Stress comparison on

support bracket

These forces create stress on the attachment points of support – Alternator assembly to bracket
bracket. The stress levels must remain low with respect to the – Hydraulic steering pump assembly to bracket
aluminum yield strength to have a robust design. Another – Bracket assembly to engine block
static scenario is passing on bumps and holes. When the ve- – Bracket layout verification with other components
hicle is crossing over bumps and holes on the road, vertical
accelerations will be transmitted to alternator and pump over
support bracket and creates stress on attachment points. These Related critical distances and dimensions on layout verifi-
scenarios can be realized by applying measured ± g loading on cations have been resolved by updating 3D models and final
the engine. Another important scenario on the support bracket design has been frozen.
is the load transfer during the crash. Measurements from full- Additive manufacturing (AM) is a joining process to build
scale crash tests showed that 27 g loading will be enough to objects from a 3D mathematical data created from computer-
simulate the rear crash case in static loading. Stress compari- aided design (CAD) tools. Usually, these 3D models are com-
sons can be seen in Table 5. According to the simulation posed by individual and separated by thin layers. Direct metal
results, stress values on optimized bracket are less than the laser sintering or DMLS is an AM technology especially has
original component and optimized design could be evaluated been used in medical, aeronautical, and aerospace industries.
as robust in terms of operating loads. This technique allows to manufacture parts by fusing thin
layers of metal powder, such as stainless steel, titanium, alu-
minum, nickel, or cobalt alloys, among others (López-Castro
et al. 2017). Prototypes are produced by utilizing metal
3 Prototype verifications sintering with 3D printing method using AlSi10Mg material.
Although 3D printed prototype components are not so signif-
After reaching the feasible lightweight support bracket design, icant for material from production point of view, preliminary
prototypes have been produced to check packaging status and verifications for layout, assembly, and modal tests could be
assembly problems. Packaging analysis which can be seen in realized before production phase. Prototypes can be seen in
Fig. 20 have been performed to verify the proper distances Fig. 21.
between: Production tools and fixing elements have been verified by
assembly of support bracket on engine which is shown in
Fig. 22. Physical layout packaging verifications have been
Table 5 Stress comparison at static loading cases also completed and frozen on prototypes.

Load case Actual bracket Optimized bracket

Max belt load (2kN) 95 MPa 74 MPa 4 Experimental verifications

Belt load + 4 g up 95 MPa 94 MPa
Belt load + 3 g down 93 MPa 90 MPa As soon as preliminary verifications finish on prototype
Crash (27 g rear) 91 MPa 87 MPa components, definitive aluminum injection mold tooling
phase has been started. Mold filling, solidification,
2522 E. Talay et al.

Fig. 20 Packaging verifications

temperature distribution, porosity, and velocity of the liq- such as ejection time, die opening sequence, delay time,
uid metal during high-pressure die casting were investi- die closing sequence, less time until the starting of the
gated using the model through numerical simulation. All- next cycle, individual control of each cooling or temper-
important parameters affecting heat flow and the thermal ing channel, illustration of die spraying procedure,
balance of the die can be regarded during the simulations, venting and vacuum, and local squeezing description

Fig. 21 Prototype production ACTUAL OPTIMIZED SHAPE INDUSTRIALIZATION

DESIGN
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2523

Fig. 22 Prototype assembly

verification

(Boydak et al. 2015). Various loops of mold flow analysis conditions, this phenomenon had to be improved before
were realized by the supplier to check porosity and hot continuing with further test steps. After effective actions
spot formation during the filling process. First topology taken by the supplier on production parameters, second
optimization output design was further improved within batch of components is tested radioscopically. During sec-
these loops to reach a producible design as close as pos- ond examinations, the level of porosity cavities has been
sible to the optimized shape by fine tuning of dimensions reported as 1–2 grades which is in acceptable limits with
and thicknesses of mathematical model and re-designing thickness of 5/8″ according to radiographic standard
of cooling jackets of the mold. After production approval ASTM E-505 which is shown in Fig. 24.
phase, first definitive production components were exam- Frequency analysis is realized on the components com-
ined for porosity in material lab. paring the actual and new design brackets. As seen in
Radioscopic examinations were realized on components Fig. 25, the results are reported as comparable not signal-
to verify if effective injection production phase was com- ing any critical issue.
pleted. First verifications signaled grade 3 shrinkage cavity An interior noise level test has been performed on test track
which can be seen in Fig. 23. It was above acceptable to validate new alternator bracket design from acoustics point
limits according to FCA internal standards. This level is of view. Noise levels have been measured by the microphones
informed as critical point which could result in component positioned at driver’s ear positions. The comparison of noise
failure during operation on the engine. In order to not risk levels between actual production (red line) and new design
any breakage during normal operation or fatigue (green line) is shown in Fig. 26. No critical difference was
observed in overall level up to 3200 rpm. The booming noise
has shifted to a low engine speed (from 3500 to 3300 rpm)
which can cause an issue for customers’ interior noise
perception.
Recorded interior noise level values were used to calculate
the articulation index values to be able to quantify the wors-
ening effect which can be seen at Fig. 27. The articulation
index (AI) gives a measure of the hearing speech in a given
noise environment. The basic interpretation of the AI value is
the higher the value, the easier to hear the spoken word (Talay
and Altınışık 2019).
According to the evaluations of articulation index re-
Fig. 23 Radioscopic examination results (1st loop) sults, 2% worsening has been calculated which has been
2524 E. Talay et al.

Fig. 24 Radioscopic examination

results (2nd loop)

evaluated as a minor effect (not audible in real-life con- problems and evaluate the effectiveness of corrective re-
ditions) on SAE assessments. Therefore, the mentioned design actions. The primary purpose is to improve reli-
modification has been evaluated as acceptable in terms ability performance rather than measure the reliability.
of NVH. The most important actions are to log all failures gener-
A final set of tests was realized on components to ated during testing and assure corrective actions are taken.
check the reliability status. There are two unique test This is verified through testing to ensure that failure-
phases within reliability development process: reliability prone aspects of the design are fixed. Reliability quantifi-
development testing and reliability growth testing. cation during this development phase is intent on measur-
Reliability development testing is performed to identify ing improvement. The purpose of a reliability growth test
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2525

Fig. 25 FRF comparison. a

Vertical. b Transversal. c
Longitudinal

is to determine whether a designed product meets or ex- between 7 and 20 °C. No difference reported from cus-
ceeds the established minimum reliability requirement tomer point of view by achieving SAE 6 performance
(Yung et al. 2015). Comparative subjective evaluation is which is the same with the actual bracket. In order to
realized with actual and optimized brackets on the same validate the performance and reliability of the new design
vehicles for reliability purpose. Vehicle has been driven bracket in long-term usage, reliability growth test has
200 km in urban, extra urban, and highway road condi- been performed. A new component has been assembled
tions while the external temperature was changing on a test vehicle and the vehicle has been driven for
2526 E. Talay et al.

Fig. 26 Overall sound pressure level comparison

6.000 km during different traffic, temperature, and road According to the modal analysis results, the minimum fre-
conditions using different driver profiles. The component quency of the alternator support bracket assembly has been
did not end up with any defects after the tests. calculated as 176 Hz with 1.9 kg support bracket. Another
important force for designing of support bracket is belt forces
which is explained in Section 2.2. When the engine is work-
5 Discussions ing, the serpentine belt drives the pulleys for the alternator and
power steering pump. Static displacements of alternator and
Alternator and steering pump are mounted on support bracket pump centers have been defined as constraint in topology
and subjected to dynamic loading that could make the system optimization. Topology optimization model, new design
oscillate at its resonance frequencies. To avoid resonance- space, and constraints with mass reduction target have been
originated problems, minimum frequency of the base design explained in Section 2.3. Topology optimization simulations
should be kept same in the optimized solution. In Section 2.1, showed that manufacturing constraints are important to reach
the details of the created finite element model are described. a feasible design after the optimization iterations. 1.2 kg

Fig. 27 Articulation index comparison

Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2527

Fig. 28 Structural design, optimization, and validation process of engine assembly support bracket
2528 E. Talay et al.

weight decreased from support bracket after iterations without study also proved that topology optimization method cre-
violating the targets, but it is understood that the result of the ates lightweight and feasible solutions even for casting
optimization is not producible and needs improvement. After parts in production. It has been also understood that to reach
taking into account the high-pressure die casting requests from feasible design solutions, manufacturing constraints must
production side, new solution was created and 0.7 kg (37%) be added within optimization problems. 0.7 kg (37%)
weight reduction was achieved with respect to the base design. weight reduction from just a support bracket showed the
After reaching an optimum and producible design, stress dis- big potential in powertrain components. Using optimiza-
tribution at different load cases has been investigated at tion methods for components in serial production is possi-
Section 2.4. In Section 2.4.1, fatigue life estimation of support ble but it will add additional constraint since the fixture and
assembly has been explained with FRF-based dynamic simu- manufacturing tools are already defined. If the optimization
lation. Dynamic loads have been measured with 3 axial accel- methods are used in concept phase, more innovative solu-
erometers from cylinder heads and applied in simulation mod- tions can be created. Bringing simulation early into the
el to be able to calculate the stresses at attachments. Since the design cycle reduces costs, helps to find mistakes early,
support bracket is aluminum, its safe stress limit for fatigue and allows for more exploration of the design space. This
simulations is near 60 MPa. According to the dynamic exci- produces far better products with less spending.
tation simulation, the maximum stress level has been calculat-
ed at 18 MPa at 167 Hz for actual production part and stress Compliance with ethical standards
levels have been increased to 19.6 MPa with optimized brack-
et which shows that optimized solution is in safe region with Conflict of interest The authors declare that they have no conflict of
interest.
respect to fatigue life. Some of the dynamic events can be
performed with static loading approximations to check if the
Replication of results Mass and inertia properties of alternator and
maximum loading amplitude can create deformation on at- steering pump are given in Tables 1 and 2. Upon request, the authors
tachments or not. In Section 2.4.2, structural simulations with will provide contact details for modal analysis and input parameters for
maximum belt load, rear crash, and bump passing scenarios the topology optimization problem presented in this paper. The finite
element model of the engine and alternator support geometry are confi-
have been investigated. According to the simulation results,
dential data of FCA and protected. Any optimization study by using the
base design was in safe region since the aluminum material introduced inputs on similar geometry will produce similar results like
ultimate tensile stress is 190 MPa. Material void examinations seen in numerical analysis section.
and experimental performance validations have been ex-
plained at Section 4. After material examinations showed void
defects on prototype components, pressurized casting param-
References
eters were optimized to get rid of the manufacturing defects.
Vehicle level comparative FRF, interior sound pressure, and Bartosz K, Podgórski J (2017) Numerical results quality in dependence on
reliability tests proved that the optimized design proposal will abaqus plane stress elements type in big displacements compression
not create any performance loss. test. Appl Comput Sci. https://doi.org/10.23743/acs-2017-29
The design process from topology optimization to the pro- Beikmann RS, Perkins N-C, Ulsoy A-G (1997) Design and analysis of
automotive serpentine belt drive systems for steady state perfor-
duction and validations can be seen at Fig. 28. mance. J Mech Des 119(2):162. https://doi.org/10.1115/1.2826231
Bendsøe MP (1995) The homogenization approach to topology design.
In: Optimization of structural topology, shape, and material.
6 Conclusions Springer: Berlin. https://doi.org/10.1007/978-3-662-03115-5
Boyce MP (2011) Rotor dynamics. Gas Turbine Engineering Handbook,
p. 244. https://doi.org/10.1016/C2009-0-64242-2
Automotive industry is getting more and more competitive
Boydak Ö, Savaş M, Ekici B (2015) A numerical and an experimental
nowadays and companies even have to prefer merging to investigation of a high-pressure die-casting aluminum alloy. Int J
decrease their development costs. It is very important to use Met 10(1):56–69. https://doi.org/10.1007/s40962-015-0004-4
new technologies to keep up competitiveness. Topology Chang K-H (2015a) Structural design sensitivity analysis. Design Theory
optimization method has become one of the common tools and Methods Using CAD/CAE, p.301. https://doi.org/10.1016/
B978-0-12-398512-5.00004-9
which generates innovative and lightweight solutions. That
Chang K (2015b) Structural design sensitivity analysis. Design Theory
is why using optimization methods more on vehicle devel- and Methods Using CAD/CAE, p. 178. https://doi.org/10.1016/
opment process have become a crucial tool to survive. B978-0-12-398512-5.00004-9
Typical cast designs have hollow structure with cross- Chao L, Kim IY, Jeswiet J (2014) Conceptual and detailed design of an
bracing ribs to connect attachment points to mounting lo- automotive engine cradle by using topology, shape, and size opti-
mization. Struct Multidiscip Optim 51(2):547–564. https://doi.org/
cations with a direct load path. Using casting components 10.1007/s00158-014-1151-6
in vehicles reduces vehicle weight and very complicated Christensen PW, Klarbring A (2009) Introduction. In: An introduction to
shapes can be produced with respect to sheet design. This structural optimization. Solid mechanics and its applications, vol
Designing lightweight diesel engine alternator support bracket with topology optimization methodology 2529

153. Springer, Dordrecht. https://doi.org/10.1007/978-1-4020- Novoselac S, Kozak D, Ergić T, Damjanović D (2016) Fatigue damage
8666-3 assessment of bolted joint under different preload forces and vari-
Hirsch J (2004) Automotive trends in aluminium-The European able amplitude eccentric forces for high reliability. Lect Notes Mech
Perspective, Mater. Forum, p. 17 Eng 239–268. https://doi.org/10.1007/978-3-319-32634-4_13
Leamy M (2003) Dynamics analysis of the time-varying operation of Rakotondrainibe L, Allaire G, Orval P (2020) Topology optimization of
belt-drives. Volume 5: 19th Biennial Conference on Mechanical connections in mechanical systems. Struct Multidiscip Optim 61:
Vibration and Noise. https://doi.org/10.1115/DETC2003/VIB- 2253–2269. https://doi.org/10.1007/s00158-020-02511-9
48346 Talay E, Altınışık A (2019) The effect of door structural stiffness and
Lee D-C, Lee J-I (2003) Structural optimization concept for the design of flexural components to the interior wind noise at elevated vehicle
an aluminium control arm. Proc IME D:J Automob Eng 217(8): speeds. Appl Acoust ISSN: 0003-682X 148:86–96. https://doi.org/
647–656. https://doi.org/10.1243/09544070360692041 10.1016/j.apacoust.2018.12.005
Lee K-H, Shin J-K, Song S-I, Yoo Y-M, Park G-J (2003) Automotive Wallentowitz H (2004) Structural design of vehicles. Aachen, Institut fur
door design using structural optimization and design of experiments. Kraftfahrwesen Aachen
Proc IME D J Automob Eng 217(10):855–865. https://doi.org/10. Wang L, Burger R, Aloe A (2017) Considerations of vibration fatigue for
1243/095440703769683261 automotive components. SAE Int J Commer Veh 10. https://doi.org/
Li C, Kim I-Y, Jeswiet J (2015) Conceptual and detailed design of an 10.4271/2017-01-0380
automotive engine cradle by using topology, shape, and size opti- Wang B, Zhou Y, Tian K (2020) Novel implementation of extrusion
mization. Struct Multidiscip Optim 51:547–564. https://doi.org/10. constraint in topology optimization by Helmholtz-type anisotropic
1007/s00158-014-1151-6 filter. Struct Multidiscip Optim 62:2091–2100. https://doi.org/10.
Lima L, Pereira A, Cavalcanti J, Kawano M, (2015) Topology and fre- 1007/s00158-020-02597-1
quency response function analysis applied to structural parts aiming Yung L, Sandeep M, Sean M, Ming-Wei L (2015) Durability reliability
mass and cost reduction. SAE Tech Pap 2015-36-0286. https://doi. demonstration test methods. Procedia Eng ISSN: 1877-7058 133:
org/10.4271/2015-36-0286 31–59. https://doi.org/10.1016/j.proeng.2015.12.62
López-Castro JD, Marchal A, González L, Botana J (2017) Topological Zhong W, Su R, Gui L (2016) Multi-objective topology and sizing opti-
optimization and manufacturing by Direct Metal Laser Sintering of mization of bus body frame. Struct Multidiscip Optim 54:701–714.
an aeronautical part in 15-5PH stainless steel. Procedia Manuf 13: https://doi.org/10.1007/s00158-016-1431-4
818–824. https://doi.org/10.1016/j.promfg.2017.09.121 Zhou M, Shyy Y-K, Thomas H-L (2001) Checkerboard and minimum
Mayer H, Papakyriacou M, ZETTL B, VACIC S (2005) Endurance limit member size control in topology optimization. Struct Multidiscip
and threshold stress intensity of die cast magnesium and aluminium Optim 21(2):152–158. https://doi.org/10.1007/s001580050179
alloys at elevated temperatures. Int J Fatigue 27(9):1076–1088.
https://doi.org/10.1016/j.ijfatigue.2005.02.002 Publisher's note Springer Nature remains neutral with regard to
MIT (2008). On the road in 2035: reducing transportation’s petroleum jurisdictional claims in published maps and institutional affiliations.
consumption and GHG emissions. Mass Inst Technol 2008; p.45