Documente Academic
Documente Profesional
Documente Cultură
net/publication/320239639
Article in Proceedings of the Institution of Mechanical Engineers Part D Journal of Automobile Engineering · October 2017
DOI: 10.1177/0954407017727165
CITATIONS READS
0 234
3 authors:
Ajay Kapoor
Swinburne University of Technology
26 PUBLICATIONS 262 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Effect of wear on underhead radius stresses under high axle load conditons View project
Multiaxial fatigue analysis of rail under high axle load conditions View project
All content following this page was uploaded by Sagheer A. Ranjha on 06 November 2017.
Abstract
Electric vehicles (EVs) are an alternative architecture in the automotive industry that provide reduced emissions. This
research has developed a switch reluctance motor (SRM) in-wheel drivetrain for an EV. SRM drivetrains are cheaper and
do not use rare earth elements unlike a permanent magnet motor (PMM). Conversely, the in-wheel SRM has a drawback
of an increased mass on the suspension when compared with an equivalent power output PMM drivetrain. This situation
results in an increased mass at the wheels; hence, a suspension analysis is required. This paper discusses the suspension
dynamics evaluated using a quarter-car simulation of an in-wheel SRM EV and compares it to the internal combustion
engine (ICE) vehicle. The simulation used step loads derived design scenarios, namely (1) sprung, (2) unsprung and (3)
driver’s seat. Further Bode plot analysis techniques were used to determine the ride comfort range for the developed
EV.
Keywords
Quarter-car model, suspension, simulation, Bode plot analysis, transient response and electric vehicle
Figure 2. Flow chart for methodology (comparing an internal combustion engine (ICE) vehicle with an electric vehicle (EV) for
increased mass in the research).
FEA: finite element analysis.
addition due to the driver’s seat placed on top. These Spark. This weight on each wheel was 240 kg calculated
two components were insulated by the spring cushion based on kerb weight of the Barina Spark (960 kg). Of
that was, in the subsequent analysis, represented as a the ~240 kg total weight, 90 kg was considered to be the
spring with a spring index Kds and damper with the coef- weight of the driver and the seat. In addition, the mea-
ficient Cds . Underneath the sprung mass was the surements performed during the experiments indicated
unsprung mass, which represented the mass of the single that each set of existing wheels, brakes and suspension
wheel, suspension, tyre, shock absorbers and suspension weighed approximately 40 kg, which was the unsprung
knuckles. In the case of the EV, the unsprung mass also mass of the ICE (Mu ). This led to the estimate of 110
included the mass of the in-wheel SRM, which was kg for the ICE vehicle sprung mass.
placed inside the wheel. In the subsequent analysis, the The sprung mass for the EV (Mse ) changed due to
EV suspension was considered to have a spring stiffness the ICE being removed and weight of the motor placed
of Ks and the damping coefficient of the damper Cs . inside the wheel. The conversion involved changing the
The tyre was also considered to act like a spring with a full wheel and brake system with the new proposed in-
spring coefficient Kt . The model for the passive suspen- wheel SRM and wheel. In the process, 17 inches wheels
sion system of a quarter-car was thus ready for analysis. with the SRM fitted inside were used as a replacement
Figure 3 below provides the free body diagram of the for the two rear wheels. Based on the overall weight;
quarter-car model. the sprung mass of the EV was calculated with the
In Figure 3, Xr , Xu , Xs and Xds correspond to the ver- aforementioned methods as (Mse ) 130 kg. The unsprung
tical displacement of the road, unsprung mass, sprung mass of an EV (Mue ) was 80 kg in total calculated from
mass and the seat, respectively. In Table 1, the equation (i) 40 kg as the weight of the in-wheel SRM, (ii) 25 kg
coefficients used in the subsequent text are provided. as the weight of the 17 inches wheel (7 kg for rim and
The variables given above were assigned values for 18 kg for tyre nuts and bolts) and (iii) 15 kg as the
the corresponding car configuration. The driver’s seat weight of a new brake system (5 kg for disc brake and
mass (Mds ) was considered to be a uniform 90 kg for 10 kg for the calliper, hydraulics, nuts and bolts). The
both car types, as it was identical in both experimental driver’s seat mass of the EV remained at 90 kg.
arrangements. The vehicle sprung mass (Ms ) changed
for the two configurations, as the vehicle kerb weight
was modified. In the case of the ICE setup, the sprung Suspension modelling
mass was considered to be 110 kg, which was the The parameters required for performing the analysis
default load on each wheel for the Holden Barina were the spring constants and the damping coefficients
4 Proc IMechE Part D: J Automobile Engineering 00(0)
p d3 t
F= ð1Þ
16 r
64 n r3 F
f= ð2Þ
d4 G
K = F=f ð3Þ
where d is the spring coil diameter (14 mm), r is the
spring radius (60 mm), n is the number of coils (five)
and G is the shear modulus of the spring—79:33109 Pa.
The above equations (1) to (3) were used to calculate
the spring constant of the MacPherson strut suspension,
which was 44,070 N/m. Next, the damping ratio of the
suspension damper was calculated using equation (4)
given below, based on the value of the spring constant
and the mass of the ICE vehicle
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðKs Ms Þ
Cs = pffiffiffi ð4Þ
2
Figure 3. Free body diagram of quarter-car model for the where Ks is a spring constant for the MacPherson strut
electric vehicle (EV). and Ms is a sprung mass of the ICE vehicle. The damp-
ing coefficient was calculated as 2969 N/m and the
spring coefficient and damping coefficient for the seat
Table 1. Descriptions of variables used in the equations. were calculated as 176,280 N/m and 8372 Ns/m. The
road excitation for a passive suspension was considered
Variables Description using a road bump in the form of transient force. These
conditions were simulated by non-periodic, determinis-
Mds Driver and seat mass
Ms Vehicle sprung mass tic excitation, or random excitation, with unpredictable
Mu Vehicle unsprung mass time described using probability.24 A road bump in the
Cs Damping ratio of the vehicle suspension form of a high-risk pothole was considered as a source
Cds Damping ratio of the seat of excitation for the quarter-car model,15 using the
Kt Tyre stiffness depth of 100 mm, leading to Xr equal to 1. The key
Ks Vehicle suspension spring stiffness
Kds Seat suspension spring stiffness parameters used to perform the quarter-car analysis are
Xr Road input given in Table 2.
The quarter-car model equations are based on the
model depicted in Figure 3. The block diagram in
Figure 4 shows the various forces acting on the differ-
of the spring–mass systems in the model. As previously ent masses on the vehicle suspension.
noted, the model with three degrees of freedom was Equations of motion are established for the three
employed in the analysis, whereby the two spring different masses, namely the sprung mass (Ms ), the
Table 2. Internal combustion engine (ICE) and electric vehicle (EV) specifications considered for the quarter-car modelling.
unsprung mass (Mu ) and the driver’s seat mass (Mds ), The state-space equation was established using the
whereby the latter includes the various forces acting on state-space variables given by equations (11) to (16)
the driver’s seat mass. This is given by and was written as a column matrix x(t), n 3 1, which
represented the current state of the system, and matrix
Mds X€ds + Kds ðXds Xs Þ + Cds X_ ds X_ s = 0 ð5Þ u(t), which represented the system input vector. When
the current state and the system input were available in
Equation (5) represents the various forces acting on the
the state-space equation, the system output (y) was
mass in an equilibrium position. This enables the calcu-
obtained from the state-space equation. Here, the state-
lation of driver’s seat mass acceleration, as given by
space variables were differentiated with respect to time
X€ds = Kds ðXds Xs Þ=Mds Cds X_ ds X_ s =Mds and formed into matrices. Equation (17) shows the
state-space equation for the quarter-car suspension sys-
ð6Þ tem. This state-space equation, which formed a mathe-
The motion of the sprung mass of the system in the matical model from a set of inputs and state-space
position of equilibrium is given by variable, was written on the basis of the state-space
variables, as mentioned above.
Ms X€ds Kds ðXds Xs Þ Cds X_ ds X_ s 2 3
x_ 1
+ Ks ðXs Xu Þ + Cs X_ s X_ u = 0 ð7Þ 6 x_ 2 7
6 7
6 x_ 3 7
As above, the acceleration of the sprung mass was 6 7=
6 x_ 4 7
6 7
obtained from equation (7), and is given below as 4 x_ 5 5
x_ 6
Kds ðXds Xs Þ Cds X_ ds X_ s
X€s = + 2
0 0 0 0 0 0
3
Ms Ms 6 Kds =Mds Cds =Mds Kds =Mds Cds =Mds 7
6 0 0 7
_
Ks ðXs Xu Þ Cs Xs Xu _ 6
6 0 0
Kds
0 1 0 0 7
7
ð8Þ 6 Kds =Ms Cds =Ms Ms Ms
Ks
0 Ks =Ms Cs =Ms 7
6 7
Ms Ms 4 0 0 0 0 0 1 5
Ks
0 0 Ks =Ms Cs =Mu Mu Mu
Kt
Cs =Mu
Similarly, the equation of motion in the equilibrium 2 3 2 3
position was written for the unsprung mass x1 0
6 x2 7 6 0 7
6 7 6 7
Mu X€u Ks ðXs Xu Þ Cs X_ s X_ u + Kt ðXu Xr Þ = 0 6 x3 7 6 7
6 7+ 6 0 7u ð17Þ
6 x4 7 6 0 7
ð9Þ 6 7 6 7
4 x5 5 4 0 5
Once again, the acceleration of the unsprung mass was x6 Kt =Mu
obtained from equation (9), and is given below as
The above state-space equation was used to derive the
ðXs Xu Þ X_ s X_ u output equation for displacement and velocity, given
X€u = Ks + Cs below in the matrix form
Mu Mu
2 3 2 32 3
ðX u X r Þ Xds 1 0 0 0 0 0 x1
Kt ð10Þ 6 Xs 7 6 0 0 1 0 0 0 76 x2 7
Mu 6 7 6 76 7
6 Xu 7 6 0 0 0 1 0 0 76 x3 7
The three equations of second order (equations (6), (8) y=6 7 6 76 7
6 X_ ds 7 = 6 0 1 0 0 0 0 76 x4 7 + u
6 7 6 76 7
and (10)) were used as input into the system. Next, a 4 X_ s 5 4 0 0 0 1 0 0 54 x5 5
state-space equation was established using the equa- X_ u 0 0 0 0 0 1 x6
tions of motion, for which six different state-space vari-
ables were defined, representing the displacement and ð18Þ
velocity of the sprung, unsprung and driver’s seat mass Clearly, y on the left side of the equality in equation
of the quarter-car suspension system, respectively. (18) represented the system output. The state-space
Hence, six state-space variables were derived from equation given above was later solved using
equations (6), (8) and (10). These were represented by MATLABÒ to obtain the system output in the form of
equations (11) to (16) below displacement and velocity for the suspension analysis,
which are discussed in the next section.
x1 = X_ ds ð11Þ
x2 = Xs Xds ð12Þ
Results and discussions
x3 = X_ s ð13Þ
x4 = Xu Xs ð14Þ
Suspension dynamics
The quarter-car model equations derived in the preced-
x5 = X_ u ð15Þ
ing section were used as inputs to develop the suspen-
x6 = Xr Xu ð16Þ sion analysis program. As noted above, equation (18)
6 Proc IMechE Part D: J Automobile Engineering 00(0)
Figure 7. Velocity and displacement of the unsprung mass for Figure 9. Velocity and displacement of the driver’s seat of the
the internal combustion engine (ICE) vehicle. internal combustion engine (ICE) vehicle.
Figure 8. Velocity and displacement of the unsprung mass for Figure 10. Velocity and displacement of the driver’s seat of the
the electric vehicle (EV). electric vehicle (EV).
the two compared vehicles (from 0.6 s in an ICE to 0.8 was similar to the other two simulation results, suggest-
s in an EV). These results suggest that this behaviour ing direct impact on both the driver and the passengers
affects the ride comfort for an EV, since instability was in the vehicle. The velocity fluctuation of an EV was
observed due to the increase in the unsprung mass. The more than 16% of an ICE (from 1.5 s in an ICE to 1.8 s
ride quality was enhanced due to decreased velocity in an EV). The displacement fluctuation was 20%
changes for the same amplitude of displacement. The greater in an EV when compared to the ICE (from 1.0 s
time required for the EV to stabilise to its normal posi- in an ICE, to 1.25 s in an EV). This change in velocity
tion was longer than that of the ICE vehicle. In the EV, and displacement affected the driver’s comfort range,
the damper in the vehicle was able to handle the extra with additional oscillations as observed in an EV.
load due to the unsprung mass, which led to slight velo- This situation had a direct effect on the damper han-
city fluctuations over a longer time period. Finally, the dling due to the increase in both the unsprung and sprung
impacts of these loads were tested for the driver’s seat mass. The increased oscillations led to an uncomfortable
mass. The driver’s seat of the vehicle was the most criti- ride, thus reducing the quality of the experience for the
cal part of the quarter-car model analysis, since it had passengers. In other words, although the increased spring
a direct effect on the driver and passengers travelling in stiffness of the seat and driver mass minimised the prob-
the vehicle. lem, it still led to reduced ride quality. It can be concluded
that the increase in damping coefficient benefitted han-
dling and drive conditions of the vehicle. In an EV with
Driver’s seat mass. In Figures 9 and 10, it can be observed the in-wheel SRM, the ride quality was also slightly
that the response of the seat and driver to the step input affected. However, the simulation results indicated
8 Proc IMechE Part D: J Automobile Engineering 00(0)
model, when compared to that exhibited by the 5. Verros G, Natsiavas S and Papadimitriou C. Design
ICE. This variation was due to the variation in the optimization of quarter-car models with passive and
velocity plots, whereby those for the EV exceeded semi-active suspensions under random road excitation. J
those of the ICE by (i) 16% in displacements and Vib Control 2005; 11: 581–606.
21% in velocity amplitude at the sprung mass, (ii) 6. Stein GJ and Múčka P. Theoretical investigation of a lin-
ear planar model of a passenger car with seated people.
20% in displacements and 25% in velocity ampli-
Proc IMechE Part D: J Automobile Engineering 2003;
tude at the unsprung mass and (iii) 16% in displa-
217: 257–268.
cements and 20% in velocity amplitude at the 7. Firat M and Kocabicak U. Analytical durability model-
driver’s seat mass. This large variation at the ing and evaluation-complementary techniques for physi-
unsprung mass was due to the increased mass of cal testing of automotive components. Eng Fail Anal
the vehicle caused by the in-wheel SRM at the 2004; 11: 655–674.
wheels. 8. Sohn H-C, Hong K-S and Hedrick JK. Semi-active con-
4. Following the mathematical analysis, further study trol of the MacPherson suspension system: hardware-in-
was performed in order to determine the ride safety the-loop simulations. In: Proceedings of the 2000 IEEE
of the EV suspension in comparison to that of the international conference on control applications, Ancho-
ICE. Bode plot analysis techniques were used to rage, AK, 27 September 2000. New York: IEEE.
determine the ride comfort range for the developed 9. Georgiou G, Verros G and Natsiavas S. Multi-objective
optimization of quarter-car models with a passive or
EV. Generally, a system is considered unstable if it
semi-active suspension system. Vehicle Syst Dyn 2007;
has a phase of 2180º at its cross-over frequency,
45: 77–92.
which should be below 1–1.5 Hz. A system is also 10. Metered H, Bonello P and Oyadiji SO. An investigation
potentially unstable if the magnitude exceeds 10 into the use of neural networks for the semi-active control
dB when the phase is at its cross-over frequency of of a magnetorheologically damped vehicle suspension.
1–1.5 Hz. In the experiments performed here, it Proc IMechE Part D: J Automobile Engineering 2010;
was seen that both EV and ICE performed in a fre- 224: 829–848.
quency range of 1–1.5 Hz. Thus, suspension design 11. Gao W, Zhang N and Dai J. A stochastic quarter-car
was safe for the developed EV design. model for dynamic analysis of vehicles with uncertain
parameters. Vehicle Syst Dyn 2008; 46: 1159–1169.
12. Chan BJ and Sandu C. A ray-tracing approach to simula-
Acknowledgements tion and evaluation of a real-time quarter car model with
semi-active suspension system using MATLAB. In: 2003
The authors thank i-move CRC Auto CRC C2-25 for
ASME design engineering technical conferences and comput-
funding and supporting the development of the EV dri-
ers and information in engineering conference, Chicago, IL,
vetrain. They also thank other consortium members for 2–6 September 2003, pp.2129–2134. New York: ASME.
assisting with the required information. 13. Cao JT, Liu HH, Li P et al. A study of electric vehicle
suspension control system based on an improved half-
Declaration of conflicting interests vehicle model. Int J Autom Comput 2007; 4: 236–242.
14. Cao J, Liu H, Li P et al. State of the art in vehicle active
The authors declare that there is no conflict of interest. suspension adaptive control systems based on intelligent
methodologies. IEEE Trans Intell Transp Syst 2008; 9:
Funding 392–405.
15. Cao D, Song X and Ahmadian M. Editors’ perspectives:
This research received no specific grant from any fund- road vehicle suspension design, dynamics, and control.
ing agency in the public, commercial or not-for-profit Vehicle Syst Dyn 2011; 49: 3–28.
sectors. 16. Huang S-J and Chao H-C. Fuzzy logic controller for a
vehicle active suspension system. Proc IMechE Part D: J
Automobile Engineering 2000; 214: 1–12.
References
17. Cao J-T, Li P and Liu H-H. An extended fuzzy controller
1. Kulkarni A, Ranjha SA and Kapoor A. Fatigue analysis for a vehicle active suspension system. Proc IMechE Part
of a suspension for an in-wheel electric vehicle. Eng Fail D: J Automobile Engineering 2010; 224: 717–733.
Anal 2016; 68: 150–158. 18. Karnopp D, Crosby MJ and Harwood RA. Vibration
2. Lovatt HC, Elton D, Cahill L et al. Design procedure for control using semi-active force generators. J Eng Ind
low cost, low mass, direct drive, in-wheel motor drivetrains 1974; 96: 619–626.
for electric and hybrid vehicles. In: IECON 2011 – 37th 19. Poussot-Vassal C, Sename O, Dugard L et al. A new
annual conference on IEEE industrial electronics society, semi-active suspension control strategy through LPV
Melbourne, VIC, 7–10 November 2011. New York: IEEE. technique. Control Eng Pract 2008; 16: 1519–1534.
3. Novellis LD, Sorniotti A and Gruber P. Design and 20. Agharkakli A, Sabet GS and Barouz A. Simulation and
comparison of the handling performance of different analysis of passive and active suspension system using
electric vehicle layouts. Proc IMechE Part D: J Automo- quarter car model for different road profile. Int J Eng
bile Engineering 2014; 228: 218–232. Trends Tech 2012; 31: 636–644.
4. George AD and Besselink I. Rear suspension design for 21. Shin D, Lee G, Yi K et al. Motorized vehicle active sus-
an in-wheel-drive electric car. Proc IMechE Part D: J pension damper control with dynamic friction and actua-
Automobile Engineering 2016; 230: 147–159. tor delay compensation for a better ride quality. Proc
10 Proc IMechE Part D: J Automobile Engineering 00(0)