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

net/publication/305759137

Analysis and optimization of air suspension system with independent height

and stiffness tuning

Article  in  International Journal of Automotive Technology · October 2016

DOI: 10.1007/s12239-016-0079-9

CITATIONS READS

5 1,510

4 authors, including:

Peyman Karimi Eskandary

McGill University
3 PUBLICATIONS   7 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Peyman Karimi Eskandary on 04 June 2018.

The user has requested enhancement of the downloaded file.

International Journal of Automotive Technology, Vol. 17, No. 5, pp. 807−816 (2016) Copyright © 2016 KSAE/ 092−07
DOI 10.1007/s12239−016−0079−9 pISSN 1229−9138/ eISSN 1976−3832

ANALYSIS AND OPTIMIZATION OF AIR SUSPENSION SYSTEM WITH

INDEPENDENT HEIGHT AND STIFFNESS TUNING

P. KARIMI ESKANDARY*, A. KHAJEPOUR, A. WONG and M. ANSARI

Department of Mechanical and Mechatronics, University of Waterloo, Waterloo, N2L 3G1, Canada

(Received 15 June 2015; Revised 3 February 2016; Accepted 18 February 2016)

ABSTRACT−Suspensions play a crucial role in vehicle comfort and handling. Different types of suspensions have been
proposed to address essential comfort and handling requirements of vehicles. The conventional air suspension systems use a
single flexible rubber airbag to transfer the chassis load to the wheels. In this type of air suspensions, the chassis height can
be controlled by further inflating the airbag; however, the suspension stiffness is not controllable, and it depends on the airbag
volume and chassis load. A recent development in a new air suspension includes two air chambers (rubber airbags), allowing
independent ride height and stiffness tuning. In this air suspension system, stiffness and ride height of the vehicle can be
simultaneously altered for different driving conditions by controlling the air pressure in the two air chambers. This allows the
vehicle’s natural frequency and height to be adjusted according to the load and road conditions. This article discusses
optimization of an air suspension design with ride height and stiffness tuning. An analytical formulation is developed to yield
the optimum design of the new air suspension system. Experimental results verify the mathematical modeling and show the
advantages of the new air suspension system.

KEY WORDS : Air suspension systems, Ride height control, Stiffness tuning, Natural frequency control

1. INTRODUCTION be decreased to improve ride comfort. Likewise, in flat

roads, handling can be improved using a stiff damper.
One of the main purposes of any suspension system is to Additionally, active suspension systems could eliminate a
isolate the vehicle's body from the vibration induced from compromised design of suspensions by actively providing
road unevenness. There are many standards to define the desired force between the wheel and chassis. However,
appropriate measurable criteria for the performance of these suspensions are more expensive and complicated.
suspension systems (Els et al., 2007; Standardization, In an air suspension, the compressibility of air in a
1997; Giacomin and Hacaambwa; Irvine, 2000). High pneumatic spring or air spring is used to provide the
stiffness and damping coefficients are required for better suspension stiffness. Normally, static pressure is about 3
vehicle handling and stability, while a soft spring and a low bars, and will not reach more than 7 ~ 8 bars (100 ~ 120
damping coefficient are more appropriate for ride comfort psi) (Matschinsky, 2000). The force exerted by the airbag is
(Ahmadian and Pare, 2000; Cao et al., 2010). Therefore, the effective area of airbag (spring) multiplied by the air
the design of a suspension is a compromise between the pressure. The effective area of an airbag is not completely
vehicle’s comfort level and handling (Cole, 2001; Cao et constant over its compression and extension (Chang and
al., 2011). For example, Gobbi and Mastinu (2001) applied Lu, 2008). The pressure in the air spring depends on the
numerical optimization to find the best values for a effective cross section area of the air spring and the load.
conventional passive suspension by using a quarter car The suspension stiffness is a function of the air pressure
model. The conclusion was that semi-active elements and air spring volume. As the load increases the pressure
(Swevers et al., 2007) like a variable damper (Sharp and also increases and as a result the suspension stiffness
Hassan, 1986) could improve suspension performance by increases almost linearly by the load. This provides a
proper adaptation of the suspension damping to road nearly constant bounce natural frequency for the vehicle
conditions or load. Semi-active suspension systems have regardless of the load that is unique to air suspensions
been studied in detailed (Chien et al., 2008; Song et al., (Matschinsky, 2000).
2005; Bolandhemmat et al., 2010; Fu and Mills, 2006; To change the stiffness of an air suspension for a
Ahmadian and Blanchard, 2011; Liu et al., 2005, 2008; constant load, several studies have been conducted (Deo,
Eltantawie, 2012; Balamurugan et al., 2014) in the 2006; Giliomee and Els, 1998; Schisler, 2005; Lee, 2010;
literature. In off-road conditions, the damping coefficient can Easter, 1995; Brookes et al., 2008; Toyofuku et al., 1999).
These systems work based on changing the gas volume of
*Corresponding author. e-mail: [email protected] the air springs. For instance, Hrishikesh (Deo, 2006)

807
808 P. KARIMI ESKANDARY, A. KHAJEPOUR, A. WONG and M. ANSARI

proposed an air suspension system with some extra tune two parameters (height and stiffness) independently
volume. By connecting each or some of these extra for different driving conditions, it needs to control the air
volumes to the main air spring, the stiffness of the pressure in two air chambers. There are various
suspension can be adjusted at certain levels. Toyofuku et geometrical configurations for the two air chambers of this
al. (1999) developed an analytical model for air suspension suspension, which will be also discussed in this paper.
with auxiliary volumes, which considers the effect of pipes Experimental results will be provided for a designed
and connections. Their research showed that the auxiliary suspension.
chamber had a smaller effect on the system at high
frequencies than at low frequencies. 2. MATHEMATICAL MODELING
The concept of air suspension systems with independent
stiffness and height tuning was introduced in Yin et al. The Different geometrical configurations can be considered
(2012) and Khajepour et al. (2012, 2014). In this air for the two air springs needed for this suspension. Type 1
suspension system, two air bags are used to change the air suspension system shown in Figure 2, works based on
suspension stiffness through the adjustment of the air bags’ the motion of a four-bar linkage mechanism with extended
air pressure. Despite the nonlinear behavior of air bags, the trailing arm. Increasing the air pressure in both chambers
stiffness of air bags can be approximated constant in their makes the system stiffer. In order to change the ride height,
operational range. They developed a comprehensive model the air volume of the air springs needs to be adjusted
of the new air suspension with an accumulator and an accordingly. The force exerted by air chambers should
orifice valve, shown in Figure 1. They modeled the system clearly satisfy the force and moment equilibrium equations
based on a solid cylinder with two chambers separated by a for the trailing arm at any time. Therefore, the pressure of
piston. A pressure regulator valve can adjust the pressure in the air springs should be set properly to obtain the desired
each chamber independently. Each side of the piston is ride height and suspension stiffness. Normally, heavy trucks
connected to an accumulator. An orifice valve is also have enough space around their axles and underneath of
considered for additional damping. The experimental their chassis. Therefore, air springs are placed on both sides
results reported in Yin et al. (2012) and Khajepour et al. of the axle (Type 1). Passenger vehicles, however, have
(2012) proved the concept and the ability of the new several critical geometrical constraints such as ground
suspension in tuning stiffness and ride height independently clearance that it is preferred to combine the two air
for a given load. chambers in a single cylinder as seen in Figure 2 − Type 2.
Modeling and experimental results on performance of The modeling and formulation in this section is first
air suspension systems with independent stiffness and ride
height tuning are provided in Yin et al. (2012) and
Khajepour et al. (2012, 2014). The focus of this paper
however, is on design and design optimization of the air
suspension system considering various constraints and
desired criteria. In general, this article studies the effect of
different design parameters in the performance of the new
suspension and develops an optimization tool for design of
the new suspension. Since such a system should be able to

Figure 1. Schematic of quarter car model with the Figure 2. Different configurations of air suspension
proposed air suspension system (Yin et al., 2012). systems.
ANALYSIS AND OPTIMIZATION OF AIR SUSPENSION SYSTEM WITH INDEPENDENT HEIGHT AND STIFFNESS 809

suspension. Also, K1(z) and K2(z) represent the stiffness of

air chamber 1 and air chamber 2 respectively, which are
approximated constants in the operational range of air
chambers around the equilibrium point.
The most important part of the air suspension modelling
is the evaluation of P1(z) and P2(z) by solving a set of
equations for desired natural frequency and force
equilibrium at a desired height (z = Heq);
⎧⎪ F(at z=Heq ) = mg
Figure 3. Schematic of an air suspension with dual ⎨ (4)
⎪⎩ f (at z=H eq ) = f s
chamber.
where fs is the desired natural frequency and Heq is the
desired piston position (ride height). As a result, Peq1 and
developed based on the combined chambers in a single Peq2, which are the equilibrium pressures of air chambers at
cylinder (Type 2) and then it is extended to Type 1. the desired height and frequency using Equation (1) by
Figure 3 illustrates the detailed schematic of the air setting F(z) = F(Heq) = mg and Equation (3) by setting f(z)
spring with independent ride height and stiffness tuning. It = fs can be calculated as;
consists of a pneumatic cylinder, two pressure regulators (P
regulator) which control the two chambers’ pressure, and Peq1 =
(4 f s
2
π 2 m(V2 + A2 H eq ) − A2γ ( ( A2 − A1 ) Pa + mg ) (V1 + A1 ( H − H eq ) )
)
an air tank which supplies the high pressured air for the A1γ ( AV
1 2 + A2V1 + A1 A2 H )

system. There are two valves that separate chambers from

Peq2 =
( 4 f π m (V + A ( H − H ) ) + A γ ( ( A − A )P + mg )) (V + A H )
s
2 2
1 1 eq 1 2 1 a 2 2 eq

the tank and pressure regulators during the operation of A2γ ( AV

1 2 + A2V1 + A1 A2 H )
suspension, and two accumulators are connected to the
(5)
upper and lower chambers of cylinder.
The force exerted by the air spring can be written as; The pressures in Equation (5) will balance the load while
satisfying the desired frequency at the desired height, Heq.
F ( z ) = ( P2 ( z ) − Pa ) A2 − ( P1 ( z ) − Pa ) A1 (1)
As previously mentioned, heavy trucks have adequate
where P1(z) and P2(z) represent the absolute pressures of space around their axles and underneath of their chassis but
air chamber 1 and air chamber 2 respectively, as a function not vertically. As a result, the Type 1 in Figure 2 is
of the piston position (z). A1 and A2 are effective areas of considered. The air springs are placed on both sides of the
corresponding chambers, and Pa is atmospheric pressure. axle as seen in Figure 4. The pivot point (connection
Due to the fast motion of suspension, the compression between the trailing arm and the hanger) is considered as
and expansion is rapid and there is not enough time for heat the coordinate origin. L1 and L2 are horizontal distances of
transfer; therefore, it can be assumed that the gas process is air chambers 1 and 2 from the pivot point, respectively. Heq1
isentropic. As a result, P1(z) and P2(z) are; and Heq2 represent the height of air chambers when the air
suspension system is at the nominal position (Z=0).
C1 C2
P1 ( z ) = P2 ( z ) = (2)
(V1 + A1 ( H − z )) γ (V2 + A2 z ) γ

where H is the total length of cylinder, z is the piston

position measured from the bottom of the cylinder, and C1
and C2 are constant values. γ is the heat capacity ratio of
gas.
The stiffness can be derived by calculating the partial
derivative of F(z) with respect to the piston position (z). In
other words, the equivalent stiffness of the system is a
summation of the stiffness of each air chamber. Then, the
natural frequency of air suspension, regarding the fact that
tire stiffness is much larger than suspension system is;

P2 ( z )γ A22 P1 ( z )γ A12
+
K 2 ( z ) + K1 ( z ) (V2 + A2 z ) (V1 + A1 ( H − z ))
ωn
(3)
m m
f (z) = = =
2π 2π 2π

where m is a part of vehicle sprung mass on each

suspension and f(z) is the natural frequency of air Figure 4. Heavy truck air suspension system.
810 P. KARIMI ESKANDARY, A. KHAJEPOUR, A. WONG and M. ANSARI

Total vertical forces exerted to the solid axle by the air optimum design for a given vehicle (load and load
suspension and stiffness at the axle point are derived using variation, nominal height, nominal natural frequency) and
same method as Equations (1) ~ (3); also natural frequency tuning range, and ride height tuning
−γ
range. Design constraints such as maximum pressure and
L2 ⎛ ⎛ ⎛ L ⎞⎞ ⎞
F ( z) = A2 ⎜ C2 ⎜⎜ V2 + A2 ⎜ H eq2 + 2 z ⎟ ⎟⎟ − Pa ⎟
space for accumulators are also considered.
Lax ⎜ ⎝ ⎝ L ⎠⎠ ⎟
⎝ ax
⎠
⎛ ⎛ −γ
⎞
3.1. Design Criteria
L ⎛ L ⎞⎞
+ 1 A1 ⎜ C1 ⎜⎜ V1 + A1 ⎜ H eq1 − 1 z ⎟ ⎟⎟ − Pa ⎟ (6) The most essential unknown parameters are the accumulator
Lax ⎜ ⎝ ⎝ Lax ⎠ ⎠ ⎟ volumes. The range of natural frequency depends on the
⎝ ⎠
L2 2 2
A2 γ P2 ( z ) L1 2
A12γ P1 ( z ) accumulator volumes. For packaging and also limited
Kt (z) = 2 + 2
Lax ⎛ ⎛ L2 ⎞ ⎞ Lax ⎛ ⎛ L1 ⎞ ⎞ space especially in passenger vehicles, the accumulator
⎜ V2 + A2 ⎜ H eq2 + z ⎟⎟ ⎜ V1 + A1 ⎜ H eq1 − z ⎟⎟ volumes should be minimized. The following constraints
⎝ ⎝ Lax ⎠⎠ ⎝ ⎝ Lax ⎠⎠
are crucial criteria for the evaluation of optimum volumes;
where A1 and A2 are effective areas of chamber 1 and
chamber 2, respectively. P1(z) and P2(z) represent the 3.1.1. Minimum equilibrium pressure
absolute pressures of corresponding chambers, as a The equilibrium pressure of the chambers should be greater
function of axle vertical position (z). They can also be than the atmospheric pressure (Pa). All of the evaluated
written as a function of initial equilibrium pressures of pressures in the modeling are in absolute scale. This design
chambers (Peq1 and Peq2); criterion can be written as;

⎛
−γ Peq1 ≥ Pa Peq 2 ≥ Pa (10)
⎛ L ⎞⎞
P1 ( z ) = Peq1 (V1 + A1 ( H eq1 ) ) ⎜ V1 + A1 ⎜ H eq1 − 1 z ⎟ ⎟
γ

⎜ Lax ⎠ ⎟⎠ The equilibrium pressure of chamber 1 (the upper

⎝ ⎝
−γ
(7) chamber) is less than the second chamber; therefore, by
γ⎛ ⎛ L ⎞⎞
P2 ( z ) = Peq 2 (V2 + A2 ( H eq 2 ) ) ⎜ V2 + A2 ⎜ H eq 2 + 2 z ⎟ ⎟
satisfying the minimum pressure condition for Peq1, it is
⎜ L ⎟ assured that both equilibrium pressures are higher than the
⎝ ⎝ ax ⎠⎠
atmospheric pressure. The new set of equations that satisfy
where the minimum pressure condition is;

Lax =
(Y2 X 3 − Y3 X 4 ) Cosθ + X 3 ( X 2 − X 4 ) Sinθ ⎧ F(at z=Heq ) = mg
(Y2 − Y3 ) Cosθ + ( X 2 − X 3 ) Sinθ (8) ⎪⎪
⎨ f (at z=Heq ) = fs (11)
⎛ Y −Y ⎞
−1 ⎪
θ = tan ⎜ 1 2 ⎟
⎝ X 2 − X1 ⎠ ⎪⎩ Peq1 ≥ Pa
Solving the set of equations in order to find the relation
Lax is the effective horizontal distance of axle from the
between V1 and V2, derives an inequality rectangular
pivot point, which is calculated based on the kinematic and
hyperbola equation;
dynamic analysis of the four-bar linkage mechanism
attached to the solid axle. A2 ( a1V1 + b1 A1 )
Due to different configurations of air chambers and the V2 ≥
c1V1 + d1
geometry of mechanisms compared to the model discussed
in the previous section (Figure 3), the force, stiffness and a1 = γ mg + γ A2 Pa − 4 f s2π 2 mH eq
pressure equations are slightly different from those in b1 = γ Pa A1 H eq + ( H − H eq )(γ mg + γ A2 Pa − 4 f s2π 2 mH eq ) (12)
Equations (1) ~ (3). By substituting the newly defined 2
c1 = 4 f s π m 2

parameters in Equation (26) and in all of the equations in

Sections 2 and 3, the equations developed for mathematical d1 = 4 f s2π 2 mA1 ( H − H eq ) − γ Pa A12
modeling and optimum design can be used for this
Equation (12) represents all the acceptable answers for
suspension system as;
V1 and V2 satisfying the minimum equilibrium pressure
L1 L2 condition, which is the area above the rectangular
A1* = A1 A2* = A2 hyperbola on the V1 – V2 graph.
Lax Lax
(9)
L
* *
Lax ( L1 H eq2 + L2 H eq1 )
H = ax H eq2
eq H = 3.1.2. Minimum natural frequency
L2 L1 L2 Having a certain minimum natural frequency ( fmin) is
necessary to cover the desired range of stiffness. As
3. OPTIMUM DESIGN previously mentioned, decreasing the pressure of both air
chambers will decrease the stiffness of air suspension.
In this section a procedure is developed to arrive at an Therefore, the minimum natural frequency occurs when
ANALYSIS AND OPTIMIZATION OF AIR SUSPENSION SYSTEM WITH INDEPENDENT HEIGHT AND STIFFNESS 811

pressure of the chamber 1 (lower pressure) is at atmospheric

pressure, and the natural frequency is in the lower bound of
desired natural frequency range. By solving the equations
for the minimum natural frequency, the solution for V1 and
V2 is;
A2 ( a2V1 + b2 A1 )
V2 ≥
c2V1 + d 2
2
a2 = γ mg + γ A2 Pa − 4 f min π 2 mH eq
2
b2 = γ Pa A1 H eq + ( H − H eq )(γ mg + γ A2 Pa − 4 f min π 2 mH eq ) (13)
2
c2 = 4 f min π 2m
2
d 2 = 4 f min π 2 mA1 ( H − H eq ) − γ Pa A12

3.1.3. Maximum natural frequency

Having a certain maximum natural frequency ( fmax) is Figure 5. Minimum and maximum natural frequency
required to cover the desired range of the stiffness tuning. curves on the V1 – V2 graph.
As with the minimum natural frequency constraint, by
increasing the pressure of both air chambers, the suspension As Figure 5 shows, light gray region and blue region
system will become stiffer. The maximum working pressure represent Equations (13) and (14), respectively. Equation
of an air suspension system (the air compressor maximum (16) refers to the dark gray region between two rectangular
pressure, Pmax) is known. The maximum natural frequency hyperbolas (Minimum and maximum natural frequency
occurs when the pressure in chamber 2 (higher pressure) is curves) before their intersection point which satisfies
equal to Pmax. The solution for V1 and V2 will then become; minimum and maximum natural frequency conditions
simultaneously.
A2 ( a3V1 + b3 A1 ) In this work, we assume the optimum choice for V1 and
V2 ≤
c3V1 + d 3 V2 is when their summation is at the minimum. Therefore,
2
a3 = A2γ Pmax − 4 f max π 2 mH eq this point on the V1 – V2 graph will be on the minimum
b3 = A2γ Pmax H − γ H eq ( mg + ( A2 − A1 ) Pa ) − 4 f max
2
π 2 mH eq ( H − H eq ) (14) natural frequency curve because it is in the lower boundary
2 2
of the acceptable region for V1 and V2 and it has the lowest
c3 = 4 f max π m
value of summation. A derivative of (V1 + V2) function
2
d 3 = 4 f max π 2 mA1 ( H − H eq ) + A1γ ( mg + ( A2 − A1 ) Pa − A2 Pmax )
(based on the minimum natural frequency condition) with
respect to V1 or V2 is used to find the optimum values of V1
3.2. Optimum Accumulator Volumes and V2;
By comparing Equations (12) and (13) based on the first
d d a V +b
and second criteria, it can be seen that minimum natural (V1 + V2 ) = (V1 + 2 1 2 ) = 0 ⇒
dV1 dV1 c2V1 + d 2
frequency condition satisfies the minimum pressure
condition. Horizontal and vertical asymptotes of inequality ⎧
⎪V ( 2
A1 γ Pa A1 − 4 f min π 2 m( H − H eq ) + A2γ 2 Pa (mg + A2 Pa ) )
rectangular hyperbola Equation (12) are smaller than ⎪ 1−min = 2
4 f min π 2m
hyperbola Equation (13) asymptotes, thus fmin is smaller ⎪ (17)
⎪
( )
−1
2 2 2 2
than fs; ⎨V2− min = A1 A2γ Pa (mg + A2 Pa ) 4 f minπ m A2γ Pa (mg + A2 Pa )
⎪
( )( )
−1
a2 a ⎪ + A2 γ (mg + A2 Pa ) − 4 f min2
π 2 mH eq 4 f min
2
π 2m
Horizontal asymptotes ( and 1 ): ⎪
c2 c1 ⎪
⎩ However, there is no guarantee that the optimum point
⎛ γ mg + γ A2 Pa ⎞ ⎛ γ mg + γ A2 Pa ⎞
A2 ⎜ − 4π 2 mH eq ⎟ A2 ⎜ − 4π 2 mH eq ⎟ will be above the intersection of minimum and maximum
⎝
2
f min ⎠< ⎝ f s2 ⎠
4π 2 m 4π 2 m natural frequency curves (acceptable region). If the
d2 d1 optimum point is below the intersection, the optimum
Vertical asymptotes ( − and − ): (15)
c2 c1 answer will be the intersection of the curves, as;
4π 2 mA1 ( H − H eq ) − γ Pa A12 f min
( ))
−2
4π 2 mA1 ( H − H eq ) − γ Pa A12 f s−2
2
< ⎧ A kP A f 2 ( H − H eq ) − k1 − k3 kf max
⎪V1−int = 1 max 2 min
( 2
( H − H eq ) + γ k2
4π m 4π 2 m
⎪
⎪ k ( k f 2
3 max − P A f
max 2 min
2
)
Hence, the accepted V1 and V2 will be found based on ⎨
Equations (13) and (14);
⎪
⎪V2−int =
(
2
− A2 kf max ( 2
H eq A1 Pa − k1 + k 2 kf min H eq − γ k3 )) (18)
⎪⎩ ( 2 2
k k2 f min − Pa A1 f max )
a2V1 + b2 a V +b 2
≤ V2 ≤ 3 1 3 (16) k = 4π m k1 = A1 A2γ Pmax Pa
c2V1 + d 2 c3V1 + d3 k2 = mg + ( A2 − A1 ) Pa − A2 Pmax k3 = mg + A2 Pa
812 P. KARIMI ESKANDARY, A. KHAJEPOUR, A. WONG and M. ANSARI

To conclude, it can be said that the optimum accumulator Peq2-new are the new equilibrium pressure of chambers based
volumes are; on the isothermal process;

Peq1−new = Peq1 (V1 + A1 ( H − H eq ) )(V1 + A1 ( H − H eq −new ) )

−1
a2V1−opt + b2
V1−opt = min (V1− min ,V1−int ) V2−opt = (19) (24)
c2V1−opt + d 2
Peq2−new = Peq2 (V2 + A2 H eq )(V2 + A2 H eq −new )
−1

3.3. Effect of Load Variations The new equilibrium position will be;
The suspension system design is based on the nominal b − b 2 − 4ac
vehicle mass, which includes the net mass of the vehicle H eq −new =
2a
plus half of the maximum allowed load of the vehicle. In
a = Peq2 A22 A1 − Peq1 A12 A2 + ml gA2 A1
the maximum load situation, the effect of the extra load on (25)
the behavior of the air suspension should be studied as well b = Peq1 A12 ( A2 H eq − V2 ) − Peq2 A22 (V1 + A1 ( H + H eq ))
as the minimum load situation, which is the net mass of the + ml g ( AV
1 2 − A2V1 − A2 A1 H )

vehicle. In this section, the maximum load situation is c = Peq2 A2 H eq (V1 A2 + A1 A2 H ) + Peq1 A12 H eqV2 − ml gV2 (V1 + A1H )
studied and all of the formulations are applicable to
minimum load situation, assuming the extra load is The new equilibrium ride height of the air suspension
negative. As expected, after adding extra load, the sprung due to extra load (based on the isothermal process) is
mass and piston moves down and will not be in the desired calculated. The normal vibratory motion of the suspension
equilibrium position and extra air is needed to increase the is based on the isentropic process and new chamber
pressure to keep the chassis at the equilibrium point. In pressures and equilibrium height found in Equations (24)
practice, a leveling valve is used to keep the ride height and (25) will be valid. The pressure of the chambers can
fixed. Since this process is slow, an isothermal process will then be changed according to Equation (13) if the
be considered to examine the behavior of the air suspension suspension stiffness or natural frequency of the vehicle
for the addition of extra load. In this case, the equilibrium needs to be changed.
force equations become;
3.4. Ride Height Tuning
C
PV = C ⇒ P = As discussed before, another advantage of the proposed air
V
suspension system is ride height tuning. It is possible to
C1 C2
P1 ( z ) = P2 ( z ) = (20) achieve the same desired natural frequency in a different
V1 + A1 ( H − z ) V2 + A2 z
equilibrium height (ride height), by changing the air mass
⎛ C2 ⎞ ⎛ C1 ⎞ of the chambers. Since height change will modify the
F ( z) = ⎜ − Pa ⎟ A2 − ⎜ − Pa ⎟ A1
V
⎝ 2 + A2 z ⎠ V +
⎝ 1 1 A ( H − z ) ⎠ chambers’ volumes and since the suspension stiffness
depends on the chambers’ volumes, the pressure in the
and the stiffness becomes; chambers should also need to be modified to achieve the
∂F ( z ) P2 ( z ) A22 P1 ( z ) A12 same stiffness. The new equilibrium pressures for a new
K ( z) = − = + (21) ride height are a function of equilibrium height, which is
∂z V2 + A2 z V1 + A1 ( H − z )
obtained by solving Equation (4) respect to equilibrium
The relation between natural frequencies of a system height (Heq);
with the isothermal process, defined by Equation (21), and
⎛a⎞ ⎛b⎞ ⎛c ⎞
isentropic process is; Peq1 = ⎜ ⎟ H eq2 + ⎜ ⎟ H eq + ⎜ 1 ⎟
⎝ d1 ⎠ ⎝ d1 ⎠ ⎝ d1 ⎠
f Isentropic = γ f Isothermal ⇒ f Isothermal = 0.845 f Isentropic (22) ⎛ a ⎞ ⎛ b ⎞ ⎛c ⎞
Peq 2 = ⎜ ⎟ H eq2 + ⎜ ⎟ H eq + ⎜ 2 ⎟
where γ is the heat capacity ratio of gas and it is 1.4 for air. ⎝ d2 ⎠ ⎝ d2 ⎠ ⎝ d2 ⎠
For instance, the isothermal natural frequency of air a = 4 fs2π 2 mA1 A2
suspension is 1.05 Hz at the equilibrium position where the (26)
b = 4 fs2π 2 m( A2 ( A1 H + V1 ) − V2 A1 ) + A1 A2γ ( Adamper Pa + mg )
isentropic natural frequency is 1.25 Hz.
The force equilibrium equation should be satisfied in the c1 = (4 fs2π 2 mV2 − A2γ ( Adamper Pa + mg ))(V1 + A1 H )
new equilibrium position (Heq-new) after adding the extra c2 = V2 (4 fs2π 2 m( A1 H + V1 ) + A1γ ( Adamper Pa + mg ))
load;
d1 = A1γ ( AV
1 2 + A2V1 + A1 A2 H )

⎧ F(at z=Heq ) = mg ⇒ mg = ( Peq 2 − Pa ) A2 − ( Peq1 − Pa ) A1 d 2 = A2γ ( AV

1 2 + A2V1 + A1 A2 H )
⎪⎪
⎨ F(at z=Heq-new ) = ( m + ml ) g ⇒ (23)
⎪
(m + ml ) g = ( Peq 2− new − Pa ) A2 − ( Peq1− new − Pa ) A1 4. DESIGN AND EXPERIMENT
⎪⎩

where ml is the extra load mass and Heq > Heq-new. Peq1-new and 4.1. System Design
ANALYSIS AND OPTIMIZATION OF AIR SUSPENSION SYSTEM WITH INDEPENDENT HEIGHT AND STIFFNESS 813

Table 1. Heavy truck air suspension specifications. suspension with tunable stiffness and ride height and
specifications listed in Table 1. The solid axle is attached to
Description Symbol Value Unit
an axle adapter, which transfers the force and motion of the
Truck sprung mass on each axle to the trailing arm or lever as a connection to the air
m 5,216 kg
suspension chambers. The upper control arm controls the motion of
Air chamber 1 nominal height Heq1 0.25 m axle and axle adapter according to the designed four-bar
Air chamber 2 nominal height Heq2 0.33 m linkage. Finally, the hanger is rigidly connected to the
chassis and fixes the trailing arm pivot point.
Air chamber 1 area A1 507 cm2 A test fixture was designed for evaluating the
Air chamber 2 area A2 730 cm2 performance of the air suspension system. As shown in
Air chamber 1 horizontal location L1 0.24 m Figure 6, three hydraulic cylinders were used on each side
to simulate the longitudinal, lateral and vertical forces
Air chamber 2 horizontal location L2 0.71 m
applied to the axel. Longitudinal and lateral forces are
Axle effective horizontal location Lax 0.47 m applied exactly at the tire contact patch location by cylinder
Minimum natural frequency fmin 1.4 Hz 3 and 2 respectively. Due to spatial constraints, a bigger
Maximum natural frequency fmax 1.6 Hz hydraulic cylinder for vertical force (cylinder 1) is mounted
on the top of axle point. A control system developed in
Heat capacity ratio of air γ 1.4 - dSpace was used to control the hydraulic servo valves
according to the different test scenarios.
To measure the stiffness of the air suspension, only the
New parameters in Equation (9) (starred) will be used in vertical actuator was activated and moved the axle to its
order to determine the optimum accumulator volumes and nominal position (Z = 0). The longitudinal and lateral
the initial equilibrium pressures of air chambers. Table 1 cylinders are free to move and do not exert any force on the
represents the specifications of the air suspension system system. The force applied by the vertical cylinder is
for a heavy truck (class 8 truck). measured through a sensor. The test is done for low-
Optimum accumulator volumes should now be calculated stiffness (minimum desired natural frequency, fmin) and
using Equation (19) with values obtained experimentally in high-stiffness (maximum desired natural frequency, fmax)
Table 1. According to the manufacturer recommendation, conditions. To perform the test, the air pressure in chamber
the desired natural frequency range is chosen to be between 1 (Peq1) was set to the theoretical evaluated pressure. Then,
1.4 Hz to 1.6 Hz (Chalmers Suspensions International Inc., the air pressure in the second chamber was increased until
2015). After substituting Table 1 parameters into Equation the measured vertical force at axle point became equal to
(19), the optimum accumulator volumes are found to be the expected truck weight on the suspension. The determined
less than zero. It means that trapped air in each air chamber air pressure at chamber 2 was in good agreement with the
(A1Heq1 and A2Heq2) can cover the whole range of the desired calculated pressure (Peq2) in Equation (5). The experimental
natural frequency and the current proposed system can measured and theoretical calculated equilibrium pressures
support a wider range of stiffness. Considering Equation of chambers are;
(5) at the nominal position of system (Z=0) and zero The measured air pressures of chamber 2 from the
volumes for accumulators (V1=V2=0), the pressure of air
chambers 1 and 2 for different desired natural frequencies
are shown in Table 2; Table 3. Comparison of experimental chamber pressures
and theoretical calculated chamber pressures for low and
4.2. Experiment Results high stiffness scenarios.
The method discussed above was used to develop a truck
Experimental Theoretical
Descrip- measured calculated Error
Chamber
tion pressure pressure (%)
Table 2. Chamber pressures for different desired natural (psi) (psi)
frequencies.
Chamber 1
0 0 -
Desired natural Chamber 1 Chamber 2 Low (Peq1)
frequency (Hz) pressure Peq1 (psi) pressure Peq2 (psi) stiffness Chamber 2
69 67 2.9
fmin = 1.4 (Peq2)
0 67
(Low stiffness) Chamber 1
74 74 -
F = 1.5 35 75.5 High (Peq1)
fmax = 1.6 stiffness Chamber 2
74 85 88 85 3.4
(High stiffness) (Peq2)
814 P. KARIMI ESKANDARY, A. KHAJEPOUR, A. WONG and M. ANSARI

Table 4. Comparison of experimental stiffness and

theoretical calculated stiffness for low and high stiffness
scenarios.
Experimental Theoretical
Error
Description measured calculated
(%)
stiffness (kN/m) stiffness (kN/m)
Low stiffness 576 591 2.6
High stiffness 740.5 776 4.8

process for a gas trapped in the air chambers are valid for
very slow and very fast motion of air suspension,
respectively. Since the experiments were conducted at low
frequency, an isothermal condition was used for comparing
the calculated stiffness of the system based on Equation
(21) with the experimental results. The results are presented
Figure 6. Test setup for air suspension system: (1) Vertical in Table 4;
actuator; (2) Lateral actuator; (3) Longitudinal actuator. It can be seen that theoretical calculated stiffness has less
than five percent discrepancy with the experimental results.
This difference is mostly due to friction of joints, calibration
experiment compared to calculated pressures in Table 2 of pressure transducers and measurement of air chambers
based on Equation (5), have less than four percent effective area. The results however, confirm the validity of
discrepancy. This verifies the accuracy of the mathematical the mathematical modeling approach.
formulation developed for calculation of the initial
equilibrium pressures. 5. CONCLUSION
Figure 7 depicts the results of different experiments
performed on the air suspension system in order to find the Mathematical model of the air suspension system with
actual stiffness of the system. The axle vertical force is independent height and stiffness tuning was developed. An
slightly different for loading and unloading the air optimum design for this air suspension system was found
suspension system. This force discrepancy is related to and can be used for any vehicle specifications and air
damping and friction forces in the system, mainly in the suspension geometry. Several design constraints were
bushings, which their direction changes over the extension considered in the development of the model and design
and retraction. procedure. The optimum accumulator volume equations
The stiffness of the new air suspension system for low were derived based on the minimum and maximum desired
and high stiffness scenarios was measured based on the natural frequencies in maximum working pressures. After
experiments. The assumptions for isothermal and isentropic analyzing the air suspension design, corresponding
equations were derived for setting the air chamber
equilibrium pressures required to keep the vehicle at any
desired height with extra load.
The approach was used in the design and fabrication of
the new air suspension for a heavy truck. Experiments were
conducted to measure the pressure of the air chambers at
the equilibrium point and stiffness of the air suspension.
The results corroborate the estimates obtained from the
models.

ACKNOWLEDGEMENT−The authors would like to

acknowledge the financial support of the Natural Sciences and
Engineering Research Council of Canada (NSERC), and also
financial and technical support of Chalmers Suspensions
International Inc. in this work.

REFERENCES
Figure 7. Vertical force vs. displacement of axle for low-
stiffness (Peq1 = 0 psi and Peq2 = 69 psi) and high-stiffness Ahmadian, M. and Blanchard, E. (2011). Non-
(Peq1 = 74 psi and Peq2 = 88 psi). dimensionalised closed-form parametric analysis of
ANALYSIS AND OPTIMIZATION OF AIR SUSPENSION SYSTEM WITH INDEPENDENT HEIGHT AND STIFFNESS 815

semi-active vehicle suspensions using a quarter-car 42, 3-4, 328−347.

model. Vehicle System Dynamics, 49, 219−235. Giacomin, J. and Hacaambwa, T. Performance of ISO2631
Ahmadian, M. and Pare, C. A. (2000). A quarter-car and BS6841 Comfort Criteria for Evaluating Automobile
experimental analysis of alternative semiactive control Road Vibrations. The University of Sheffield. Sheffield,
methods. J. Intelligent Material Systems and Structures UK.
11, 8, 604−612. Giliomee, C. and Els, P. (1998). Semi-active hydropneumatic
Balamurugan, L., Jancirani, J. and Eltantawie, M. A. spring and damper system. J. Terramechanics 35, 2,
(2014). Generalized magnetorheological (MR) damper 109−117.
model and its application in semi-active control of Gobbi, M. and Mastinu, G. (2001). Analytical description
vehicle suspension system. Int. J. Automotive Technology and optimization of the dynamic behaviour of passively
15, 3, 419−427. suspended road vehicles. J. Sound and Vibration 245, 3,
Bolandhemmat, H., Clark, C. M. and Golnaraghi, F. 457−481.
(2010). Development of a systematic and practical Irvine, T. (2000). Power Spectral Density Units: [G2/Hz].
methodology for the design of vehicles semi-active Vibration Data Press. Pennsylvania, USA.
suspension control system. Vehicle System Dynamics 48, Khajepour, A., Yin, Z., Cao, D. and Ebrahimi, B. (2012).
5, 567−585. Adaptive Gas and Fluidic Suspension Systems with
Brookes, G. R., Mickler, D. J., Monroe, C. R., Niaura, W. S. Independent Stiffness and Height Tunings. Lancaster,
and Tener, D. R. (2008). Adjustable Vehicle Suspension UK.
System with Adjustable-rate Air Spring. US Patents. Khajepour, A., Yin, Z., Cao, D. and Ebrahimi, B. (2014).
Patent No. 7,380,799. Suspension Systems and Methods with Independent
Cao, D., Rakheja, S. and Su, C.-Y. (2010). Roll-and pitch- Stiffness and Height Tuning. US Patents. Patent No.
plane coupled hydro-pneumatic suspension, Part 1: 8,672,335.
Feasibility analysis and suspension properties. Vehicle Lee, S. J. (2010). Development and analysis of an air
System Dynamics 48, 3, 361−386. spring model. Int. J. Automotive Technology 11, 4, 471−
Cao, D., Song, X. and Ahmadian, M. (2011). Editors’ 479.
perspectives: Road vehicle suspension design, dynamics, Liu, Y., Matsuhisa, H. and Utsuno, H. (2008). Semi-active
and control. Vehicle System Dynamics 49, 1-2, 3−28. vibration isolation system with variable stiffness and
Chalmers Suspensions International Inc. (2015). http:// damping control. J. Sound and Vibration 313, 1-2, 16−
www.chalmerssuspensions.com 28.
Chang, F. and Lu, Z. (2008). Dynamic model of an air Liu, Y., Matsuhisa, H., Utsuno, H. and Park, J. G. (2005).
spring and integration into a vehicle dynamics model. Vibration isolation by a variable stiffness and damping
Proc. Institution of Mechanical Engineers, Part D: J. system. JSME Int. J. Series C: Mechanical Systems,
Automobile Engineering 222, 10, 1813−1825. Machine Elements and Manufacturing 48, 2, 305−310.
Chien, T., Chen, C., Chiu, H. and Cheng, H. (2008). Almost Matschinsky, W. (2000). Road Vehicle Suspensions.
disturbance decoupling control of nonlinear MIMO Transportation Research Board, The National Academies
uncertain system and application to half-car active of Sciences, Engineering, and Medicine, Washington,
suspension system. Int. J. Vehicle Design 46, 4, 367− DC, USA.
392. Schisler, R. C. (2005). Dual Height Airspring Having
Cole, D. J. (2001). Fundamental issues in suspension design Adjusted Spring Rate. Google Patents. Patent No.
for heavy road vehicles. Vehicle System Dynamics 35, 4- 6,923,434.
5, 319−360. Sharp, R. and Hassan, S. (1986). An evaluation of passive
Deo, H. V. (2006). Axiomatic Design of Customizable automotive suspension systems with variable stiffness
Automotive Suspension Systems. Ph. D. Dissertation. and damping parameters. Vehicle System Dynamics 15,
Massachusetts Institute of Technology. Cambridge, MA, 6, 335−350.
USA. Song, X., Ahmadian, M., Southward, S. and Miller, L. R.
Easter, M. R. (1995). Adjustable Rate Air Spring. US (2005). An adaptive semiactive control algorithm for
Patents. Patent No. 5,413,316. magnetorheological suspension systems. J. Vibration
Els, P. S., Theron, N., Uys, P. and Thoresson, M. (2007). and Acoustics 127, 5, 493−502.
The ride comfort vs. handling compromise for off-road Standardization, I. O. F. (1997). Mechanical Vibration and
vehicles. J. Terramechanics 44, 4, 303−317. Shock-evaluation of Human Exposure to Whole-body
Eltantawie, M. A. (2012). Decentralized neuro-fuzzy control Vibration - Part 1: General Requirements. The
for half car with semi-active suspension system. Int. J. Organization. ISO 2631, 1.
Automotive Technology 13, 3, 423−431. Swevers, J., Lauwerys, C., Vandersmissen, B., Maes, M.,
Fu, K. and Mills, J. K. (2006). Integrated design of a Reybrouck, K. and SAS, P. (2007). A model-free control
quarter-car semi-active suspension system using a structure for the on-line tuning of the semi-active
convex integrated design method. Int. J. Vehicle Design suspension of a passenger car. Mechanical Systems and
 816 P. KARIMI ESKANDARY, A. KHAJEPOUR, A. WONG and M. ANSARI

Signal Processing 21, 3, 1422−1436. Yin, Z., Khajepour, A., Cao, D., Ebrahimi, B. and Guo, K.
Toyofuku, K., Yamada, C., Kagawa, T. and Fujita, T. (2012). A new pneumatic suspension system with
(1999). Study on dynamic characteristic analysis of air independent stiffness and ride height tuning capabilities.
spring with auxiliary chamber. JSAE Review 20, 3, 349− Vehicle System Dynamics 50, 12, 1735−1746.
355.

View publication stats