A universal suspension test rig for electrohydraulic active and passive automotive suspension system

A universal suspension test rig for electrohydraulic active and passive automotive suspension system Alexandria Engineering Journal (2017) xxx, xxx–xxx HO ST E D BY Alexandria University Alexandria En[.]

ORIGINAL ARTICLE

A universal suspension test rig for electrohydraulic

active and passive automotive suspension system

Mechanical Department, Faculty of Engineering, Alexandria University, Egypt

Received 29 October 2016; revised 24 December 2016; accepted 16 January 2017

KEYWORDS

Active suspension;

Passive suspension;

Servo;

Hydraulic;

Control;

PID

Abstract A fully active electro-hydraulic and passive automotive quarter car suspensions with their experimental test-rigs are designed and implemented Investigation of the active performance compared against the passive is performed experimentally and numerically utilizing SIMULINK’s Simscape library Both systems are modeled as single-degree-of-freedom in order to simplify the val-idation process Economic considerations were considered during the rig’s implementation The rig consists of two identical platforms fixed side by side allowing testing two independent suspensions simultaneously Position sensors for sprung and unsprung masses on both platforms are installed The road input is introduced by a cam and a roller follower mechanism driven by 1.12 kW single phase induction motor with speed reduction assembly The active hydraulic cylinder was the most viable choice due to its high power-to-weight ratio The active control is of the proportional-inte gral-differential (PID) type Though this technique is quite simple and not new, yet the emphasis

of this paper is the engineering, design and implementation of the experimental setup and con-troller A successful validation process is performed Ride comfort significantly improved with active suspension, as shown by the results; 24.8% sprung mass vibration attenuation is achieved The details of the developed system with the analytical and experimental results are presented

Ó 2017 Faculty of Engineering, Alexandria University Production and hosting by Elsevier B.V This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

1 Introduction

The three available classifications of the suspension system

(Fig 1) are passive, semi-active and active suspension systems,

and this classification depends on the ability of the system to

absorb, add or extract energy The passive suspension

(Fig 1a) is the most commonly used due to its simplicity,

robustness and low price It has limited performance because

its components can only store or dissipate energy and can never create energy which cannot satisfy both the comfort and handling requirements under varying road conditions Most passive suspension systems employ spring with hydraulic

or pneumatic shock absorber The damping force created by shock absorbers is based on converting vibration energy into heat, then dissipating it to surroundings This leads to change

in oil viscosity which influences the damping characteristics[1] Traditional automotive suspension designs have been a compromise between three conflicting criteria which are road handling, load carrying, and passenger comfort Good ride comfort requires a soft suspension but it will be sensitive to changes in applied loads Good handling requires a suspension

* Corresponding author.

E-mail address: mahmoudomar91@gmail.com (M Omar).

Peer review under responsibility of Faculty of Engineering, Alexandria

University.

H O S T E D BY

Alexandria University Alexandria Engineering Journal

www.elsevier.com/locate/aej www.sciencedirect.com

http://dx.doi.org/10.1016/j.aej.2017.01.024

setting neither stiff nor soft The conventional passive

suspen-sion involves spring and damper with constant coefficients[2]

Due to these conflicting demands, suspension design should

compromise between these two problems as shown inFig 2

[3]

Nowadays picking a car became such a prolonged and

tir-ing process Cars were previously chosen accordtir-ing to their

size and power but now as people spend a considerably long

time in their cars, comfort became one of the major aspects

of choosing one’s vehicle Hence, all of car manufacturers

are competing in providing the utmost level of comfort by

modifying their suspension systems to cope up with the road

bumps and potholes Although the excitations arising from

road roughness primarily affect the vehicle ride comfort, it is

the input over which vehicle design engineers and vehicle

dri-vers have the least amount of control There are three different

models for potholes which are smooth, non-smooth and

statis-tical potholes[4] Thus automotive manufacturers started to

explore alternatives for the passive suspension to eliminate

the above mentioned compromise and that is when the

princi-ple of active and semi-active suspensions started to be

increas-ingly employed in high-end luxury cars as they improve

comfort and stability despite their high price and power

con-sumption[5]

The semi-active (Fig 1b) was first introduced by Karnopp

and Crosby in the early 1970s, [6]based on the well-known

skyhook control The damping coefficient is varied by variety

of methods but still the suspension system can only dissipate

the road forces and can’t add additional force to the system

With the right control system, the passive suspension’s

com-promise can be reduced resulting in a smart system making

cars comfortable regardless of the road they are driven on

Choi et al.[7]and Yao et al.[8]discussed the design and

con-trol of the Magnetorheological (MR) dampers via several tech-niques while utilizing Hardware-in-the-loop-Simulation (HILS) methodology Another type of semi-active suspension utilized the Electro-rheological (ER) damper system Choi

et al.[9]performed field test to evaluate performance charac-teristics of a semi-active ER suspension system associated with skyhook controller They demonstrated that ride comfort and steering stability of the vehicle were improved Cao et al [4] showed that semi-active systems have advantages over active systems, including low power requirements, simplicity, ease

of implementation and low-cost

Active suspension systems (Fig 1c) employ a controllable actuator between the sprung and unsprung masses This actu-ator is able to both add and dissipate energy to and from the system The early studies on active suspensions performed by Hrovat [10] included numerous approaches such as modal analysis, eigenvalue assignment, model order reduction, non-linear programming, multi-criteria optimization, and optimal control Classic control methods have also been considered,

Figure 1 Suspension system classifications

Low Damping -> High Damping

Vehicle Handling Ride Comfort

Figure 2 Damping compromise for passive dampers

Nomenclature

LPT linear position transducer

PID proportional integral derivative

such as root locus, Bode diagrams, and Nichols plots Before

applying any of those control techniques well-defined, linear

model for the system is a necessity To design the controller,

linearization of the system is a must The main obstacle for

commercialization of such systems is the significant power

requirement In order to reduce the cost associated with the

required power, practical active suspension designs generally

function as a low-bandwidth system that requires 4 kW of

peak power for road vehicle applications

Chandekar et al [11] investigated the servo controlled

hydraulic suspension The hydraulic pressure is supplied by a

radial piston pump Body movement and vehicle level are

monitored in real time by the controller which operates the

hydraulic servos, mounted beside each wheel

Bello et al.[12]and Venkateswarulu et al.[13]designed a

PID control for a 4 DOF, nonlinear, half vehicle active

suspen-sion system model A comparison was made between nonlinear

passive and the developed active system which showed a better

active performance The constructed model ignored

nonlinear-ities in the hydraulic actuator as their effect was minimal and

was created numerically using Matlab/Simulink Both

intro-duced a sinusoidal road input disturbance and the developed

system by Bello performed a 52.29% and 57.47% reduction

in front and rear suspension deflection respectively It was

con-cluded that regardless of the power consumption of the active

system, it has better performance Bello et al.[14]constructed a

state space model for 2 DOF quarter car using full

state-feedback controller numerically via Simulink For step input

of 0.1 m, the sprung mass acceleration and displacement of

the active system has been reduced by 80% and 11%

respec-tively compared to the passive system which shows an

improvement in the ride comfort, also, the rattle space usage was reduced by 92.5% compared to the passive suspension sys-tem The settling time in all cases was about 2 s

Fayyad [15], Kumar et al [16] and Elattar et al [17] designed a PID controller for a QC model to improve the ride comfort and road holding ability Fayyad showed numerically that for the step input of 80 mm, the sprung mass displacement has been reduced by 25% while Kumar found that ride com-fort is improved by 78.03% and suspension travel has been reduced by 71.05% with active system compared to passive one both experimentally and numerically Elattar et al com-pared between PID and PDF controllers and showed that although both showed improved performance, PDF has more potential

2 Experimental setup

A detailed layout of the test rig is shown inFig 3with all com-ponents specified During the design stage, a reasonable factor

of safety was employed while designing the tables and the cam-shaft to ensure durability of the test rig and its ability to with-stand large hydraulic and inertial forces safely

The test rig shown inFig 3 consists of, two testing plat-forms fixed side by side Each of them consists of 0.5 m steel square table 0.6 m high (1 & 10), two journal bearings (14)

to hold the camshaft, a bottom base plate (6), 4 steel guide bars (8), a spring and damper shock absorber assembly with its fix-ation points (21), a sprung masse divided into two plates (11,12

& 2,3), an unsprung mass (4 & 5) and upper plate to hold the assembly together (7) Both platforms are almost identical and

Figure 3 Detailed schematic with components

assembled together side by side in a way allowing one camshaft

(15) to drive both platforms to ensure that the road input is

identical for both suspension systems under test The assembly

is lifted on wheels in case that it is needed to be transported

and also equipped with round base plates for ground fixation

to secure the assembly in position while operating

As seen fromFig 3the left platform is for the passive

sus-pension and the right one is equipped with universal fixation

points to accommodate a spring and damper assembly, an

active hydraulic cylinder or even both of them side by side

Both sprung and unsprung mass plates are equipped with a

fix-ation point for a linear position transducer of type (OPKON

LPT200)[18]for the closed loop feedback control system

The camshaft is supported by four journal bearings for

uni-form weight distribution, and driven by 1.12 kW induction

motor (19) through a 1:7 reduction gear box and 1:5 reduction

chain (17) to reduce the rpm of the motor from 1450 rpm to

41.4 rpm as a camshaft rotation speed, and this speed

corre-sponds to a vehicle’s linear velocity of 1.56 km/h which is slow

but was chosen to initially test the control system’s response A

schematic of the test rig is shown inFig 4indicating important

sections in the design

Fig 5 shows the mechanism for road disturbance input,

which is developed on the basis of vertical movement of the

suspension system produced by a cam (3) and follower (4)

mechanism driven by an induction motor (7) through a

gear-box (2) and chains (5) for speed reduction and torque

augmen-tation The desired input is sinusoidal which is achieved by

using a circular cam 190 mm in diameter and its driving shaft

is offset from its center a distance OQ = 50 mm (Fig 5) The

cam profile (X) can be calculated from Eq.(1), where X is in

mm.[19]

where OQ is the offset distance between the cam and rotating centers

The cam had 100 mm peak to peak stroke (calculated at

b = 180°) and it generated a waveform modeled as rðtÞ ¼ 100 sin xt where x ¼ 2pf and f ¼ 0:7Hz There was no need for a return spring on the follower because the heavy weight of the system, about 80 kg, above each follower is enough to always ensure surface contact between the cam and the follower

The test rig allows configurations for 1DOF (Fig 6(a)) and 2DOF (Fig 6(b)) systems by removing or installing the springs representing the tire stiffness

2.1 Hydraulic power unit

A hydraulic power unit is used to drive the active suspension system A schematic for it with the active suspension system

is designed by Automation Studio Program as shown in Fig 7 It consists of a 60-liter hydraulic oil reservoir (1) and

a pressure compensated axial piston pump of 14 cm3/rev (2) mounted on top of the reservoir and driven by a 2.23 kW single phase induction motor (3) running at 1450 rpm The maximum supply pressure of the pump is 8 MPa A medium pressure filter (4) is of filtrations 10lm filters the oil supplied

by the pump The pump delivery line includes a check valve

Figure 4 Schematic layout of the test rig

(5), an accumulator (6) of volume 0.6 liter and a pressure

gauge (7)

The pressure line feeds through two lines, the first for the

active suspension’s cylinder servo valve and the second for

the road input disturbance cylinder servo valve Both lines

are connected to a specially designed servo valve adapter block

according to the DIN standard porting pattern to adapt a size

6 Rexroth single stage servo valve and a size 6 MOOG single

stage servo valve

2.2 Electronic control panel

The electric and electronic control panel is developed especially

for the test rig including the following major components: An

Arduino Mega 2560 used for processing and control of the

dig-ital and analogue I/O signals via serial communications, a

uni-versal multioutput power supply with input 220 VAC and

MOOG NG6 Driver Card with integrated Signal Generator

and On-Board Relay Module, circuit breakers used as

protec-tion for the main supply and the two inducprotec-tion motors, control

relays and cooling fan The human interface is through start

and stop push buttons on the front panel, two potentiometers for manual control of the two servo valves and a main switch 2.3 Testing parameters

Both numerical and experimental results were performed with the same parameters which are stated inTable 1

All parameters were calculated experimentally to be applied

to the Simulink models Springs were calibrated by a hydraulic cylinder equipped with one position transducer and two pres-sure sensors for the piston and rod prespres-sures, and the force was calculated from Eq.(2)

The distance and force were recorded for the used suspension (Fig 8) and tire helical springs (Fig 9) and the average linear trend line slope represented the stiffness For the twin-tube type dampers (Fig 11) the distance from the sensor was differenti-ated to get the velocity and then it was plotted against the force (Fig 10) and the slope represented the damping coefficient[20] The pump[21], Rexroth[22]and Moog[23]valves param-eters were obtained from their manufacturer’s datasheets

Figure 5 Cam profile and cam drive

Tire Springs Rigid Columns

Figure 6 (a) 1DOF, (b) 2DOF

3 Numerical results 3.1 DOF passive suspension

A Simulink model has been created to simulate a 1 DOF pas-sive suspension system using the Simscape library within the

(1) Hydraulic oil Reservoir (40 l) (6) Accumulator (2) Variable Displacement Pump (21.17 LPM) (7) Pressure Gauge (3) Electrical Motor (2.27 kW) (8) 4/3 Servo Valve (4) Medium Pressure Filter (9) Double Acting Cylinder (5) Check Valve

Figure 7 Hydraulic circuit diagram of the hydraulic power supply unit

Table 1 Testing parameters

Suspension parameters

Active cylinder parameters

Piston side effective area Apiston 0.001134114 m 2

Pump parameters

Rexroth size 6 servo valve

Moog size 6 servo valve

y = 14029x - 345.28

0

500

1000

1500

2000

2500

0 0.05 0.1 0.15 0.2

Displacement (m) Figure 8 Suspension spring calibration

y = 30012x - 321.51

0

500

1000

1500

0 0.01 0.02 0.03 0.04 0.05 0.06

Displacement (m) Figure 9 Tire spring calibration

Simulink as seen inFig 12 The road input here is either a step

input created by a fast ramp in order to limit the derivative

ini-tial overshoot or the actual cam profile of a peak amplitude of

100 mm and a frequency of 0.7 Hz and the output is viewed on

the scope and recorded in the Matlab workspace for the

vali-dation process

3.2 1DOF active suspension

To improve the performance, the active suspension system had

to be incorporated to absorb the shock A Simulink model was

created in the Simscape environment to model the active

sus-pension system incorporating a double acting cylinder, as the

main actuator, installed parallel to the passive suspension and driven by a servo valve as seen inFig 13

To make the active suspension effective a control system must be incorporated to control the output signal to the servo valve according to the reference input, which in our case a ref-erence sprung mass height from the ground to be tracked regardless of the disturbed input road signal One of the robust control systems which is utilized here is the PID controller Using the auto tuning feature of the Simulink, the system was linearized and the correct PID gains were obtained to enhance the system performance The servo valve’s parameters were optimized so that the model perfectly simulates the actual valve’s performance according to its datasheet

Both the numerical active and passive sprung mass dis-placements against the road input were recorded (Fig 14) for the half sine wave input For a half sine wave input of

100 mm the passive suspension system reached a maximum

of 101 mm Also when the input dropped to zero the suspen-sion went down for only 0.872 mm which is almost zero The overall amplitude achieved by the passive suspension is equal to 101.87 mm which neither attenuated nor augmented the road input, because both the damping coefficient and spring stiffness are relatively high compared to the low sprung mass value of 50 kg

y = 5008.1x + 1036.3

0

500

1000

1500

-0.1 -0.05 0 0.05 0.1

Velocity (m/s) Figure 10 Damper calibration

Figure 11 Section view of the Twin-tube shock absorber[20]

Figure 12 Simscape 1DOF passive modeling

The Active suspension had a maximum peak value of

23.6 mm, which is a 76.4% reduction in amplitude resulting

in improved ride comfort However, after the peak value of

the input road signal has passed the system took a dive of

-36.7 mm which is much higher than the dive level of the

pas-sive system The system has an overall displacement of 60 mm

which means that 40% of the bump’s input and 41% improve-ment from the passive suspension behavior were successfully achieved

The obvious drawback of the suspension system is the dive that the sprung mass endures when the bump ends, which is justified because it is related to the hydraulic characteristics

Figure 13 Simscape 1 DOF active suspension model

-40 -20

0

20

40

60

80

100

120

0 0.5 1 1.5 2 2.5 3

Time(sec)

Road input Sprung mass displacement (Acve) Sprung mass displacement (Passive)

Figure 14 Numerical active, passive and road input vs time

of the system When the bump is introduced to the system the

suspension response is to retract the piston via the servo valve

so pressurized oil flows into the rod side of the cylinder

How-ever, when the peak tip of the bump passes and the sprung

mass starts falling, the system’s response is to extend the

cylin-der thus supplying the pressurized oil to the piston side of the

cylinder

Taking into consideration that the maximum flow of the

pump Qmax is constant then the relation between extension

and retraction speeds can be obtained from Eqs.(3)-(5).[24]

which means that the cylinder needs more time to extend than

to retract and while the time for the rising and falling ramps of the bump is the same then the active system dives but settles and reaches steady state after 1 s from passing the bump

4 Experimental results The two used position sensors are of the linear resistor type OPKON LPT200 Position sensor 1 has been mounted on the sprung mass and the second one on the unsprung mass plate The position sensor only outputs the displacement but utilizing the Simulink resources, the signal is differentiated once for velocity and twice for acceleration if needed

Figure 15 Experimental data acquisition for passive suspension system

Figure 16 Simulink program for experimental active suspension

4.1 DOF Quarter car passive suspension system

The camshaft motor is started, and the output signals from the

sensors are transferred to the Simulink via the analogue input

pins of the Arduino Mega 2560 through serial communication The transferred data are processed by the Simulink interface as the Arduino board serves the function of input and output interface and data acquisition card as seen inFig 15

-60 -40 -20

0

20

40

60

80

100

120

0 0.5 1 1.5 2 2.5 3

Time (sec)

Road input Sprung mass displacement (Acve) Sprung mass displacement (Passive)

Figure 17 Experimental active, passive and road input vs time

-150 -100 -50

0

50

100

150

0 0.5 1 1.5 2 2.5 3

Time (sec) Figure 18 Applied PID output signal to the servo valve from the PWM pin from the Arduino

-60 -40 -20

0

20

40

60

80

100

120

0 0.5 1 1.5 2 2.5 3

Time (sec)

Sprung mass displacement Exp (Passive) Sprung mass displacement Num (Passive) Sprung mass displacement Exp (Acve) Sprung mass displacement Num (Acve)

Figure 19 Comparison between experimental and numerical results