Académique Documents
Professionnel Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/308535976
Article in Proceedings of the Institution of Mechanical Engineers Part K Journal of Multi-body Dynamics
September 2016
DOI: 10.1177/1464419316671025
CITATIONS READS
0 199
3 authors:
Kjell Andersson
KTH Royal Institute of Technology
52 PUBLICATIONS 127 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Ulf Sellgren on 26 September 2016.
Abstract
A forwarder is an off-road working machine that is used to transport logs from logging sites to a landing area that is
accessible by trucks. Soil damage and operator comfort, especially whole-body vibrations when operating on hard and
rough terrain, are crucial issues when developing novel forest machines. Most forwarders on the market are heavy
machines with articulated steering and they are equipped with pairs of wheels mounted on bogies. For such bogie
machines, only the flexibility and the dynamic dissipation in the tyres contribute to the chassis damping. The roll and
lateral motions are the most severe components of the whole-body vibrations. So, developing new traction units, chassis
suspensions and/or cabin suspension are in focus. Model-based design relies on focused models that are as simple as
possible, but not too simple. This paper presents a 12 degrees-of-freedom multi-body dynamics simulation model of a
standard eight-wheeled bogie type of medium-sized forwarder. The presented model is targeted for assessing and
comparing different design solutions. It is shown that a configuration of seven rigid subsystems and eight flexible
tyres represented with the simple and computer efficient Fiala tyre model enables the forwarder dynamic simulation
model to be used to predict the roll and lateral motions of a forwarder operating on hard and rough terrain.
Keywords
Bogie, cabin roll angle, forwarder, hard ground, rough terrain, wheelsoil interaction
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
2 Proc IMechE Part K: J Multi-body Dynamics 0(0)
view was presented on how forestry machines and the terrains, climbing steep slopes, and traversing areas
soil preferably should be modelled in order to enable covered with boulders and stones, i.e. hard rough ter-
model-based and simulation-driven design with rain. Research with the aim to identify critical design
respect to design issues as operator comfort, trac- issues for forestry machines traveling in a sloped ter-
ability, mobility, and traction. rain was presented in Ismoilov et al.7
Most forwarders on the market today have articu- The main objective of this study is to build a
lated steering and they are equipped with pairs of computationally ecient multi-body dynamics base
wheels mounted on bogies. These machines are not model of a standard eight-wheeled and medium-size
equipped with any suspension system that can isolate bogie type of forwarder and to verify the simulation
the machine from vibrations caused by ground dis- model with measured data from a eld tests that was
turbances. Furthermore, most forestry machines are carried out with the actual machine operating on a
heavy and, consequently, the risk for rutting and standardised hard ground test track. The validated
soil compaction during operation is considerably simulation model is targeted to be used as a design
high. There are, thus, engineering challenges for base model that can be congured with dierent types
both manufacturers and engineers to nd means for of traction units to predict traction unit and soil inter-
reducing soil damage, rolling resistance, and the daily action and machine dynamics, with specic focus on
vibration dosages for the machine operators, i.e. the the roll behaviour of the front wagon, when operating
level of whole-body vibrations. on hard and rough terrain.
In previously published work,5 based on multi- The rest of the paper is structured as follows. In the
body dynamics simulations, a number of passive for- Field test section, we will review the actual eld test
warder chassis suspension concepts were presented environment, sensors and test condition. The MBD
and their performances were compared from three simulation model section presents the MSC Adams-
perspectives: their gentleness to the terrain and to based dynamic simulation model. A verication of
the operator, and their energy eciency. A represen- the MDB model and the model-based simulation
tative medium-sized bogie type of forwarder was results with actual test result is presented in the
modied into a virtual pendulum arm suspended for- Model verication and discussion section. The results
warder and its performance was compared with the are discussed in the Discussion section, and conclu-
bogie suspended forwarder. The simulation model of sions of the presented research and the planned
the bogie type of forwarder was veried with full-scale future work are given in the Conclusions and future
eld measurements.6 By including the FTire tyre work section.
model that takes into account the tyre shape, the air
volume, the stiness of the tyre ring, and the inter-
action with large terrain features, the simulated Field test
dynamic behaviour correlated well with the results
measured in eld tests.6 The drawback of using this
Test objective
detailed tyre model is the signicant computer The main objective of the eld test is to gather import-
resources required to simulate motion in rough and ant data on the dynamic behaviour of a traditional
varied terrain. medium-size bogie-suspended forwarder, operating
Dynamic stability reects the ability of a working on hard rough terrain. The purpose of the acquired
machine to traverse uneven terrain at a feasible range data is to enable model-based development of the next
of speed and acceleration without causing high stres- generation of forestry machines. The measured data
ses in the chassis and without causing excessive vibra- assist proper tuning and verication of the multi-body
tions in the operator cabin. Forestry machines must dynamic simulation model, which is targeted to be
be capable of traveling highly challenging and varying used as a reference model for future research on,
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
Ismoilov et al. 3
Figure 2. Jalla test field overall view (left), and the bogie forwarder at the starting point of the Skogforsk test track.
e.g., new modular chassis designs, cabin suspension The tyre ination pressure was 350 kPa in the four
systems and tracked traction units. front tyres and 500 kPa in the four rear tyres.
The eld test was carried out with an unloaded
forwarder on the standardised 28 m long and horizon-
The modelled forwarder
tal Skogforsk test track (see Figure 2) in Jalla, located
approximately 60 km north from Stockholm. The test The forwarder under consideration has a machine
track was targeted for repeatable studies of the vibra- weight around 16 tons and a loading capacity of
tion levels on dierent machines and to evaluate the 12 tons. It is an articulated, eight-wheel machine
performance of dierent suspension solutions. The with no primary chassis suspension system. The gen-
test track, which was developed by Skogforsk and eral dimensions of the forwarder can be seen in
Hultdins AB, is paved with concrete, and equipped Figure 3. The forwarders width, called A in Figure 3,
with a number of unevenly distributed metal obstacles varies between 2760 and 2990 mm, depending on the
of the same shape but dierent height. The obstacles tyre size. The forwarder that was used in the
are attached to the ground in two rows, one for each eld test was equipped with Trelleborg 710/50
side of the forwarder. The obstacles in each track are 26.5 T428 163A8 forestry tyres, giving a total
uniform across the wheel width, i.e. each wheel will width of 2990 mm. The propulsion is mechanically
meet a row of two-dimensional obstacles. transferred to all eight wheels from a 140 kW diesel
In the test, a number of machine parameters were engine through a hydrostatic transmission system.
measured when the machine was traversing the test Inside each bogie box, the tractive power is trans-
track: mitted to the wheel pair through a chain-sprocket
system, i.e. there is all-wheel traction but no traction
Translational vertical, lateral and longitudinal control for the individual wheels in a bogie.
accelerations in front and rear wagon
Angular pitch and roll rate in front and rear wagon
Bogie angles relative to frame
Sensors
Machine speed In order to capture the translational and angular
Vertical static load on each wheel. motions of the forwarder, two gyroscopes
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
4 Proc IMechE Part K: J Multi-body Dynamics 0(0)
(see Figure 4) were mounted on the front and rear In order to capture all necessary frame motion sig-
wagons, respectively. Each gyroscope was capable of nals from the front and rear wagons, the rear gyro-
measuring translational acceleration in all three direc- scope sensor was attached with screws to the rear
tions (vertical, lateral, and longitudinal) as well as the axles lid. The second gyroscope unit was placed
pitch and roll angular motions. under the operator seat inside the cabin, as shown
When the module is placed on a horizontal at in Figure 7.
surface, acceleration (gravity) is only detected in the To get information on all bogie angle motions, four
Z-direction, as shown in Figure 5. angle sensors were mounted on each bogie, close to
If the module is tilting backwards, the value for the the pivot joint (see Figure 8, left). During the test, the
Z-axis decreases, the value for the Y-axis increases, machine speed was measured in two independent
and the gyro1 value becomes negative. When the ways: an external sensor that was mounted in front
module tilts left or right, its Y-axis value increases of the forwarder (see the middle part of Figure 8), and
or decreases, respectively. Gyro2 value has negative from the machine ECU.
value while tilting left and positive value while tilting Before starting the tests, the vertical load on each
right (see Figure 6). wheel mas measured with a weight measurement
scale, as was shown in the rightmost picture in
Figure 8. The total machine weight was obtained by
adding all the eight static wheel loads.
Figure 6. Gyroscope tilted backward (left), and gyroscope tilted left and right (right).
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
Ismoilov et al. 5
Figure 7. Locations for the gyroscopes: the front one is located under the seat (left), the rear one is mounted on top of the rear axle
(right).
Figure 8. Angle sensor location on front bogie (left), speed sensor (middle), and weight measurement (right).
throughout mechanical systems.8 But, this requires a that was further translated to a 3D Adams road le
thoroughly veried model. (see Figure 10).
In this study, an MBD model of the studied for-
warder was created with the commercial MBD tool
Adams from MSC software. A core package (Adams/
Importing forwarder CAD model
View, Adams/Solver, and Adams/PostProcessor) In Adams/View, there is a possibility to import geo-
enables import of geometry from most major CAD metric models in many dierent CAD formats10 and
systems. In Adams, a model of a system is built by get them converted into Adams proprietary geometric
creating and assembling (rigid/exible) parts, connect- elements. In this study, the forwarder CAD model was
ing them with joints, and driving them with motion translated from the ISO STEP format to the Solid
generators and/or forces. Edge proprietary format and the parts were re-
assembled in Solid Edge. It is important to save each
subsystem in the global coordinate system before
Creating road file
importing the model to Adams. In this way, it is easy
The test track has obstacles with the same shape to match the orientation as in the full CAD assembly.
but dierent heights, 150 mm, 250 mm and 350 mm, A detailed list of the subsystems of the studied for-
respectively, as shown in Figure 9. The lower part of warder, with data of their mass property data and their
Figure 9 shows the top view of the test track with the location in the global coordinate system, is presented
driving direction from right to left. The widths and in Table 1. The detailed CAD model of the forwarder
heights of the obstacles, as well as the irregular dis- parts is shown as an exploded view in Figure 11.
tance between them are designed to meet the terrain After importing the geometric subsystem models to
Class 2.9 Adams, the next step was to dene the mass properties
In this model-based study, all three dierently sized of the imported bodies and dening constraints
obstacles were dened in Matlab. In the next model- imposed by joints between pairs of interacting bodies.
ling step, the geometry of each obstacle was repre- Constraints remove degrees of freedom (DOF) from
sented by discrete nodal coordinates that were used the model by adding algebraic constraint equations
to dene facet surface elements. The nodes and facet to the governing system of dierential and algebraic
elements were then translated to a road template le equations (DAEs). A joint removes from one to six
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
6 Proc IMechE Part K: J Multi-body Dynamics 0(0)
Figure 9. Dimensions and configuration of the three types of test track obstacles.
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
Ismoilov et al. 7
Number of Number of
Letter Subsystem subsystems Letter Subsystems subsystems
Table 2. List of bodies in the MBD forwarder model,18 Ftire model, etc. The drawback of these
simulation model. models is that they require considerable computer
resources. So, there must be a trade-o between simu-
Number of
Bodies subsystems Weight (kg) lation accuracy and simulation wall-clock time.
In the presented study, the Fiala tyre model, with a
Front frame 1 5860 mass of 170 kg, a width of 710 mm and an unloaded
Front bogie 2 2 1206.5 radius of 670 mm was used to represent the 710/
Front wheel 4 4 170 45-26.5 tyre from Trelleborg. The reason for chosing
Articulation part 1 250 the Fiala tyre model was because the main interest in
Rear frame 1 6060 this study was the vertical forces and the lateral and
Rear wheel 4 4 170 roll motions of the cabin, when the forwarder was
Rear bogie 2 2 1126.5 operating on hard rough terrain, i.e. not on soft and
fairly at soil. The Fiala tyre model is a physical tyre
Total: 15 18196
model and the carcass is modelled as a beam on an
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
8 Proc IMechE Part K: J Multi-body Dynamics 0(0)
elastic foundation in the lateral direction. Elastic elem- Table 4 summarises the input that the Fiala tyre
ents provide the contact between carcass and the model uses to calculate ground contact forces.
road.10 Table 3 gives detailed information of that tyre.
The Fiala model is the standard tyre model in all
Adams/Tyre modules. The input to the Fiala tyre Model verification and discussion
model originates from two sources:
Global coordinate system
. Input parameters from the tyre property le (.tir), In order to get the location of centre of mass for the
such as tyre unloaded radius, that the tyre model front and rear wagons with the required accuracy, it is
references. important to choose a correct global coordinate
. Tyre kinematic state variables, such as slip angle system for whole system. The location of the global
(a), which Adams/Tyre calculates. coordinate system should match the original (0, 0, 0)
position in the CAD model. In the developed Adams
model, the location of the global coordinate system is
Table 3. Tyre properties.23
where the articulation joint connects the front fame
Section width with the articulation part, the positive vertical Z-axis
Diameter 1340 of the system is pointing upwards, and the posi-
Rolling circumference 4020 tive longitudinal Y-axis is pointing backwards
Static loaded radius 625 towards the rear part of the forwarder, as shown in
Speed radius index 625 Figure 12. After performing the virtual test in Adams,
Nominal rim AG24.0000 which replicates the physical eld test, the transla-
Permitted rim(s) tional and angular motions of the model can be com-
Pressure For Tracks (bar) 6.00
pared with the measured data from the eld test, and
the model can thus be veried.
Nominal pressure (bar) 5.00
Tread type LS-2
Tube type Tubetype Machine weight and wheel weight distribution
Tube 700/50-26.5 In order to get good correlation between the simula-
Steelbelted Steelbelt tion results and the results from the eld tests, it is
important to have a good match of the total machine
Table 4. Explicit input state parameters for the Fiala tyre model.11
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
Ismoilov et al. 9
Table 5. Machine weight distribution on the eight wheels. However, during the test, recording of the signals
started when the machine began to move, i.e. before
Weight (kg) Weight (kg)
it reached the actual test track. To enable comparison
Wheel position Actual test Simulation model
of measured and simulated dynamic behaviour, there
Front wagon, Left bogie, 2580 2571 is consequently a need to lter the measured data and
Front wheel (FLF) to identify the point in time when the machine enters
Front wagon, Left bogie, 2570 2571 the test track and when the machine leaves the track.
Rear wheel (FLR) The average speed on the track must correlate
Front wagon, Right bogie, 2580 2562 with the average speed in the simulation model. In
Front wheel (FRF) Figure 14, a comparison between simulated speed
Front wagon, Right bogie, 2510 2562 and eld test speed is given.
Rear wheel (FRR)
Rear wagon, Left bogie, 2000 1972
Front wheel (RLF)
Bogie angles
Rear wagon, Left bogie, 1990 1985 In the eld tests, four angle sensors were placed on the
Rear wheel (RLR) middle top surfaces of each of the four bogie units
Rear wagon, Right bogie, 1990 1986 (see Figure 8). These sensors were used to measure
Front wheel (RRF) the bogie angles when the machine was traversing
Rear wagon, Right bogie, 1970 1986 the test track. The measured bogie angles and the
Rear wheel (RRR) bogie angles from the virtual test, as a function of
Total mass (kg) 18,190 18,195 time, are presented in Figure 15.
Forestry tyres when used in hard and rough terrain
are relatively sti compared to tyres used in other
weight and its distribution. It might be necessary to working machines and there is not much deformation
adjust, or ne-tune, the centre of mass of the MBD of a wheel when it is rolling over an obstacle.
system model, to get the correct weight distribution Therefore, a simple way to verify the bogie angle in
between the wheels. The total machine weight and the a simulation is to perform a kinematic analysis of
individual static wheel loads, as measured in the eld bogie and obstacle dimensions. As can be observed
test, and the weight of the MBD submodels are given in Figure 15, the rst high peak in front bogie angle
in Table 5. Figure 13 shows the acronyms for the eight (Front left and Front right) occurs when the front
wheels. wheels are interacting with the rst obstacles on the
test track (see Figures 9 and 10). The distance between
the mounting points of the two bogie wheels is clearly
Machine speed visible in Figure 3. The rst angle peak for the front
The target speed for the machine, when it was tra- bogie, visible in Figure 15, should correlate with the
versing the test track, was 1.8 km/h, i.e. 0.5 m/s. smallest obstacle and the highest peaks must correlate
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
10 Proc IMechE Part K: J Multi-body Dynamics 0(0)
Figure 14. Simulated machine speed and speed measured at the field test.
with the highest obstacles (see Figure 9). A rst simulation response in Figure 16, and the same com-
approximation of the bogie angle is that it should be parison can be seen for the rear wagon in Figure 17.
an inverse sine of obstacle height divided by the wheel The gyroscope gives us the rate of change of the
base, as shown in equation (1). angular position over time (angular velocity) with a
unit of (deg/s). This means that the sensor gives the
L derivative of the angular position over time, as shown
arcsin 1
h in equation (2).
d
Table 6 compares eld measurements, simulation _ 2
results and analytical results for the front bogie peak dt
angles. Rate feedback is extremely useful in control engin-
eering. It is usually used in combination with position
feedback. To obtain the angular position, the integra-
Angular speed
tion of the angular velocity numerically is required, as
The angular motions of the front and rear wagons are shown in equation (3).
measured with the two gyroscope sensors placed under
the operators seat, and on top of the rear axle, respect- Z t X
t
ively (see Figure 7). Roll and pitch rate signals from the t _ tdt _ tTs 3
0 0
sensor located under the seat are compared with the
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
Ismoilov et al. 11
5 5
Angle [deg]
Angle [deg]
0 0
-5 -5
-10 -10
-15 -15
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
time [sec] time [sec]
5 5
Angle [deg]
Angle [deg]
0 0
-5 -5
-10 -10
-15 -15
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
time [sec] time [sec]
Figure 15. Comparison of simulated bogie angles and the angles for the four bogies measured in the field test.
Table 6. Bogie peak angles. model for virtually assessing and comparing dierent
design means, such as passive, active and semi-active
h, Obstacle L bogie angles
chassis damping concepts and also dierent types
height Wheel
(mm) base (mm) Measurement Simulation Analytical of ground contact units, e.g. wheels, tracks, and
combinations of wheels and tracks. The interface
150 1520 5.33 5.11 5.66 between forwarder and hard ground in the presented
350 11.40 12.14 13.31 base model was represented as a Fiala wheel model
running on a hard test track with irregularly placed
obstacles of three dierent sizes. The simulation
Figure 18 shows the integrated cabin roll rate and results, with a focus on roll, were compared with
rear wagon roll rate from the simulation. The peak dynamic machine responses measured at full-scale
roll angle for the two wagons is around 6 . eld tests.
A standard FFT transform with a rectangular In the rst stage of developing the multi-body
lter, i.e. no ltering, shows the measured and simu- dynamic simulation model, the road le coordinate
lated roll angle speed of the front wagon in the fre- nodes and elements that represent the Skogforsk test
quency domain, see Figure 19. track were calculated in Matlab. These data were then
translated to a road template le that was used in the
multi-dynamic simulation model.
Discussion
Then, the detailed CAD geometry model of the
Forwarder operators are exposed to large whole studied forwarder was imported to Adams, and the
body vibrations in all six directions when machine weight and its centre of gravity are tuned
operating in rough terrain, but most severe is the with the actual measured wheel weights.
roll motion and the large lateral motions caused In the simulation model, all wheels in the machine
by the roll. It is consequently of high importance model are driven by a constant angular velocity to
to develop durable and eective means for reducing represent the constant longitudinal speed in the eld
the roll and lateral motions that present operators are test. The machine speed in the simulations is very
exposed to. similar to the average speed measured in the eld
In the presented study, a multi-body dynamics test. However, the simulation result shows that the
simulation model of a standard medium-size eight- variation in speed during an overpass diers signi-
wheeled bogie type of forwarder was developed in cantly compared to the measured speed. In future
MSC Adams and veried with eld measurements. studies, it is recommended to use torque-control on
The purpose for the model is to serve as a base each wheel as a means to keep a constant speed.
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
12 Proc IMechE Part K: J Multi-body Dynamics 0(0)
3
2
1
0
-1
-2
-3
-4
-5
0 10 20 30 40 50 60 70 80
time [sec]
15
Measurement data
Simulation data
10
roll rate [deg/sec]
-5
-10
-15
0 10 20 30 40 50 60 70 80
time [sec]
Figure 16. Simulated and measured pitch and roll angular speed history of the front wagon.
3
pitch rate [deg/sec]
2
1
0
-1
-2
-3
-4
-5
0 10 20 30 40 50 60 70 80
time [sec]
20
Measurement data
Simulation data
15
roll rate [deg/sec]
10
-5
-10
-15
0 10 20 30 40 50 60 70 80
time [sec]
Figure 17. Simulated and measured pitch and roll angular speed of the rear wagon.
An extension of the purpose for the base model, caused by the simplied tyre model on the transla-
e.g. for investigating the power performance on and tional, vertical, lateral, or longitudinal motions.
rutting of soft soil, most likely would require a more These motions all contribute to the whole-body vibra-
detailed tyre model, such as the signicantly more tion dosages for the machine operator, and also on
computer demanding FTire model. This study does soil damage. Studies with such general scope would
not present any analysis of the simulation error also require a more sophisticated tyre model.
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
Ismoilov et al. 13
Figure 18. Simulation roll angle history for the front and rear wagons.
1.5
Measurement data
Simulation data
roll rate [deg/sec]
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Frequency [Hz]
Figure 19. Measured and simulated front wagon roll angle speed in the frequency domain.
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
14 Proc IMechE Part K: J Multi-body Dynamics 0(0)
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016
Ismoilov et al. 15
Appendix 1
There are 15 bodies in the multibody simulation model of the forwarder in Adams and they are connected to each
with 14 revolute joints, as shown in Figure 20. Without any constrains, each body has six degrees of freedom
(DOF), which means 15 moving parts have 90 DOF. However, each revolute joint removes 5 DOF and each
motion removes 1 DOF. That brings this particular simulation model has 12 DOFs.
The centre of mass position in the X and Z directions for front and rear frame is shown in Figure 21. Both front
and rear wagons centre of mass position is aligned with the global coordinate system in Y direction.
Figure 21. The centre of mass position for the front and rear frames of the forwarder model.
Downloaded from pik.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on September 26, 2016