ACTIVE DAMPING OF A FORWARDER CABIN

(1)

ACTIVE DAMPING OF A FORWARDER

CABIN

BİRCAN UZUNOĞLU

(2)

(3)

ACTIVE DAMPING OF A FORWARDER

CABIN

BİRCAN UZUNOĞLU

Master of Science Thesis MMK 2016:13 MDA 514 KTH Industrial Engineering and Management

(4)

(5)

Examensarbete MMK 2016:13 MDA 514 Aktiv dämpning av en skotarhytt

Bircan Uzunoğlu Godkänt 2015-12-15 Examinator Jan Wikander Handledare Björn Möller Uppdragsgivare Komatsu Forest Kontaktperson Erik Nilsson

Sammanfattning

Skogsindustrin har en viktig roll inom den svenska ekonomin genom att skapa många jobbtillfällen samt att den gör Sverige till en av världens största exportörer av träprodukter. För att behålla den här positionen på en tuff marknad tvingas företagen att kontinuerligt förbättra sin produktivitet. Skogsindustrin förlitar sig i huvudsak på två maskiner; en skördare som fäller och kapar träden samt skotaren som transporterar träden från avverkningsplatsen. Ett sätt att öka effektiviteten i denna process är att köra fortare med skotaren utan att försämra komforten för föraren när fordonet kör genom den tuffa terrängen i skogen.

Det här examensarbetet undersöker design och implementation av en robust kontrollstrategi för att aktivt dämpa vibrationer i kabinen på en skotare för att förbättra komforten för operatören. Arbetet har begränsats till att använda en befintlig kabinram och hårdvara som tillhandahålls av Komatsu Forest samt tester i labmiljö.

Initiala tester gjordes på en cylinder med syftet att bygga upp en förståelse för den avancerade ventilens funktion samt att bestämma de nödvändiga parametrarna hos cylindern. Resultaten från dessa tester gav information om hur ventilen beter sig vid för vissa specifika insignaler samt gjorde det möjligt att modellera friktionen hos cylindern.

(6)

(7)

Abstract

Forest industry has an important role in Sweden’s economy by creating employment within the country and making it one of the world’s biggest exporters of wood products. Keeping this place in such a competitive market forces companies to increase productivity continuously. The forest industry mainly relies on two machines; a harvester that fells and cuts the trees and a forwarder that carries logs from forest to processing facilities. One way of making this process more efficient is to increase the speed of the forwarder without deteriorating the operator’s comfort while driving over the rough forestry terrain.

This thesis investigates the design and implementation of a robust control strategy to actively dampen out vibrations on a forwarder cabin frame and provide better ride comfort for the operators. The scope is limited to using the existing mechanical structure and hardware provided by Komatsu Forest and performing the tests in a lab environment.

Initial tests were done using a single cylinder test rig with the purpose of understanding valve’s working principle as well as identifying hydraulic cylinder parameters. The results from these tests provided information about the valve response for a specific set of inputs and also cylinder friction forces were modelled.

After that an active damping test rig consisting of a lower and an upper platform together with four hydraulic cylinders was used for testing. Valve model was obtained with parameter identification tests. Based on the model, a linear velocity controller was designed for cylinders. The disturbance introduced to the system was inversed to be given as a reference to the rig and then this reference was mapped to cylinder reference velocities using geometrical relations. Due to a hardware problem encountered towards the end of the project, whole test rig couldn’t be used. Instead two cylinders were tested and the data collected from these were used for the other two, therefore it became possible to simulate the upper platform motion. Only disturbances in the roll direction between 0.5 Hz to 2 Hz were given to the system. The tests on the active damping rig showed promising results in terms of vibration reduction especially for relatively low frequencies. For 0.5 Hz, the rig managed to reduce the disturbance amplitude by 63% whereas for 1 Hz, the damping ratio decreased to 48%.

Master of Science Thesis MMK 2016:13 MDA 514

Active damping of a forwarder cabin

(8)

(9)

Acknowledgement

I would like to express my sincere thanks to my supervisor Björn Möller for his help and guidance during this thesis work. I would also like to thank Bengt Eriksson for his valuable support.

Bircan Uzunoglu

(10)

(11)

Nomenclature

Abbreviations

CAN Controller Area Network

cRIO Compact Reconfigurable Input Output

DOF Degrees of Freedom

FPGA Field-Programmable Gate Array

IMU Inertial Measurement Unit

LQR/LQG Linear-Quadratic Regulator/Linear-Quadratic Gaussian

MIMO Multiple-Input Multiple-Output

NI National Instruments

RT Real-Time

SISO Single-Input Single-Output

Notations

α Ratio between rod side and piston side area

∆p Pressure drop across the orifice

˙z Piston velocity

φ Pitch angle

σ Viscous friction parameter

τ Time constant

θ Roll angle

Aa Piston side area, chamber A

Ab Rod side area, chamber B

Ap Piston side area

cs Static friction parameter

Cv Valve discharge coefficient

E0 Bulk modulus

Fc Coulomb friction force

Fs Static friction force

Fv Viscous friction force

Fc0 Coulomb friction parameter

Fext External force

Ff riction Friction force inside cylinder

Fs0 Static friction parameter

k Gain

l Length of the rig

m External load

pA Pressure inside chamber A

pB Pressure inside chamber B

QA Flow into piston side chamber, chamber A

QB Flow into rod side chamber, chamber B

QLe External leakage flow

QLi Internal leakage flow

(12)

VB Volume of rod side chamber, chamber B

w Width of the rig

xp Piston displacement

Xv Valve opening

z Heave displacement

(13)

List of Figures

1 World leading exporters of paper, pulp and sawn wood products,

2012 . . . 1

2 Komatsu forwarder . . . 2

3 Suspension types, q(t):displacement of the base, x(t):displacement of the mass . . . 5

4 An overview of the mechanical frame placement . . . 7

5 Mechanical structure consisting of two frames, connected with three steel bars . . . 8

6 Single cylinder test rig . . . 11

7 Active damping test rig,1:cylinder 1, 2:cylinder 2, 3:cylinder 3, 4:cylinder 4, 5:lower frame, 6:upper frame . . . 12

8 Valve-cylinder combination . . . 13

9 Velocity dependent friction force, Stribeck friction curve . . . 15

10 Position coordinate for the cylinders . . . 15

11 Asymmetric cylinder, showing piston side chamber and rod side chamber . . . 16

12 Model comparison, yellow line: input command, blue line: simulation, red line: real system (mean value of all four cylinders) 17 13 Sketch of the simplified system for the roll motion . . . 18

14 Output feedback controller scheme, r:reference, y:output, u:control signal, S/R:feedback, T/R:feedforward, B/A:plant (Gpe and Gpr) . . . 20

15 Simulink model of the controller, 1:feedforward, 2:feedback, 3:delay, 4:actuator limit, 5:integral gain, 6:plant model, 7:sign function . . . 21

16 Overall control structure . . . 21

17 Hardware overview . . . 22

18 Model comparison for valve 4 - cylinder 4 . . . 25

19 Model comparison for valve 3 - cylinder 4 . . . 26

20 Low flow commands . . . 27

21 Measured data points (blue circles) and fitted Stribeck friction curve (red line) on the left, estimated friction parameters on the right . . . 28

22 Simplified sketch of the rig showing roll direction of motion . . . 29

23 Active damping test rig, red circles: cylinders used during tests, blue circles: cylinders blocked during tests . . . 30

24 Overall motion sketch, red circles: measured variables, blue circles: simulated variables . . . 30

25 0,5 Hz roll disturbance on lower platform . . . 31

26 1 Hz roll disturbance on lower platform . . . 32

27 1,5 Hz roll disturbance on lower platform . . . 32

(15)

1 Introduction

This section describes the background, purpose and delimitations of this thesis project together with the method to be followed.

1.1 Background

Forest industry is one of the Sweden’s most important industries for its role in creating employment and maintaining the economy. Relatively small import of forest industry products compared to the export in this industry shows that it has a critical place in Sweden’s trade balance. According to Figure 1 below, Sweden was the third largest exporter of paper, pulp and sawn wood in the world in 2012 [1].

Figure 1: World leading exporters of paper, pulp and sawn wood products, 2012

(16)

Figure 2: Komatsu forwarder

During the operation of the forwarder, depending on the speed of the vehicle and the ground conditions, operator is exposed to whole body vibrations which have negative effect on his/her health condition. Therefore vibration reduction is of interest to improve the comfort for the operator and increase the productivity. For this purpose generally passive suspension systems consisting of a spring and/or a damper are used; however these systems are not able to damp all the vibrations, especially the ones with low frequency and high amplitude which the forwarder is mainly exposed to. At this point, active suspension systems are proposed to be implemented considering their effectiveness in the low frequency domain [4]. Suspension systems can be developed for the seat, cabin or chassis of the forwarder.

1.2 Purpose

This thesis investigates the improvement on an active cabin suspension system to dampen out vibrations and achieve better ride comfort for the forwarder.The research question can be stated as: What robust control strategy can be designed and implemented on an active damping test rig in order to reduce vibrations on a forwarder cabin?

1.3 Delimitations

The delimitations of the project that are specified by the company can be seen below:

• The scope of the thesis is limited to using the available mechanical structure (see Figure 5) and its physical envelope limitations.

(17)

• Sauer Danfoss hydraulic valves and four hydraulic cylinders with integrated position sensors provided by Skogforsk will be used.

• National Instruments Compact Reconfigurable Input Output (NI cRIO) 9076 chassis will be used as a control unit.

1.4 Method

To gain an understanding of the plant to be controlled and design a controller for it, an important step is to have a model that provides insight into the plant behaviour. The model should also be verified with the data gathered from the system and after that a controller can be synthesized and tested in a simulation environment first, before implementing on the real system [6]. Therefore the methods that will be used in this work are simulation and experimental verification.

In the previous project, two test rigs were built; a single cylinder test rig and an active damping test rig which has four cylinders. To start simple it has been decided to work with one cylinder first. With the purpose of investigating the properties of the valve, identifying its model parameters and performing the initial tests of cylinder actuation, the single cylinder setup will be used.

(18)

(19)

2 Frame of reference

This section of the report explains the frame of reference for the project. First part presents the differences between the basic suspension systems, passive, semi-active and active, and how they are realized. Second part starts with the state of the art in active cabin suspensions and summarizes their implementation. After that the most common control strategies in active vehicle suspensions are mentioned. Last part describes the mechanical structure and the hardware used in the project.

2.1 Passive/semi-active/active suspensions

Passive suspension systems generally have a spring and a damper connected in parallel (see Figure 3a) [6]. Currently they represent the majority of the suspension systems due to their relatively low cost, small size and high reliability. However since the elastic and damping characteristics are constant, passive suspensions are most effective over a narrow range of disturbances [7]. The second category, semi-active suspensions typically consist of a spring and an adjustable damper as can be seen in Figure 3b below. No active force is applied to the system, however by controlling the damper characteristics, β, the suspension can be switched to a softer or a stiffer one depending on the road inputs [7].

Active suspension systems include an actuator that can supply an active force to the system and be controlled by a control algorithm using the sensor data. Although theoretically both the spring and the damper can be replaced with the force actuator, this is not common in practical applications, some type of spring is usually used to provide support [6]. A typical active suspension system is shown in Figure 3c. Damping constant bs here is not a

separate damper but a representation of the inherent system damping. Active suspensions are less common than the passive and semi-active ones in the industry because of their complexity and high cost [6] [7].

(a) Passive (b) Semi-active (c) Active

(20)

2.2 Active cabin suspensions

(21)

reference positions. The local controller for each cylinder position is designed in a cascaded loop featuring an inner loop for the valve position control using error feedback [12].

Mantaras and Luque [13] evaluate the ride comfort performance of the different active suspension control strategies in the simulation environment using a two degrees of freedom quarter car model and the simulation results show that fuzzy logic [9] performs better in vibration isolation. Fuzzy logic is based on the rules that relate controller inputs to desired outputs and those rules are developed through relevant experiences and heuristics instead of accurate mathematical model of the system. Due to their nonlinear characteristics, it becomes difficult to have a good model of the suspension systems; at this point fuzzy approach is regarded as a suitable control strategy [9] [14]. Another control method worth to be mentioned here is linear-quadratic regulator/linear-quadratic gaussian (LQR/LQG) pointed out as one the first strategies that have been applied in the design of the active vehicle suspensions [13] [15]. The controller tries to minimize a performance index consisting of different performance parameters such as tyre load, ride comfort, suspension deformation or vehicle body attitude. Weighting factors are used for trade-off between these parameters [15].

2.3 Mechanical structure

The mechanical structure studied in this thesis is designed to be placed between the cabin and the chassis of the forwarder. An overview of the placement is shown in Figure 4 [5]:

Figure 4: An overview of the mechanical frame placement

(22)

constrain the relative motion between the frames to the roll, pitch and heave directions.

Figure 5: Mechanical structure consisting of two frames, connected with three steel bars

The upper frame is connected to the cabin while the lower frame is connected to the chassis and four hydraulic cylinders are situated on the corners between these frames. Each cylinder has 0.2 m stroke and an integrated position sensor which was interfaced with the controller via a current measuring unit (see Appendix A for cylinder drawing).

Sauer Danfoss hydraulic valves were used to control the cylinders. The communication with the valve is done through a Controller Area Network (CAN) bus interface which is split into two: an external and an internal CAN bus. The external CAN bus is the connection between an external controller and the valve inlet module whereas the internal one is the connection between the external CAN bus and the connected four valves. Basically, the user sends flow commands and gets informed about the system through the external CAN bus. The info that is present on the bus for the user is pump pressure, tank pressure, pressures inside the cylinder chambers, oil temperature and error state of the valves.

(23)

(24)

(25)

3 Implementation

This part of the report has the implementation of the project. It starts with describing the test rigs and continues with the basic hydraulic equations, modelling of the cylinder, valve and mechanical structure. Following that, controller design is explained and lastly an overview of the hardware and software is given.

3.1 Test rigs

Two test rigs were used in the project: a single cylinder test rig and an active damping test rig.

3.1.1 Single cylinder test rig

For this rig, one of the cylinders was disassembled from the active damping test rig and put into another frame as can be seen in Figure 6 below:

Figure 6: Single cylinder test rig

(26)

direction of motion which would represent the actual working configuration of the cylinder. This might have helped in checking the valve response under the load and designing the control algorithms based on that; however due to the limited time and the company’s focus on the active damping test rig, this modification was cancelled and the single cylinder test rig was used as it was.

3.1.2 Active damping test rig

After the tests with the single cylinder test rig were completed, the disassembled cylinder was put back into the rig and the picture of the active damping test rig can be seen in Figure 7 showing the cylinder numbers and the coordinate system:

Figure 7: Active damping test rig,1:cylinder 1, 2:cylinder 2, 3:cylinder 3, 4:cylinder 4, 5:lower frame, 6:upper frame

The lower frame was bolted to the Stewart rig which was fixed to the ground. A seat undercarriage was mounted on the upper frame for another project.

3.2 Modelling

(27)

expressed. It also describes the way that the different components of the system are modelled.

3.2.1 Basic hydraulic equations

Basic components of a hydraulic system include a hydraulic power supply, control elements (valves, sensors,etc.) and actuating elements. A pump converts the available mechanical power to the hydraulic power. Valves control the direction of pump flow as well as the amount of fluid and pressure to the actuator. A cylinder (linear actuator) or a motor (rotary actuator) converts the hydraulic power to the mechanical power to be applied at the point of interest. A typical valve-cylinder configuration is shown in Figure 8 [16]:

Figure 8: Valve-cylinder combination

To start with the cylinder behaviour, the equations that govern the cylinder motion and the pressure dynamics inside the chambers are:

QA− QLi= Apx˙p+ VA(xp) E0_(p A) ˙ pA (1) QB+ QLi− QLe = −αApx˙p+ VB(xp) E0_(p B) ˙ pB (2)

m ¨xp = AppA− αAppB− Fext− Ff riction (3)

where VA is the volume of the piston side chamber (A) and VB is the volume

of the rod side chamber (B). QLi and QLeare the internal and external leakage

flows respectively. Apx˙p and −αApx˙p denote the flow displacement due to the

piston motion, with α being the ratio between the rod side and the piston side areas AB/AA. In order to simplify these nonlinear equations, the chamber

volumes, VA(xp) and VB(xp), and the pressure dependent bulk modulus values,

E0(pA) and E0(pB), can be replaced with constants. The internal and external

(28)

QA = Apx˙p (4)

QB= −αApx˙p (5)

meaning that the flow is proportional to the piston velocity which constituted the cylinder model in this project.

To continue with a typical valve model, flow goes though the valve orifices depends on the valve opening (Xv), discharge coefficient (Cv) and

pressure drop across the orifice (∆p) [16]:

Q = CvXvp|∆p|sign(∆p) (6)

In traditional valves, mostly the valve opening is used to control the flow going into the cylinder. However the valve used in this project works directly with a flow command instead of the valve opening. Pressure compensation is implemented into the positioning of the spool in an onboard computer. It monitors the pump pressure, tank pressure and pressures of the cylinder chambers and positions the spool to provide the demanded flow by the user. Therefore, instead of modelling the valve with this basic equation, a parameter identification process which is described later in Section 3.2.3 was used. Friction force inside the cylinder, being one of the main nonlinearities in hydraulic motion, appears in Equation 3, the piston equation of motion. Here a common way to model the friction is as a function of velocity, known as Stribeck curve (see Figure 9) [16]:

(29)

Figure 9: Velocity dependent friction force, Stribeck friction curve The curve has three parts: viscous friction (Fv) with the parameter σ,

Coulomb friction (Fc) with the parameter (Fc0) and static friction (Fs) with

the parameters Fs0 and cs. This model was applied to determine the friction

inside the cylinder.

3.2.2 Hydraulic cylinder model

As mentioned earlier, instead of controlling the spool position as in a traditional valve, the hydraulic valve employed in this thesis work controls the input flow to the cylinder with a flow command and the pressure compensation is handled in an integrated controller. Due to this fact, instead of force, velocity was decided to be controlled and therefore the cylinders were modelled using the simplified equations, Equation 4 and 5, which relate the flow going into the cylinder to the piston velocity.

The coordinate of the cylinder position has an origin at the mid-stroke and it is denoted by z, see Figure 10:

(30)

˙z(t) = Q(t) A(sign(Q(t))) (8) A(sign(Q(t))) = ( Aa , Q(t) > 0 Ab , Q(t) < 0

Here ˙z(t) denotes the piston velocity while Q(t) is the volumetric flow rate and A is the area. The cylinders are double-acting asymmetric cylinders meaning that the piston side area (Aa) and the rod side area (Ab) are different (see

Figure 11) and which one to be used in the equation depends on the sign of the flow. Positive sign means that the flow is entering into the piston side chamber, resulting in the extension of the cylinder, while negative sign means that the flow is entering into the rod side chamber, resulting in the retraction of the cylinder.

Figure 11: Asymmetric cylinder, showing piston side chamber and rod side chamber

3.2.3 Hydraulic valve model

Since the physical parameters of the valve and working principle of its internal controller were not known to the user, an identification process for the parameters was required to model the valve.

(31)

the valve behaviour:

Gv(s) =

k

τ s + 1 (9)

In the equation above, k represents the gain and τ represents the time constant of the valve. The input to the transfer function is the input flow command given to the valve and the output is the actual flow the valve delivers. Based on the data collected from the system, the obtained transfer functions are given in Equation 10 and 11: Gve(s) = 1, 1 0, 09s + 1 (10) Gvr(s) = 0, 9 0, 08s + 1 (11)

Gve is the transfer function for the extend direction and Gvr is the transfer

function for the retract direction.

Combining the valve models with the cylinder models, an overall plant transfer function from the input flow to the output velocity for each direction can be written as below:

Gpe(s) = 1, 1 0, 09s + 1 · 1 Aa (12) Gpr(s) = 0, 9 0, 08s + 1· 1 Ab (13) These models were simulated with the same input commands as in the real system and the comparison is shown in Figure 12:

Time [s] 290.5 290.6 290.7 290.8 290.9 291 291.1 Velocity [m/s] 0 0.05 0.1 0.15 0.2 0.25 0.3

Comparison between simulation and real system velocities

simulation real system input command

(a) Extend direction

Time [s] 163.1 163.2 163.3 163.4 163.5 163.6 163.7 163.8 Velocity [m/s] -0.2 -0.15 -0.1 -0.05 0

(b) Retract direction

(32)

line) is plotted which is basically calculated by dividing the input flow with the corresponding area of the cylinder. Red line is the mean response of all four cylinders and the blue line is the simulation output. One important thing has been noticed during this open loop identification process is the time delay between the input and output. It takes approximately 80ms ± 10ms for the cylinder to start moving after the command is sent. Probably multiple factors contribute to this delay like the communication time on the CAN bus or the time required to overcome the cylinder friction; however the big part of it originates from the fact that it takes a while for the spools to start moving after the input is received and additionally, it takes some time to pass the deadband limit of the spools and open up the valve.

3.2.4 Mechanical structure model

The mechanical structure has the degrees of freedom for the heave (z), roll (θ) and pitch (φ) directions and four actuators to create these motions. A coordinate transformation was needed to establish a relation between the cylinder movements and the upper platform’s relative displacement. To simplify the equations, it was assumed that the motions are decoupled and the cylinders are parallel to each other.

Figure 13: Sketch of the simplified system for the roll motion

A sketch of the simplified system for the roll motion is given in Figure 13. Based upon this sketch, the relative displacement of the upper platform with respect to the lower platform can be transformed into the cylinder positions:

(33)

z4 = z −

w

2 sin(θ) − l

2sin(φ) (17)

In these equations z denotes the heave displacement from the origin O, θ is the roll angle between the frames, φ is the pitch angle between the frames, zi denotes the corresponding cylinder position, l is the length of the rig, w is

the width and the dashed line shows the neutral position of the upper frame when all the cylinders are at the middle of their stroke. Using the first three equations, the inverse transformation was obtained:

z = z1+ z3 2 (18) θ = arcsin(z2− z3 w ) (19) φ = arcsin(z2− z1 l ) (20)

In order to get a linear set of equations, the small angle approximation was utilized which is reasonable here since the angular displacements are quite small. Maximum roll angle of 11 degrees and maximum pitch angle of 6 degrees can be reached within the physical limits of the system and even these values are highly unlikely to have during the regular operation of the forwarder. The final set of equations are:

z1 = z + w 2θ − l 2φ (21) z2 = z + w 2θ + l 2φ (22) z3 = z − w 2θ + l 2φ (23) z4 = z − w 2θ − l 2φ (24) z = z1+ z3 2 (25) θ = z2− z3 w (26) φ = z2− z1 l (27)

(34)

3.3 Control

As it is pointed out earlier, velocity control, instead of force, seemed to be possible to implement with the valve and this situation removed the options for the force based control strategies. Considering the valve’s linear response for each direction, it has been decided to try a linear controller to control the piston velocity. By using the coordinate transformation between the platform and the cylinder motions, reference velocities for the system can be mapped to reference velocities for the cylinders, and each cylinder can be controlled with a single-input single-output linear controller.

An output feedback controller scheme is given in Figure 14. Pole placement theory was used to determine the values of T and S depending on the desired system characteristics.

Figure 14: Output feedback controller scheme, r:reference, y:output, u:control signal, S/R:feedback, T/R:feedforward, B/A:plant (Gpe and Gpr)

(35)

Figure 15: Simulink model of the controller, 1:feedforward, 2:feedback, 3:delay, 4:actuator limit, 5:integral gain, 6:plant model, 7:sign function Simulink model of the controller is shown in Figure 15. Based on the sign of the reference signal, the controller parameters are chosen. If the sign is zero or positive, the extend direction controller values are used whereas if the sign is negative, the retract direction controller values are used.

In the Figure 16 below, the overall control structure is presented. Lower frame disturbances are measured, multiplied by -1 and given as a reference to the system which is supposed to counteract these disturbances. Desired relative motion between the frames is then mapped to the each cylinder’s reference velocity.

(36)

3.4 System overview

To start with the hardware overview, a general picture including the control unit, sensors and actuators can be seen in Figure 17. National Instruments Compact Reconfigurable Input Output (NI cRIO) 9076 chassis was used as the control unit. It has a 400 MHz industrial real-time (RT) processor and a reconfigurable field-programmable gate array (FPGA). Together with the cRIO, two more modules were employed, a CAN communication module and an analog input module. The CAN module (NI 9853) has two different ports, one used for communication with the valve and the other one used for communication with the IMU which was mounted on the lower frame. The current measurement module (NI 9203) was used to read the position sensors that are already built in inside the cylinders.

Figure 17: Hardware overview

(37)

actuation commands to the valves. It ran five loops simultaneously, one for sampling the current measurements and two for each CAN bus communication consisting of one sending and one receiving loop. The measured data from the valve, position sensors and IMU were saved in global variables and the RT layer could read them any time. When the actuation commands were needed to send to the valves, the RT layer updated the data and gave the commands to the FPGA to send them on the CAN bus.

(38)

(39)

4 Results

Results of the tests will be presented here. First part explains the observations from the tests performed on the single cylinder test rig while the second part has the results of the tests on the active damping test rig.

4.1 Single cylinder test rig results

The first tests of cylinder actuation were done with the single cylinder test rig in order to understand especially the valve properties and try working with one cylinder as a start. In this part, observations from these are given.

4.1.1 Valve parameter identification

At the beginning, the cylinder number 4 and the valve number 4 were used in the single cylinder test rig and an identification process similar to the one explained in Section 3.2.3 was performed to obtain the valve parameters for this single cylinder setup. The cylinder was horizontal to the ground without an external load attached (see Figure 6). The valve was given the different input flow commands and the cylinder velocity was logged. The result is shown in Figure 18: Time [s] 469.2 469.4 469.6 469.8 470 470.2 Velocity [m/s] 0 0.05 0.1 0.15 0.2 0.25 0.3

Comparison between simulation and real system velocities simulation

real system input command

Time [s] 368.4 368.6 368.8 369 369.2 369.4 369.6 369.8 370 370.2 Velocity [m/s] -0.2 -0.18 -0.16 -0.14 -0.12 -0.1 -0.08 -0.06 -0.04 -0.02 0

Figure 18: Model comparison for valve 4 - cylinder 4

In the figures, the input signal is plotted as an expected velocity based on the given input flow command. It was observed that the valve response was approximately linear for the each direction but with different gain and bandwidth parameters; therefore it was modelled with two first order transfer functions representing each direction, see Equation 28 and 29:

Gve(s) =

1, 15

(40)

Gvr(s) =

1

0, 25s + 1 (29)

These models were given the same input flow commands as in the real rig and the comparison between the simulation output (blue line) and the real cylinder response (red line) is plotted in Figure 18 above.

After this, the valve 3 was connected to the cylinder 4 in order to check that whether this valve has similar parameter values with the valve 4 or not, Figure 19 shows the results of the identification process on the valve 3:

Time [s] 334.55 334.6 334.65 334.7 334.75 334.8 334.85 334.9 334.95 335 Velocity [m/s] 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45

Time [s] 206.6 206.8 207 207.2 207.4 207.6 207.8 208 Velocity [m/s] -0.25 -0.2 -0.15 -0.1 -0.05 0

Figure 19: Model comparison for valve 3 - cylinder 4

The characteristics of it were the same, meaning that the response was almost linear in each direction, and the gain and time constant parameters changed when the flow changed direction. However the parameter values of the valve 3 were quite different than the valve 4. Comparing Figure 18 and Figure 19, it was seen that, for the extend direction, the valve 4 has a time constant of 50 ms whereas the valve 3 has a time constant of 100 ms. Similarly for the retract direction, the valve 4 has a time constant of 250 ms whereas the valve 3 has a time constant of 550 ms. Although both of the valves are the same, their responses for the same inputs are different and the valve 3 is almost twice as slow as the valve 4.

(41)

4.1.2 Low input flow commands

During these single cylinder tests, it was also observed that the flow accuracy suffered for the low flows especially lower than approximately 2,5 L/min. For a constant flow command, instead of getting a smooth motion, the cylinder oscillated with quite big amplitudes which made the control difficult for this range. The figures below belong to the tests done with the valve 4 and the cylinder 4: Time [s] 136 138 140 142 144 146 Velocity [m/s] -0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06

(a) 0,84 L/min flow command in both directions Time [s] 176 177 178 179 180 181 182 183 184 185 186 Velocity [m/s] -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08

(b) 2,04 L/min flow command in both directions

Figure 20: Low flow commands

As it was approved by the manufacturer, the spool capacity is rated to 90 L/min and it is not unusual that the flow accuracy suffers at such low flows for this type of valve.

4.1.3 Cylinder friction identification

Nonlinear friction is one of the main nonlinearities of the hydraulic cylinder model and apart from the piston velocity ( ˙xp) it actually depends on many

factors such as the pressure difference across the piston (∆p), position of the piston (xp), temperature (θ) and time (t) [16]:

Ff riction = Ff riction(xp, ˙xp, ∆p, θ, t) (30)

Due to the complexity of taking all these factors into account, the Stribeck model as introduced in Section 3.2.1 (Equation 7) is preferred to quantify the friction depending on the velocity:

Ff riction( ˙xp) = Fv( ˙xp) + Fc( ˙xp) + Fs( ˙xp) = σ ˙xp+ sign( ˙xp) Fc0+ Fs0exp −| ˙xp| cs ₍₃₁₎

which divides the friction forces into three: viscous friction (Fv( ˙xp)), Coulomb

(42)

(σ, Fc0, Fs0 and cs) must be identified from the experimental data. For this

purpose, the total actuator force equation was utilized (see Equation 3 in Section 3.2.1):

m ¨xp = AppA− αAppB− Fext− Ff riction (32)

When the cylinder was horizontal to the ground without any external load (Fext = 0) and the piston reached to a constant velocity ( ˙xp constant, m ¨xp =

0), Equation 32 became:

Ff riction = AppA− αAppB (33)

Under these conditions, measuring the piston velocity and the load force when the cylinder was extended and retracted gave the data points (plotted as blue circles) in Figure 21: Velocity [m/s] -0.15 -0.1 -0.05 0 0.05 0.1 0.15 Friction force [N] -2000 -1500 -1000 -500 0 500 1000 1500

2000 Velocity vs Friction Force

fitted curve

Figure 21: Measured data points (blue circles) and fitted Stribeck friction curve (red line) on the left, estimated friction parameters on the right In order to obtain the parameters of the Stribeck friction curve, Matlab Curve Fitting Tool was used and the red line provided the best fit to these data points with the estimated parameters given on the right. The result showed that the friction forces differ in each direction as this is generally the case in the differential cylinders. One thing that needs to be pointed out might be the difference in the static friction, the Fs0 for the extend direction is almost twice

as much as the Fs0 for the retract direction. Although friction compensation

was not a part of the control since the control strategy was not force based, this analysis gave an understanding of how the friction can be modelled in the hydraulic cylinders and also it might be used in further projects that evaluate different control strategies.

4.2 Active damping test rig results

(43)

with sinusoidal disturbances at different frequencies and only in one direction of motion at a time to eliminate any sort of complexity or disturbance that might come from the rig.

At this stage a hardware problem was encountered, one of the valves (the valve 4) did not respond as expected. Therefore the whole rig couldn’t be tested for all kinds of motion. Only the sinusoidal disturbances in the roll direction ( ˙θ) were given (see Figure 22).

(44)

Figure 23: Active damping test rig, red circles: cylinders used during tests, blue circles: cylinders blocked during tests

Figure 24: Overall motion sketch, red circles: measured variables, blue circles: simulated variables

(45)

show the measured variables from the test rig and the blue circles show the simulated variables. The IMU placed on the lower platform measured the rate of rotation around the x axis, ˙θlower (see Figure 22), then this was multiplied

by -1 to get the desired relative motion between the platforms. Using the coordinate transformation (Equation 21, 22, 23 and 24), this desired relative motion was mapped to the cylinder reference velocities. Only the cylinder 1 and the cylinder 2 were given the reference velocities. Since the cylinder 3 and the cylinder 4 couldn’t be used, the actual data from them were not available; however the data collected from the cylinder 1 and 2 were used for these cylinders as well in order to get an idea of how the platform would move if all the cylinders would be working. After that, by employing the inverse of the coordinate transformation (Equation 25, 26 and 27), the relative roll motion between the platforms were calculated. To get the upper platform disturbance, the relative motion was added to the lower platform motion as described in Figure 24.

Below, the results for different frequencies are presented. The figures have two different graphs: the ones on the right side show the controller performance for the cylinder 1 and 2, the reference and measured velocities are plotted. The ones on the left show the roll disturbance of the upper and lower platforms. The damping was measured based on how much of the peak-to-peak amplitude of the disturbance was decreased. The Stewart rig was used to introduce the disturbances to the lower frame. It works by setting the amplitude, frequency and direction of the motion.

For an amplitude of 6 degrees and a frequency of 0,5 Hz roll disturbance in the Stewart rig, the measured ˙θlower is plotted as the blue signal in Figure 25a.

The system was able to damp 63% of the disturbance. Also the controller performance with the reference and measured velocities are shown in Figure 25b.

Time [s]

66 67 68 69 70 71 72

Angular velocities [rad/s]

-0.3 -0.2 -0.1 0 0.1 0.2 0.3

Lower and upper platform, roll disturbance 0.5 Hz lower platform upper platform

(a) ˙θlower and ˙θupper, 0.5 Hz

Time [s] 68 69 70 71 72 73 Velocities [m/s] -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2

Reference vs. measured velocities for cylinder1 and cylinder2 RefVel1 RefVel2 Vel1 Vel2

(b) Reference and measured velocities

(46)

The Stewart rig was set to give a roll motion which has 4 degrees amplitude and 1 Hz frequency. According to Figure 26a below, the disturbance was decreased by approximately 48%. The controller performance started to deteriorate for the positive reference velocities.

Time [s]

42 42.5 43 43.5 44 44.5 45 45.5 46

-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4

0.5 Lower and upper platform, roll disturbance 1 Hz lower platform upper platform

(a) ˙θlower and ˙θupper, 1 Hz

Time [s] 42 42.5 43 43.5 44 44.5 45 45.5 46 Velocities [m/s] -0.2 -0.1 0 0.1 0.2 0.3

Figure 26: 1 Hz roll disturbance on lower platform

After that, the Stewart rig roll motion was set for 2 degrees amplitude and 1,5 Hz frequency and by looking at the Figure 27, it can be said that the disturbance was damped by approximately 19%.

Time [s]

57 57.5 58 58.5 59

-0.3 -0.2 -0.1 0 0.1 0.2 0.3

Lower and upper platform, roll disturbance 1.5 Hz lower platform upper platform

(a) ˙θlower and ˙θupper, 1.5 Hz

Time [s] 51.5 52 52.5 53 53.5 54 Velocities [m/s] -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2

Figure 27: 1,5 Hz roll disturbance on lower platform

(47)

Time [s]

29 29.5 30 30.5 31

-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4

0.5 Lower and upper platform, roll disturbance 2 Hz lower platform upper platform

(a) ˙θlower and ˙θupper, 2 Hz

Time [s] 29 29.5 30 30.5 31 Velocities [m/s] -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25

0.3Reference vs. measured velocities for cylinder1 and cylinder2 RefVel1 RefVel2 Vel1 Vel2

Figure 28: 2 Hz roll disturbance on lower platform

One thing to be noted here as the frequency increases, the amplitude of the Stewart rig’s roll disturbance was decreased to keep the reference velocities to the cylinders at a certain level. If the amplitude would have been kept the same, the reference velocities would be out of the range that the cylinders can reach with the current flow limitations.

(48)

(49)

5 Discussion and conclusion

This section starts with a discussion of the active damping and the single cylinder test rig results and also some observations regarding the valves are explained.

The active damping rig was tested by running two cylinders only, the cylinder 1 and 2. The velocities of these cylinders were used for the other ones with the assumption that the controller would behave in the same way for these cylinders as well and therefore it became possible to simulate the motion of the upper frame. Only the disturbances in the roll direction of motion were created in the lower frame using the Stewart rig placed under the system. The upper and lower frame roll disturbances ( ˙θlower and ˙θupper)

are plotted in the figures in previous section together with the controller performances for the cylinders. For the roll disturbance with 0,5 Hz frequency and 6 degrees amplitude, the test rig managed to dampen out 63% of the disturbance. When the frequency was set to 1 Hz and the amplitude was set to 4 degrees, 48% of dampening was achieved. At this point the controller performance had started to deteriorate for the positive reference velocities meaning that in the extend direction of the cylinders, the response had overshoot. When the frequency was increased to 1,5 Hz and the amplitude was set to 2 degrees in the Stewart rig, the response became worse and the rig dampened out approximately 19%. Lastly when the frequency was set to 2 Hz, the controller performance became poorer in both directions. The rig amplified the disturbance by 6% rather than dampening it. The peaks in the upper platform’s response come from the overshoots in the cylinders’ responses; with a better tuning of the controller parameters, for instance slowing down the controller, these overshoots can be eliminated. However due to the existing phase shift between the reference velocity and output velocity, it would not be possible to damp all of the disturbance.

The single cylinder rig tests showed that although the valves are the same, they might have different responses to the same inputs. The valve 3 and 4 were both tested with the cylinder 4 and it turned out that the valve 3 is almost two times slower than the valve 4. Additionally the response differs in the presence of the load. The flow accuracy suffers for the flow commands approximately lower than 2,5 L/min. This makes it difficult to have a reliable model of the valves for the entire working range as well as control the system around 0 m/s reference velocity.

(50)

seems highly unlikely. The reason for the difference in static friction can be caused by the internal construction of the cylinders. However this friction identification requires a deeper investigation. Considering the mentioned flow inaccuracy, more data points can be collected especially for the cylinder velocities between 0 - 0,05 m/s in order to have more reliable values for the static and Coulomb friction. For the viscous friction, it gives relatively more reliable result and this friction can be considered as the most important one for higher velocities.

As it was indicated earlier in Section 3.2.3, there is a delay of 80 ms ± 10 ms between the input flow command to cylinder movement. After the input command is sent, it takes a considerable amount of time for the spools to start moving as well as reach the dead band limit and open up the valve. Probably the communication delay on the CAN bus and the time required to overcome the cylinder friction contribute to this delay. It has also been observed that it is not constant; for instance after the dead band limit, it decreases or when the cylinder changes direction, it increases naturally. Since this type of valve was used in open loop applications before, the amount of delay was not even considered as a problem. On the other hand, from a control perspective, this delay is considered as huge and deteriorates the controller performance by introducing additional phase shift and as a result creating oscillations and overshoots in the system response. In order to see the effect of delay on the damping performance, the controllers were simulated with a 40 ms delay which is approximately the half of its current value and the roll disturbances at different frequencies were tested. For an input disturbance at 0,5 Hz, when the delay length was changed from 80 ms to 40 ms, the damping on the upper frame was increased from 70% to 78%. For a disturbance at 1 Hz, the result was improved from 40% to 56%. When the disturbance frequency was set to 1,5 Hz, an increase from 5% to 34% in the damping was measured. For 2 Hz, instead of increasing the input disturbance, the rig managed to damp 13% of it. Although the simulation output might not exactly match the test rig result, it was possible to have an idea about how much the damping would have been improved if there would be less time delay in the valve response.

When the rig stands still, in other words when the reference velocity of the cylinders is 0 m/s, the controller has problems with keeping the cylinders stable. The rig starts to shake and sometimes it makes sudden small motions up and down. The reason for this is most probably the flow inaccuracy that the valve has for the low flows. One way to work around this might be to have a hysteresis around 0 m/s velocity.

(51)

the cylinders were given the same input flow command and the responses were logged. They didn’t give the exact same response however the cylinder number 1 and 2 had troubles with following the others, especially the cylinder 1. All cylinders started moving as expected but the cylinder 1 and 2 slowed down in the middle of the motion for a while causing an unwanted oscillation in the rig. In order to understand the reason of this behaviour, the cylinders were run in combinations of two. The ones on the left were locked and the ones on the right were given the same input and vice versa. After that the cylinders at the back were locked while the ones at the front were given the same input. In these tests, the responses were better. Although the cylinder 1 still slowed down in the middle a bit, both of them didn’t have much lagging behaviour. The reason of this can be a friction somewhere in the mechanical structure causes disturbances in the heave direction of motion and slows down the cylinders on the left side of the rig or the cylinder 1 or its valve has some sort of friction inside.

When the valves are closed and the pump is turned on, there is a pressure build up inside the cylinder chambers and the cylinders move upwards with time. It was discussed with the manufacturer and informed that due to the processing and machining tolerances of the spools, there is a leakage present in the system from the pump to cylinders as well as from the cylinders to tank. To put this into numbers, each cylinder moves approximately 15 mm in 55 seconds.

(52)

(53)

6 Future work

As a future work, the first step can be the replacement of the broken valve and running all the cylinders for the same disturbance inputs as above so that the simulation results can be verified. Skogforsk has a test track that represents the forest terrain that the forwarder typically drives over at its regular operation. To evaluate the damping performance for the real operating condition, the data collected from this test track can be given as an input to the rig.

Current solution to the active damping problem gives promising results within its limitations and there are many improvements to be made, for instance the issues discussed in the previous section can be addressed. As it is mentioned that when all the valves were given the same input, some of them had trouble with following the others and the valve model was developed based on the mean response. Since all the valves don’t have the exact same response, the controller for each of them can be tuned separately for a better reference following. Additionally, when the valve was tested with and without the load, it has been observed that the response was different although it controls the flow based on the input commands and handles the pressure compensation itself. At the moment, the active damping test rig has a seat undercarriage mounted on it for another project (see Figure 7). To test the rig under a more realistic external load condition, a mock up cabin can be mounted on top of it and the valve response can be checked. The disturbances should be measured at the center of the gravity of the cabin, which is approximately 70 cm above the upper platform in the forwarder. The controllers on the cylinders are P velocity controllers with a slow integral action added to the control signal. For the main loop, the references to these controllers are provided by inversing the lower frame disturbance and mapping this into the cylinder reference velocities by the coordinate transformation. The velocity controller can be tuned and improved and also a better control allocation can be investigated that considers different aspects such as minimizing the actuator losses or the physical envelope limitations. For the main loop design, most of the articles that has been reviewed in the frame of reference section are based on the force actuation such as the skyhook controllers. If the pressure can be controlled instead of the flow and the system dynamics can be modelled in an accurate way, these models are worth implementing considering their proven effectiveness in vibration reduction. For this purpose, the working principle of the valve’s embedded pressure compensation should be investigated deeply.

(54)

with the processes that have considerable time delays is to use predictors. Future changes in the measurement signal are predicted by feeding the control signal through a model of the plant, however this method is quite sensitive to mismatches between the model and process. Another way to decrease the delay in the spool reaction is to impose a dither signal on the input command to create small vibrations around the spool’s desired position. In this way the static friction acting on the spool can be broken constantly and it reacts even to the small input changes. Another option can be the replacement of these valves with the ones that have less communication delay.

(55)

References

[1] Skogs Industrierna. Facts and figures 2013. http://www.

forestindustries.se/documentation/facts_and_figures. (Accessed on 22/09/2015).

[2] Skogforsk. http://www.skogforsk.se/english/about-skogforsk/. (Accessed on 22/09/2015).

[3] Björn Möller. Desing, development and implementation of a mechatronic log traceability system. PhD thesis, KTH Royal Institute of Technology, 2011.

[4] Grzegorz Tora. The active suspension of a cab in a heavy machine. In Grzegorz Tora, editor, Advances on Analysis and Control of Vibrations-Theory and Applications. Intech, 2012.

[5] Mechatronics Advanced Course. Active cabin suspension for a forwarder, December 1993.

[6] William Daniel Robinson. A pneumatic semi-active control methodology for vibration control of air spring based suspension systems. PhD thesis, Iowa State University, 2012.

[7] X. D. Xue, K. W. E. Cheng, Z. Zhang, J. K. Lin, and D. H. Wang. Study of art of automative active suspensions. Power Electronics Systems and Applications (PESA), 2011 4th International Conference, June 2011. [8] C. Graf, J. Maas, and H. C. Pflug. Concept for an active cabin suspension.

IEEE International Conference on Mechatronics, April 2009.

[9] Shahriar Sarami. Development and Evaluation of a Semi-active Suspension System for Full Suspension Tractors. PhD thesis, Technical University of Berlin, 2009.

[10] S. M. Savaresi, C. Poussot-Vassal, C. Spelta, O. Sename, and L. Dugard. Semi-Active Suspension Control Design for Vehicles. Elsevier Ltd., 2010. [11] W. J. Evers, I. Besselink, A. Teerhuis, A. Knaap, and H. Nijmeijer. Controlling active cabin suspensions in commercial vehicles. American Control Conference, June 2009.

[12] A. Liu-Henke, J. Lückel, and K.P. Jaker. An active suspension/tilt system for a mechatronic railway carriage. Control Engineering Practice 10, January 2002.

(56)

[14] Zheng Yinhuan. Research on fuzzy logic control of vehicle suspension system. Mechanic Automation and Control Engineering (MACE) 2010 International Conference, 2010.

[15] Lei Zuo and Samir A. Nayfeh. Structured h2 optimization of vehicle suspensions based on multi-wheel models. Vehicle System Dynamics, 40(5), 2003.

(57)

Appendices

(58)

Appendix B: IMU CAN specification

(59)

Appendix C: Pump specification

Catalogue HY30-3243/UK April 2006

Axial Piston Pump

Series PV

(60)

Axial Piston Pump Series PV

Parker Hannifin

Pump and Motor Division Chemnitz, Germany

(61)

3 Parker Hannifin

Pump and Motor Division Chemnitz, Germany Catalogue HY30-343/UK Table of contents Contents Page Introduction ...4 Characteristics ...5 Ordering Code ...6 Noise Levels ...8 Efficiency and Case Drain Flows ...9 Dimensions ...1 Mountingkits ...0

Pump combinations

Thru Drive, Shaft Load Limitations ...1

Compensators

(62)

Catalogue HY30-343/UK

Pump with Standard Pressure Compensator, code F*S Pump with Load-Sensing Compensator, code FFC

Pump with Power Compensator, code *LB Pump with Electrohydr. Displacement Control, code FPV With thru drive for single and multiple pumps

Swash plate type for open circuit

p Q P L S L A P1 PM L P1 A = included p Q P L S L A P1 PM L P1 A PF Ø0,8 = included p Q P L S L A P1 PL PM L P1 A PP = included P L S L A P1 PM s U L P1 A UQ p Q UQ = included Technical Features

• Mounting interface according to VDMA-standards sheet 4560 part 1

• Standard: 4-hole flange ISO 3019/ (metric) • Large servo piston with strong bias spring achieves

fast response; e.g. for PV04 upstroke < 75 ms

downstroke < 45 ms

Note: Follow installation instructions.

• Reduced pressure peaks due to active decompression of system at downstroke

• Also at low system pressure reliable compensator operation. Lowest compensating pressure 1-15 bar • 9 piston and precompression technology

(precom-pression volume) result in unbeaten low outlet flow pulsation.

• Rigid and FEM-optimized body design for lowest noise level

• Complete compensator program • Thru drive for 100% nominal torque

(63)

Displacement [cm3_{/rev] from 16 to 70}

Operating pressures

Outlet [bar] nominal pressure pN 350

[bar] max. pressure pmax. 40 1)

[bar] drain port )

Inlet min. [bar] 0.8 (absolute) max. [bar] 16 Minimum speed [min-1_{] 300 min}-1

Mounting interface 4-hole flange ISO 3019/ Installation drain port as high as possible

Pump with Standard Pressure Comp. Pump with Power Comp.

Combination PV/PV Combination PV/PGP

Selection table

Model Max. displacement Output flow Input power at Max speed * Moment of Weight [cm3_/rev] _{at 1500 min}-1 _{[l/min] 1500 min}-1_{and 350 bar [kW]} _[min-1_] _{inertia [kgm²]} _[kg]

* The maximum speed ratings are shown for an inlet pressure of 1 bar (absolute) and for a fluid viscosity of ν = 30 mm²/s.

Technical data

1)_{max. 0% of working cycle} )_{peak pressure only}

(64)

Code nd pump option7)

1 Single pump, no 2nd pump and coupling

PV140 or PV180 mounted 3 PV or PVM pump mounted 4 Gear pump series PGP mounted Code Variation 1 Standard 9 Reduced ment adjusted ) Code Displacement 016 16 cm³/rev 020 20 cm³/rev 023 23 cm³/rev 063 63 cm³/rev 080 80 cm³/rev 092 92 cm³/rev 140 140 cm³/rev 180 180 cm³/rev 270 270 cm³/rev PV Code Rotation 1) R Clockwise L Counter clockwise

1)_{When looked on shaft}

7)_{Specify nd pump with full model}

code

8)_{Only for PV016 - PV03} 9)_{Only for PV03 and larger} 10)_{Only for PV063 and larger} 11)_{Only for PV70} 1)_{Only for PV016 - PV09} )_{For order specify}

displace-ment.

Code Port 4) _Threads5)

1 BSPP Metric

3 UNF UNC 46) _BSPP _{Metr. M14}

7 ISO 6149 UNC 8 ISO 6149 Metric1)

Code Mounting interface Shaft D 4-hole flange Cylindric, key E 4-hole flange Splined, SAE F3) _{4-hole flange} _{Cylindric, key}

G3) _{4-hole flange} _{Splined, SAE}

K 4-hole flange Cylindric, key L 4-hole flange Splined, DIN 5480

SAE ISO 3019/1

metr. ISO 3019/2

3)_{Codes F and G only for PV140/180, see dimension sheet}

Code Thru drive option no adaptor for nd pump

T Single pump

prepared for thru drive

With adaptor for nd pump

Y8) _{SAE AA, Ø 50.8mm} A SAE A, Ø 8.55mm B SAE B, Ø 101.6mm C9) SAE C, Ø 17mm D10) SAE D, Ø 15.4mm E11) SAE E, Ø 165.1mm G1) Metric, Ø 63mm H Metric, Ø 80mm J Metric, Ø 100mm K9) _{Metric, Ø 15mm} L10) _{Metric, Ø 160mm} M11) _{Metric, Ø 00mm} Seals Axial piston pump variable displacement high pressure version Size and displacement 2nd pump Compensator

see opposite page

Variation Threads Design series:

Mounting

code Thru drive Rotation

4)_{Drain, gauge and flushing ports} 5)_{All mounting and connecting threads}

6)_{For PV063-PV180 only: pressure port 1 1/4” with 4 x M14 instead of 4 x M1}

Code Seals

N NBR

V FPM E EPDM W NBR with PTFE shaft seal P FPM with PTFE shaft seal S EPDM with PTFE shaft seal

Compensator Pump

(65)

Pump and Motor Division Chemnitz, Germany Catalogue HY30-343/UK Function L x x x x x Power compensator C x x x x x Power compensator and load-sensing Variation

A x x x x x NG6 interface top side B x x x x x No pressure compensation C x x x x x Adjustable

pressure compensation D x x x x x Proportional pilot valve PVACPP* mounted Z x x x x x Accessories mounted 14)

13) _{Not for two-valve-compensator} 14) _{Accessories not included, please}

specify on order with full model code.

15)_{Only for Codes *FR* and *FT*} PV

Compensator Pump

Compensator

Power compensator

Code Displacement Compensator option 016 063 140 180 70 Nom. power [kW] Nom. torque 03 09 at 1500 min-1 _[Nm] B x 3 19.5 C x 4 6 D x 5.5 36 E x 7.5 49 G x x 11 71 H x x 15 97 K x x x 18.5 10 M x x x 14 S x x x 30 195 T x x x x 37 40 U x x x x 45 90 W x x x x 55 355 Y x x x 75 485 Z x x x 90 585 x x 110 715 3 x 13 850

Standard Pressure Compensator Code Compensator options

0 0 1 No compensator 1 0 0 With coverplate, no control function

F D S 10 - 140 bar, spindle + lock nut F H S 40 - 210 bar, spindle + lock nut F W S 70 - 350 bar, spindle + lock nut Remote Compensator options

F R Remote pressure compensator

F S Variation R, for quick unload valve

F F Load-Sensing compensator

F T Two valve load-sensing compensator Variations for Remote Compensator

C External pressure pilot 13) 1 NG6/D03 interface top side 2 Like 1 but with ext. pilot port 15)

P Pilot valve PVAC1P* mounted D Proportional pilot valve type PVACPP* mounted L Pilot valve with DIN lock mounted Z Accessory mounted 14)

Design series:

not required for order

16) _{not for *UPV}

Note

Compensator differential ∆p is to be adjusted: Remote pressure comp., power comp. 15 ± 1 bar (Codes FR*, FT*, *L*, *C*, FPR, FPD, FPZ, FPG) With quick unload manifold 1 ± 1 bar (Codes FS*, FPS, FPT, FPP, FPE)

Load-Sensing comp. (not power comp.) 10 ± 1 bar (Codes FF*)

The ordering code PVACPP* correspond to the DSAE1007P07*

Electrohydraulic compensator Code Compensator option

Pilot pressure supply F Standard (internal), no shuttle valve U Elbow manifold, compensator horizontal 16)

Function

P Proportional displacement control Variation

V Standard, no pressure compensation R Remote pressure comp. NG6 interface Variation R, Pressure sensor and G proportional pilot valve mounted for pressure resp. power control D Variation R, Proportional pilot valve PVACPP* mounted Z Variation R, accessories mounted 14)

S Remote pressure comp., NG6 interface top side, for quick unload valve Variation S, pressure sensor and T proportional pilot valve mounted for pressure resp. power control Remote pressure comp., NG6 interface top P side, for preload and quick unload manifold Variation P, pressure sensor and E proportional pilot valve mounted for pressure resp. power control

(66)

Pump and Motor Division Chemnitz, Germany Catalogue HY30-343/UK PV016 - PV023 PV140 PV180 PV063 - PV092 PV270

Typical sound level for single pumps, measured in unechoic chamber according to DIN 45 635, part 1 and 6. Microphone distance 1m; speed: n = 1500 min-1.

All data measured with mineral oil viscosity 30 mm²/s (cSt) at 50°C.

(67)

Case drain flows PV063-092 Efficiency, power consumption

PV063

Efficiency and case drain flows PV063, PV080, PV092

The efficiency and power graphs are measured at an input speed of n = 1500 min-1_{, a temperature of 50 °C}

and a fluid viscosity of 30 mm_/s.

Case drain flow and compensator control flow leave via the drain port of the pump. To the values shown are to be added 1 to 1. l/min , if at pilot operated compensators (codes FR*, FF*, FT*, power compensator and p-Q-con-trol) the control flow of the pressure pilot valve also goes through the pump.

Please note: The values shown below are only valid for

static operation. Under dynamic conditions and at rapid compensation of the pump the volume displaced by the servo piston also leaves the case drain port. This dynamic control flow can reach up to 80 l/min! Therefore the case drain line is to lead to the reservoir at full size and without restrictions as short and direct as possible.

14 0 28 42 56 70 0 100 200 300 0 20 40 60 80 100 0 20 40 60 80 100 overall efficiency output flow input powe r at fu ll flow

input power at deadhead vol. efficiency Pressure [bar] Efficiency [%] (o ve rall, v olumetr ic) Input po wer [kW] Output flo w [l/min] 16 0 32 48 64 80 0 100 200 300 0 20 40 60 80 100 0 24 48 72 96 120

overal effl iciency

output flow, vol. efficiency

input powe

r at f ullflo

w

input powerat deadh ead Pressure [bar] Efficiency [%] (o ve rall, v olumetr ic) Input po wer [kW] Output flo w [l/min] 20 0 40 60 80 100 0 100 200 300 0 20 40 60 80 100 0 30 60 90 120 150

overa ll efficien yc

output flow input powe r at fu ll flow vol. efficiency

input power atdeadhead

Pressure [bar] Efficiency [%] (o ve rall, v olumetr ic) Input po wer [kW] Output flo w [l/min] PV080 PV092

(68)

PV063 - 092, metric version

Shown is a clockwise rotating pump. Counter clockwise rotating pumps have inlet, outlet and gauge ports reversed. For further information about flanges see catalogue No. 4039/UK „Pressure Hydraulic Flanges“ (on request).