1

Programmable Valves Enable Both Precision Motion Control and Energy Saving

Dr. Bin Yao

Purdue UniversityWest Lafayette, IN 47907, USA

2

Outline

Development of Programmable Valves

Control of Programmable Valves

On-board Modeling of Valve Flow Mapping

Programmable Valves Bypass Sandwiched Deadband Problem

Summary

3

Outline

Development of Programmable Valves

Control of Programmable Valves

On-board Modeling of Valve Flow Mapping

Programmable Valves Bypass Sandwiched Deadband Problem

Summary

4

E/H precision motion control system

Objective:Performance Energy efficiencyCost

To reduce energy usage:Reduce the supply pressure Ps(t)Reduce the pump flow rate Qs(t)

Energy usage:

5

To reduce supply pressure …Must break the mechanical linkage between the meter-in and meter –out orifices

Arne Jansson and Jan-Ove Palmberg, "Separate controls of meter-in and meter-out orifices in mobile hydraulic systems", SAE Transactions, Vol.99, Sect 2, pp377-383, 1990

Meter-in and meter-out orifice areas are coupled in 4-way directional control valves

Cannot control all cylinder states (pressures of both chambers)

Deadband for PDC valves

Leakage for servo valves

desired

motion

6

Dual-Valve Meter-in and Meter-out

J.A. Aardema, Caterpillar Inc., "Hydraulic circuit having dual electrohydraulic control valves", United States Patent 5,568,759, 1996

Two valves:Patented by:

J. Ardema, 1996

Uses two directional control valves to meter flows

7

Four-Valve Meter-in and Meter-out

Aardema and Koehler, Caterpillar Inc., "System and method for controlling an independent metering valve", United States Patent 5,947,140, 1999

Four valves

J. Ardema and D. Koehler

Uses four poppet valves to independently control meter-in and meter-out flows

8

Five-Valve Meter-in and Meter-out

Ruth Book and Carrol E. Goering , "Programmable electrohydraulic valve", SAE Vol. 1, No. 2, pp28-34, 1999

Same functionality as IMVs only

9

Use Regeneration Flow

Garnjost, K. D., "Energy-conserving regenerative-flow valves for hydraulic servomoters", United States Patent 4,840,111, 1989

Regeneration Valve

Patented by:K. Garnjost, 1989.

Uses one additional valve to provide regenerative flow for energy saving but cannot control both chambers independently

10

Energy Saving Programmable Valves

Purdue Energy SavingProgrammable Valves

Developed by:Bin Yao, 2000

Take advantages of four valve configuration to control meter-in/meter-out flows independently for precise cylinder positioning

Use an additional valve to precisely control cross-port flow (or regenerative flow) for energy saving

Overcome the sandwiched deadband control problem of conventional PDC valves through the use of cheap but fast acting cartridge valves

11

Programmable Valves Advantages

Fully decoupled meter-in and meter-out flowFully controlled true cross port regeneration flowMore flexibility and controllabilityCompletely solve/bypass the sandwiched deadband problem of EH systems controlled by closed-center valvesVirtually eliminate leakageFaster response than PDC valvesLow costLow maintenance cost

12

OutlineDevelopment of Programmable ValvesControl of Programmable Valves

ChallengesTwo-level control systemComparative experimental results

On-board Modeling of Valve Flow MappingProgrammable Valves Bypass Sandwiched Deadband ProblemSummary

13

Challenges

Multi-input dual-objective systemLack for accurate mathematical model of cartridge valvesCoordination of the five cartridge valvesHighly nonlinear hydraulic dynamicsLarge parameter variationsUncertain nonlinearities such as external disturbances, flow leakage and seal frictions

14

Nonlinear Valve Flow Mapping

),(1vivivivi QPfu ∆= −

15

Control of Programmable Valves

Schematic of the two-level Control System

16

Two-level Control System

Task level (Working Mode Selection)Proper hydraulic circuitryOptimal valve configuration for maximal energy saving

Valve level (ARC Control)Nonlinear model based design to take the system nonlinearities into account explicitlyAdaptive model compensation to reduce the effects of parametric uncertainties and disturbancesRobust feedback to guarantee stability

17

Working Mode Selection ---Tracking Task

18

Working Mode Selection ---Regulation Task

19

Mode T1

Desired Cylinder Velocity > 0Desired Cylinder Force >0

Uses Valve #2 and Valve #5Maintain P2 as Low as Possible

Cylinder

Valve #2

Load

Valve #5

Valve #4

Valve #3

Valve #1

XL

Q 2

Q 1

P1

P2

PS PT

Q V2

Q V5

Extend Resistive-FLoad

20

Mode T3

Desired Cylinder Velocity < 0Desired Cylinder Force >0P1 >P2

Uses Valve #3 and Valve #5Maintain P2 as Low as Possible

Cylinder

Valve #2

Load

Valve #5

Valve #4

Valve #3

Valve #1

XL

Q 2

Q 1

P1

P2

PS PT

Q V1

Q V2

Q V5

Q V4

Q V3

21

Valve-Level Control --- Adaptive Robust Controller

Nonlinear model based controller designCompensate known nonlinearities and disturbancesGuarantee stability and prescribed transient performance in the presence of unmodeleduncertainties and external disturbancesAchieve asymptotic stability in the absence of disturbance

22

Adaptive Robust Control Structure

Plant

ControlledParameterEstimation

ModelCompensation

RobustFeedback

+ u x

xd

...

.

23

Valve-Level Control

Off-side pressure regulatorMaintain the off-side chamber at a constant low pressure

Working-side motion controllerControl the working-side pressure so that the desired trajectory is followed as close as possible

Flow distributionconvert flow commands to valve voltage input signals

24

Flow distribution

mdmdvmdv

mdv

v

v

v

QQQQQQQ

QQQ

12325

13

4

2

1

000

−=+=−=

===

25

Flow distributionInverse flow mapping to convert flow commands to valve voltages input signals

26

Experimental Setup

27

Smooth Working Mode Switching

28

Comparative Experiment•PDC valveVickers KBFDG4V-5-2C50N-Z-PE7-H7-10

•Servo valveParker BD760AAAN10

•Programmable valves Vickers EPV10-A-8H-12D-U-10

29

Comparison of Performance

1e

2e e

∞

0.00360.00100.0007Prog

0.00510.00180.0014Srv

0.01460.00460.0030PDC

30

Comparison of Cylinder Force

31

Comparison of Cylinder Pressures

32

Comparison of Energy Usage, Constant Supply Pressure Ps=1000psi

32.4 KJ

32.7KJ

21.3KJ34% less than PDC35% less than Servo

33

Comparison of Energy UsagePs=Working Pressure + 500KPa

20.9 KJ

19.3KJ

6.4KJ69% less than PDC67% less than Servo

34

OutlineDevelopment of Programmable ValvesControl of Programmable ValvesOn-board Modeling of Valve Flow Mapping

Problem formulationLocalized basis functions and smooth blendingOn-board modeling Comparative experimental results

Programmable Valves Bypass Sandwiched Deadband ProblemSummary

35

Cartridge Valve Flow Mapping

Individually calibrated flow mapping (2002-2003)AccurateOff-board test, time consuming, additional calibration equipments, not suitable for industrial application

Manufacture supplied flow mapping (2004)Ready to useInaccurate, large modeling error exists, system performance compromised

Automated on-board modeling of cartridge valve flow mapping (2004-2005)

Accuracy & easy implementation

36

Modeling of Cartridge Valve Flow Mapping

Unknown nonlinear function

37

Flow Rate Observer based on Cylinder Pressure Dynamics

On-board Modeling of Cartridge Valve Flow Mapping --- Observer Approach

),( vivivi PufQ ∆=

Known Control Signal

Measurable

38

On-board Modeling of Cartridge Valve Flow Mapping

Huge number of basis functions and weighting factorsLimited experimental dataDynamic flow rate unavailable

39

Cylinder Dynamics Based Estimation

Cylinder dynamics:

40

Why localized estimation?

Invertibility

41

Localized Basis

42

Localized Estimation

PE condition is easily satisfied in small local region rather than globallyEstimation error can be controlled by checking the condition number instead of invertibilityDiscontinuity may happen at the block borders

Extrapolation

43

Smooth Blending

44

Automated On-board Modelling of Valve Flow Mapping

45

On-line EstimationOn-line, Swiped sinusoidal trajectory, 0.1Hz~0.5Hz over 80sec.

0 1 1.51.7522.252.52.7533.253.53.754 5 6 7 8 9 100

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

6

6.5

7

7.5

Pre

ssur

e D

rop

−−

− M

Pa

Control Input −−− Volt

Blocks and Locations of Simulation Data

Estimated

Calibrated

46

Off-line Estimation

02

46

810

0

2000

4000

6000

80000

5

10

15

20

25

30

35

Valve Input − Volt

Estimated Flow Mapping

Pressure Drop − KPa

Vol

umat

ric F

low

Rat

e −

LP

M

Adjust

47

Comparison of Different Flow Mappings

0 2 4 6 8 10 12

0

0.2

0.4

0.6

0.8

1

Des

ired

Mot

ion

− r

adComparative Tracking Results with Different Flow Mappings

Disired Motion Trajectory

0 2 4 6 8 10 12−0.02

−0.01

0

0.01

0.02

Tra

ckin

g E

rror

− r

ad

Time − sec

Calibrated Flow Mapping Estimated Flow Mapping Manufacture Flow Mapping

48

Outline

Development of Programmable Valves

Control of Programmable Valves

On-board Modeling of Valve Flow Mapping

Programmable Valves Bypass Sandwiched Deadband Problem

Summary

49

EH system controlled by a closed-center valve

50

What is sandwiched deadband

PlantDynamics

Valve Dynamics

Control Input

Spool Displacement

Actual Valve Orifice Opening

System Output

Actuator dynamics can

not be neglected

51

SD is hard to deal with ---Feedforward Compensation

Use feed-forward controller to increase the actuator dynamics so that it is sufficiently high to be neglectedUsed the inverse deadband function to compensate the nonlinear deadbandDepends on the accuracy of the valve dynamics, can only achieve limited improvements in practice due to the unavoidable uncertainties in the valve model

PlantDynamics

ControlLaw

D(s)N(s)F(s)

N(s)D(s)

52

SD is hard to deal with ---Feedback Compensation

Use local high gain feedback control to attenuate the deadbandRequire actuator output/state feedback. Significantly increases system cost.Sometimes actuator output/state signals are too noisy to help increasing the bandwidth significantly.

PlantDynamics

ControlLaw

N(s)D(s)

ActuatorControl

53

SD simple solution ---Direct Compensation

Simply neglect valve (spool) dynamicsEconomical, easy to implementLimit cycle may happen if closed loop bandwidth is not low enough Closed loop system is usually conservative, i.e., bandwidth is limited quite low in order to safely neglect valve dynamics

PlantDynamics

Valve Dynamics

ControlLaw

54

Deadband of poppet-type cartridge valves

The input signal has to be large enough to overcome the spring force and static friction.

55

Deadband Compensation of Poppet Valve

Input deadband is easier to compensateCartridge valve dynamics is much more faster than PDC valves.

56

Outline

Development of Programmable Valves

Control of Programmable Valves

On-board Modeling of Valve Flow Mapping

Programmable Valves Bypass Sandwiched Deadband Problem

Summary

57

Conclusions

Eliminate the mechanical linkage between the meter-in and meter-out orifices and enable accurately using the cross port regeneration flow for a controlled motion.Energy saving can be achieved without sacrificing precision motion performance.The two-level control system successfully control the system to achieve the dual objectives.Sandwiched deadband in EH systems can be bypassed by the programmable valves.Promising alternatives of conventional four-way valves in precision motion control.

58

Acknowledgement

Formal Students:

Dr. Song LiuChris Deboer

Past Caterpillar Collaborators:

John LitherlandDoug KoehlerJ. Ardema

59

AcknowledgementsAcknowledgementsSponsors:

National Science FoundationCAREER Grant CMS-9734345Regular Grant CMS-0220179

Purdue Electro-Hydraulic Research Center

Valve Donations by:

Department of Mechanical Engineering, Purdue UniversityDepartment of Mechanical Engineering, Purdue UniversityDepartment of Mechanical Engineering, Purdue University

60

ReferencesBin Yao and C. Deboer, “Energy-saving adaptive robust motion control of single-rod hydraulic cylinders with programmable valves”, the American Control Conference, pp4819-4824, Alaska, May, 2002

S. Liu and Bin Yao, “Energy-saving control of single-rod hydraulic cylinders with programmable valves and improved working mode selection”, the SAE Transactions – Journal of Commercial Vehicle, SAE 2002-01-1343, pp51-61, 2002

S. Liu and Bin Yao, "Programmable Valves: a Solution to Bypass DeadbandProblem of Electro-Hydraulic System", the American Control Conference, pp4438-4443, Boston, 2004.

Liu and Bin Yao, "Characterization and Attenuation of of Sandwiched Deadband Problem of Electro-Hydraulic Systems Controlled by Closed-Center Valves using Describing Function Analysis", the ASME International Mechanical Engineers Congress and Exposition (IMECE), IMECE2004-60946, pp1-8, 2004

S. Liu and Bin Yao, "Adaptive Robust Control of Programmable Valves with Manufacture Supplied Flow Mapping Only", the IEEE Conference on Decision and Control, pp1117-1122, 2004.

61

ReferencesS. Liu and Bin Yao, "Automated Modelling of Cartridge Valve Flow Mapping", the IEEE/ASME Conference on Advanced Intelligent Mechatronics (AIM'05), pp789-794, Monterey, California, 2005.

(Winner of the Best Student Paper Competition)