Вы находитесь на странице: 1из 12

A METHOD FOR EVALUATING THE STIFFNESS OF A HEXAPOD MACHINE TOOL SUPPORT STRUCTURE

Erik Rebeck and Guangming Zhang Department of Mechanical Engineering and Institute for Systems Research University of Maryland College Park, MD 20742

ABSTRACT: This paper describes a new method for evaluating the contribution of a hexapod frame structure to the machines overall stiffness. A stiffness model that includes the contributions of the actuators is first presented. Use of such a model to obtain a good estimate of the stiffness contribution of the support structure to the overall machine is then discussed. Finally, the procedure is demonstrated using a finite element analysis on a portion of a prototype milling machine tool located at the National Institute for Standards and Technology (NIST)1.

INTRODUCTION A hexapod machine tool is a Stewart platform [12] based machine. It consists of six actuated serial chains, or struts, connected directly to both the base and end-effector in such a manner that the endeffector can be controlled in all six degrees of freedom. The parallel nature of a hexapod provides it with several potential advantages over serial mechanisms, such as conventional machining centers or industrial robotic arms. These advantages include high stiffness to weight ratio, small moving mass, good positioning accuracy and symmetric design. The primary disadvantages of these parallel mechanisms are their relatively small and complex workspace and the complexity in their control and operation. Because of its potential advantages, hexapods are currently being developed as high performance / high speed machining centers. The work presented in this paper is based on the analysis of a prototype Ingersoll Octahedral Hexapod milling machine located at the National Institute for Standards and Technology (NIST). The structure of this machine is pictured in Figure 1. This is a fairly large machine tool approximately 4 meters in height. The milling tool is mounted in a spindle at the bottom of the movable spindle platform, which is the hexapods end effector. Movement and orientation of the spindle platform is accomplished with the extension and retraction of the actuated struts connecting the spindle platform to the support structure (base). The struts are essentially telescoping arms, and are connected to the spindle platform and support structure through spherical joints. The support structure consists of an octahedral space frame and a worktable.

Certain commercial equipment, instruments, or materials are identified in this paper to specify the experimental procedure adequately. Such identification is not intended to imply recommendation or endorsement by the National Institute of Standards and Technology, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose.

Spindle Platform Struts Worktable Space Frame

Figure 1. NIST Ingersoll Octahedral Hexapod Structure

Machine tool stiffness refers to the stiffness between the tool point and the workpiece [1]. In this paper, applied loads, measured deflections, and stiffness values of the spindle platform are relative to the tool point position. The stiffness of a machine tool is critical because it determines the ability of the machine to maintain its desired position under the loads applied during material removal, and thus the machining accuracy. This paper focuses on analyzing the contributions of a hexapods support structure to its overall stiffness. Typically a machine tools support structure is very rigid. However, designers are always interested in reducing the amount of material used, and thus total cost, without sacrificing the performance of the machine. A method for evaluating the stiffness of a particular support structure design is therefore required to obtain a good stiffness to weight ratio. Support structures are often complex. Because of this, they are often evaluated using finite element analysis. Displacement of the joints attached to support structure relative to the base coordinate system due to platform loading can be used to determine the resulting deflection of the platform. This would typically require a forward kinematic solution [14]. In this paper a method of analyzing the support structure joint deflections through the struts is presented that only requires an inverse kinematic solution is presented.

KINEMATICS OF A HEXAPOD When using an inverse kinematic solution the platform position and orientation are given and used to determine the lengths of the struts. With a forward kinematic problem the strut lengths are given and the platform position and orientation are determined. The use of forward kinematics for the evaluation of parallel mechanisms is much more difficult to formulate and poses undetermined solutions [3, 6, 14]. Inverse kinematics provides a more convenient means for analysis. Inverse kinematic procedures are well established and have been used by researchers in the study of parallel mechanisms [3-7, 10]. Coordinate systems are attached to the fixed base and movable platform. In the study of hexapods, support structure joint locations are defined in the base coordinate system while platform joint coordinates are defined in the platform coordinate system. Platform joint coordinates are then converted into base coordinates using a transformation matrix. Strut vectors are then determined by subtracting the platform joint coordinates (in base coordinates) from the support structure joint coordinates. Base and platform coordinate systems can be chosen arbitrarily. In this study the base coordinate system was attached to the center of the worktable surface and the platform coordinate systems was attached to the position of the tool post (Figure 2).

Support Joints

w u z x y v
Spindle Platform Platform Joints

Figure 2. Inverse Kinematic Coordinate Systems

STRUT BASED STIFFNESS MODEL A schematic of a hexapod stiffness model that accounts for the contributions of the struts is shown in Figure 3. Such a model is used to determine the stiffness of the platform in the direction of an applied load P. P can be either a directional force with an equivalent directional stiffness or a moment with an equivalent angular stiffness. Several methods can be used to determine the stiffness of the hexapod machine tool in terms of the stiffness of the struts. The stiffness model formulated in this research is based on the principle of virtual work. Use of virtual work for analysis of hexapods and similar mechanisms has been used in [3, 8, 10, 13, 14]. Hexapod stiffness problems are typically formulated using the Jacobian matrix. In this research however a more classical structural analysis approach was taken [2]. This was done to provide a clear insight into the parameters involved.

ks,1

ks,2

... ks,6 kh,s P P

Figure 3. Schematic of a Strut-Based Hexapod Stiffness Model

Virtual work equates the external work done on a structure to the internal work done in the structure. External work (WE) done on a structure is created when an external load (Pi) moves due through a displacement (di) in the direction of the applied load.

W E = Pn d n = Pd
n

(1)

where P is the force applied to the spindle platform and d is the displacement of the spindle platform in the direction of the applied force P at the point of force application. Internal work (WI) done on a structure is created when stresses in the structure move through their corresponding strains. To develop this model, assume that the struts can be modeled as simple truss elements with a constant cross sectional area. Truss elements can only transmit axial loads. Bending is assumed negligible. If strains in the struts are due to externally applied loads the internal work done by a single strut is given by:

F2 F2 F F = A dl = WI , Strut = ( )dV = (A)dl = (2) k A AE AE V l l l where is the axial stress, is the axial strain, V is the volume of the strut, A is cross sectional area, l is length, F is the axial load, and E is the modulus of elasticity. Noting that the quantity (k =AE/l) is equal to the stiffness of a truss element, the internal work in the structure is given by:
WI =
i =1 6

Fi 2 ki

(3)

where the subscript i refers to the strut number. Note that because the quantity of stiffness (k) was identified, the assumption of truss elements can now be dropped. Any value or equation can be used to define the axial stiffness, but the condition that negligible moments can be transferred through the elements still applies. By equating internal and external work and specifying P as a unit force: 6 F2 Pd = d = i i =1 k i Because the stiffness of the hexapod is given by:
P 1 = d d and combining equations (4) and (5), the stiffness of the hexapod is given by: kp =

(4)

(5)

1 6 (6) Fi 2 k i =1 i Where Fi is the axial load transmitted through the ith strut due to a unit force in the direction of the applied load P and ki is the stiffness of the ith strut. It is noted that joint friction can produce moments that can result in some bending of the actuators. This is certainly true of the spherical joints used in the prototype machine at NIST. However, moments transferred by joint friction are typically small. It can also be shown that bending of a telescoping actuators due to applied moments does not significantly affect the length of the actuator, and therefore has a negligible effect on the position, and thus stiffness, of the machine [11]. kp =
This model requires that the forces transmitted through the struts due to an applied unit load at the tool point be determined. Strut forces can be determined using a free body diagram of the spindle platform and are solved by equating forces and moments in the x, y, and z directions to obtain static equilibrium (Figure 4). Forces are always in the direction of the strut unit vector, positive when the strut is in tension.

Fi

Ball Joints

ri Spindle Platform

Tool Post P

Figure 4. Static Equilibrium of the Spindle Platform

s i , x Fi = Fi s i = Fi s i , y s i ,z

(7)

In the above equation si is the unit vector of the ith strut in the base coordinate system, which is determined using inverse kinematics. Summing forces in the x, y, and z directions produces the following equations.
Px = s i , x Fi
i =1 6 6

Py = s i , y Fi
i =1 6

(8)

Pz = s i , z Fi
i =1

Moments are calculated about the position of the tool post where the force or moment P is applied. The vectors from the position of the tool post to the ball joint centers (ri) in the base coordinate system are first calculated using a rotation matrix. Moments about the tool point position are then given by:
M x = (F r ) = ( si , y ri , z s i , z ri , y ) Fi = J i , x Fi i
i =1 6 i =1 6 i =1 6 6 6 6

M y = (F r ) = ( s i , x ri , z s i , z ri , x ) Fi = J i , y Fi j
i =1 6 i =1 6 i =1

(9)

M z = (F r ) k = ( s i , y ri , x s i , x ri , y ) Fi = J i , z Fi
i =1 i =1 i =1

Equations (8) and (9) provide the six equations that are needed to solve for the six strut forces.

s1, x Px s P 1, y y s1, z Pz = J 1, x M x J 1, y M y M z J 1, z

s2, x s2, y s 2,z J 2, x J 2, y J 2,z

s 3, x s 3, y s 3, z J 3, x J 3, y J 3, z

s 4, x s 4, y s4, z J 4, x J 4, y J 4, z

s 5, x s 5, y s5,z J 5, x J 5, y J 5, z

s 6, x F1 s 6, y F2 s 6, z F3 J 6, x F4 J 6, y F5 J 6, z F6

(10)

{P} = [A]{F}
{F} = [A]1 {P}
With the strut forces known the strut stiffness model is complete. Recall that under the conditions used to develop the stiffness model using virtual work, P must be either a unit force or moment. Stiffness of a hexapod varies both with position and orientation of the platform. This is due to the reconfiguration of the struts and the fact that stiffness of the six struts typically depends on their length, and thus the position and orientation of the spindle platform. The forces acting on the individual struts depend both on the position and orientation of the spindle platform and on the direction of loading (P). As a result, for a given position and orientation of the platform, stiffness will vary with direction. A thorough stiffness study of the support structure therefore requires analysis be performed at several locations within the workspace and in multiple directions at these locations. Stiffness mapping across sections of the workspace is typically used to visualize stiffness variations [4, 5, 6, 8, 10].

SUPPORT STRUCTURE ANALYSIS The deflected position of the tool post due to the support structure requires that the deflections of the six ball joints attached to the support structure, which are 3rd order vectors, be known. These deflections are measured in the base coordinate system relative to the measurement location. Determining the resulting deflection at the tool point due to these joint deflections would require a forward-kinematic solution. The actuator lengths and deflected base joint locations are known, but the position of the platform is not. However, only the deflections in the direction of the strut axis directions are important for this mechanism [11], reducing the information required for a good approximate solution to six scalar quantities. With this simplification an inverse kinematics solution can be used to obtain a good approximation of the hexapods stiffness. The claim that only support joint deflections in the direction of the struts are of importance can be demonstrated in a fairly simple manner. For a static problem, only changes in length of the six struts affect the position of the platform. This is illustrated in Figure 5. If the strut has a constant length the base joint can be moved along a spherical surface centered at the platform joint center without affecting the position of the small ball joint, and thus tool point. When small deflections are assumed, deflections of the large ball joint in a direction perpendicular to the strut axis (r) nearly follow this path, producing a negligible effect on the length of the strut.

Support Joint

Platform Joint

Figure 5. Effect of Support Joint Deflections Perpendicular to the Strut Axis Changes in the orientation of the struts due to deflections of the support joints are negligibly small considering the relative magnitudes of the strut length (meters) and joint deflections (-meters). The deflection of the support joints in the direction of the strut axis (a) can therefore be estimated by taking the scalar product of the support joint deflection (sj=[sj-x sj-y sj-z]T) and the undeflected strut unit vector: (11) = S
a ,i sj ,i i

In the above equation the subscript i refers to the ith strut/support joint. When only the deflections of the large ball joint centers (due to loads applied to the space frame) in the direction of the strut axis are considered, these deflections can be described by equivalent strut stiffness values. This is accomplished by dividing the force applied through the strut by the strut axial deflection of its ball joint center: Fi (12) a ,i These equivalent strut stiffness values can then be inserted into the strut based hexapod stiffness model to determine the hexapod stiffness contributed by the space frame. k p ,i =

ks,1

ks,2

... ks,6

kh,s

a) Struts Only

kp,1

kp,2

...

kp,6

kh,p

b) Space Frame Only

kp,1

kp,2

...

kp,6

ks,1

ks,2

... ks,6 kh,sp

c) Struts and Space Frame Figure 6. Incorporation of Strut and Space Frame Stiffness The stiffness of the hexapod due to the struts and support structure can be combined using the principle of superposition. This is illustrated in Figure 6. Since the deflections of the space frame and struts are additive they can be treated as springs in series. The hexapod stiffness due to the struts (kh,s) is combined with the hexapod stiffness due to the support structure (kh,p) to produce the overall hexapod stiffness using the following relationship:
1 1 kh = + k h,s k h, p
1

(13)

ANALYSIS OF THE PROTOTYPE NIST INGERSOLL SUPPORT STRUCTURE A finite element model was created to analyze the contributions of part of the prototype NIST Ingersoll hexapods space frame to its overall stiffness. The model used in this analysis is pictured in Figure 7. This model focuses on the upper portion of the space frame. The worktable and tubes connecting the feet of the hexapod have been removed and replaced with boundary conditions that prohibit the feet from translation. This simplification is equivalent to stating that the removed parts are infinitely rigid. While this is certainly not true, these components are believed to be sufficiently rigid to neglect, allowing the complexity of the model to be greatly reduced. Another modification made to the actual geometry is that cylindrical surfaces have been extended where the center of the ball joints would be. This provides a convenient surface for force application.

Upper Space Frame

Joint Center Surfaces

Strut Forces

Boundary Conditions

Legs Feet

Figure 7. Space Frame Finite Element Model Finite element trials were performed for loading in the x, y, and z directions at 14 locations for a total of 42 trials. At all locations the spindle platform was in a horizontal orientation. The locations chosen were based on previously performed experiments of the hexapods stiffness [4, 5]. Measurement positions were performed within a 60 wedge on the horizontal plane to take advantage of z-axis symmetry (Figure 8). Half the measurements were taken at a height of z = 1.268 m and half were taken at a height of z = 0.353 m. The procedure used for each trial was as follows: An inverse kinematic solution was used to calculate the un-deflected strut vectors (Si) and platform joint vectors (ri) in base coordinates for the particular location. z
1 7 5 3 14 12 13 11 9 10 8 6 4 2

Z= 1.268 m

Z= 0.353 m

60

y
X= -0.150 m X= -0.300 m

-x Figure 8. Hexapod Stiffness Measurement Positions

1) Strut axial loads for the particular position and unit force loading in a particular direction were then calculated. 2) Strut axial loads were applied to the nodes of the model around the location of the ball joint centers. 3) The analysis was run and deflections of the ball joint center nodes (w2bw-x w-y w-z]T) were recorded. 4) Measured deflections were then analyzed using the procedures outlined in this paper.

Figure 9. Magnified Deflection of Space Frame Under a Load Applied in the x Direction Results of this finite element analysis are shown in Table 1. In this table the hexapod stiffness due to deflections of the space frame (kh,p) are shown. The data is presented in two groups. In the first group, directional stiffness values in the x, y, and z directions for positions at a height of z = 1.268 m are shown. In the second group, directions stiffness values in these directions for positions at a height of z = 0.353 m. Two observations can be made from examining this data: 1) The directional stiffness along the z-axis is greater in the lower portions of the workspace. It can be seen that the directional stiffness ranged from 2405 to 2502 N/m at a height of z = 0.353 m and ranged from 2268 to 2353 N/m at a height of z = 1.268 m. 2) When comparing data at a specific platform position (data within a row) the space frame is nearly three times stiffer in the z-direction and the x and y directions. This is observed at both tested heights (z = 1.268 and 0.353 m). The reason stiffness is greater in the z-direction is because loads applied in horizontal direction (x or y for instance) subject the legs of the structure to cantilever bending. This is illustrated in Figure 9. When forces are applied to the spindle platform in the z-direction, cantilever bending of the legs is greatly reduced

CONCLUSIONS A new method for estimating the contributions of a hexapods support structure to its overall stiffness is presented. The advantage of the new approach is that it eliminates the need for a forward kinematic solution, simplifying support structure stiffness analysis. This was accomplished by recognizing that only the support joint deflections in the direction of the strut vectors are significant. Under conditions where support joint deflections are relatively small, particularly those

perpendicular to their respective strut axes, this method provides an effective means of accurately evaluating the stiffness of a hexapod support structure. Table 1 Hexapod Stiffness Attributed to Space Frame Position 1 2 3 4 5 6 7 8 9 10 11 12 13 14 x 0.000 -0.150 -0.300 -0.130 -0.260 -0.075 -0.150 0.000 -0.150 -0.300 -0.130 -0.260 -0.075 -0.150 Location (m) y 0.000 0.000 0.000 0.075 0.150 0.130 0.260 0.000 0.000 0.000 0.075 0.150 0.130 0.260 Stiffness (N/m) Y 719 661 657 663 647 666 658 695 682 668 686 669 690 682

z 1.268 1.268 1.268 1.268 1.268 1.268 1.268 0.353 0.353 0.353 0.353 0.353 0.353 0.353

x 678 698 704 699 690 690 688 698 705 715 703 710 718 700

z 2352 2324 2277 2324 2268 2337 2282 2502 2469 2422 2468 2405 2468 2409

ACKNOWLEDGMENTS The authors would like to thank the National Institute for Standards and Technology and the National Science Foundation for financial support of this work. Additionally, the Algor Finite Element Analysis software used in this research was provided by a Society of Manufacturing Engineers Education Foundation grant. We would also like to thank Johannes Soons, Albert Wavering, and Fred Rudder from NIST for their help and comments. Joe Falco (also from NIST) created the Pro/ENGINEER model pictured in Figure 1 and used to develop the space frame FEA model. REFERENCES 1. 2. 3. ANSI/ASME B5.54-1992, Methods for Performance Evaluation of Computer Controlled Machining Centers, ASME, New York, 1993. Armenakas, A.E., Modern Structural Analysis: the Matrix Method Approach, Mcgraw-Hill, Inc., New York, 1991. Bryfogle, M.D., Nyguyen, C.C., Antrazi, S.S., Chiou, P.C., Kinematics and Control of a Fully Parallel Force Reflecting Hand Controller for Manipulator Teleoperation, Journal of Robotic Systems, Vol. 10, 1993. Clinton, C., Zhang, G., Wavering, A., Stiffness Modeling of a Stewart Platform based Milling Machine, Transactions of NAMRC XXV, 1997. Clinton, C., Characteristics of a Hexapod Machine Tool, M.S. Thesis, University of Maryland, College Park, 1998. El-Khasawneh, B.S., Ferreira, P., On Using Parallel Link Manipulators as Machine Tools, Transactions of NAMRI/SME, Vol. 25, 1997. Fichter, E.F., A Stewart Platform Based Manipulator: General Theory and Practical Construction, International journal of Robot Research, Vol. 5, 1986. Gosselin, C., Stiffness Mapping for Parallel Manipulators, IEEE Trans. On Robotics and Automation, Vol. 6, 1990.

4. 5. 6. 7. 8.

9. 10.

11. 12. 13. 14.

Luh, C., Adkins, F., Haung, E., and Qiu, C., Working Capability Analysis of Stewart Platforms, ASME Journal of Mechanical Design, Vol. 118, 1996. Rudder, F.F., Static Structural Analysis of a Reconfigurable Rigid Platform Supported by Elastic Legs, National Institute of Standards and Technology Internal Report, NISTIR 5885, 1996. Soons, J.A., On the Geometric and Thermal Errors of a Hexapod Machine Tool, First European-American Forum on Parallel Kinematic Machines, 1998 (in press) Stewart, D., A Platform with Six Degrees of Freedom, Proc. Inst. Mechanical Engineering, 371-386, 1965. Tahmasebi, F., Tsai, L.W., On the Stiffness of a Novel Six-Degree-of-Freedom Parallel Minimanipulator, Journal of Robotic Systems, Vol. 12, 1995 Tsai, L.W., Robotic Analysis: The Mechanics of Serial and Parallel Manipulators, JohnWiley & Sons, New York, 1999

Вам также может понравиться