Impact Driver Simulation

  • Upload
    snn77

  • View
    218

  • Download
    0

Embed Size (px)

Citation preview

  • 8/7/2019 Impact Driver Simulation

    1/15

    1

    Simulating an Impact Driver with ABAQUS/Explicit

    Viktor Wilhelmy

    Manta Corporation

    1000 Ford Circle

    Milford, OH 45150, USA

    www.iti-oh.com

    [email protected] [email protected]

    Abstract: A battery powered impact driver is capable of driving a 6 screw into a solid piece of

    wood, without the need of predrilling, in less than 10 seconds. The impact unit consists of a gear

    drive, spindle, spring, hammer and anvil, to which a tool is connected to drive the screw or bolt.

    The periodic torsional impacting action of the hammer is achieved by a windup and releasemechanism.

    The dynamic interaction between these parts is simulated using ABAQUS/Explicit. With the

    model, it is possible to predict the kinematics of the impact mechanism, including torque spike

    characteristics and driving speed. Key characteristics of the model have been validated by tests.

    Thus, analysis leads the design towards finding the most efficient combination of cam lead angle,hammer release clearance, inertias, and other design variables.

    High-speed camera test video clips compare well with simulation animations.

    1. Introductory Remarks

    The current presentation is part of consulting work performed at Manta Corporation. The reader is

    requested to understand that due to confidentiality considerations, the extent and level of detail ofdisclosed material must remain limited.

    2. A battery powered impact driver

    The battery operated impact driver is a new type of hand tool for the construction industry and

    light mechanical industry (Fig. 1). It has become a highly desirable tool due to its portability and

    ability for continuous usage when alternating between two sets of batteries. Thanks to the

    impacting mechanism, it can produce torque impulses ofsufficient magnitude and duration todrive a typical wood screw without the need of pre-drilling. Contrary to a conventional driver, it

  • 8/7/2019 Impact Driver Simulation

    2/15

    2

    does not require a high pushing force on the part of the operator to keep the bit engaged while

    driving the screw.

    The key elements of the impact driver lay in the impact unit (Fig. 2). This unit is driven by a

    motor through a planetary gear and consists of a spindle with a V-grove, a steel ball, a spring

    compressed between a hammer and a disk at the top of the spindle, an anvil and a stopper (Fig. 3).

    The anvil has a chuck for attaching different drive bits.

    As the planetary gear output spindle rotates at a fairly constant speed, it winds up the spring when

    the hammer is impeded from rotation by the anvil, which feels the resistance of the screw. This

    windup happens because the steel ball is trapped in the spindle V-groove and simultaneously

    presses against an inverted V-shaped cavity on the inside of the hammer. The hammer rises and

    gathers momentum, so it eventually clears the top of the anvil. At this point, the spring has

    accumulated a large amount of elastic energy and wants to unwind. The only way for this to occur

    is by causing the hammer to rotate, as it starts moving down, guided by the ball in the groove. It

    will accelerate forward to maintain position with respect to the spindle, and eventually strike the

    anvil again as it lowers towards the next impact position. Because of the high velocity and kinetic

    energy of the hammer upon impact, the anvil will undergo a finite amount of rotation before it

    stops. At this point, the process is repeated and the next impact cycle is initiated.

    In the design of an impact driver, several design variable combinations may have to be considered

    to increase operation efficiency. For example, the spring stiffness and preload, hammer/anvilclearance, inertia and mass, V-groove cam lead angle, etc., all affect performance, i.e. driving

    speed.

    The purpose of simulation is to capture the effects of these design parameters and to b e able to

    help predict tool efficiency. A simulation model (Fig. 4) is described in detail below. Selected

    stages of the operation described in this section can be observed in a series of high-speed camera

    frames in Fig. 5, together with equivalent simulation animation frames which will be discussed in

    later sections.

    3. Model description

    The anvil and hammer are modeled with solid C3D8R bricks (Figs. 4 and 5). The spindle is

    modeled with rigid elements and kept rotating at constant speed. The hammer spring is modeled

    with a series of preloaded springs whose upper end rotates with the hammer but is kept restrained

    vertically. The steel ball is not modeled explicitly for several reasons. The primary reason is that

    this would require a prohibitive level of detail throughout the model. Preliminary 2D studies have

    also suggested potential difficulties in contact stability because of the extremely small mass of this

    body in comparison to hammer and anvil.

    The kinematic constraints imposed between the hammer and spindle by the steel ball riding in the

    V-groove are instead modeled with constraint equations. This would be straight forward if the

  • 8/7/2019 Impact Driver Simulation

    3/15

    3

    bottoming out of the hammer in the V-groove could be neglected. A single constraint equation

    would have to reverse its sign in this region, which of course, is not possible. This problem was

    addressed by defining two simultaneous constraint equations, which differed only by their

    opposite signs and by governing the vertical motion of two separate parallel penalty springs (Fig.

    6). The two equations will cause the upper node of each spring to always travel in equal and

    opposite directions, in dependence of the rotation differences between spindle and hammer.Because the penalty stiffness in compression is zero, only the particular spring that tends to keep

    the hammer lifted will actually transmit a force and cause hammer motion. This model also

    permits the ball from departing the roof of the hammer V-cavity, as may physically happen (the

    ball is trapped both ways in the spindle V-groove, but only one way against the hammer cavity

    top). Today, connector capabilities are available in ABAQUS/Explicit level 6 version releases.

    These capabilities might be advantageous in this area, as they have proven to be in our current

    work with other power tool simulations.

    Woodscrew driving tests revealed that, after an initial rise period, the torque-rotation

    characteristics vary only moderately, or remain flat with depth over most of the driving range

    process. For design purposes, this study was conducted under the assumption that the threshold

    torque between successive impacts remains constant, using a typical value for the average wood

    screw driving application. A threshold torque is that torque which is required to initiate and

    maintain rotation.

    The screw is thus suited for modeling as a nonlinear spring with elastic -plastic characteristics

    (Fig.7). The elastic characteristics were estimated from screw shank dimensions and the plastic

    threshold was set to correspond to test measurement levels . In release 5.8 of ABAQUS/Explicit,

    no torsional nonlinear spring was available, so a grounded translational spring was coupled to the

    anvil rotation with the *EQUATION option. In addition, a parallel nonlinear dashpot was defined,

    in order to match measured torque spike rise and decay characteristics. This dashpot is very weak

    during rise, but stronger during the decay phase. Screw/bit backlash were similarly modeled by a

    combination of nonlinear dashpots and penalty type springs, (Fig. 8). Also, the elastic rebound

    characteristics of the anvil were controlled to match observed decay behavior using Rayleighdamping for that part.

    Friction was considered between the elastic contact surfaces of anvil and hammer. Other energydissipating mechanisms exist, for example, in the hammer spring assembly. These were modeled

    with nonlinear dashpots that have different characteristics during rise and descent, to match

    observed behavior approximately. Friction models proved more involved to apply in this area.

    The potential for connector alternatives for this, as well as for all of the fairly complex mechanism

    behaviors described in the preceding paragraphs, will be the subject of future investigations.

    4. Solution stability

    The model was subjected to approximate initial velocity conditions in order to reduce severe

    startup response effects and to shorten the time to steady state. Energy dissipation mechanisms are

  • 8/7/2019 Impact Driver Simulation

    4/15

    4

    present in most parts, helping solution stability and controlling noise. Examples of these

    mechanisms, as mentioned above, are the nonlinear dampers in the hammer spring area and the

    screw model, Rayleigh damping for the anvil, as well as friction between anvil and hammer.

    Severe stress pulses created by impact could create some noise in the solid element response of

    hammer and anvil. For this reason, an additional parameter was set to control hourglassing to

    levels beyond the default.

    Shell and solid elements in most applications undergo a limited amount of rigid body rotation. For

    computational efficiency, certain second order computations are not performed by default. In

    rotating bodies, where elements may be subjected to several revolutions, these effects become

    important. In version 5.8 of ABAQUS/Explicit, an undocumented option had to be specified for

    second order effects to be included, as follows:

    *SECTION CONTROLS, NAME=XYZ,2ndorder=yes

    If these controls are not set, accumulative model inaccuracies will be introduced with each

    element rotation. The error is inversely proportional to the number of increments per revolution, so

    it becomes less severe for small time steps. For version 6.2, the options can be found in the

    pertinenet documentation.

    5. Selected model results

    In Fig. 5, a sequence of simulation animation frames are compared side by side with the high

    speed camera images of the impact unit in operation.

    Fig. 9 shows the effect of the constraint equations and nonlinear springs of the model on the

    vertical hammer displacement. CE 1 represents the motion the first constraint equation imposes on

    node 1 of the first penalty spring (Fig.6). Similarly, CE 2 represents the exact opposite motion,

    which the second constraint equation imposes on node 2 of the second penalty spring. These are

    the two conflicting motions that each constraint equation would impose to the hammer if they

    were driving it directly. But the penalty springs are in between. The penalty springs can exert a

    force only upon positive relative displacement. For this reason, most of the time, equation CE1

    and its associated penalty spring are in control. Only if the displacement governed by this equation

    becomes negative, the spring controlled by CE2 steers the hammer motion, since the displacementgoverned by it becomes positive. Thus, the hammer bottoms out and is always kept in the positive

    displacement domain. This is also shown in more detail in Fig. 10.

    Note that the steel ball is allowed to descend away from the inverted V-surface inside the hammercavity. The behavior is indeed replicated computationally by the tension-only penalty springs,

    which tie hammer motion with the motion imposed by the constraint equations effectively only in

    one direction. The effect of the stopper, which limits hammer vertical upward travel, is also

    shown.

    Fig. 11 shows typical rotational velocities for hammer and anvil before and after an impact. It can

    be observed that the hammer gathers speed steadily, until reaching a maximum upon impact,

  • 8/7/2019 Impact Driver Simulation

    5/15

    5

    causing a rapid decrease. At this point, the motion is transferred to the anvil, which accelerates

    sharply to a peak. As is characteristic for ABAQUS/Explicit, first derivatives, such as velocities,

    tend to be noisier than displacements.

    Fig. 12 shows a comparison of the same hammer velocity pulse of Fig. 11 with a test

    measurement. Considering the high speed sequence and short duration of events like this one, a

    remarkable overall agreement can be observed, aside from some noise, which is also noticeable inthe test data.

    Another comparison with test measurements is shown for a single screw torque pulse in Fig. 13.

    The magnitude and duration of the pulse, which determine the amount of transmitted energy, and

    thus of the amount of screw rotation, agree well between analyses and test. The local shape (rise

    slope, decay slope, ring out) can also be matched closely to test by adjusting the characteristics of

    the nonlinear springs and dashpots of Fig. 8.

    Fig. 14 shows a series of analysis torque impulses, as seen by the screw. In this case, the

    simulation reaches steady state after a few initial impacts. When each of the large torque spikes

    reaches the torque threshold, screw rotation is initiated. For smaller torque spikes, the anvil-bit-

    screw torsional system responds elastically only and no permanent screw rotation occurs.

    Further insight into the behavior of the impact driver and the capabilities of the model is gained

    from Fig. 15. It shows the rotational displacement of the three essential parts: spindle, hammer and

    anvil. While the spindle moves on at constant speed (straight line), the hammer motion follows

    and oscillates as it slows down and accelerates in each impact cycle. Overall, it must travel the

    same amount as the spindle, since the steel ball in the cam system assembly ties the parts together.

    The anvil moves much slower, making progress with each impact. The progress is directly

    dependent on screw and wood characteristics and would be faster for a thin screw and softer

    wood, and slower for a bigger screw or harder wood. However, it can also depend on the

    efficiency of the particular design, as discussed below.

    6. Modeling performance

    Fig. 16 shows an example of a design study where a key design parameter was varied. The screw

    angular rotation is shown versus time for three cases. It can be seen that a small elastic rebound

    occurs after each impact, which rings out quickly. This is observed in practice.

    The model correctly predicted the relative effect of the design changes, as verified by tests. Thus,

    advanced analysis techniques were applied to help predict the product performance, in addition to

    the more conventional purposes of determining stresses or vibration.

  • 8/7/2019 Impact Driver Simulation

    6/15

    6

    7. Conclusions

    Some simplifications were made in modeling the behavior of the power tool. The torque spike and

    hammer velocity characteristics nevertheless came remarkably close to measured behavior, but it

    must be made clear that driving speed prediction precision can still be improved, especially in

    absolute terms. In relative terms, the simulation is very useful for predicting which design change

    is expected to be the more favorable regarding performance.

    The improvement of model accuracy will be the subject of future activities. Model expansions can

    include interaction with the motor and may also include the tool body, for evaluation of the effects

    of vibration on human response, or feel.

    In the product development environment, the modeling of product performance and efficiency is

    an interesting application extension of advanced FEA analysis tools such as ABAQUS/Explicit.

    The benefit is that a consistent model can be used for this purpose as it is fo r the other, more

    traditional objectives of stress and vibration evaluation that are also part of the development

    process. This stands in contrast to conventional practice of applying an array of unrelated

    individual tools and is a step forward in the Analysis Leads Design (ALD) product development

    approach.

    8. Acknowledgements

    Without Manta Corporations client product knowledge and expertise, in particular in the

    performance of advanced and accurate test work, this project would not have been possible. Their

    help and cooperation and their funding of this effort is duly acknowledged. The technology

    developed in this team environment is shared by all of its members.

  • 8/7/2019 Impact Driver Simulation

    7/15

    7

    Figure 1. CAD model of a battery powered impact driver

    Figure 2. Impact unit and key elements

    Impact unit

    Battery

    Motor

    Anvil

    Hammer

    Motor Shaft

    Planetary

    Gear

    Stopper

    Chuck

    Spring

  • 8/7/2019 Impact Driver Simulation

    8/15

    8

    Figure 3a. Hammer with invertedV-shaped cavity

    Figure 3. Impact unit assembly(stopper not shown)

    Figure 4. Simulation model ofimpact unit

    Spindle

    Hammer

    V-groove

    Anvil

    Spring

    Steel ball

    V-shaped

    cavity

    Stopper

  • 8/7/2019 Impact Driver Simulation

    9/15

    9

    (1)Cycle start.

    Hammer against anvilcauses rotating spindleto wind up the spring

    (2)Hammer rise

    (3)Hammer clears anvil

    Figure 5. High speed camera frames comparing various stages of the impactmechanism operation with simulation

  • 8/7/2019 Impact Driver Simulation

    10/15

    10

    (4)Hammer starts descent

    and accelerates,moving to next impact

    position

    (5)Hammer impacts anviland initiates anvil

    rotation

    (6)

    Anvil rotation iscompleted

    Hammer reacheslowest position and anew cycle is started

    Figure 5 (Continuation). High speed camera frames comparing various stages ofthe impact mechanism operation with simulation

  • 8/7/2019 Impact Driver Simulation

    11/15

    11

    Figure 6. Penalty springs and constraint equations to characterize spindle and

    hammer interaction

    Figure 7. Translational elastic-plasticspring representing main screw

    characteristics

    Figure 8. Additional elements tocharacterize screw behavior, such asbacklash and torque spike rise and

    decay

    Uzn

    = vertical displacement, node n

    = hammer rotation

    S

    = spindle rotation

    = cam lead angle

    Kball = steel ball stiffness

    Kball C ball

    100 * Kball

    1

    2100 * K ball

    0

    Kball C ball

    100 * Kball

    1

    2100 * K ball

    0

    tan)(1 SHzu =

    tan)(2 SHzu =

    Constraint equation 1 (CE1):

    Constraint equation 2 (CE2):Force

    Relative displacement

    Centerline

    Hammer

    Uzn

    = vertical displacement, node n

    = hammer rotation

    S

    = spindle rotation

    = cam lead angle

    Kball = steel ball stiffness

    Kball C ball

    100 * Kball

    1

    2100 * K ball

    0

    Kball C ball

    100 * Kball

    1

    2100 * K ball

    0

    tan)(1 SHzu =

    tan)(2 SHzu =

    Constraint equation 1 (CE1):

    Constraint equation 2 (CE2):Force

    Relative displacement

    Centerline

    Hammer

    Screw Bit backlash

    Springs

    Dashpots

    Screw Bit backlash

    Springs

    Dashpots

    Threshold

    *EQUATION

    Relates Spring Extension with AnvilRotation

    Threshold

    *EQUATION

    Relates Spring Extension with AnvilRotation

  • 8/7/2019 Impact Driver Simulation

    12/15

    12

    Figure 9. Effect of constraint equations and penalty springs on hammer verticaldisplacement

    Figure 10. Detail of hammer vertical displacement

    CE2

    See detail area below

    CE1

    0

    0

    0

    0

    0

    0

    0

    0

    0

    0

    0

    0 0 0 0 0

    Time, [s]

    VerticalDisplacemen

    t,[mm]

    Hammer

    Zero penalty force region

    CE2

    See detail area below

    CE1

    0

    0

    0

    0

    0

    0

    0

    0

    0

    0

    0

    0 0 0 0 0

    Time, [s]

    VerticalDisplacemen

    t,[mm]

    Hammer

    Zero penalty force region

    0

    0

    0

    0

    0

    0

    Time, [s]

    VerticalDisplacement,[mm]

    Hit stopper

    CE2

    CE1

    Bottoming out

    Hammer

    Zero penalty force region

    0

    0

    0

    0

    0

    0

    Time, [s]

    VerticalDisplacement,[mm]

    Hit stopper

    CE2

    CE1

    Bottoming out

    Hammer

    Zero penalty force region

  • 8/7/2019 Impact Driver Simulation

    13/15

    13

    Figure 11. Hammer and anvil velocity before and after impact

    Figure 12. Hammer rotational velocity comparison of analysis versus test

    -200

    -100

    0

    100

    200

    300

    400

    500

    600

    0 0 0 0 0 0

    Time, [s]

    RotationalVelocity,

    [rad/s]

    Hammer

    Anvil

    -200

    -100

    0

    100

    200

    300

    400

    500

    600

    0 0 0 0 0 0

    Time, [s]

    RotationalVelocity,

    [rad/s]

    Hammer

    Anvil

    Test

    -200

    -100

    0

    100

    200

    300

    400

    500

    600

    0 0 0 0 0 0 0

    Time, [s]

    Rotatio

    nalvelocity,

    [rad/s]

    Analysis

    Test

    -200

    -100

    0

    100

    200

    300

    400

    500

    600

    0 0 0 0 0 0 0

    Time, [s]

    Rotatio

    nalvelocity,

    [rad/s]

    Analysis

  • 8/7/2019 Impact Driver Simulation

    14/15

    14

    Figure 13. Screw torque pulse comparison of analysis versus test

    Figure 14. Screw torque signal versus time

    -1

    -1

    0

    1

    1

    2

    2

    3

    3

    4

    4

    0 0.02 0.04 0.06 0.08 0.1 0.12

    Time, [s]

    T

    orque,

    [Nm]

    -1

    0

    1

    2

    3

    4

    0 0

    Time, [s]

    Torque,

    [Nm

    ]

    Test

    Analysis II

    Analysis I

    -1

    0

    1

    2

    3

    4

    0 0

    Time, [s]

    Torque,

    [Nm

    ]

    Test

    Analysis II

    Analysis I

  • 8/7/2019 Impact Driver Simulation

    15/15

    15

    Figure 15. Rotation of spindle, hammer and anvil

    Figure 16. Screw rotation versus time for various design parameters

    Spindle

    Hammer

    Anvil

    0

    2

    4

    6

    8

    10

    12

    14

    16

    18

    20

    0 0.02 0.04 0.06 0.08 0.1 0.12

    Time, [s]

    Rotation,

    [ra

    d]

    Spindle

    Hammer

    Anvil

    0

    2

    4

    6

    8

    10

    12

    14

    16

    18

    20

    0 0.02 0.04 0.06 0.08 0.1 0.12

    Time, [s]

    Rotation,

    [ra

    d]

    0

    0.5

    1

    1.5

    2

    2.5

    3

    0 0.02 0.04 0.06 0.08 0.1 0.12

    Time, [s]

    R

    otation,

    [rad] Change I

    Change II

    Baseline

    0

    0.5

    1

    1.5

    2

    2.5

    3

    0 0.02 0.04 0.06 0.08 0.1 0.12

    Time, [s]

    R

    otation,

    [rad] Change I

    Change II

    Baseline