1

2006-01-0044

Electromagnetic Fully Flexible Valve Actuator

David Cope and Andrew Wright Engineering Matters, Inc.

Copyright © 2006 SAE International

ABSTRACT

An electromagnetic fully flexible valve actuator (FFVA) for internal combustion engines is described which offers the potential for significant improvements in fuel economy, emissions, and performance, especially at low end torque, in internal combustion engines. The FFVA offers variable lift and timing combined with controllable seating velocity. It operates on a design principle distinct from existing actuators: the electromagnetic actuator exerts appreciable bi-directional force throughout the device stroke mitigating the need for mechanical spring-derived resonance. The FFVA is a direct drive device with a unique magnetic structure that combines high bandwidth and strong forces to meet the engine performance requirements.

This paper presents the innovative electromagnetic design, simulation, and bench testing of the actuator on a single cylinder engine. INTRODUCTION

Variable valve actuator (VVA) strategies have been proposed for decades as a means for improving performance and efficiency while controlling emissions [1,2]. The benefits to engine performance of various valve actuation strategies, such as Early Intake Valve Closing (EIVC), Late Intake Valve Closing (LIVC), Late Intake Valve Opening (LIVO), and Variable Max Valve Lift (VMVL), have been thoroughly investigated and experimentally verified. These strategies range from cylinder deactivation to discrete step and continuously variable cam profiling, and, ultimately, to camless technologies such as electrohydraulic and electromagnetic actuation [3,4,5,6]. However, despite the performance benefits most current internal combustion engines do not take advantage of VVA. Those that do are extensions of standard cam technologies [1,3,7,8].

Camless technologies have the capability to fulfill the promise of FFVA. Specifically, the ability to fully define the motion of the engine valves combined with intelligent control enables the adoption of any valve actuation strategy achieving the above mentioned benefits. Electrohydraulic valve actuation has a long history as a research tool for quickly varying cam profiles to study valve lift vs. timing and is technically capable of

achieving all the requirements of FFVA. Recent efforts to apply electrohydraulic valve actuation to production engines have focused on reducing power consumption as well as redesigning prohibitively expensive components such as the high pressure pump [6]. Electromagnetic valve actuation potentially achieves the requirements of FFVA while avoiding the complexity of an additional hydraulic system. The potential barriers to the FFVA adoption are increased electrical power consumption, too great a valve seating velocity, unacceptable actuator failure modes, cost, and actuator packaging difficulties. Additionally, current cam valve train technology has evolved to an extremely well-developed and thoroughly tested system setting a high standard for replacement technologies [3].

The objective of the current research is to develop an actuator capable of satisfying the FFVA concept. A new, innovative, patented, and patent pending electromagnetic valve actuator for internal combustion engines is discussed. Fully flexible valve actuation is achieved through concurrent design of electromagnetic, electrical, mechanical, and thermal aspects. This actuator achieves fully flexible valve actuation through variable valve timing as well as variable lift. Among the numerous advantages of FFVA are increased engine efficiency over the engine speed and load range, and the elimination of the cam and throttle subsystems.

Demanding power, force, speed, and control requirements have prevented standard actuators from fulfilling FFVA principles, prompting the development of highly complex mechanical and electromechanical systems. The innovative actuator discussed herein consists of stationary permanent magnets, a stationary coil, and moving iron stem that transmits bi-directional forces to the valve. Under contract to the National Science Foundation, an experimental proof-of-principle actuator was developed and mounted on a single cylinder engine. Experimental data confirm basic operational capabilities, with both variable timing and lift. The projected system power requirements for a 16 valve system are low enough to be operable from a standard 12V automotive electrical system with alternator augmentation [9,10,11].

Magnetic finite element analysis and simulation of the actuator demonstrated the achievement of the desired performance objectives (power, force, and control) for a FFVA.

2

A traditional cam drive train, shown in Figure 1, acts on the valve stems to open and close the valves. As the crankshaft drives the camshaft through gears or a belt, the timing of the opening and closing of a valve is controlled by the cam design and is fixed relative to the piston position. This means that the engine performance is optimized only over a narrow range of engine speed and load. Most existing electromagnetic valve actuators focus on variable timing (the ability to open and close the valves at will) however, they do not allow for variable lift. The FFVA approach, shown in Figure 2, allows for fully flexible valve control (i.e. both variable timing and variable lift, low valve seating velocity, fast transition times (up to 6000 rpm), and full stroke force authority).

Electromagnetically-controlled valves can operate optimally at all engine speeds, torque levels, and temperatures thereby greatly improving the engine performance, including emissions. For example, improvement in fuel economy in excess of 15% is expected for FFVAs [12, 13].

Figure 1. Traditional Mechanical Cam Drive Train

Figure 2. First Generation Fully Flexible Valve Actuator

VALVE ACTUATOR STRATEGIES

With the additional degrees of freedom offered by FFVA (lift, timing, and seating velocity) several strategies for manipulating the intake flow exist. The most attractive strategies are those which eliminate the need for intake throttling and in the process, reduce pumping losses. Specifically, EIVC, LIVO, LIVC, and VMVL are the general intake valve actuator strategies which achieve this [3]. A qualitative comparison by Sellnau of the performance of these four strategies reveals EIVC to be the most favorable strategy for reducing pumping loss and LIVO to provide the best mixture motion at ignition. The other two strategies, LIVC and VMVL, employ only one of the allowed degrees of freedom and are shown to be generally less effective by comparison. LIVC changes only the duration of the intake opening process, while VMVL varies only the lift. The other two strategies, EIVC and LIVO may be accomplished with variable timing alone, but are preferentially combined with lower lift.

Two complementary methods for electromagnetic valve actuation exist: commanded holding and commanded acceleration. The holding style actuator relies on stored mechanical potential energy to be released into kinetic energy for transition, then storing once again the kinetic energy into mechanical potential energy to be held until the next desired transition. The acceleration style actuator electrically supplies the kinetic energy, and then reclaims the kinetic energy through regeneration. In theory, both methods require only the initial energy input for operation. In detail, however, the power requirements and achievable dynamics of each method differ based upon the specific actuator properties such as force per current over stroke.

In a practical embodiment, most actuators are a hybrid of these two methods. An example of the holding style actuator, the Pischinger design, uses solenoids to hold the valve either in the fully open or fully closed position while mechanical springs are in full compression [14]. When the solenoid releases the armature, the valve travels through the spring equilibrium point then decelerates to the end of the travel at which time another solenoid holds the armature in place. In this way, electromagnetic force is used only to hold the solenoid and spring force is used to accelerate/ decelerate the valve. In order to achieve the fast transition times required at high RPM, the spring constant is tuned to a resonance of the combined mass of the armature and valve. The selection of solenoid actuation is well suited to this method because of the one-way actuation (holding) and the limited stroke range of appreciable force. The disadvantage of this method is the dependence on resonant transition allows only variations in the timing, but not the lift or speed of transition. While holding force alone is adequate for actuation, the inability to push the armature requires complex control for soft valve seating [15].

In contrast, the commanded acceleration method, with force authority over the entire stroke, is capable of

3

controlling the valve lift directly. A bidirectional force approach is also better suited to controllable valve seating. Potential difficulties with this method are obtaining enough force for high speed valve acceleration and maintaining reasonable power requirements. The benefits of the commanded acceleration method include the capability of achieving the valve actuation strategies mentioned above utilizing variable valve lift, variable timing, and seating velocity control. FFVA DESIGN

The primary thrust of the actuator design is to use the best combination of methods discussed above, holding and accelerating, which will achieve the optimal valve strategies with reasonable power requirements and robust construction. The continuing development for fully flexible valve actuation is an iterative design process with the following stages: magnetic finite element analysis combined with parameter optimization routines, power minimization through a tailored current profile, thermal analysis, dynamic simulation, and failure mode effects and analysis.

ACTUATOR CONFIGURATIONS

A common issue for actuator design is which components move relative to an external body and which other components are stationary. Typically there are two possibilities: (1) moving magnets (MM) and stationary coil, or (2) moving coils (MC) and stationary magnets. As discussed in the following sections, we have developed another option, (3) moving plunger (MP) with stationary coil and magnets.

Typical operation of an engine might average 2000-3000 rpm for 15,000 miles/year, which equates to approximately 30-45 million actuation cycles/year. A MC configuration has potential issues with flexing of electrical leads, and a MM configuration has potential issues with magnet mechanical damage (fatigue or cracking) due to the constantly reversing acceleration profile. Therefore, a configuration without either of these components moving is indeed attractive. A MP configuration can be made extremely rugged and able to withstand the high acceleration cycles. Furthermore, the stationary magnets and stationary coil can be electrically, thermally and mechanically buffered to some extent within their environment. As will be seen, this is the approach used for the FFVA.

MAGNETIC FINITE ELEMENT ANALYSES (MFEA)

As mentioned above, the central concept of the actuator permits several quite different configurations [20,21]. Commercial magnetic finite element analysis software (Maxwell® 3-D from Ansoft Corp. [16]) was utilized to input the various configurations and analyze the performance of each configuration. In addition, using the Optimetrics™ component, we performed optimization of each configuration. The central idea of Optimetrics is an automated way to numerically

determine the sensitivity of a design configuration to a defined “goodness” parameter and then to continue to refine the design in the “direction” of increasing goodness. When the design space is large (many independent parameters), there are simply too many possible permutations of design variables to compute each one individually. For example, there are more than fifteen individual dimensional characteristics which collectively, with materials choices and boundary conditions, define a single design. If 6 values of each dimension were to be analyzed, over 11 million design combinations would result for each major configuration. This is too many to seriously evaluate and would represent analyst overload. Instead, Optimetrics essentially calculates the greatest slope toward increasing goodness and marches along that path. Localized extremes were encountered and dealt with by utilizing multiple starting designs to verify convergence to an optimized specific design combination.

Nine design configurations were examined. The eight unselected designs are shown in Figure 4 through Figure 11 in Table 1 and the selected design is shown in Figure 12. (Figure 3 provides a key for interpretation of the symbols in Table 1.) The configurations are compared based upon the metric of the ratio of valve acceleration and square root of dissipated input power, PA / . The best performing configuration, Figure 12, has the greatest value of this metric. The valve acceleration is computed by computing the electromagnetic force and dividing it by the sum of moving masses of the valve, the valve stem, and connecting hardware. [Note: other bases for comparison could, of course, be used. In that case, the optimized actuator design details would differ.] The MFEA simulation takes current density in the coil as an input and calculates the coil resistance and electrical power consumed. When configurations can be employed as either MM or MC, the reported figure is for the underlined configuration. There are, of course, many parameters upon which the value of PA / depends, and the utility of the actuator is not described solely by this metric. Therefore, although we considered heavily the value of PA / in determining which configuration to build for the feasibility demonstration, we also considered aspects such as expected actuator reliability and longevity. Based upon this analysis, the final FFVA design is markedly superior to the other designs. Fundamentally this is because the moving portion is simply a steel plunger of relatively small radius and relatively high magnetic saturation value. This design, once properly engineered for the environment, is expected to have an excellent reliability record because of its simple, robust nature. In addition, by use of the reluctance forces, the actuator can be designed to have the valve “closed” during power off. This will greatly reduce potential valve-piston interference events.

4

Figure 3. Key for Table 1. Table 1. FFVA Configurations Analyzed and Optimized with Magnetic Finite Element Software; Axis of Symmetry is shown dot-dashed on the left of each figure.

Coil

Coil

CL Figure 4. Either MM or MC configuration.

PA =39.1sec

1⋅kg

Figure 5. Preferred MC, but could be MM (magnets-plus-iron move); the iron mass is a significant penalty.

PA =45.6sec

1⋅kg

Figure 6. Either MM or MC configuration.

PA =28.5sec

1⋅kg

Coil

Coil

Coil

Figure 7. Either MM or MC configuration. Three coils consume significant electrical power but add substantial force capability.

PA =29.1sec

1⋅kg

Figure 8. MC configuration: iron moves with the coil and focuses the field.

PA =50.7sec

1⋅kg

Figure 9. MC configuration: iron moves with the coil and focuses the field.

PA =59.4sec

1⋅kg

Figure 10. MC configuration. Connecting to valve would be a challenge.

PA =40.8sec

1⋅kg

Figure 11. MC configuration: iron moves with the coil.

PA =60.0sec

1⋅kg

5

Figure 12. Selected FFVA Design for the demonstration engine. =PA / 75

sec1⋅kg

0 20 40 60 80

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Acceleration/Sqrt(Power) [1/sqrt(kg*sec)]

Figure 13. Comparison of FFVA Configurations: Valve acceleration per square root of dissipated power.

The preferred FFVA configuration is shown in Figure 12. Figure 13 shows a chart comparing the predicted values of PA . The selected design has a substantially greater figure of merit than the other designs. Figure 14 shows the FFVA selected design in RZ symmetry. Figure 15 shows the magnetic field vectors of the design to accelerate the valve in a downward direction. Figure 16 shows the magnetic flux lines for this scenario.

Figure 17 shows the electromagnetic force on the plunger (connected to the valve via the stem). It is seen that there is a force present even when there is no current (middle curve). This is due to the magnetic reluctance force of the ferromagnetic plunger in the permanent magnet structure. Essentially, the plunger tends to move upward if it is above the centerline or tends to move downward if it is below the centerline; it would remain in either of the extreme upper or lower positions. Since this curve of force over stroke with no current is approximately linear with distance, it can be nearly cancelled by appropriate choice of mechanical spring. The other two curves in the figure represent the electromagnetic forces associated with applying positive or negative current, respectively, to the actuator. Figure 18 shows the results of using a linear mechanical spring to cancel the reluctance force. Note the force is not quite constant over stroke and there is a slightly greater force available at the beginning and end of a valve transition to accelerate and decelerate the valve, respectively.

Fe FeCoil

Axis of Symmetry

Figure 14. Electromagnetic FFVA Geometry in R-Z Symmetry

6

Figure 15. FFVA Magnetic Field Vectors in R-Z Symmetry

Figure 16. FFVA Magnetic Flux lines in R-Z Symmetry

-200

-100

0

100

200

-4 -2 0 2 4

Positive CurrentZero CurrentNegative Current

Stroke, mm

Forc

e, N

Figure 17. Hysteretic FFVA Forces vs. Stroke and Current, uncompensated (raw) electromagnetic forces.

-200

-100

0

100

200

-4 -2 0 2 4

Positive CurrentZero Current, reluctance forceNegative Current

Decreasing Stroke

Increasing Stroke

Stroke, mm

Forc

e, N

Figure 18. Hysteretic FFVA Forces vs. Stroke and Current, compensated by a linear spring. OPTIMUM DYNAMIC ACCELERATION PROFILE

In order to accomplish the rapid valve transitions necessary for high speed valve actuation (closed to open, and open to close), various valve acceleration profiles were investigated. Increasing the applied current provides a greater acceleration force, but also increases the dissipated electrical power. It was desired to discover the acceleration profile which achieved the specified dynamic performance (essentially moving a fixed distance, d, in a time, t) at minimum electrical dissipated energy. High speed transition dynamics require valve motion of 8 mm in 3.3 ms. A symmetric acceleration/ deceleration profile requires a motion of d0=4 mm in t0=1.65 ms. An actuator force linearly proportional to the current was assumed. Figure 18 shows the force is not constant over the stroke but is greatest when accelerating the valve; for present

purpose we assume a constant IFk = (force per

current) is approximately achieved by a reluctance-compensated FFVA actuator. This assumption yields

acceleration, m

tIkxa )(⋅== && , so ∫ ∫

⋅=

0

00

)(t

dtm

tIkdtd .

The dissipated energy to be minimized is ∫0

0

2 )(t

dtRtI

and we desire to find the waveform ).(tI This minimization problem is subject to the constraint that the valve moves the specified distance in the specified time. It can be solved by calculus of variations and Lagrangian multipliers or by trial and error. In either event, it can be readily verified that a current waveform of

⎟⎟⎠

⎞⎜⎜⎝

⎛−=

01 1)(

ttItI with an acceleration profile of

⎟⎟⎠

⎞⎜⎜⎝

⎛−=

01 1)(

ttata , where

7

⎟⎟⎠

⎞⎜⎜⎝

⎛===

⋅= 2

0

002

0

011

223

233

tda

td

mIka , meets the

constraints while minimizing the energy. 0a is the constant value of acceleration required to move a distance d0 in time t0. In fact, )(ta provides an energy savings of 25% compared to a constant acceleration profile (for a constant ratio of force and current). Therefore, this is the desired acceleration waveform for the FFVA. Note, however, that this optimized waveform requires an increase in instantaneous force of 50% compared to a constant acceleration profile.

0

150

300

450

0 0.5 1.0 1.5 2.01.65

Optimized Linear ProfileConstant Acceleration Profile

Time, ms

Acc

eler

atio

n, g

ee

Figure 19. Ramped and constant acceleration profiles for the first half of the opening transition.

0

1

2

3

4

0 0.5 1.0 1.5 2.01.65

Optimized Linear AccelerationConstant Acceleration Profile

Time, ms

Valv

e D

ispl

acem

ent,

mm

Figure 20. Valve displacement vs. time for the two acceleration profiles for the first half of the opening transition.

0

0.1

0.2

0.3

0 0.5 1.0 1.5 2.01.65

Difference in Valve Displacement

Time, ms

Dis

plac

emen

t, m

m

Figure 21. Difference in valve displacement for the two profiles. The ramped acceleration profile opens the valve more quickly than the constant acceleration profile. THERMAL ANALYSIS

Most high performance electrical machines are ultimately thermally limited. This is because electrical machines, especially permanent magnet machines, improve in performance with increases in excitation current. Therefore, the common practice is to increase the current until either portions of the design magnetically saturate, or the local temperature increase required to transport the electrically dissipated power is unacceptably high. Since the neodymium-iron-boron magnets themselves have a maximum operating temperature of 150°C, they frequently provide the lowest upper limit to allowable energy dissipation. Figure 22 shows the geometry analyzed. A steady state temperature plot is shown in Figure 23 for an excitation of 80W continuous and boundary conditions of forced convection to 100°C.

Note the annular heat pipes between magnets appear to increase the magnet temperature locally, but they transport heat from the coil and ultimately reduce the magnet temperature by 5-10°C. Heat flux vectors are shown in Figure 24. As indicated in the figure, roughly one-quarter of the heat leaves the actuator through the top, one-quarter through the bottom, and one-half of the heat exits through the outer diameter of the device.

Fe FeCoil

Annular heat pipes

Annular heat pipes

Figure 22. Electromagnetic FFVA Geometry in R-Z Symmetry. The 4 annular heat pipes help cool the coil.

8

Figure 23. FFVA Steady State Temperature Plot showing the magnets do not exceed 135°C.

Figure 24. FFVA Heat Flux Q. The annular heat pipes between magnets transport significant thermal power. DYNAMIC SIMULATION

The block diagram for the FFVA simulation is shown in Figure 25. Briefly it can be described as follows: Initial valve position is determined and based upon an engine map and a current command is issued to move the valve to a position. The coil current is converted to force by the number of coil turns and look-up tables created from MFEA output, which account for hysteresis of valve position (i.e., reluctance forces) and desired direction of motion. The dynamic equations (F=ma, F = sum of external spring forces, etc.) are then computed to obtain valve acceleration, velocity and position. Force-per-current data derived from Figure 18 is compiled into a look-up table to provide accurate force and power characteristics over the full stroke. The simulation has been utilized to simulate different acceleration profiles and failure modes and effects. Friction work of 100 mJ per transition was taken into account.

Figure 25. Simplified FFVA Simulation Block Diagram

The most stressing dynamic scenario for the actuator is for high speed valve transitions. Therefore, simulations for 6000 rpm are provided below.

Figure 26 shows the forces required for three simulated acceleration profiles: Constant acceleration; Ramped acceleration; and Ramped acceleration with spring-back. Constant acceleration is the simplest case and allows for the minimum force to provide the required dynamics. Ramped acceleration is the linear decrease in acceleration with time for minimum dissipated power. Ramped acceleration with spring-back is the linear decrease in acceleration for the first half of the opening transition, followed by valve free-flight for the second half of the transition, followed by a stiff spring encounter which compresses and expands to reverse the valve velocity, finally, the valve is gradually decelerated to a low velocity seating. The rebound spring significantly reduces dissipated power for high speed operation and allows low lift operation at lower speeds since it does not contact the valve below approximately 8 mm stroke. Note that the constant acceleration profile swings between equal and opposite acceleration values. Ramped acceleration is the linear decrease in acceleration with time, followed by a linear increase in acceleration with time. Ramped acceleration with spring-back is the linear decrease in acceleration in combination with a stiff spring at the extent of the valve opening. This utilizes a slightly-reduced peak force.

Computed valve motion resulting from these three acceleration profiles is shown in Figure 27. It is seen that the constant acceleration profile provides the slowest opening characteristic while the Ramped acceleration with spring-back may involve valve over-

9

travel beyond the nominal 8 mm stroke. Alternatively, the spring-back can be designed to occur at 7.5mm-8mm and so not increase total valve displacement. This would actually reduce the dynamic requirements on the actuator (reduced force, current, and power).

Based upon the simulation, the total dissipated electrical and mechanical energy can be calculated for the different acceleration profiles. The results are shown in Table 2. In addition to the electrical energy dissipation, 0.200 J work per cycle (100 mJ per transition) has been included to account for friction and other energy loss mechanisms. Note that the Ramped acceleration profile reduces power consumption by 30% compared to the Constant acceleration profile, which is greater than the 25% predicted earlier for a constant value of k=F/I. This is because, as shown in Figure 18, the force (and hence force per unit current) is greater for valve acceleration and hence even more energy is saved. Basically, the actuator is more efficient at initial acceleration than at late acceleration. The Ramped acceleration with spring-back reduces the average power approximately 50% more since it essentially eliminates the energy for valve motion reversal.

-300

-200

-100

0

100

200

300

0 1 2 3 4 5 6 7 8 9 101.65

Constant accelerationRamped accelerationRamped acceleration with spring-back

ClosingOpening

Time, ms

Forc

e, N

Figure 26. Simulated Required Forces vs. Time for three possible acceleration profiles (6000 rpm, 3.3 ms transition).

-1

0

1

2

3

4

5

6

7

8

9

0 1 2 3 4 5 6 7 8 9 101.65

Constant accelerationRamped accelerationRamped acceleration with spring-back

ClosingOpening

Time, ms

Stro

ke, m

m

Figure 27. Simulated Valve Stroke vs Time resulting from the three acceleration profiles.

Table 2. Cycle energetics (6000 rpm, 3.3 ms transition time, 20 ms period) Profile Cycle

Energy Average Power (1 valve)

Average Power* (16 valves)

Constant acceleration 3.75 J 188 W 3008 W Ramped acceleration 2.65 J 133 W 2128 W Ramped acceleration with spring-back

1.36 J 68 W 1088 W

*Intake valves only. FAILURE MODES

A common complaint against camless valve actuation technologies is the position of the valve after a failure in the valve actuation system. If the failure is sudden, then mechanical inertia will continue to drive the pistons through their trajectory making valve-piston interference a possibility. In many systems the position of the valve is indeterminate; while in others the valve reverts to a position midway between fully open and fully closed. In either event valve-piston collision is highly likely, especially in today’s high-compression-ratio engines.

Two features of the FFVA design presented here dictate the position of the valve in the event of a failure. As shown in Figure 18, the actuator will preferentially reside in the closed position due to magnetic reluctance. Within limits, the reluctance force can be controlled by initial design, recognizing that there is a trade-off with acceleration performance. A second feature of the actuator results from its full stroke force authority. The coil can be wound “two-in-hand” meaning that electrically isolated coils excite the actuator. Hence, even if an electrical short or open occurs in one of the subcoils, then the other subcoil still has substantial force over the valve. Shown in Figure 28 and Figure 29, are two simulations of sub coil failures at different times. They show that upon failure of one subcoil, the remaining working subcoil can be excited to recall the valve home (closed position) within a cycle before the piston rises to top dead center (TDC), thereby averting an interference condition. The working subcoil can hold the valve in place, aided by the reluctance force. In Figure 28, the failure occurs at 1.0 ms and the valve opens 6 mm. In Figure 29, the failure occurs at 1.6 ms and the valve opens the full 8 mm. A key feature of the Ramped acceleration (and so ramped force) profile is evident here: due to the decreasing force profile, most of the energy has already been imparted to the valve for the transition by the time the failure has occurred.

Other failure modes are certainly possible and these are presently being investigated.

10

-50

0

50

100

150

200

250

0 2 4 6 8 10 12 14 16 18 20-2

0

2

4

6

8

10

ForceValve Displacement

Piston arrives at TDC

Valve parked home and held ClosedValve attracted homewith Subcoil #2

UsualClosedtime

Failure occurs in Subcoil #1

Time, ms

Forc

e, N

Stro

ke, m

m

Figure 28. Simulated failure of Subcoil #1 at t=1.0 ms. Subcoil #2 has sufficient attractive force to retrieve and park the valve Closed, thereby eliminating valve-piston interference.

-50

0

50

100

150

200

250

0 2 4 6 8 10 12 14 16 18 20-2

-1

0

1

2

3

4

5

6

7

8

9

10

ForceValve Displacement

Piston arrives at TDCFailure occurs in Subcoil #1

UsualClosedtime

Valve attracted homewith Subcoil #2

Valve parked home and held closed

Time, ms

Forc

e, N

Stro

ke, m

m

Figure 29. Simulated failure of Subcoil #1 at t=1.6 ms. Similar to the above figure, but the failure occurs very late in the transition. Subcoil #2 has sufficient attractive force to retrieve and park the valve Closed, thereby eliminating valve-piston interference. SINGLE CYLINDER ENGINE DEMONSTRATION

ACTUATOR CONSTRUCTION AND ENGINE PREPARATION

After optimizing the configuration and geometry of the actuator and running dynamic simulations, an intake valve actuator was fabricated and installed on a single-cylinder engine. (The exhaust valve cam was retained.) A drawing of the single-cylinder engine selected for this demonstration project is shown in Figure 30. Selection criteria for the engine were: relatively low power (4-hp), light weight, and ease of access to the valves. This engine was particularly easy to modify the valve since it has both overhead cams and overhead valves.

EXPERIMENTAL RESULTS

Feasibility of operating an internal combustion engine based upon the designed electromagnetic valve actuator was demonstrated. An increasingly difficult series of actuator tests consisted of operating the actuator under the following conditions: on a bench top,

in the unassembled cylinder head, statically on the engine (assembled), slow motion on the engine (hand crank), fast motion on the engine (pull-cord operated), and during combustion. SINGLE CYLINDER ENGINE OPERATION

Operation of an engine with the actuator controlling the intake valve was the major goal of this feasibility demonstration project. Figure 31 shows the valve displacement during the cycle as a function of crank angle from a throttled no-load run at 1500 rpm. Twin desirable attributes of high opening speed (~890 mm/sec) and low landing speed (~30 mm/sec) are evident in the figure. It should be remarked that the actuator operated the first time the engine was started and the engine runs reliably, although changing engine speeds requires a manual adjustment in valve commands. Figure 32 and

Figure 33 show the actuator mounted on the engine. Note details of the setup and diagnostics are evident in the figures.

Figure 30. Honda GC135QHA Engine [22]

0

1

2

3

4

0 90 180 270 360 450 540 630 720

Test #4

Opening speed~ 890 mm/sec

Landing speed~ 30 mm/sec

Crank Angle, deg

Lift,

mm

Figure 31. Experimental Valve Displacement vs Crank Shaft Angle

11

Figure 32. Phantom Views of Cam Magnet Path

Figure 33. FFVA mounted on Engine FUTURE WORK

The FFVA development effort is on-going under a National Science Foundation grant. Continuing improvements in magnetic configuration will influence the other aspects of the design cycle: thermal, electrical, and mechanical. In addition to refining the concurrent design parameters, an in depth study of valve control will be pursued. The current state of knowledge in variable valve control strategies is well developed and largely applicable to the developing actuator [17, 18, 19]. Further development in control is vital to realize the full capabilities of the actuator system. A position estimator of some kind is certainly necessary. We presently use a position sensor, but since the device is fundamentally a permanent magnet machine, sensorless control is entirely possible, eliminating a component of cost and failure modes. Detailed simulations will be performed to investigate integrated valve actuation strategies and control, as well as testing potential failure modes and

effects. The NSF grant will culminate with fabrication and installation of the FFVA system on an automotive engine. The valve and engine performance will be tested with dynamometer and gas analyzer to measure the FFVA-equipped engine performance and emissions.

Commercial feasibility of the actuator has not been demonstrated. With the decreasing cost of permanent magnets and the low cost of the remaining iron and copper components, the high-volume cost estimate of the actuator may approach the value of the eliminated and removed mechanical components. SUMMARY & CONCLUSIONS

The major objectives of this research were to design, build, and test an electromagnetic fully flexible valve actuator. These objectives were achieved. The design process involved many configurations with moving magnets, moving coils, and/or a moving plunger. The selected design performed significantly better than other designs based on dynamic performance (acceleration per square root of dissipated power) and had desirable characteristics such as stationary magnets and stationary coil.

A dynamic simulation was created to predict performance of the valve under a variety of conditions, especially various acceleration profiles. Valve-piston interference is avoidable during failure modes.

Detailed manufacturing drawings were made and an actuator was fabricated and assembled. A single-cylinder engine was chosen and modifications were made to the cylinder head to mount the actuator and various sensors. A series of run-up tests was performed culminating in the feasibility demonstration of engine operation under electromagnetic intake valve actuator control. In conclusion, the results show the actuator exhibits the inherent advantages of the fully flexible valve actuator such as variable timing, variable lift, and low valve-landing speed. ACKNOWLEDGMENTS

The authors would like to thank Dr. Jim Cowart of MIT and the U.S. Naval Academy, Dr. Christopher Corcoran of Corcoran Engineering, and Mr. David Fischer of DMF Associates for their input, support, and professional advice. The authors would also like to thank MIT graduate student Mr. Bernard Yen for his help with the experimental work. The authors acknowledge the support of the National Science Foundation through its SBIR program.

12

REFERENCES

1. Stone, R., and Kwan, E., “Variable Valve Actuation Mechanisms and the Potential for their Application,” SAE Paper 890673, 1989.

2. Klein, F. et al, “The Influence of the Valve Stroke Design in Variable Valve Timing Systems on Load Cycle, Mixture Formation and the Combustion Process in Conjunction with Throttle-free Load Governing,” SAE Paper 981030, 1998.

3. Sellnau, M., and Rask, E., “Two-Step Variable Valve Actuation for Fuel Economy, Emissions, and Performance, SAE Paper 2003-01-0029, 2003.

4. Ahmad, T., Theobald, M., “A Survey of Variable-Valve-Actuation Technology,” SAE Paper 891674, 1989.

5. Dresner, T., Barken, P., “A Review and Classification of Variable Valve Timing Mechanisms,” SAE Paper 890674, 1989.

6. Schechter, M., and Levin, M., “Camless Engine,” SAE Paper 960581, 1996.

7. Pierik, R., Burkhard, J., “Design and Development of a Mechanical Variable Valve Actuation System,” SAE Paper 2000-01-3307, 2000.

8. Kreuter, P. et al, “The Meta VVH System – A Continuously Variable Valve Timing System,” SAE Paper 980765, 1998.

9. Rivas, J., Perreault, D., and Keim T., “Performance Improvement of Alternators With Switched-Mode Rectifiers,” IEEE Trans. Energy Conv., Vol. 19, No. 3, Sept. 2004, p. 561

10. Tang, S.C., Keim, T., and Perreault, D.J., “Thermal Modeling of Lundell Alternators,” IEEE Trans. Energy Conv., Vol. 20, No. 1, March 2005, p. 25.

11. Liang, F., Miller, J. M., and Xu, X., “A Vehicle Electric Power Generation System with Improved Output Power and Efficiency,” IEEE Trans Ind. Appl., Vol. 35, No. 6, Nov./Dec. 1999, p. 1341.

12. Carney, D. "Internal Combustion Engineering," Automotive Engineering International, May 2004, p. 42;

13. "European Centers of Power," Automotive Engineering International, June 2004, p. 67.

14. Pischinger F., Kreuter, P., "Arrangement for Electromagnetically Operated Actuators", U.S. Patent #4,515,343, May 7, 1985, www.uspto.gov..

15. Wang, Y., Peterson, K., et al, “Modeling and Control of Electromechanical Valve Actuator,” SAE Paper 2002-01-1106, 2002.

16. Ansoft Corp., Pittsburgh, PA 412-261-3200, www.ansoft.com.

17. Ashhab, M., Stefanopoulou, A., “Camless Engine Control for a Robust Unthrottled Operation,” SAE Paper 981031, 1998.

18. Ashhab, M-S; and Stefanopoulou, A., “Control-Oriented Model for Camless Intake Process – Part I,” Transactions of the ASME Vol 122, March 2000.

19. Ashhab, M-S; and Stefanopoulou, A., “Control of a Camless Intake Process – Part II,” ASME Journal of Dynamic Systems, Measurement, and Control – March 2000.

20. Wright, A., Cope, D., “High Intensity Radial Field Magnetic Array and Actuator,” US Patent #6,876,284, April 5, 2005, www.uspto.gov.

21. Wright, A., Cope, D., “High Intensity Radial Field Magnetic Array and Actuator, US Patent #6,828,890, December 7, 2004, www.uspto.gov.

22. Honda Engines Owner's Manual GC135/GC160, Honda Motor Co., Ltd, 2001.

CONTACT

The authors may be contacted at: Engineering Matters, Inc. 375 Elliot St., Suite 130K Newton, MA 02464 617-965-8974

Dr. David Cope: dcope@engineeringmatters.com Andrew Wright: awright@engineeeringmatters.com. NOMENCLATURE

PA / Ratio of Acceleration to the square root of Power; used as a metric of valve performance

EIVC Early Intake Valve Closing (valve actuation strategy)

EVA Electromagnetic Valve Actuator FFVA Fully Flexible Valve Actuator LIVC Late Intake Valve Closing (valve actuation

strategy) LIVO Late Intake Valve Opening (valve actuation

strategy) MC Moving Coil MFEA Magnetic Finite Element Analysis MM Moving Magnet MP Moving Plunger R-Z Symmetry

Axisymmetric cross-section

VMVL Variable Max Valve Lift (valve actuation strategy)

VVA Variable Valve Actuator