IMECE DESIGN OF A HIGH-SPEED ON-OFF VALVE

Transcription

1 Proceedings o the ASME 9 International Mechanical Engineering Congress & Exposition IMECE9 ovember -9, Lake Buena Vista, Florida, USA IMECE9-89 DESIG OF A HIGH-SPEED O-OFF VALVE Allan A. Katz Mechanical Energy and Power Systems Lab Worcester Polytechnic Institute Worcester, MA, USA James D. Van de Ven Mechanical Energy and Power Systems Lab Worcester Polytechnic Institute Worcester, MA, USA ABSTRACT On-o control o hydraulic circuits enables signiicant improvements in eiciency compared with throttling valve control. A key enabling technology to on-o control is an eicient high-speed on-o valve. This paper documents the design o an on-o hydraulic valve that minimizes input power requirements and increases operating requency over existing technology by utilizing a continuously rotating valve design. This is accomplished through use o spinning port discs, which divides the low into pulses, with the relative phase between these discs determining the pulse duration. A mathematical model or determining system eiciency is developed with a ocus on the throttling, leakage, compressibility, and viscous riction power losses o the valve. Parameters aecting these losses were optimized to produce the most eicient design under the chosen disc-style architecture. Using these optimum parameter values, a irst generation prototype valve was developed and experimental data collected. The experimental valve matched predicted output pressure and lows well, but suered rom larger than expected torque requirements and leakage. In addition, due to motor limitations, the valve was only able to achieve a 64Hz switching requency versus the designed Hz requency. Future research will ocus on improving the prototype valve and improving the analytical model based on the experimental results. ITRODUCTIO The power o hydraulic pumps, motors, and actuators are traditionally controlled by one o two methods. The irst method throttles the luid through a valve until the desired output pressure is reached; however, this throttling converts excess power into heat and is very ineicient. The second method uses a variable displacement pump or motor to achieve the desired output low rate, but these units are bulky, expensive, and relatively complicated. An alternative to conventional methods is switch-mode control, Figure, where a high-speed valve is used to rapidly switch the system between eicient on and o states, creating virtually variable displacement unctionality. By varying the ratio o the on-time to the total cycle time, deined as the duty ratio, a variable output pressure is produced. On the source side o the circuit, an accumulator minimizes the pressure pulses. Figure : Switch-mode hydraulic circuit used to create a virtually variable displacement pump/motor. The design o the high-speed switching valve is the ocus o this paper. A review o the current technologies shows a wide array o methodologies or creating a high speed on/o valve. Some methods use a solenoid [-5] or piezoelectric actuator [6; 7] to oscillate a poppet or spool valve. One method uses solenoids to actuate a series o -way check valves [8]. Another attractive architecture uses a rotary spool [9-] to avoid the power loss o accelerating a control mass in the oscillatory designs. Other rotary designs use the phase angle [; ] between components to generate a variable duty ratio. This architecture Corresponding author:vandeven@wpi.edu Copyright 9 by ASME

2 is the most closely related to the proposed high-speed switching valve. PHASE COTROLLED VALVE COCEPT The novel phase-shit valve, the topic o this paper, is shown schematic in Figure. The valve consists o three subvalves labeled Section A, B, and, which all rotate at the same constant angular velocity. Each sub-valve is composed o a hal cylinder spool inside a sleeve with two radial inlet ports and one axial outlet port. The outlet port o each sub-valve is connected to each inlet port or hal o a revolution. As seen in Figure a the only open low path is rom Port P (supply pressure to inlet port a, to the outlet port o section A, then to inlet port a and inally exiting section to Port A (to a hydraulic actuator. Figure b shows the same system ater a / rotation. The previous low path is now blocked at inlet port a, but the low path rom Port T (tank to Port A is now open. By varying the relative phase between the Tier subvalves with respect to the Tier sub-valve, the duty ratio, can be continuously varied rom to. have been combined into Tier. In this setup, varying the duty ratio is achieved by changing the phase o Tier relative to Tier, which is ixed. A continuously rotating valve plate is used to produce the switching cycles. Figure : Plot o the low diversion in the valve or a given phase shit. Shaded regions denote when Section is receiving low rom Section. Labeling is consistent with Figure. Figure : Schematic o the phase shit valve or two dierent positions. In subigure a the only open low path is rom Port P to Port A. Ater a 9 degree rotation shown in b the previous path is blocked but a new low pathway is opened rom Port T to Port A. The low diversion o the valve during a single cycle is shown in Figure. ote that Section A and B remain synchronized at phase with respect to each other, while section can vary rom to phase with respect to section. This creates a continuously variable duty ratio rom to, where zero is deined as ull low rom/to tank, Port T, and a duty ratio o is deined as ull low rom/to supply pressure, Port P. A negative phase shit or a phase shit beyond will also result in a duty ratio between and A beneicial characteristic o the phase-shit architecture is that pulsed segments are generated or every rotation o the valve. This creates a switching requency that is X the operating requency o the valve, an advantage compared to other continuously rotating valve methods. Taking the kinematic inversion o the conceptual valve shown in Figure and redirecting the low to move axially, a disc style architecture is created. In Figure 4, the three discs are stacked on top o each other. ote that section A and B Figure 4: Disc architecture with Section sub-valves combined into Tier. Tier can change phase relative to Tier, which is ixed, to change the duty ratio. A new component, the continuously rotating valve plate generates the switching pulses. The disc style architecture oers several advantages over the radial low spool style architecture. The pressure orces on rotating components in the valve can be easily balanced. Manuacturing tolerances can be looser, by using axial thrust bearings to maintain valve clearance instead o spool to sleeve manuacturing tolerances. In this coniguration, the phase only needs to be maintained between two stationary components. Copyright 9 by ASME

3 4 The integration o components minimizes the switched volume, decreasing compressibility losses. 5 Flow travels axially through the valve, so there will not be a centriugal pumping eect. POWER LOSS MODELIG AD AALYSIS To better understand the pressure drops and low rates o the high-speed phase-shit valve, a mathematical model is developed. The energy losses o the valve are o primary interest and include the throttling losses o the ully open and transitioning phases, the internal leakage losses o the valve, the compressibility losses due to compliance in the luid, and the viscous riction losses between rotating components. Losses that will not be analyzed include hysteresis losses in the accumulator, ineiciencies in the pump/motor, compressibility losses due to compliance in the valve and pump/motor structure, and viscous pipe low losses. Once the equations or energy loss are developed, the operating parameters are optimized or the highest eiciency. Reerring to Figure 4, it can be seen that or the instant shown, the low path rom Tier to Tier is momentarily blocked. This occurs whenever the valve transitions rom one state to the next, which is twice per cycle. A typical application or the valve would be the control o a ixed displacement pump/motor on a hydraulic hybrid vehicle. As the valve transitions, luid low will be blocked. The motor will continue to rotate and draw a constant low, causing the motor inlet to vacuum and cavitate. I the hydraulic unit were acting as a pump, then this momentary blockage would create a large pressure spike at the outlet o the pump. To alleviate these issues, two check valves are placed in the hydraulic circuit shown in Figure 5. The right check valve prevents cavitation during motoring, and the top check valve prevents pressure spikes during pumping. Figure 5: Simpliied high-speed valve circuit used or analysis purposes. The two check valves have been added to avoid extreme pressure luctuations occurring when low is completely blocked during valve transitions. The key geometry eatures o the valve are shown in Figure 6. First, we can deine the number o replications o ports on the valve components by the variable (Figure 6 shows. The ports o the valve are deined by the inner radius R i and the outer radius R o. The ports on the valve plate span an angle o δ while the larger ports o Tier and span an angle o γ. ote that δ+γ /, thus the valve will be completely blocked twice each switching cycle. Keeping Tier ixed, the angular position o the valve plate θ is deined as zero degrees when the valve plate ports are completely blocked. Looking axially at the valve, the phase angle α is reerenced as zero degrees when Port A is aligned with Port T. The phase can vary rom to / or a to duty ratio, respectively. A is the variable oriice created by the overlap o Tier with the valve plate, while A is the variable oriice created by the overlap o Tier with the valve plate. Figure 6: Schematic o disc style valve with key eatures deined. VALVE THROTTLIG LOSSES Despite the act that a switch-mode valve was chosen to avoid the ineicient throttling loss o common valves, a signiicant source o power loss is rom throttling within the high-speed valve. The dominant throttling loss is incurred when the valve transitions rom one state to another. Since there are two ull on-o periods or each switching cycle o the valve, there are 4* switches per revolution. For each o these switches, the area o one o the internal valve ports changes rom ully open to ully closed or vice versa, creating throttling across a variable area oriice. At low or high duty ratios, throttling across two simultaneously varying area oriices occurs. Beore calculating the energy loss due to throttling, expressions or the internal port areas o the valve must be developed. Reerring back to Figure 6, the irst variable oriice area A is deined as the port area created by the overlap o Tier and the valve plate. Deining the valve plate angle θ as zero degrees when the valve plate ports are ully blocked and about to transition to Port T, A is given by: A ( θ θ ( R o R i or θ mod < δ ( A ( θ δ ( R o R i or δ θ mod < γ ( ( ( / θ A θ ( R o R i or γ θ mod < ( Copyright 9 by ASME

4 where θ modulo / maintains the evaluated angle between and / or multiple rotations. The symmetry o the valve allows the use o / instead o /. Similarly, the second variable oriice area A is deined as the port area created by the overlap o the valve plate and Tier. Creating a new variable θ * θ- α, A is given by Eqs. (- with θ replaced by θ *. ote that the valve plate has * ports, but only ports have low. Also, besides the instantaneous moment when the valve is completely blocked, ports on the valve plate will always have low. The internal port areas as a unction o rotation angle or a phase shit o º are shown in Figure 7. Figure 7: Open areas o the internal ports o the valve as a unction o rotation angle The addition o check valves to the system means that or a portion o time during each transition, low will be split between the valve and check valve pathways. For instance, reerencing Figure 5, when the high-speed valve is in motoring mode and switching rom Port P to Port T, the internal variable oriices begin to close, causing a large pressure drop. Eventually the pressure at the output o the valve reaches the tank pressure plus the cracking pressure o the tank side check valve, which causes it to open. Flow through the check valve increases until all low passes through the check valve when the high-speed valve passageways become completely blocked. As the valve plate continues to rotate, the internal variable oriices begin to open to Port T. The check valve will hold the output pressure steady as it begins to close. Eventually, the output pressure o the valve reaches the cracking pressure o the check valve and it closes. Full low rom Port T to the hydraulic motor is now going through the valve. To begin the throttling loss analysis it is necessary to irst develop an expression or the ull low pressure drop across the high-speed valve. By assuming the valve as two oriices in series the pressure drop is given by [4]: ρ ρ ΔP ΔP + ΔP + valve Tier Tier ( A + A Cd A Cd A ρ (4 C A P d A Δ valve, P Tier where Δ, and Δ P Tier are the pressure drop due to ull low through the high-speed valve, Tier o the valve, and Tier o the valve respectively, ρ is the mass density o the luid, is the low rate, C d is the discharge coeicient o the oriice, and A and A are the current area o the irst and second Tier oriices respectively. ext, it is necessary to determine when low will be split between the high-speed valve and the check valves. It is assumed that the pressure drop across the check valve is always P check. When the hydraulic unit is in motoring mode, the low pressure check valve will have low when: ΔP > P Ptan + ΔP (5 valve i k check Where P i is the input pressure o the valve, either P High or P Tank and ΔPcheck is the cracking pressure o the check valve. I this condition is met, then the pressure drop across the valve is held constant by the check valve and is given by: ΔPvalve Pi Ptan k + ΔPcheck (6 When the hydraulic unit is in pumping mode, the high pressure check valve will have low once: Δ Pvalve > PHigh Pi + ΔPcheck (7 I this condition is met, then the pressure drop across the valve is held constant by the check valve and is given by: Δ Pvalve PHigh Pi + ΔPcheck (8 I these conditions are met, note that the pressure drop across the high-speed valve remains constant while the low rate through the valve becomes variable. I these conditions are not met, then the pressure drop across the high-speed valve is given by Eq. (4. The calculation o the low through the high-speed valve is attained by rearranging Eq. (4, which gives: ΔPvalve valve Cd A A (9 ρ( A + A By assuming a constant low through the external hydraulic unit, the low through the check valve is always described by: check valve ( The pressure drop across and low through the high-speed valve when the hydraulic unit is acting as a motor are shown in Figure 8. Parameter values used to generate these plots are given in Table. In the igure, note the our large pressure drop spikes corresponding to the our transition events, along with the 4 Copyright 9 by ASME

5 decrease in low through the valve as a result o the check valves opening. Looking at the middle plot, the eect o the check valve can be seen when the valve transitions to Tank pressure around.4 radians. The output pressure is held constant until the threat o cavitation is averted. Figure 8: Pressure drop across and low rate through the valve vs. angular position Table : Values used to generate throttling pressure drop and low plots Variable Value R i.5m R o.5m δ.5 5* -4 m /s P s 7.77MPa P check.mpa.mpa P tank Once the pressure drop and low rate through the valve is determined, the instantaneous power loss due to throttling can be calculated rom: Δ + Δ ( P throttling P valve valve P check check. VALVE LEAKAGE LOSSES Another orm o energy loss is rom the internal leakage o the valve. Starting rom the high-pressure port o Tier, the two primary leakage paths are radially outward to the bore, which is held at tank pressure, and circumerentially to the tank pressure ports. These leakage paths exist between two parallel suraces, so parallel plate leakage is assumed [4]. Perimeter c leak ( μ L where Perimeter is the perimeter o the leakage path given by the average arc lengthγ ( R / bore + Ro, c is the clearance between the plates, ΔP is the pressure dierential, μ is the luid viscosity, and L is the length o the leakage path given by ( R bore Ro. The radial leakage on the ront side o the valve plate, the region between the valve plate and Tier, is given by: γ c ( Rbore + Ro leak, ( 4 μ ( Rbore Ro where R bore is the radius to the bore o the valve, and c is the clearance between the ront ace o the valve plate and the Tier sub-valve. The rear side o the valve plate, the region between the valve plate and Tier, will also experience leakage losses, but this loss is aected by the duty ratio. When the high-pressure ports are blocked, the perimeter o the rear side o the valve plate will be determined by the valve plate port angle δ, but when the ports are unblocked, the perimeter will be determined by the Tier port angle γ. This gives: α Duty (4 leak, b cb ( Rbore + Ro 4 ( Rbore Ro γ Duty + δ ( Duty μ (5 ( where Duty is the duty ratio and c b is the clearance between the back ace o the valve plate and the Tier sub-valve. Once the leakage low rate is calculated, the power loss due to radial leakage is simply: Ρ Leak, rad ( leak, + leak, b (6 The circumerential leakage analysis is complicated by the act that the rotating valve plate creates variable leakage lengths. Furthermore, at the start o the cycle when the ports are completely blocked, the leakage length L rom Equation 5 is zero, predicting ininite low. To simpliy the analysis, oriice low will be assumed around transition events and parallel plate low will be assumed once the leakage path length increases. The cycle o leakage modes is shown in Figure 9. Initially, the leakage length is near zero and oriice low is assumed. As the valve plate advances the leakage path length L becomes suiciently long or parallel plate low to be used, but L is dependent upon the valve plate position θ. Once the valve plate port advances beyond the Tier land at an angle o δ, the leakage length remains constant until the start o the next transition event. During the next transition event, the process is reversed. 5 Copyright 9 by ASME

6 Figure 9: Circumerential leakage pathways or three moments in time In order to determine the circumerential leakage, it is irst necessary to determine the leakage length that corresponds to when the parallel plate equation can be used. By setting the oriice equation equal to the parallel plate equation, an expression is obtained or the minimum leakage length and thus the angle at which to assume parallel plate low. oriice plate b c C d A (7 ρ μ L where oriice and plate are the oriice low and parallel plate low respectively, b is the width, c is the distance between the plates, ΔP P high -P tank, and L is the low length. By recognizing through geometry that A bc, b R o Ri, and L θ ( R i + Ro /, rearranging Eq. 7 gives: c ρ ( Phigh Ptan k θ trans (8 6 μ C d ( R + R i where θ trans is the rotation angle where the low is assumed to transition rom oriice low to parallel plate low. Further recognizing that the leakage cycle described above happens times per revolution per leakage path, that there are leakage paths, and using the deinition o ω d θ / dt, the volume o leakage low per revolution is: V circum, 4 4 t ω t o θ ω θ ( θ ( θ θ θ trans dt dθ o ( θ oriicedθ + plate,var dθ 4 θ θ θtrans (9 ω θ γ + dθ θ δ plate, const θ δ where is given by the let side o Eq. (7 and: oriice plate var plate ( θ ( Ro Ri c 6 μ ( Ro + Ri θ ( Ro Ri c 6 μ ( R + R δ, (, const ( o i Substituting Eq. ( and Eq. ( into Eq. (9 and integrating yields Eq. (5, the circumerential leakage volume per revolution: Cd A θtrans 4 ρ ( Vcircum, ω ( Ro Ri c ( δ γ δ + ln + 6 Ro + Ri μ θtrans δ Eq. ( only constitutes the circumerential volume on the ront side o the valve plate. The rear side circumerential leakage is the low rom the high-pressure valve plate port to the low-pressure valve plate port. Like the rear side radial leakage, the rear side circumerential leakage is urther complicated by the duty ratio o the valve. One other complication is that the leakage paths pass through variable area oriices so the pressure dierential is not constant anymore. However, as seen rom the throttling loss analysis the pressure drop remains airly constant and only varies greatly or brie instants o time. Thus it will be assumed that the pressure dierential on the rear side o the valve remains constant at ΔP P high -P tank. The derivation o the rear side circumerential leakage is similar to the ront side leakage with the addition o conditions dependent upon the phase angle α. For the range δ α δ V circum, b 4 ω plate var θ α + θtrans θ α θtrans oriice + dθ + θ / θ α θtrans θ plate,var θ α + θtrans plate,var ( θ dθ, ( θ dθ and plate ( θ dθ ( θ θ d ( where, var are the same as Equation, but with θ replaced byα θ and θ α, respectively. For the range α δ θ α + θtrans θ α θtrans oriicedθ + plate,var( θ dθ 4 θ α θtrans θ (4 Vcircum, b θ α + γ ω θ / + + plate,var( θ dθ plate, constdθ θ α + θtrans θ α + γ plate var, ( θ dθ and plate ( θ dθ Where, var are the same as Equation, but with θ replaced byα θ and θ α, respectively and plate, const dθ is the same as Equation but with δ replaced by γ. Due to symmetry, Equation 4 can also be used or the range δ α with the simple 6 Copyright 9 by ASME

7 conversion α α. ote that Equations and 4 will use the back side clearance c b instead o c. Finally, the energy loss rom circumerential leakage is: ELeak, circum ( Vcircum, + Vcircum, b (5. VALVE COMPRESSIBILITY LOSSES The next major orm o energy loss o the high-speed valve is due to the compressibility o the luid subjected to a luctuating pressure. Every time the valve switches rom low to high pressure, the luid is compressed, increasing its density. As the valve switches to the tank port, the energy put into compressing the luid is lost as it decompresses to tank pressure. The volume o luid subjected to this luctuating pressure includes the internal volume o the pump/motor, the internal volume o one path o the directional control valve, the output porting o the high speed valve, the port volume o the valve plate and Tier o the high speed valve, and the volume o any passages leading to the check valves. The bulk modulus β is deined as the pressure increase needed ΔP ΔV to cause a given relative decrease in volume, β V. The eective bulk modulus can be described by: R R + β e β oil Phigh k where β is the eective bulk modulus, e (6 β oil is the bulk modulus o air ree oil, R is the entrained air content by volume at atmospheric pressure, and k is the ratio o speciic heats or air. From the deinition o bulk modulus, the change in volume due to every switch rom the tank branch to the pressure branch can be described by: ( Phigh Ptan k Vswitch ΔV (7 β e where ΔV is the change in volume and V switch is the switched volume. Finally, the energy loss during each switch due to luid compression is: E P Ptan Δ (8 comp ( V high k.4 VALVE VISCOUS LOSSES The last orm o energy loss to be considered is caused by viscous riction, which is the riction caused by the shearing o luid. Viscous riction occurs between the ace o the valve plate and the Tier & Tier port aces. The area on the ace o the valve plate can be divided into two sections: the annular region outside the port switching area and the annular region within the port switching area. Viscous riction orces are developed on the ront and rear ace o the valve plate. From ewton s postulate, the rictional torque on the valve plate outside the switching area between the valve plate and Tier is: μ A u T, outer Fr r c θ r R b θ r R o μ ( r dr dθ ( ωr c 4 4 ( R R ωμ bore o (9 c The outer riction ace torque on the rear side o the valve plate, between the valve plate and Tier, T plate,b,o, is o the same orm o Eq. (9, but with c replaced by c b. The viscous riction within the switching area is complicated by transition events. A simpliication o Eq. (9 will be used with the limits o integration rom θ to δ or the ront ace switching torque, T,switch, and θ to or the rear ace switching torque, T b,switch. These new limits o integration are because the ports prevent a ull annular region between the valve plate and the Tier & sections in which viscous losses can have a signiicant eect. The power loss rom these rictional torques is then: T, outer + Tb, outer Ρ Loss ω ( + T, switch + Tb, switch.5 POWER LOSS SUMMARY Ater expressions or the energy losses o the high-speed valve were determined, Matlab was used to optimize critical parameters to achieve a maximum eiciency. The model parameters and resulting optimized parameters are shown in Table and Table with the estimated eiciency vs. duty ratio shown in Figure. Table : Model parameters Table : Optimized parameters Symbol Value Symbol Value R i 5mm V switch.x -5 m R o 7.6mm switch Hz δ /8 ω 5rpm γ 7/8 Duty.5 c 5.4μm P tank kpa c b 5.4μm ΔP check 8 kpa c bore mm C d.6 R bore 4.7mm ρ 85 kg/m 6.e-4 m /s μ (DTE8.875 Pa*s P high 6MPa β oil.8 GPa R % t vp 5mm r 7 Copyright 9 by ASME

8 Multiple sensors are used to measure the valve operation. A schematic o the location o each pressure sensor is shown in Figure. ote that unlike the model, the load or the experimental setup is controlled with a variable oriice. The input low is quantiied using two pressure transducers to measure the pressure drop across a ixed oriice. The downstream transducer was also used to quantiy the input pressure. A ast response time pressure transducer is used to measure the pressure output o the valve. The output low is quantiied by passing the low through a ixed displacement motor and monitoring the shat rotation with an optical encoder. Finally, a orce transducer mounted at a known radius rom the center o the reely-rotating motor measures the torque created by the valve plate motor. Figure : High-speed valve eiciency vs. duty ratio 4 PROTOTYPE DESIG AD IMPLEMETATIO Ater the valve parameters were optimized, a prototype was designed and built. Sensors were used to measure the pressure at the inlet and outlet o the valve, the low into and out o the valve, and the torque required to spin the valve plate. 4. VALVE DESIG The prototype design is shown in Figure. In the interest o brevity, a complete discussion o the mechanical design is not presented here, but can be ound in the this reerence [5]. Figure : Schematic o experimental setup showing location o sensors, oriices, and other hydraulic components 4. EXPERIMETAL SETUP The laboratory setup is shown in Figure with a close up o the prototype valve shown in Figure 4. The optimization procedure determined the sizes o all the key geometric eatures; however, limitations o the available hydraulic low bench required the optimized values to be recalculated. In particular, the available hydraulic supply is only capable o 8 liters per minute at 6.9 MPa, which is a /5th o the desired low rate and ½ o the desired pressure. Also, Mobil DTE8 hydraulic oil was not readily available so Mobil 5M was used. This oil has a viscosity roughly a third o what was desired. Optimizing with these new properties changed the inner slot radius Ri and Ro to 4mm and 8mm, respectively. These changes had the ollowing detrimental eect on the valve perormance. As can be seen in Table 4, the eiciency o the prototype valve is expected to be lower overall, especially at lower duty ratios. Table 4: Eiciency at selected duty ratios or new parameters Duty Ratio Eiciency.8% 55.% 7.% 79% Figure : Solid model o the inal prototype design. 8 Copyright 9 by ASME

9 Presure(MPa Output Pressure vs. Time or Duty Ratio Figure : Laboratory setup. Pressure(MPa Output Pressure vs Time or.67 Duty Ratio Pressure(MPa.5.5 Output Pressure vs. Time or. Duty Ratio Pressure(MPa Output Pressure vs. Time or.5 Duty Ratio Pressure(MPa Output Pressure vs. Time or.67 Duty Ratio Figure 4: Close-up o high-speed valve. 4.4 EXPERIMETAL RESULTS While certainly not complete, preliminary experimental results are presented to provide perspective on the prototype valve operation. The prototype valve was operated at varying duty ratios rom to at requencies up to 64 Hz. As seen rom the output pressure plots, Figure 5, as the duty ratio is increased, the width o the pressure pulses also increases. This unctionality demonstrates the basic desired unctionality o creating virtually variable displacement behavior. Pressure(MPa Pressure(MPa Output Pressure vs. Time or.8 Duty Ratio Output Pressure vs. Time or. Duty Ratio Figure 5: Output pressure or selected duty ratios,.5 inch clearance, clamping orce, and tank-side check valve installed. 9 Copyright 9 by ASME

10 The preliminary maximum eiciency achieved by the valve is 8% at. duty ratio and. in valve plate clearance. The discrepancy between model and experimental results is attributed to a ew sources. As observed in the zero duty ratio plot, the irst subplot o Figure 5, pressure spikes can still be seen suggesting that the valve is experiencing unoreseen leakage rom pressure to the output port. Also, in this coniguration, the low to the valve is much higher than the minimal output low, suggesting that there are signiicant unoreseen leakage losses. Physical inspection o the valve prototype reveals small gaps between the tier and tier ports and the valve plate when the valve is supposed to be blocked. Furthermore, rounds were machined on the port suraces, creating additional leakage paths. Friction losses, which were calculated to be the least signiicant o the power losses, contributed to more than hal o the total power loss in the experimental system. This discrepancy is believed to be caused by misalignment o components within the valve and greater than expected riction coeicients in the thrust bearings. 5 COCLUSIOS This paper presented the design o a high-speed phase-shit valve to enable switch-mode control o hydraulic. Switchmode control allows any ixed displacement hydraulic pump, motor, or actuator to have virtually variable displacement by rapidly switching between eicient on and o states. The novel valve design uses a phase shit between two switching tiers to create a constant requency pulsed low where the duty ratio is controlled by change the phase between the switching tiers. The design uses a disc architecture with axial low and a single constantly rotating valve plate. Through a numerical model, the various orms o energy loss in the valve were computed. Using the model and an optimization procedure valve parameters were selected to minimize the total energy loss. These results were used to develop a prototype valve. Limited experiments demonstrated the correct switch-mode unction o the valve at switching requencies up to 64 Hz. However, due to larger leakage and torque requirements than predicted, the demonstrated eiciency was signiicantly less than expected. Multiple areas o uture work exist related to the phaseshit high-speed hydraulic valve. First, the experimental device will be reined to reduce leakage losses and torque requirements. This work will hopeully result in better alignment o the predicted and experimental eiciency results. Based on the experimental results, the analytical model will be improved, with speciic ocus on the viscous orce equations and luid dynamics. A next generation prototype will then be integrated into a hydraulic pump/motor to minimize the switched volume and provide real-world results. In conclusion, many aspects o the high-speed hydraulic valve were a success, yet many areas o uture research exist. Particularly in making the valve more compact, more eicient, improve the bandwidth, and developing control systems. REFERECES [] Tomlinson, S. P., and Burrows, C. R., 99, "Achieving a Variable Flow Supply by Controlled Unloading o a Fixed- Displacement Pump," Journal o Dynamic Systems, Measurement, and Control, 4, pp [] Batdor, M. A., and Lumkes, J. H. (7. "Fast Acting Fluid Control Valve." United States Patent 7/89649 A. [] Giaardo, M. (99. "High-Speed Solenoid Valve or a Fluid under Pressure, E.G. For Pneumatic Circuits." United States Patent 5,48,564. [4] Jacobs, J.. (994. "High Speed Gate Valve." United States Patent 5,6,8. [5] Mesenich, G. (99. "Pulse Modulated Hydraulic Valve." United States Patent 4,979,54. [6] Ota, K. (99. "High Speed Flow Control Valve." United States Patent 5,48,8. [7] Yokota, S., and Akutu, K., 99, "A Fast-Acting Electro- Hydraulic Digital Transducer," JSME International Journal, 4(4, pp [8] Ibiary, Y. E., 978, "Coming: Smart Hydraulic Valves," Machine Design, 5(7, pp [9] Cyphelly, I., and Langen, H. J., 98, "Ein eues Energiesparendes Konzept Der Volumenstromdosierung Mit Konstantpumpen," Aachener Fluidtechnisches Kolloquium, pp [] Rannow, M. B., Tu, H. C., Li, P. Y., and Chase, T. R., 6, "Sotware Enabled Variable Displacement Pumps - Experimental Studies," ASME International Mechanical Engineering Congress and Exposition, Chicago, IL. [] Tu, H. C., Rannow, M., Van de Ven, J., Wang, M., Li, P., and Chase, T., 7, "High Speed Rotary Pulse Width Modulated on/o Valve," Proceedings o the ASME International Mechanical Engineering Congress, Seattle, WA, pp [] Lu, X. F., Burton, R. T., Schoenau, G. J., and Zeng, X. R., 99, "Feasibility Study o a Digital Variable Flow Divider," SAE Technical Paper Series,(986, pp [] Royston, T., and Singh, R., 99, "Development o a Pulse-Width Modulated Pneumatic Rotary Valve or Actuator Position Control," Journal o Dynamic Systems, Measurement, and Control, 5, pp [4] Cundi, J. S.,, Fluid Power Circuits and Controls: Fundamentals and Applications, CRC Press, Boca Raton, FL. [5] Katz, A., 9, "Design o a High Speed Hydraulic on/o Valve," Thesis, Worcester Polytechnic Institute, Worcester, MA. Copyright 9 by ASME