PARALLEL PUMP-CONTROLLED MULTI-CHAMBER CYLINDER

Transcription

1 Proceedings of the ASME/BATH 2014 Symposium on Fluid Power & Motion Control FPMC2014 September 10-12, Bath, United Kingdom FPMC PARALLEL PUMP-CONTROLLED MULTI-CHAMBER CYLINDER Olli Niemi-Pynttäri Department of Intelligent Hydraulics and Automation (IHA) Tampere University of Technology (TUT) Tampere, Finland Matti Linjama Department of Intelligent Hydraulics and Automation (IHA) Tampere University of Technology (TUT) Tampere, Finland Arto Laamanen Department of Intelligent Hydraulics and Automation (IHA) Tampere University of Technology (TUT) Tampere, Finland Kalevi Huhtala Department of Intelligent Hydraulics and Automation (IHA) Tampere University of Technology (TUT) Tampere, Finland ABSTRACT This study focused on the use of fixed displacement pumps in parallel connection to control the velocity of a multi-chamber cylinder piston. The system s basic principle was to combine the discrete flow supply control of parallel pumps with the discrete effective area control of a multi-chamber cylinder to produce a speed control resolution high enough for accurate velocity tracking and positioning. Some throttling was used in the return line to control the system with overrunning loads. The properties of the system were tested with a 1-DOF boom mockup mimicking a medium-sized mobile machine boom. The test system revealed a feature that caused load acceleration to drop when the effective cylinder area was reduced during movement. Additionally, some delay was observed in accelerating the piston against the load force. These two system properties along with the discrete control method resulted in mediocre speed and position tracking in the system when movement was directed against the load force. The system was able to control restricting and overrunning loads as well as a large inertia ma with a low load force. The system s energy loes were low considering that no preure accumulators were used, but the throttling loes in the return line and the lack of energy recuperation leave room for improvement. INTRODUCTION In the past few years, the energy consumption of mobile work machines has become a crucial iue owing to new emiion directives aimed at reducing the permiible exhaust emiions of internal combustion engines in non-road mobile machinery. Additionally, rising fuel prices and environmental awarene are neceitating efficient energy use. Fuel consumption and emiions can be reduced either by increasing the efficiency of the prime mover, usually a diesel motor, or by increasing the efficiency of the hydraulic system driven by the prime mover. One popular way to reduce energy loes in a hydraulic system is to use preure accumulators to recuperate energy and to generate large peak powers with small prime movers, though in some cases this is undesirable or impoible. In these systems, good energy efficiency can be achieved by adjusting the supplied hydraulic power to match the need of the actuator. If simple and robust components are a priority, supply flow control can be executed by replacing the common variabledisplacement pumps with simpler on/off components. The result is a pump unit consisting of several parallel fixed displacement pumps connected to a common axle. Parallel fixed displacement units have been previously simulated in closed circuit use [1] and as a digital pump-motor unit [2]. In the solution studied in this paper, the flow supplied by the pump unit was varied by connecting a section of the pumps to pump fluid to the supply line whereas the remaining pumps were connected straight to the tank line with a minimal preure 1 Copyright 2014 by ASME

2 difference over the pump. The resulting discrete flow supply control was combined with discrete effective area control of a multi-chamber cylinder, which in theory increases the actuator speed control resolution enough for fast and accurate piston positioning. The solution resembles the gears of a bicycle: the parallel pumps correspond to the gears connected to the pedals, and the multi-chamber cylinder corresponds to the gears on the wheel. SPEED CONTROL OF A MULTI-CHAMBER CYLINDER vv = uu PP qq PP uu cccccc aa cccccc (1) where uu PP is a vector selecting the pumps that pump to the supply line, qq PP is a vector with the flow rates of individual pumps, uu cccccc is a vector selecting the cylinder chambers that are connected to the supply line, and aa cccccc is a vector containing the effective areas of the cylinder chambers. A simplified hydraulic circuit diagram of the control method is shown in Figure 2. The multi-chamber cylinder is a hydraulic cylinder with more than two chambers. The additional chambers are used to modify the effective cylinder area, which affects the force generated by the cylinder in constant preure systems or the velocity of the piston in constant flow systems. Previously, multi-chamber cylinders have been used, for example, to realize symmetric cylinders without an extra rod or synchronizing movements, but with digital hydraulics and separate meter-in meter-out control methods, more complex control has become feasible. Figure 1 shows the structure of a multi-chamber cylinder with four concentric cylinder chambers. Chambers A and C provide an extending force and B and D a retracting force when preurized. FIGURE 1: STRUCTURE OF A FOUR-CHAMBER CYLINDER The force control of a multi-chamber cylinder in constant preure systems has been studied in [3], [4], and [5] and realized in an industrial prototype by Norrhydro [6]. Additionally, the usage of a multi-chamber cylinder as a hydraulic transformer has been studied in [7] and [8]. This paper concentrates on controlling the velocity of the piston by flow supply and effective area control whereas the supply preure is determined by the load force and the effective cylinder area. The flow supplied by the pump unit was controlled by connecting some of the pumps outlet ports straight to the tank line via free-flow valves with low preure loes whereas the remaining pumps were connected to the supply line. The effective cylinder area was selected by connecting the supply line to the desired cylinder chambers via ON/OFF valves whereas the remaining chambers were connected to the tank line. The steady state piston velocity vv is then defined by the equation FIGURE 2: SIMPLIFIED HYDRAULIC CIRCUIT DIAGRAM OF A PARALLEL PUMP-CONTROLLED MULTI-CHAMBER CYLINDER Because the pumps of the pump unit cannot function as motors, some throttle control is needed in the tank line to control overrunning loads. In theory, a hydraulic system consisting of three parallel, binary-coded fixed displacement pumps (2 cmᶟ, 4 cmᶟ, and 8 cmᶟ at 1500 rpm) and a binary-coded four-chamber cylinder (85/63/40/28, where the dimensions relate to D1/D2/D3/D4 in Figure 1) is capable of producing 106 different combinations of pump unit and cylinder states (Figure 3). The cylinder state with all chambers connected to the tank line was removed because the piston velocity could not be controlled by altering the supply flow. The state in question was replaced with a stop state with every cylinder chamber valve closed (shown in the origin of the figure). Each cylinder state was capable of producing a different maximum force at the provided maximum supply preure, which is why the operating points appear at 16 different 2 Copyright 2014 by ASME

3 maximum force levels. The three parallel pumps could provide seven different supply flows, which resulted in seven different piston velocities in each poible cylinder state. FIGURE 4: PISTON VELOCITY DROP DURING A CHANGE IN THE CYLINDER STATE FIGURE 3: STATES OF A COMBINATION OF THREE PARALLEL PUMPS AND A FOUR-CHAMBER CYLINDER AT 18 MPA SUPPLY PRESSURE The figure shows that the system could not generate a negative force when the speed was positive or a positive force when the speed was negative. This means that the system was unable to control overrunning loads without throttling. The speed control resolution was significantly higher at low speeds than at high speeds because a pump unit supply flow step of 3 L/min produced a large velocity step with small effective cylinder areas. Furthermore, the cylinder could generate large maximum forces only at slow piston velocities whereas large velocities were poible only at low load forces. Cylinder state change with constant flow supply When an actuator with step-like changes in the effective area was used in a system with a constant flow supply, a new problem arose. When the effective cylinder area was decreased to increase the piston speed, the force generated by the cylinder dropped. At the same time, extra force is required from the cylinder to accelerate the load to the desired velocity. The supply preure does not rise immediately to the desired level and the piston velocity drops momentarily. Figure 4 illustrates this phenomenon. When changing cylinder states requires preurizing cylinder chambers from tank line preure to supply preure, the energy needed for preurization is taken from the supply line and the supply preure drops momentarily, which further intensifies the velocity drop. In the studied system, the velocity drop effect of changing states was reduced slightly by introducing delays to some changes in the valve state. When the piston was in motion and a contracting cylinder chamber was switched from tank line to supply line, the controller delayed opening the chambers supply line valve whereas the tank line valve closed immediately. This resulted in a brief full closure of the cylinder chamber and a rise in cylinder preure from the continued motion of the piston. When the chamber preure reached the supply preure, the supply line valve opened, and the operation continued normally. This method accelerates the rise in supply preure in changing cylinder states, but it applies only to small cylinder chambers. With large chambers, preure rising inside a contracting chamber would generate such a decelerating force that the piston velocity would drop even more than without delayed valve switching. Control strategy The controller of the system was designed to minimize changes in cylinder state to reduce energy loes originating from chamber depreurization and the velocity drop effect when the effective cylinder area was decreased. The controller was designed to find the largest effective cylinder area that satisfied the velocity reference at the maximum supply flow of the pump unit and to produce enough force to overcome the load force at maximum system preure. Then it chose the pump unit state that produced a flow to generate a piston velocity with the lowest velocity error. Valve switching penalties were added to the cost function that selected the states of the pump unit and the cylinder to reduce unneceary state switching. 3 Copyright 2014 by ASME

4 When the piston was accelerated from a standstill, the controller used at first the slowest cylinder state with all the expanding chambers connected to the supply line and all contracting chambers connected to the tank line. At slow speeds, speed control was executed only by varying the fluid flow supplied by the pump unit whereas the cylinder state remained the same. When the speed reference rose to a point where the maximum flow of the pump unit was not enough to satisfy the velocity reference with the slowest cylinder state, the controller started decreasing the effective cylinder area. Four cylinder states that produced the fastest piston extension velocities were disabled because with them the net flow of the tank line would have been directed to the cylinder from the tank and the expanding chambers of the cylinder might have cavitated. With all the other states, the net flow of the tank line was directed from the cylinder to the tank, which means that the expanding chambers received their fluid flow from the contracting chambers while the surplus was routed to the tank. Furthermore, the two next fastest extending states were disabled because they required the fairly large B chamber to be switched from tank line to supply line, which would have caused a significant drop in the piston velocity before acceleration, as discued in the previous section. Figure 5 shows a flow chart of the control strategy. DFCU controller Because the combined pump unit and multi-chamber cylinder cannot alone control an overrunning load, some throttling must be added. One option is to use a proportional valve or a Digital Flow Control Unit in the tank line, as shown in Figure 2. In this study, however, throttling was implemented by using two DFCUs to connect chambers A and B to the tank line because that reduced the hydraulic capacitance of the controlled volumes and thus improved control performance. Because the fastest extending modes were disabled, the A and B chambers are never both connected to the supply line at the same time and the flow from the contracting cylinder chamber is always throttled through a DFCU. The DFCUs were controlled by creating preure references for the two chambers from the estimated load force. Preure references were then used to calculate flow rates for the DFCUs in different states. The flow rates of all states were compared to a theoretical flow rate through the DFCUs, determined from pump unit and cylinder states. Figure 6 shows a flow chart of the DFCU controller. FIGURE 6: DFCU CONTROLLER Preure references for chambers A and B were calculated by using the equations FIGURE 5: CONTROL STRATEGY The estimated load force was calculated by using the equation pp rrrrrrrr = max (pp mmmmmm AA BB + FF llllllll ) AA AA, pp mmmmmm pp rrrrrrrr = max (pp mmmmmm AA AA FF llllllll ) AA BB, pp mmmmmm (3) FF llllllll = pp AA AA AA pp BB AA BB + pp CC AA CC pp DD AA DD, (2) where pp AA,BB,CC,DD are the measured preures of the cylinder chambers, and AA AA,BB,CC,DD are the effective chamber areas. where pp mmmmmm is the desired minimum preure of the controlled chambers, AA AA,BB are the effective areas of chambers A and B, and FF llllllll is the load force estimated from the chamber preures by using equation 2. Preure references were selected so as to have enough desired chamber preure to negate the load force, 4 Copyright 2014 by ASME

5 except when the load force was very small or negative. In that case, the desired preure saturates to the desired minimum preure reference pp mmmmmm. The A chamber DFCU was controlled when the cylinder was retracted and the B chamber DFCU when the cylinder was extended. Flow rates through the DFCUs were calculated by using the equation for turbulent flow through an orifice: valves were controlled with booster electronics that reduced valve delays to 5.3 ms for opening and 7.7 ms for closing [11]. Figure 7 shows a hydraulic circuit diagram of the test system. nn QQ DDDDDDDD = uu vvvv KK vvvv (pp iiii pp oooooo ) xx i ii=1 (4) where QQ DDDDDDDD is the flow rate through one valve of the DFCU, uu iiii a binary value indicating whether the valve ii is open or closed, KK vvvv is the flow coefficient of the valve ii, pp iiii is the preure at the inlet port, pp oooooo is the preure at the outlet port, and xx ii is the exponent term of the valve ii, which replaces the square root of the basic flow equation to improve the accuracy of the model. [9] Flow coefficients and exponent terms were determined by measurements for each valve separately. Modelbased control of a DFCU is explained in detail in [10]. The DFCU controller included also cavitation avoidance because decelerating a large inertia makes the supply line preure drop and eventually cavitate. Cavitation avoidance monitors the rate of deceleration of the velocity reference, the measured velocity of the piston, and the estimated load force and generates a flow rate that is subtracted from that of the control edge calculated from cylinder and pump unit states. When the velocity reference decelerates, cavitation avoidance reduces the flow reference from the cylinder to the DFCU, making the DFCU controller choose smaller states, thus producing a large preure in the controlled cylinder chamber. Rising preure in the contracting cylinder chamber generates a decelerating force and prevents the preure level of the supply line declining to cavitation. TEST SYSTEM The pump unit of the test system consisted of three Rexroth AZP external gear pumps with a volumetric displacement of 2, 4, and 8 cmᶟ/rev and a VEM K21R electric motor with 7 kw maximum output power and 1500 rpm rotational speed. The supply flow of the pump unit was controlled by connecting selected pumps to free flow mode by means of three Parker D1VW030DNL 4/2 spool valves. The valves and their control electronics were modified to decrease the response time to 7.5 ms for opening and closing the valve. The preure transients in the state changes of the pump unit were dampened by using a 5-L rigid wall damping volume with a 1-mm orifice parallel to the supply line. Cylinder chambers were connected to the supply line and to the tank line via Hydac WS08W-01 2/2 poppet valves. The DFCUs controlling chambers A and B of the cylinder used the same valves with added orifices in the flow paths to achieve binary coded valve flow rates in the DFCUs. All Hydac poppet FIGURE 7: HYDRAULIC CIRCUIT DIAGRAM OF THE TEST SYSTEM The hydraulic system was used to drive a 1-DOF boom mockup mimicking a medium-sized mobile machine boom. The boom mockup is described in detail in [12] and shown in Figure 8. FIGURE 8: MOBILE BOOM MOCKUP The load force of the mobile boom mockup was varied by changing the configuration of the load maes at the opposite ends of the boom. The studied load cases are given in table 1, 5 Copyright 2014 by ASME

6 where F load is the reduced load force on the cylinder when the boom is in the horizontal position. TABLE 1: LOAD CASES OF THE BOOM MOCKUP Load case m 1 [kg] m 2 [kg] m 3 [kg] m 4 [kg] F load [kn] A B C The load case A was an almost balanced load with a low load force, B was a highly restricting load with a large load force, and C was an overrunning load with a moderate load force. RESULTS The test system was driven with reference trajectories, which produced a slow extending and retracting motion followed by a fast extending and retracting motion. The desired maximum piston velocity during a movement cycle was 105 mm/s. Figures 9, 10, and 11 show measurement results for load cases A, B, and C. FIGURE 10: MEASUREMENT RESULTS WITH RESTRICTING LOADING (B) FIGURE 9: MEASUREMENT RESULTS WITH ALMOST BALANCED LOADING (A) FIGURE 11: MEASUREMENT RESULTS WITH OVERRUNNING LOADING (C) When the piston was stopped after a movement directed against the load force, the preure in the supply line remained on the level generated by the load force. If the piston was then driven in the direction of the load force, a high preure in the supply line made the piston briefly surge forward, as can be seen in Figure 10 at 3.5 s and in Figure 11 at 1 s. This behavior 6 Copyright 2014 by ASME

7 was problematic if small movements of the piston were attempted in these situations. Boosting chamber preure by delaying the opening of the supply line valves can be seen in preure measurements for chamber D in Figure 9 at 6 s and in Figure 10 at 6.3 s. The effect of the drop in speed due to a change in the effective cylinder area can be seen in Figure 10 at 6.3 s. The operation of the control strategy can be seen in the states of the cylinder and pump unit during the extending motions in all load cases. When the cylinder was extended slowly, it remained in the slowest state, and the pump unit took over velocity control. During fast extension, the controller first gradually increased the pump unit state to a maximum and then began to lower the effective cylinder area to meet the velocity reference. Energy loes Figure 12 shows the energy loes in the studied system compared to other solutions tested with the boom mockup. Since the loading cases and reference trajectories are nearly identical in all systems, straightforward comparison makes sense. In the figure, Prop 12 is a proportional valve system with 12 MPa supply preure [10], Prop LS is the same system with load sensing supply preure [10], ResDigiCyl is a resistance controlled multi-chamber cylinder system with 18 MPa supply preure [4], DVS is a Digital Valve System controlled system with load sensing supply preure [10], PumpDigiCyl is the parallel-pump-controlled multi-chamber cylinder system, SecDigiCyl is a secondary-controlled multichamber cylinder system with 16 MPa supply preure [3] and DVS+pT is a DVS-controlled system with load sensing supply preure and a preurized tank line [13]. Energy loes were estimated by subtracting mechanical output power from hydraulic input power. Mechanical output power was determined by multiplying the measured piston velocity with the cylinder load force whereas hydraulic input power was determined by multiplying the measured supply preure with the measured supply flow. Because only hydraulic input power was measured, the results do not take into account the method by which this power was produced. This weakens the comparability of the results because a variable displacement pump becomes clearly le efficient with small displacements. Had the total input power of the electric motor been used in the energy lo calculations, the energy loes in the multi-chamber systems would have been reduced slightly when compared to the other systems because they used fixed displacement pumps instead of variable displacement pumps. CONCLUSIONS The results show that the studied system has low energy loes even without energy recuperation. The geed energy efficiency derives mostly from the preure level in the supply line being determined by the load force with little throttling. This also poses challenges to system control. For example, when a restricting load is lifted and the preure in the supply line drops below what is needed to accelerate the load, the controller must wait for the preure to build up before initiating a movement. This causes a delay at the beginning of the movement. The delay is considerably reduced by the momentary pumping of the largest pump to preurize the supply line to the desired level as fast as poible. Because the level of the supply line preure depends on the load force, using multiple actuators with one pumping unit would markedly increase the energy loes in the system. Speed reference tracking was only moderate due to the discrete nature of the control principle and preure oscillations, though in theory it could be improved by increasing the number of pumps. Additionally, removal of the six fastest cylinder states reduced the maximum piston speed in the pump configuration we used, and the desired speed of 105 mm/s was not reached when the cylinder was extended in almost balanced or restricting loading. ACKNOWLEDGMENTS This study is a part of EFFIMA/DigiHybrid (Energy and Life Cycle Cost Efficient Machines) project which was coordinated and funded by FIMECC Ltd. (Finnish Metals and Engineering Competence Cluster). FIGURE 12: ENERGY LOSSES OF DIFFERENT SOLUTIONS WITH THE BOOM MOCKUP 7 Copyright 2014 by ASME

8 NOMENCLATURE AA AA,BB,CC,DD AA ccyyyy AA eeeeee Effective areas of the cylinder chambers A, B, C and D Vector with the effective areas of the cylinder chambers Effective area of the selected cylinder state DDDDDDDD Digital Flow Control Unit FF cccccc Cylinder force NN KK vvvv Flow coefficient of the valve ii mm 2 mm 2 mm 2 PPPP mm 1,2,3,4 Load maes of the boom mockup kkkk pp AA,BB,CC,DD Preures of the cylinder chambers A, B, C and D PPPP pp iiii Preure on the input port of the DFCU PPPP pp mmmmmm pp oooooo Minimum preure difference over the DFCU Preure on the output port of the DFCU pp rrrrrrrr A chamber preure reference PPPP PPPP PPPP pp SS Supply line preure PPPP pp TT Tank line preure PPPP QQ DDDDDDDD qq PP QQ SS QQ TT Theoretical flow rate over the DFCU Vector with the flow rates of individual pumps of the pump unit Supply line flow Tank line flow UU AAAA Control signal of the AT-DFCU UU BBBB Control signal of the BT-DFCU UU cccccc State of the cylinder uu cccccc Vector with the control signals of the cylinder valves UU PP State of the pump unit uu PP Vector with the control signals of the pumps uu vvvv Control signal of the valve ii vv vv rrrrrr vv Piston velocity Piston velocity reference Steady state piston velocity mmmm mmmm mm xx Piston position mmmm xx ii Exponent term of the flow equation for the valve ii xx rrrrrr Piston position reference mmmm REFERENCES [1] Heitzig, S. & Theien, H. Aspects of Digital Pumps in Closed Circuit. The Fourth Workshop on Digital Fluid Power, September Linz, Austria 2011, pp [2] Linjama, M. & Tammisto, J. New Alternative for Digital Pump-motor-transformer. Proceedings of The Second Workshop on Digital Fluid Power, November Linz, Austria 2009, pp [3] Linjama, M., Vihtanen, H., Sipola, A. & Vilenius, M. Secondary Controlled Multi-Chamber Hydraulic Cylinder. The 11th Scandinavian International Conference on Fluid Power, SICFP'09, June 2-4, [4] Huova, M. & Laamanen, A. Control of Three- Chamber Cylinder With Digital Valve System. The Second Workshop on Digital Fluid Power, November , pp [5] Dell'Amico, A., Carlon, M., Norlin, E. & Sethson, M. Investigation of a Digital Hydraulic Actuation System on an Excavator Arm. Proceedings of the 13th Scandinavian International Conference on Fluid Power, June 3-5, Linköping, Sweden [6] Sipola, A., Mäkitalo, J. & Hautamäki, J. The Product Called NorrDigi<SUP>TM</SUP>. Proceedings of The Fifth Workshop on Digital Fluid Power, October 24-25, Tampere, Finland 2012, pp [7] Bishop, E. Digital Hydraulic Transformer - Approaching Theoretical Perfection in Hydraulic Drive Efficiency. The 11th Scandinavian International Conference on Fluid Power, SICFP'09, June 2-4, Linköping, Sweden [8] Bishop, E. Digital Hydraulic Transformer - Efficiency of Natural Design. Proceedings of The 7th 8 Copyright 2014 by ASME

9 International Fluid Power Conference, IFK'10, March 22-24, Aachen, Germany [9] Linjama, M., Huova, M. & Karvonen, M. Modelling of Flow Characteristics of ON/OFF Valves. Proceedings of The Fifth Workshop on Digital Fluid Power, October 24-25, Tampere, Finland 2012, pp [10] Linjama, M., Huova, M., Boström, P., Laamanen, A., Siivonen, L., Morel, L., Waldén, M. & Vilenius, M. Design And Implementation of Energy Saving Digital Hydraulic Control System. The Tenth Scandinavian International Conference on Fluid Power, SICFP 07, May Tampere, Finland [11] Mikkola, J., Ahola, V., Lauttamus, T., Luomaranta, M., Linjama, M. & Vilenius, M. Improving Characteristics of ON/OFF Solenoid Valves. The Tenth Scandinavian International Conference on Fluid Power, SICFP'07, May [12] Linjama, M. & Vilenius, M. Digital Hydraulic Control of A Mobile Machine Joint Actuator Mockup. Proceedings of The Bath Workshop on Fluid Power Transmiion & Motion Control (PTMC'04), September , pp [13] Huova, M. & Linjama, M. Energy efficient digital hydraulic valve control utilizing preurized tank line. Proceedings of the 8th International Fluid Power Conference Dresden, March 26-28, 201, March 26-28, Dresden 2012, Technische Universutät Dresden, Dresdner Verein zur Förderung der Fluidtechnik. pp Copyright 2014 by ASME