vor 2 Jahren

O+P Fluidtechnik 10/2016

  • Text
  • Fluidtechnik
  • Sensor
  • Predictive
  • Maintenance
  • Industrie
  • Spool
  • Valve
  • Controller
  • Anwendungen
  • Produkte
O+P Fluidtechnik 10/2016


FORSCHUNG UND ENTWICKLUNG STEUERUNGEN UND REGELUNGEN Notice that the damping term is not as critical for this case since the meter-in case is less sensitive to dropping a heavy load. This control is only necessary for a passive load. For an overrunning load, the meter-in spool can be fully opened to tank and the actuator speed controlled with the meter-out valve, as in the conventional case. 3 EXPERIMENTAL RESULTS The fail operational controller was implemented on the actuators of a backhoe loader. While the boom, arm, and bucket were all successfully tested, the boom provided the heaviest load, with the most potential to fall if there were any controller errors. Thus, the boom was used for demonstration. Note that the sensors did not actually fail, so their readings are included in the plots. Internally to the valve controller, the feedback from the faulty sensors was disconnected. 3.1 TRACES OF A BOOM LOWERING WITH A FAILED METER-OUT PRESSURE SENSOR Figure 1 shows an example of a system working with a failed meterin pressure sensor. In this example, the load is overrunning, which 03 Traces of a boom lowering with a failed meter-out position sensor 04 Traces of a boom raising with a failed meter-in position sensor 74 O+P Fluidtechnik 10/2016

STEUERUNGEN UND REGELUNGEN means the meter-out valve is holding the load. The set point for the meter-in pressure was 15 bar, and, after an initial transient, the pressure settles to that value. In the initial transient, the pressure climbs higher than 15 bar, which means there is more flow coming in the meter-in side than is leaving the faulty meter-out side. This is due to the fact that, at the start of motion, the estimate of the load pressure is initialized to the maximum possible load pressure of 200 bar instead of the actual value of around 70 bar. Thus, the meter-out valve does not open far enough. However, in response to the meterin pressure being higher than 15 bar, the controller in (4) adjusts the estimate to open the valve further. The duration of the initial transient can be manipulated by the design of the cross-port pressure controller (4). As the magnitude of the flow demand is increased, both spools open further in response, maintaining the meter-in pressure near its set point. 3.2 TRACES OF A BOOM RAISING WITH A FAILED METER-IN PRESSURE SENSOR In Figure 2, the actuator is moved in the opposite direction, raising the boom using a faulty pressure sensor on the meter-in side. Note that the meter-in (MI) and meter-out (MO) pressure changed places from the previous case. In this figure, the meter-out pressure is controlled to be 15 bar, despite a varying flow demand. This test was run with an initial guess of the meter-in pressure of 50 bar, which was close to the actual value. As a result, the start of motion did not have a large transient. The estimated meter-in pressure is shown. Note that the pressure estimate is below the true value, which makes the pressure margin lower than the typical 10 bar. However, the controller in (6) will adjust itself to ensure that there is sufficient margin to achieve the desired flow, and thus maintain the meter-out pressure at 15 bar. The faulty spool (spool 2) is responsive to changes in the magnitude of the flow demand as it is increased and then decreased. 3.3 TRACES OF A BOOM LOWERING WITH A FAILED METER-OUT POSITION SENSOR The case of a faulty position sensor on the meter-out side of the boom is shown in Figure 3. Lowering a heavy load without position feedback on the mainstage spool is the most challenging case for the fail operational controller. The faulty spool opens out of the deadband in about 120 ms, which is slower than the 30-50 ms that is typical on a spool with feedback, but is still barely perceptible to an operator. While the PID terms in (7) are active during this time, they are tuned to be fairly slow in order to maintain stability. The large movement needed to escape the deadband is significantly improved by the feedforward term. Initially, the meter-in pressure is higher than the commanded 15 bar, indicating that the meter-out valve is not open enough to allow as much flow out of the actuator as is being put in on the meter-in side. However, the spool quickly adapts, and the meter-in pressure converges to its set point. With step changes in the flow demand, the faulty spool opens and closes accordingly in order to maintain the proper cross-port pressure. 3.4 TRACES OF A BOOM RAISING WITH A FAILED METER-IN POSITION SENSOR Finally, in Figure 4, an example of a failed meter-in position sensor is given. In this example, a controller like in (9) is used to maintain the meter-in spool in a partially open state by trying to maintain the meter-out pressure. Notice that the meter-in spool opens up a bit slower than the meter-out spool due to the delay in moving the spool out of the dead band using direct current control. Since the service speed is controlled by the meter-out valve, the controller in (9) does not need to be as responsive as the controller in (7). Thus, the feedforward term that helped the spool exit the dead band quickly in the failed meterout case was not tuned to be as aggressive. This could be adjusted if desired. 4 CONCLUSION In this paper, a method for utilizing the inherent information redundancy in an independent metering valve with position and pressure sensors to create a fault tolerant system was described. While a working position and pressure sensor are needed to accurately control the flow across a spool that uses electronic pressure compensation, the controller can always be re-structured so that the flow is controlled on the spool with two working sensors, with the other spool attempting to regulate the pressure on the non-faulty side of the work port to a constant value. Maintaining a constant pressure ensures a balance between the flow in and out of the actuator. Experimental results demonstrate that this approach can be tailored to a failure in any one of the four sensors on the valve. This creates a valve that is tolerant to sensor faults, which improves the reliability and uptime for the system. Author: Michael Rannow, EATON Corporation Nomenclature ∆P P in, P out x in , x out, x des Q des f() β V A in, A out P K p, K i, K d, K ff i in, i out φ -1 () g(P in, P out ) i deadband P is,des P margin Pressure difference: Psupply – Pin or Pout – Ptank on in or out side Pressure, Pressure on input side, Pressure on output side Desired spool position, position on input side, position on output side Desired flow rate – note any variable with a _des is a desired value Function describing relationship between pres sure, flow, and position Bulk modulus of the fluid Volume of the fluid Actuator Area on the input side, actuator Area on the output side Estimated pressure on the inlet side, estimated pressure on outlet side Proportional, Integral, Defivative, and Feed forward controller gains Current sent to the pilot spool actuator in the input and output side Inverse of a function that relates the input current to the spool velocity Damping function that uses actuator pressures to smooth the control Current needed to move the pilot spool to the edge of its deadband Desired load sense pressure (desired supply pressure minus margin) Pressure margin between Pls and supply pressure O+P Fluidtechnik 10/2016 75


© 2018 by Vereinigte Fachverlage GmbH. Alle Rechte vorbehalten.