Optimal Control of a New Hydraulic Elevator Architecture

This study presents application of Dynamic Programming towards finding globally optimal control policies of a new, highly-efficient hydraulic elevator architecture. The architecture employs an actuator, three accumulators, and two hydraulic circuits bridged by a hydraulic transformer. An auxiliary electric motor exists only to compensate for system losses. Prior research identified a multitude of heuristic control strategies for commanding two pump/motor displacements, the choice of which was shown to have significant impact on the overall system efficiency. The prior research stopped short of identifying which of the control policies are optimal under what conditions. In this study, Dynamic Programming is implemented with backwards-looking simulation and elevator usage schedules to develop optimal control policies that minimize energy consumption. Results from Dynamic Programming are then studied to develop real-time implementation strategies for controlling the two pump/motor displacements and the electric motor. Follow-on research will assess these strategies on a small hydraulic elevator prototype. Introduction A new, highly-efficient hydraulic elevator architecture was introduced in [1]. The architecture uses 7 primary components: two variable-displacement pump/motors, two accumulators, a small auxiliary electric motor (not a motor/generator), an actuator, and a reservoir or low-pressure auxiliary accumulator. The main accumulator (Accumulator 1) serves as the main source of fluid for actuation. This connects to the actuator via a main pump/motor (PM1) that shuttles fluid between the actuator and Accumulator 1. A second pump/motor (PM2) connects to PM1 through a shaft and shuttles fluid from a reservoir or low pressure accumulator (Accumulator 3) to a secondary accumulator (Accumulator 2). A small electric machine (EM) serves as a supplemental power source which operates either PM1, PM2, or both to restore lost energy in the system due to system losses such as hydraulic friction, fluid leakage in PM1 or PM2, etc. Control of the system can be achieved by varying the displacements in the PM1-shaft-PM2 assembly (the hydraulic transformer). A visual depiction of the architecture can be seen in Figure 1. FIGURE 1: Proposed Architecture 2015 Fluid Power Innovation & Research Conference (FPIRC15) Research Presentation Extended Abstract Submission This architecture uses the concept of displacement control to actuate the hydraulic system. Furthermore, energy regeneration functionality is achieved through the use of the hydraulic accumulators and therefore avoid an electric generator and its associated inefficiencies and costs. Displacement control avoids fluid throttling, a substantial source of loss in hydraulic systems [2-4]. Together with the use of hydraulic energy regeneration (in lieu of electric energy regeneration), the architecture achieves a highly efficient operation. The variable displacement nature of the hydraulic transformer adds a highly versatile control spectrum to the system. Previous work developed multiple control strategies for the same prescribed motion; the choice of which has been shown to further and substantially impact the overall efficiency [1]. Theory The system in Figure 1 above has three independent control inputs: 1) PM1 displacement ( ), 2) PM2 displacement ( ), and 3) EM torque ( ); the combination of which can be used to control the system satisfactorily.