6th International DAAAM Baltic Conference INDUSTRIAL ENGINEERING 24-26 April 2008, Tallinn, Estonia MODELLING AND SIMULATION OF A HYDRAULIC LOAD- SENSING SYSTEM IN THE COCOVILA ENVIRONMENT Grossschmidt, G. and Harf, M. Abstract: In this paper we discuss the design of a hydraulic-mechanical loadsensing system using modelling and simulation in an intelligent programming environment. The approach is based on using multi-pole models and signal-flow graphs of functional elements, enabling methodical, graphical representation of mathematical models of large and sophisticated fluid power systems. Modelling and simulation of separate objects, subsystems and the whole system are discussed. Keywords: Fluid power system, multi-pole modelling, intelligent programming environment, simulation. 1. INTRODUCTION Fluid power systems, in which working pressure (pressure in pump output) is kept proportional to load, are called hydraulic load-sensing systems. Such systems are mainly used in mechanisms containing numerous drives to run with the purpose to save energy. Hydraulic load-sensing systems are automatically regulating systems with number of components and feedbacks. A very precise parameter setting is required to make the system function. An intelligent programming environment CoCoViLa [ 1 ] providing automatic program synthesis and visual programming features is used as a tool in modelling and simulation. In the paper we consider modelling and simulation of steady state conditions. This enables to set up system configuration and parameters. The results of the simulations are as initial data for further simulations of dynamics. 2. MULTI-POLE MODELS Generally, multi-pole models [ 2, 3 ] represent oriented mathematical relations between pairs of input and output potential and flow variables of functional elements. The multi-pole models take into account the signal propagation in both directions as it occurs in hydraulic and mechanical systems. When composing a model of the fluid power system, multi-pole models of necessary components must be connected through poles. Composing of multi-pole models enables one to find the best model, where all the oriented causalities of objects are graphically settled. The mathematical dependences between variables of functional elements (e.g. various hydraulic flow control and pressure control valves) are convenient to express as oriented graphs. 3. HYDRAULIC-MECHANICAL LOAD-SENSING SYSTEM In Fig. 1, the variable displacement axial piston pump is driven by an electric motor M. Hydraulic-mechanical control of the pump volumetric flow is performed by control valve and positioning cylinder. The feeding chain of the hydraulic motor R Verbr contains tube R L-zu, measuring valve R VW with pressure compensator R IDW and check valve, meter-in throttle edge R SK-zu and connection elements.
R IDVW p 0 = const Fig. 1. Scheme of the hydraulic mechanical load-sensing system The output chain of the hydraulic motor R Verb contains a meter-out throttle edge R SK-r, and tube R L-ab. The device contains load-sensing pressure feedback. Feedback pressures have been taken directly from the measuring valve with pressure compensator R IDVW. 4. COCOVILA PROGRAMMING ENVIRONMENT CoCoViLa is an intelligent programming environment, which supports declarative programming in a high-level language, automatic program synthesis and visual programming. CoCoViLa is implemented in the Institute of Cybernetics at the Tallinn University of Technology, since 2005. The CoCoViLa environment is Java based, free and platform-independent. CoCoViLa supports a language designer in the definition of visual languages, including the specification of graphical objects, syntax and semantics of the language. CoCoViLa provides the user with a visual programming environment, which is automatically generated from the visual language definition. When a visual scheme is composed by the user, the following steps parsing, planning and code generation are fully automatic. The compiled program then provides a solution for the problem specified in the scheme, and the results it provides can be fed back into the scheme, thus providing interactive properties. Automatic synthesis of programs is a technique for the automatic construction of programs from the knowledge available in specifications. Having a specification of a class, we are, in general, interested in solving following problems: find an algorithm for computing the values of components y1,..., yn from the given values of components x1,..., xm; find an algorithm for computing the values of all components that can be computed. The automatic synthesis of programs is based on proof search in intuitionistic propositional logic. From a user s point of view the CoCoViLa framework consists of two components: Class Editor and Scheme Editor. The Class Editor is used for defining models of components of schemes as well as their visual and interactive aspects. The Scheme Editor is a tool for the language user. It is intended for developing schemes and for compiling (synthesizing) programs from the schemes according to the specified
semantics of a particular domain. The environment generated for a particular visual language allows the user to draw, edit and compile visual sentences (schemes) through language-specific menus and toolbars. Having developed the visual language we are able to load it in the Scheme Editor and build schemes by putting visual objects on the drawing canvas and connecting them through ports. The scheme editor is fully syntax directed in the sense that the correctness of the scheme is forced during editing: drawing syntactically incorrect diagrams is impossible. When the visual classes have been built by software developers who must understand the problem domain as well, the language user need not be a software expert, but can work on the level of visual programming, arranging and connecting objects to create a scheme. Fig. 2. Multi-pole model of the hydraulic-mechanical load-sensing system Load-sensing system components: VP - Displacement of the control valve; resh, resg Hydraulic resistors; RVP - Meter-in throttle edge of the control valve; IEH1-3, IEH2-2m - Hydraulic interface elements; ZV - Positioning cylinder; REL Constant resistor; RVT - Meter-out throttle edge of the control valve; PV - Variable displacement pump; ME - Electric motor; tubeh, tubeg - Tubes; RIDVWlin - Measuring valve with pressure compensator; RSKZ - Meter-in throttle edge for hydraulic motor; MH - Hydraulic motor; RSKA - Meter-out throttle edge for hydraulic motor. Simulation task components: dynamic Process Process organizer; constant Source, dynamic Source Sources; Graph drawings.
5. COMPOSING MODEL OF THE FLUID POWER SYSTEM Simulation of steady state conditions in the NUT environment has been considered in [4]. Here, the advanced simulation of steady state conditions in the CoCoViLa environment is considered. The multi-pole model (Fig. 2) represents the scheme of the load-sensing system (Fig. 1). To build up the multi-pole model it is necessary to decompose the scheme of the load-sensing system into logical components and subsystems. Models of the following components of the load-sensing system have been developed: electric motor, variable displacement axial piston pump, hydraulic-mechanical controller, valve block, hydraulic motor, tubes and multiple tube connection elements. Model of the hydraulic-mechanical controller includes models of control valve, meter-in throttle edge, meter-out throttle edge, constant resistor and positioning cylinder with swash plate. Model of the valve block includes a measuring valve with pressure compensator and check valve, meter-in and meterout throttle edges of the hydraulic motor. The multi-pole model of the whole loadsensing system has been built up using the components models. First, necessary components have been connected through poles. Second, variables of connection poles have been defined as inputs or outputs for each component depending on required causalities [ 2 ]. 6. SIMULATION AND DESIGN STEPS The most important stage in design of a hydraulic load-sensing system is parameters specification. In order to help designer to find better solutions, computer simulation technology is proposed that includes following steps. First, hydraulic motor and hydraulic pump parameters must be chosen. Second, the fluid and its properties must be chosen. Cinematic viscosity and density depending on the pressure are calculated at each step. Third, initial approximate values of pressures and pressure drops for pump control must be set up. In our case maximum working pressure p max =250 bars, pressure drop in measuring valve Δp IDW = 5 7 bars, pressure drop in measuring valve with pressure compensator Δp IDVW = 14 19 bars, pump control system feeding pressure p 0 = 60 bars and pump control pressure p CP = 7 20 bars have been taken. Fourth, maximum displacements of the valves must be set up. Fifth, all the models of components must be tested separately. For each component the simulation problem must be composed and input signals must be chosen. Behaviour of the component must be simulated. Initial approximate components parameters values (stiffness of springs, geometry of working slots of valves, etc.) must be refined as a result of simulations. Sixth, the separately tested components models must be connected into more complicated subsystems, tested in behaviour and refined if necessary. Seventh, model of the whole load-sensing system must be built up and simulation tasks must be solved. If possible, steps 5 7 should be supported by testing on pilot models and using the results for refining the parameters and models for further simulations. 7. SIMULATION OF A SUBSYSTEM The hydraulic-mechanical controller includes constant pressure feeding that enables to make feedback independent of hydraulic pump pressure. The simulation task description of steady state conditions of the hydraulic-mechanical controller is shown in Fig. 3.
8. SIMULATION OF THE LOAD-SENSING SYSTEM Fig. 3. Simulation task description of the hydraulic-mechanical controller Notations: VP - Displacement of the control valve; RVP - Meter-in throttle edge of the control valve; ZV - Positioning cylinder; REL Constant resistor; RVT - Meter-out throttle edge of the control valve; IEH1-3 - Hydraulic interface element; p1p2 Pressure difference calculator; static Process Process organizer; constant Source, static Source Sources; Graph drawing. Simulation results are shown in Fig. 4. The hydraulic-mechanical controller has been simulated in the case the pressure difference applied to the control valve changes from 14.1 to 19.1 bars. The simulation task for steady state conditions of the load-sensing system is shown in Fig. 2. Dependences on the displacement of the directional valve from 0.0024 to 0.0068 m (the load moment of the hydraulic motor is constant 65 Nm) are calculated (Fig. 5, 6). In Fig. 5, displacement of the control valve (3) drops from 0.003 to 0.00025 m, pump control pressure (1) drops from 19 to 5 bars and pump volumetric flow (2) increases from 0 to 0.0015 m 3 /s. Fig. 5. Simulated graphs of the system In the ideal case all the dependences should be linear. As the shape of the graphs depends mostly on the passage areas of the throttle edges of the valves it is very difficult to achieve the exact linearity. Fig. 4. Simulated graphs of the hydraulicmechanical controller Then the control valve moves from 0 to 0.00275 m (2). Pressure interval for pump volumetric flow regulation from max to min (3) is 4 18 bars. Volumetric flow to the control valve (1) increases from 0.000025 to 0.0000425 m 3 /s. Position angle of the pump swash plate (4) drops from maximum 0.3264 to 0 rads. Fig. 6. Simulated graphs of the system In Fig. 6, angular velocity of the hydraulic motor (3) increases from 0 to 224 rad/s. Output power of the electric motor (4) in the case of zero volumetric flow (see (2) of Fig. 5) of the pump is ~600 W. For pump control we need ~300 W electric motor power. Increasing the volumetric flow cause the output power of the electric motor to grow until 22400 W.
The efficiency coefficient of the hydraulic pump (2) rapidly rises to 0.90. The efficiency coefficient of the whole loadsensing system (1) rises rapidly to 0.55 and then rises slightly to 0.70. 9. SIZE AND COMPLEXITY The package for modelling and simulation of steady state conditions of the loadsensing system contains: 34 classes, including 24 load-sensing system component classes; more than 1000 variables; 11 variables that have to be iterated during the computations; 52 links between system components. Java code of the automatically constructed algorithm for solving the simulation task of the load-sensing system contains 2892 lines and includes 4 algorithms for solving subtasks. Simulation time is 2.84 s. 10. CONCLUSIONS The main features of the approach proposed in the paper are as follows. Mathematical models of the functional elements are composed as multi-pole models taking into account signal propagation in both directions. Used multi-pole models can have various causalities. The mathematical model of the fluid power system contains models of components and carries the full information about connections of input/output variables, which express the considered mathematical causalities and guarantees the completeness of the model. Using the CoCoViLa programming environment enables one in flexible way compose and experiment with various large and complicated models. Automatic program synthesis allows fast describing and solving a great number of simulation tasks. Simulation is performed step by step, starting from simulation of components and moving to more complicated systems. Iteration methods are used in cases of loop dependencies that may appear between component models when they are connected together into more complicated ones. As a result of the current research, a simulation system is proposed that enables one to perform fast computer experiments at the first stage of design. The approach is original and there is difficult to find similar works. Acknowledgement: This research was in part supported by the Estonian Science Foundation (Grant No. 7091). 11. REFERENCES 1. Grigorenko, P., Saabas, A. and Tyugu, E. COCOVILA Compiler-Compiler for Visual Languages. Proc. of the 5th Workshop on Language Descriptions, Tools and Applications, Edinburgh, 2005, v. 141, n. 4 of Electron. Notes in Theor. Comput. Sci., pp. 137-142. Elsevier. 2. Grossschmidt, G. and Vanaveski, J. Causality of Mathematical Models of Technical Systems. Proc. of the 12th European Simulation Multiconference ESM 98, June 16-19, 1998, Manchester, United Kingdom, pp. 191-195. 3. Grossschmidt, G., Vanaveski, J. and Harf, M. Simulation of Hydraulic Chains using Multi-pole Models in the NUT Programming Environment. Proc. of the 14th European Simulation Multiconference ESM 2000, May 23-26, 2000, Ghent, Belgium, pp. 709-713. 4. Grossschmidt, G. and Harf, M. Design of a Hydraulic-Mechanical Load-Sensing System using Object-Oriented Modelling and Simulation. In: 21st European Conference on Modelling and Simulation ECMS 2007, June 4-6, 2007, Prague, Czech Republic, pp. 383 390.