The industrial application of the ESTOMAD methodology is feasible only if the time and effort needed to develop a reliable Energetic Virtual Prototype are compatible with current product development, which can be very fast, especially for customized production in small-medium enterprises. Since machine productivity, eco-performance and cost are all strongly influenced by the commercial components employed (e.g. motors, sensors, controllers, transmissions, guide ways, …), their numerical models must be available. It is not, however, realistic to expect them to be developed by the machinery builders themselves. ESTOMAD proposes a “co-modeling” approach, coordinated by CNR-ITIA and involving component and subsystem manufacturers in the modeling process: component suppliers furnish, together with their Physical Components, the corresponding Virtual Components, adopting a business model already applied in automotive and military fields. Virtual Component services constitute a powerful added value for European suppliers, advancing in the direction of a knowledge-based economy.
Fig. 1: Virtual compoent business model.
Ball bearings for linear actuators produce high friction losses, as illustrated in the course of the project for FMTC’s badminton robot. The company Leuven Air Bearings (LAB) was therefore contacted with the question whether an alternative bearing technology could provide an efficiency improvement. As a result of these contacts, hydrostatic bearings were proposed. A schematic drawing of one rectangular hydrostatic bearing pad is shown below.
Fig. 2: Schematic drawing of a rectangular hydrostatic bearing pad.
To evaluate (in a virtual way) the energetic effect of this bearing, a virtual component was developed in cooperation with LAB. The analytical formulas that exist for the design and dimensioning of a hydrostatic bearing were started from, to describe how the component works and behaves (specifically with respect to energy consumption). Some simplifying assumptions were made: (1) the pressure of the supplied fluid is constant, (2) the atmospheric pressure is 0, (3) the pressure in the bearing pad is constant over the surface of the pad, (4) the restrictor is a capillary restrictor, and (5) the bearing was dimensioned for maximal stiffness and minimal energy consumption.
Using these assumptions and based on interaction with LAB, the formulas for the two loss components of the bearing could be derived:
(1) the pumping power, that is needed to bring the supplied fluid under pressure:
(2) the friction loss, that opposes the bearing motion:
where q stands for the flow, η for the pump efficiency, h for the distance between the bearing and the surface, O for the total circumference of the bearing pads, C for the bearing land width (C=Cx=Cy), ηvisc for the viscosity of the fluid and V for the velocity of the linear motion. Ffric is the friction force that, based on this equation, can be considered as a purely viscous friction.
The hydrostatic bearing can thus be represented as a combination of a constant pumping power and a viscous friction with a constant friction coefficient. The implementation of this virtual component in LMS.AMESim is shown below.
Fig. 3: Virtual component for a hydrostatic bearing, implemented in AMESim (red: constant pumping power / green: viscous friction).
In the figure below, the evolution of the total energy consumption during a typical linear motion of the badminton is compared (in simulation) for the hydrostatic bearings versus the ball bearings. It can be seen that at the end of the run, the linear motor with the ball bearings has consumed more than twice the amount of energy that it would have used with the hydrostatic bearings.
Fig. 4: Total energy consumption [J] during a typical linear motion of the badminton robot with the original ball bearings (red curve) vs. the hydrostatic bearings (green curve), based on simulation in LMS.AMESim.
The next figure presents a more detailed overview of the contribution of the different loss components to the total energy consumption for both cases. As can be seen, the friction losses are more than 30 times smaller for the hydrostatic bearings, while the energy that the pump consumes is negligible with respect to the other components. A small reduction of the copper losses can also be noticed (lower friction requires lower motor force).
Fig. 5: Comparison of the diferent energy loss storage components [J] during a typical linear motion of the badminton robot with the original ball bearings (left) vs. the hydrostatic bearings (right), based on simulation in LMS.AMESim.
The proposed virtual component can therefore support the optimal selection of the bearing system. The best solution will also depend on the use scenario, given that the hydrostatic solution is penalized by long stand still periods, where the pump still consumes energy.
The virtual component approach has been applied to the industrial Chiller component (refrigerating machine used in very wide range of industrial application). A collaboration with RITTAL (component supplier) has been defined and implemented by CNR-ITIA in order to associate an energetic simulation model to the physical object chiller (Fig. 6). The case of an industrial inverter chiller has been considered.
Fig. 6: Chillers.
The modeling approach for chiller energy consumption estimation is centered on the model of the refrigerant thermodynamic cycle, that provides a representation of the thermal and physical behavior of the refrigerant fluid. The refrigerant flow toward the coolant depends from the refrigerant mass flow rate into the circuit, that is determined starting from the volumetric flow rate at the aspiration of the compressor. A PID controller leads the compressor according to the error between the temperature value at the outlet of the pump of coolant and the chiller set point, imposing the revolutions per minute. The heat absorbed from the cooled application (that represent the refrigerant power of the chiller) is calculated considering its thermal capacity and temperature, and the temperature of the cooling at the inlet. An experimental phase for model characterization is required and can be performed in the component supplier’s laboratory emulating the thermal behaviour of the reference application using variable thermal resistances. Just to recall the Virtual Component philosophy, the idea is to provide a model that can be useful to the user in the use phase of the product: the interesting aspect is not how the component is built, but how the component works. At this level, the constructor parameters cannot be modified (the final product already exists). According with the aim of developing an energetic model, some elements of the chiller and the relative cooling system are represented as thermal entities (e.g. resistances, capacities) and the model results simple in terms formulation and utilization.
Industrial suppliers that are interested in promoting eco-efficient solutions through a Virtual Component approach to the co-modeling of components with their customers, and want to get involved in the Virtual Component approach, can contact us at: