This page summarizes the results of the ESTOMAD project. These results can be divided in three parts:
- The developed methodology
- The software developments (in AMESim)
- The practical application on a number of benchmark cases
The overall scheme proposed by ESTOMAD to carry out and energy efficient design is given in Fig. 1. It has been recognized quite quickly that it is not realistic to model always the complete machine from an energetic point of view, as such modelling can be very time consuming and some machine components might have little effect on the energetic behaviour. Therefore, it has been proposed to perform an initial assessment of the contribution of the different machine modules to the energy consumption of the complete machine and only model those components that significantly contribute to the losses. Such initial assessment can be performed using time-energy measurements on prototypes or using operational machine variants if they exist, measurement data from the field or insights from literature. The detailed analysis is than focused on most energy loosing components. In further steps the remaining loss sources will be addressed. That is why the design process will often be iterative.
Fig. 1: Overview of the ESTOMAD methodology.
More in detail, the following steps can be identified:
- Identification of most relevant phenomena: The first step is to understand and characterize the current situation and to identify most relevant phenomena, i.e. how energy efficient is the actual machine, how much energy is used during the process, how much energy is not crucial for the process.
Based on an evaluation of the energy losses, one can decide on which element the design or re-design should focus. It is good practice to start with the largest losses and the elements that are easy to modify.
- Modeling and analysis of most relevant components: Starting from a rough analysis the modules that mainly contribute to the losses are identified. These modules are than further analyzed in detail, for which more detailed models are used.
- Identification of losses: When models are constructed, the energy flows can be analyzed in detail. These energy flows typically have to be analyses under different usage scenarios, as machines are typically used for various purposes.
- Improved design: When the main loss sources have been identified, ways to minimize the losses can be thought of. Alternative designs can emerge from the experience of the machine designer and ideas for improvements can also come from descriptions of best practices found in literature. The effect of different design modifications can then be analyzed using models of the design alternatives.
- Energy consumption in operation: The preferred way of reducing the energy losses on the machine is then implemented on the machine and its effectiveness can be evaluated.
The AMESim software, developed by LMS Imagine, already existed at the start of the ESTOMAD project. This software provides a complete environment to perform 1D functional numerical simulation using the multiport approach. AMESim is composed on the one hand of a platform, including the User Interface, the solvers and analysis tools, and on the other hand of libraries, which are groups of components each simulating a given physical behaviour (such as a mass) or a subsystem (like a pump or an electric motor). See LMS Imagine.Lab AMESim for more details on AMESim.
Performing an energetic analysis was not possible in AMESim at the start of the ESTOMAD project. To be able to perform such analyses library submodels had to be extended with capabilities (i) to compute power and energy flows between different components, (ii) to compute power and/or energy storage and losses in the components. Also, construction of new models might be needed to be able to implement the benchmarks foreseen in the project. To provide an answer to (i), so called energy and power sensors have been developed and most existing AMESim libraries have been extended with these sensors. Fig. 2 shows the AMESim block of the energy sensor.
Fig. 2: Energy sensor block in AMESim.
To provide an answer to (ii), new Green Design analysis tools have been developed and integrated in the Graphic User Interface (GUI) of AMESim. These tools comprise the following:
- The Energy Management window which enables to quickly visualize and identify all power, energy and activity variables.
- The 2D plot which allows analysing the evolution of powers and energies over time.
- The Animated bar chart which enables to compare different powers or energies in a clear graph (see Fig. 3).
Fig. 3: Animated bar chart in AMESim.
- Power/energy flow charts that provide a clear graphical representation of the powers or energies present in the system during all the operational phases (see Fig. 4).
Fig. 4: Energy flow chart in AMESim.
Various examples and benchmarks have been used (i) to fine tune and evaluate the above listed functionalities and (ii) to show how these capabilities can be used for locating and reducing losses in machine drive trains.
The ESTOMAD methodology and software have been developed and tested, using a number of benchmark cases. Below you can find a selection of the project results that have been achieved on these cases:
- Badminton robot
- Bearing frictional power loss models
- Worm gear: testing and modeling results
- XY positioning system
- Machine tool
One of the benchmark cases is a badmintonrobot. FMTC constructed this badmintonrobot to demonstrate its mechatronic competences (Fig. 5).
Fig. 5: Badminton demonstrator. A movie of the badminton robot in action can be found here.
A visual system with two high-definition black and white cameras is used to localize the fast moving shuttle. The robot itself consists of one linear motor and two rotational motors. The robot was designed in 2009 following a standard engineering approach, that is, without taking into account energy consumption considerations.
In the ESTOMAD project, this high dynamic motion system is being modeled and analyzed with respect to energy consumption to identify the main sources of energy loss and find ways of reduce losses. Fig. 6 shows the electric drive configuration of the robot.
Fig. 6: Electrical drive configuration of the badminton robot.
The energetic behavior of the linear motion has been modeled in Matlab/Simulink and AMESim (see Fig. 7) and the model parameters were obtained from suppliers or experimentally identified on the robot.
Fig. 7: AMESim model of the badminton robot's linear axis.
The badmintonrobot is typically performing point-to-point motions. From the figures below, it can be seen that the friction and the copper losses are the main loss components, with the copper losses being clearly the larger of the two.
Fig. 8: Energy loss components during a standard movement of the linear motor (E_grid indicates the total energy consumed, while E_cu indicates the copper losses).
Fig. 9: Energy loss components during a standard movement of the linear motor (E_cu indicates the copper losses).
Energy efficiency can be increased by taking measures in the design and operation of the robot that work on the following aspects:(1) reduce the friction losses by reducing the friction parameters and the velocity, and (2) reduce the copper losses that are proportional to the square of the current by reducing the acceleration. These reductions can be achieved in the following way:
- reduce the friction parameters by using more efficient bearings;
- adapt the acceleration and velocity by adapting the parameters of the controller; analysis using the AMESim model showed that a reduction of 32% of energy losses can be realized for an increase of response time of 14% only and these findings were experimentally verified
Building further on the second aspect, the idea arose to change the controller strategy from time-optimal to energy-optimal without changing the controller hardware, but just by changing the software settings. In order to do thisthe trajectory parameters in the controller (max. velocity Vmax and max. acceleration Amax) were selected for each motion in such way that the robot reaches the desired end-point just in time. This way the time-optimal robot controller can be transformed in and energy-efficient just-in-time robot controller. To obtain the energy optimal Vmax and Amax parameters as function of the desired travel distance and travel time, multi-objective optimisation optimization algorithms has have been used. Fig. 10 shows the result of this optimalisation.
Fig. 10: Squared control signal as energy measure for comparing time-optimal and energy optimal approaches for identical test motions..
More details on this research can be found in following paper:
- J. Stoev, G. Pinte, and W. Symens, Time-constrained energy-optimal motion control - Application to a Badminton Robot. The 13th Mechatronics Forum. Linz, Austria, September 17-19, 2012.
In ESTOMAD, a methodology has been developed to verify and update the different loss models to make them suitable for accurate estimation of frictional power loss. On this page you can find a brief description of the developed methodology applied to the analysis of a needle roller bearing.
Recent decades have witnessed growing global energy concerns. In industry, higher energy efficiency helps to save non-renewable energy, reducing production costs, thereby making finished goods cheaper. Every machine or equipment is designed with several bearings. Although bearings are known as anti-friction elements, a typical bearing power loss of over 100W is quite common and as such, bearings are often key contributors to the overall system frictional power loss.
One approach to increase the energy efficiency of the overall system is hence to reduce the power loss in key machine components such as bearings. This can be achieved by accurate estimation of the losses followed by improved design. Estimation of the bearing frictional power loss can be achieved relatively quickly by considering advanced bearing power loss models, instead of time consuming and costly measurements.
In literature, a number of state-of-the-art models for bearing loss prediction are available. The most commonly know ones are the Palmgen and SKF® models. Within ESTOMAD, a methodology has been developed to verify and update these models to make them suitable for accurate estimation of bearing friction power loss.
The methodology is briefly described below:
- Conduct experiments to obtain the bearing frictional power loss information using a dedicated set-up
- Asses the existing state-of-the-art bearing frictional power loss models by a correlation study between data obtained by simulations using models and those obtained by experiments
- Update the model parameters using a model updating technique to make the models more suitable for the bearing under investigation
- Validate the developed model with different test cases to check its robustness
This methodology is verified for the needle roller bearing as an application case. Some key results are shared below:
Fig. 11 (a) shows the bearing test setup. It consists of an AC motor, which rotates the test bearing via a coupling with an electro motor. The test bearing is placed on an experimental head supported on a hydrostatic bearing. The experimental head is located between two support bearings. A radial load is applied to the test bearing, acting through the hydrostatic bearing. Fig. 11 (b) shows the obtained power loss (W) in function of speed (rpm) for different constant loading conditions between 1-8 kN.
Fig. 11: (a) Test Setup [Courtesy EC Engineering], (b) Experimental results.
Fig. 11 (a) shows that there is a quite big difference between the measurement results (dotted surfaces) and the simulation results (coloured surface) using a bearing frictional power loss model with initial, catalogue parameters. In this case the Palmgren model is shown. The model is then updated, based on a non linear least square surface fitting technique to make it more suitable for the considered bearing. A data set is used of constant load and constant speed measurements. The results with the updated model are shown in Fig. 12 (b). Very similar results are obtained when updating the SKF® model (see Fig. 13).
Fig. 12: Correlation between experimental results and results obtained by simulations using (a) Initial model (b) Updated model.
Fig. 13: Frictional power loss for standard cylindrical roller bearing estimated by updated (a) Palmgren and (b) SKF® models.
To validate the updated Palmgren and SKF® models and to check their robustness, the models are compared with a second set of experimental results with a varying speed profile (while the models themselves were updated with measurements at constant speed levels) as shown in Fig. 14 (a). Fig. 14 (b) shows that a good correlation is achieved between the simulation results using the updated models and the measurements.
Fig. 14: (a) Correlation between experimental results (varying speed profile) and results obtained by updated models (b) Varying speed profile.
The updated models can be used for energy efficiency analysis of systems and sub-systems such as drive trains and gearboxes. Also, the developed methodology is applicable for other bearing types/components.
More details on this research can be found in following paper:
- Shoaib Iqbal, Jan Croes, Farid Al-Bender, Bert Pluymers, Wim Desmet; Frictional power loss in solid grease needle roller bearing, Wiley: Lubrication Science, doi: 10.1002/ls.1195 (June 2012).
The energy consumption of a worm gear mechanism, which is a widely used component in industry, is analyzed. Compared to traditional gearboxes, a worm gear mechanism is in general characterized by a lower value of efficiency and is, according to manufacturers, situated between 30 and 90% (see Nord catalogue). However, the efficiency level can differ depending on the specific conditions of operation (output torque, input rotational velocity, ambient temperature and type of lubrication, etc.). In order to characterize the efficiency of worm gear mechanism in mechanical systems, it is therefore essential to establish the actual efficiency value for the conditions of a specific application.
Efficiency tests have been performed for a worm gear mechanism that is part of a tram pantograph system, where such gearbox contributes mostly in the total power loss. The worm gear has been tested in an open loop configuration, and the efficiency has been determined based on input and output torque and rotational velocity measurements. The tests have been performed for various output torques and input velocities to obtain a view on the losses in the worm gear mechanism under different operational conditions. The results of measurements are for modeling the worm gear component in a 1D simulation environment.
The experimental campaign was performed on a test rig specially built for this particular application according to the diagram shown in Fig. 15. Fig 16 shows a picture of the set-up.
Fig. 15: Test rig elements.
Fig. 16: Complete test setup used for worm gear experiments.
In this configuration the energy is transferred from the electric motor to the powder brake through the worm gear and a system of couplings. Within the test the following quantities have been measured: rotational velocity on the input shaft (obtained directly from the VFD controller), torque on the input and output shaft and oil temperature in the worm gear housing.
A part of measurement results are presented in Fig. 17.
Fig. 17: Efficiency of worm gear as a function of output torque for different rotational velocities.
Losses in the worm gear mechanism can be divided into three main types: tooth engagement losses, oil churning losses and bearing losses. The overall power loss is equal to the sum of these components. All loss types have been modeled in the AMESim environment based on mathematical models available in literature and experimental data obtained from measurements. This resulted in a reliable numerical model that was used for analyzing the energetic behavior of the tram pantograph system. Fig. 18 represents the model in the form of an efficiency map.
Fig. 18: Efficiency map representing the efficiency of the worm gear.
A pantograph system, which is constructed by EC Engineering has been used as a test case within the ESTOMAD project.
The complete system is quite complex from a kinematic point of view. The initial drive mechanism of the pantograph consists of a dc motor, a worm gear and a lead screw (see Fig. 19). This mechanism is analyzed from an energetic point of view as problems were encountered with the drivetrain during long life time test (thousands of up/down motion cycles are carried out during these tests). The motor gets overheated due to high losses induced in the transmission components.
Fig. 19: 3D visualization of pantograph system.
Modeling was done in the ESTOMAD 1D environment (AMESim), and results of a multibody analysis were used as inputs. The numerical model of the pantograph is presented in Fig. 20.
Fig. 20: AMESim pantograph model.
The analysis of the pantograph drive system revealed that high power losses occur mainly in the gear reduction (worm gear) and the transformation from rotary to linear motion (lead screw), as can be seen in Fig. 21. The overall efficiency for this system is about 20 %.
Fig. 21: Components contribution in the overall power loss.
Measurements on system level and component level (worm gear) were carried out to validate the simulation results.
The results of the pantograph mechanism investigation led to the development of a new actuation mechanism for the system. In the new concept, the lead screw and worm gear have been replaced with a ball screw and a spur gear reducer respectively, since these components were identified as most inefficient. Compared to the initial prototype, the principle of operation of the new drive is similar. The rotary motion is generated in the dc motor and transmitted through the reducer to the ball screw. The rotational motion is transformed to a linear one and outputted to the pantograph.
The choice of the optimal drive solution was supported by the results of a model-based analysis that was carried out for this solution in AMESim. Fig. 22 shows the AMESim model of the new system.
Fig. 22: AMESim model of the altered drive system.
The efficiency parameters for the new components of the drive system have initially been taken from manufacturers’ catalogues, available literature and general experience. For this initial model of the new concept the following parameters have been applied:
- dc motor efficiency: 90%
- reducer efficiency: 70%
- ball screw efficiency: 65%
TThe results from initial simulation (with assumed parameters) revealed a system efficiency of about 40%. As a conclusion, the modifications introduced to the system proved to be beneficial.
The pie chart below (Fig. 23) represents the contribution of the individual components of the new drive system in the overall power loss.
Fig. 23: Components contribution in the overall power loss (for the new drive system).
A prototype of the new concept has been made (see Fig. 24 for a close up).
Fig. 24: New concept of a tram pantograph system.
A measurement campaign was performed to verify the energy consumption of the new design and validate the outcome of the model-based analysis. These measurements revealed an efficiency of approximately 50%. The difference between the simulation and experimental results mainly originates from simplifications of the numerical model (i.e. neglected dynamic behavior of the system and friction in the kinematic joints). Updating of some model parameters using the experimental data however allowed improving the simulation results.
Table 1 shows that the new concept is a great improvement in terms of energy efficiency (over 50% drop of power consumption). The table contains both measured and simulated results (coming from the initial and the correlated model).
Table 1: Comparison of energy efficiency of the initial and the new pantograph system for the lowering phase.
This system is broadly used in manufacturing processes for tasks like e.g.: pick and place, palletizing, indexing, positioning for sorting, applying glue, welding or part inspection. Some examples of XY positioning systems are shown in Fig. 21.
Fig. 21: Some examples of XY positioning systems.
This system has been used as a demonstrator example for the ESTOMAD project results, during the final ESTOMAD workshop. A detailed text and accompanying movie have been prepared for this system: detailed text XY positioning system, movie XY positioning sytem.
Machine tools represent one of the main reference applications inside the ESTOMAD project.
In the industrial sector and producing machinery, machine tools are high energy consuming products. Increasing attention is focused on:
- design of energy efficiency machines;
- optimization of existing machine from the energy point of view;
- improvement of strategies at the operation level for minimizing the energy consumption during the use phase.
The machine tool can be seen as a modular entity, an assembly of elements (drive units, cooling systems, hydraulic unit, electrical cabinet, etc.) that all contribute to the global energy consumption of the machine.
Firstly, analyses based on experimental measurements have been executed, providing an energetic characterization (map) of two different typologies of machine tools: Jobs LINX machining center, equipped with electrical linear motors (Fig. 25), and Jobs GRANDSPEEDER machine center, equipped with electrical traditional rotation motorization (Fig. 26).
Fig. 25: LINX machine.
Fig. 26: GRANDSPEEDER machine.
The measurement setup and the physical measuring points selected on the machines are shown in Fig. 27.
Fig. 27: Measuring point for the machine tool measurement campaign.
The energy consumption measured on the LINX machine during a test cycle is reported in Fig. 28. It can be noted that more than 50% of the energy is required by the auxiliary subsystems.
Fig. 28: Energy consumption measured on LINX during a test cycle.
An energetic model of the machine tool has been made in AMESim: it consists of an assembly of the machine element (subsystem and component) energy models. A graphical representation of part of the proposed model is shown in Fig. 29. By simulation of this model, the energy consumption of the machine tool can be predicted.
Fig. 29: Energetic model of the machine tool.
To validate the model, various validation have been conducted: the simulation results have been compared with results from a measurement campaign on the machine executing a test cycle. The good fit between the model and the experiments can be noted from the results for LINX machine shown in Fig. 30.
Fig. 30: Measured and simulated power and energy at the Main measure point for LINX standard.
In order to reduce the energy consumption, various improvement actions have been implemented on the machines, mainly by substituting a number of auxiliary system component with more efficiency ones. In particular:
- the use of an inverter chiller (inverter-compressor + fan with inverter-motor + high efficiency coolant pump);
- use of high efficiency hydraulic unit equipped with an accumulator.
Resulting from this, an eco-efficiency configuration of the LINX machining center (the so-called ‘ESTOMAD version’) has been defined and realized. The energy saving obtained using the LINX machine with eco-efficiency configuration with respect to the LINX machine in standard configuration is shown in Fig. 31.
Fig. 31: 'Eco-LINX' energy saving.