Final Report Summary - ASTERICS (Ageing and efficiency Simulation & TEsting under Real world conditions for Innovative electric vehicle Components and Systems)
Executive Summary:
In September 2009, both the European Union (EU) and G8 leaders agreed that CO2 emissions must be cut by 80% by 2050 if atmospheric CO2 is to stabilize at 450 parts per million – and global warming stay below the safe level of 2°C. But 80% decarbonisation overall by 2050 may require 95% decarbonisation of the road transport sector. With the number of passenger cars set to rise to 273 million in Europe and 2.5 billion worldwide by 2050, this is not achievable by improvements of traditional internal combustion engines or alternative fuels: the traditional combustion engine is expected to improve by 30% in comparison to 2009, so achieving full decarbonisation is not possible through efficiency alone. It is also uncertain if large amounts of (sustainably produced) biofuels - i.e. more than 50% of the demand - will be available for passenger cars, given the potential demand for biofuels from other sectors, such as goods vehicles, aviation, marine, power and heavy industry. Combined with the increasing scarcity and cost of energy resources, it is therefore vital to develop a range of alternative technologies that will ensure the long-term sustainability of mobility in Europe. Electric Vehicles (EVs) are the most promising technology for a drastic reduction of the environmental burden of road transport, i.e. emissions of CO2, air pollutants and noise of particularly passenger cars and light commercial vehicles. In December 2008, a new regulation was adopted which nominally strives to reduce the average CO2 emissions from new cars to 130 g/km by 2015, which nowadays can be said is satisfied with huge efforts on the combustion engine, weight optimization and first mild hybrid concepts. Significantly the law added a 95 g/km target for 2020. There were several concepts for PEV (pure electric vehicle) and HEV (hybrid electric vehicle) at project start available that support this “clean mobility” demand. However, the development and improvement of the different concepts required and still require a huge effort in analysis, design, implementation and testing, not to forget feeding back experience, results and knowledge to design new generations of such cars. Advanced modeling tools and testing procedures going from one-dimension to three dimensional approaches played a fundamental role in the optimization process during the earliest project phases for the energy dimensioning of PEV & HEV as well as their “energy management strategies”, while reducing the project’s development lead-time as well as to build up requirements for subsystems and their related control units. Deep and thorough investigation of the ASTERICS’ simulation tool suite and testing procedures were done to deliver these functionalities. ASTERICS hence contributes to a better performance and thus customer acceptance of PEV’s by improving modeling and testing tools needed for the development of future PEVs throughout Europe. Primary goal was to develop a “full next generation” virtual vehicle model taking different existing vehicles and their use by customers as a reference. All subsystems and components of the EV are the base for other ‘next generation’ vehicle archetypes, also different from the one studied within ASTERICS. In this way, ASTERICS supports the competitiveness of the automotive sector in all its aspects: basic components, integrated components, sub-systems, algorithms, systems and OEMs applications.
Project Context and Objectives:
The concept of the ASTERICS project comprises a systematic and comprehensive approach for the design, development and testing phases of E-drivelines in Pure Electric Vehicles, which is based upon four major building blocks. The four building blocks, their features and the major innovations are as follows:
1. Real world environment and conditions based drive cycles
Real world environment, conditions and fleet data from other projects set the baseline by defining drive cycles, usage profiles, the real world use of the EVs and customer demands. This information was used to identify and specify representative operating conditions, performance requirements and constraints for EV components, as well as stress scenarios, use cases and assessment criteria.
2. Advanced testing methodologies and models for E-driveline components
Enhanced test methodologies using Software, Model and Hardware in the Loop (SiL, HiL, MiL) environments and advanced models capable of simulating the real world behavior with the required accuracy and calculation speed together with sophisticated test approaches (like DoE or online-adaptions) enable standardized, accurate and time-efficient testing of E-driveline components in different environments (Office, test-bench, chassis-roller, ...) and different development phases.
3. Descriptive/ predictive models for battery subsystem, power electronics and electric motor
Standardized procedures for battery ageing and models describing ageing effects and energetic approaches enable the comparison of batteries of the same type and models, LCA (Life Cycle Analysis) , LCC (Life Cycle Costing) as well as the identification of second life applications. Tests are done on test benches rather than in the field in order to determine models that describe ageing and other effects regarding mechanical, thermal and electrical behavior.
Standardized test procedures for power electronics and enhanced inverter models (e.g. based on MOS-FET) were developed which include thermal simulation of the environment and different materials of inverter components accurately that describe a wide range of power classes and efficiency of the components (Si, SiCa, GaN ). Advanced testing approaches for e-motor ageing (e.g. windings) and models describing mechanical, thermal and electric behavior in combination with environmental models were developed - 1D models for fast real-time simulation and 3D for accurate drive cycle in-depth timescale studies as well as linkage between construction parameters and model parameters was created.
4. Total system (e-driveline and EV)
Standards for the connection/integration of high-fidelity subsystem models by means of components and interfaces, which take into account existing standards (like Modelisar-FMI), enable accurate simulation, calibration and optimization of the entire drive train in terms of vehicle performance and energy efficiency, and the usage/ verification together with real world drive cycles in early stages of the development process were adapted.
The proposed approach a seamless model based development process from design to validation, has the largest potential to arrive at optimal Electric Vehicles, in terms of performance (range, speed, reliability, durability and efficiency), reduces development and testing times and related costs.
The overall objectives of the ASTERICS project were:
• to develop a systematic and comprehensive approach for the design, development and testing phases of E-drivelines in Fully Electric Vehicles;
• to reduce the overall development time and testing efforts for EV and EV components by 50% compared to the current time and efforts;
• to enable improvement and optimization of the overall efficiency and performance of electric vehicles by at least 20% compared to existing/known concepts.
Scientific and technological objectives of ASTERICS were:
1. Real world environment and condition based drive cycles
To define a set of real-world drive cycle usage profiles, the real world use of EVs and customer demands for EVs based on analyses of existing fleet data of electric vehicle prototypes. Since existing driving cycles are conceived to assess the characteristics of ICE/HEV vehicles, their capability to assess the performances (e.g. energy consumption, durability) of EV was investigated. Such analyses take into account specific EV issues, such as regenerative braking capabilities and the relationship with driver skills in order to analyze component stress (e.g. effects of peak currents on batteries) and evaluate the overall performance (e.g. range, energy consumption). Analyses further identified clues for advanced vehicle design (e.g. advanced power systems including flywheel/ super capacitors/ range extender devices) which in turn helped to build up a complete “use case” scenario, including information about charging strategy during vehicle life. Such assessments are needed for the evaluation of typical battery Depth-of-Discharge (DoD) and the related battery State-Of-Health (SoH) expected during vehicle life.
To describe real-world contexts, different driving cycles were provided, including urban/extra-urban/motorway cycles or specific applications, such as freight delivery and passenger public services.
2. Advanced Testing methodologies
Test procedures that automatically populate the component simulation models for Battery, Inverter and E-Motor were developed, which can be used in all phases of the development process in MIL (Model in the Loop), SIL (Software in the Loop), HIL (Hardware in the Loop) and other test environments.
Focus in the development of these test procedures was:
a) on autonomous running testing procedures, which take into account limitations of battery and/or other involved units without interruption.
b) on automatically adaptive procedures for batteries with respect to the need of the model building algorithms.
c) on finding appropriate combinations of test conditions and stress factors influencing the battery’s characteristic (including tolerances between individual cells, self-discharge rate, open-circuit voltage, isolation resistance, mechanical deformation/expansion) and their changes over life-time – especially on cell-level.
d) on the identification of stress factors that influence the ageing behavior of inverters (thermal, electric factors,..) and set up tests with sophisticated methods (e.g. DoE).
e) on the identification of components that have an influence on ageing effects of Inverters (e.g. MOS-FET, DC-link capacitor) for different topologies and compositions.
f) on the definition of influencing factors, needed measurement accuracy and dynamic resolutions (e.g. phase current, torque, efficiency, fluxes...) for different E-Motor types and models (SRM – switched reluctance machine, PMSM – permanent magnet synchronous machine, IM – induction machine).
g) to develop procedures for accelerated ageing of Battery, Inverter and E-Motor based on the knowledge of S/T objective 1, in order to shorten the testing time for this kind of tests.
Main aspects of these test procedures lay in the identification of stress factors and their relevant contribution to the accelerated ageing (plus degradation and abuse) of components as well as in the definition of standard accelerated ageing tests for EV components based on the knowledge about ageing/stress influencing factors. The development and set-up of a test system that is able to perform parallel (multi-channel) battery cell tests, in order to run different test procedures with different test excitation sequences on the same type of cells with higher accuracy and less energy consumption as known today is another aspect here.
3. Electric Sub-system models developments
Several accurate high fidelity model for Batteries, Inverters and E-Motors considering all relevant dependences from inputs/parameters were developed, valid for the whole life-cycle of the E-driveline components. Main aspects that were covered in the course of the battery model development were
a) to identify the influence of parameters like capacity, impedance, tolerances between the individual cells, self-discharge rate, temperature, open-circuit voltage, isolation resistance, mechanical deformation/expansion, etc...
b) to describe the dependence of battery performance, electrical range, internal resistance, power acceptance from “age” of the battery.
c) to establish a standard definition of “battery age” by means of an appropriate set of parameters to enable the comparison to other batteries of the same type.
d) to integrate the battery model into a thermal sub-system describing the cooling/heating transfer-rate with the sub-system.
e) to design a real-time capable representation of the battery model needed for real-time purposes like usage in battery simulator or entire vehicle models.
Concerning Inverter models emphasis was put on the development of a computationally fast model, covering different inverter topologies and compositions and that they can be used in higher power ranges (150kW) and higher speed drives (22000 rpm) as usual. The evaluation and identification of thermal and ageing effects as well as influences of component tolerances to the life-time behavior of Inverters was another important aspect that was covered.
A major aspect in the development of E-Motor models was to identify/ parameterize E-Motor models by transferring construction parameters automatically into the model description. Further the influence of combined stress factors and the subsequent identification of expected loss-of-life damaging models was determined and the dynamic behavior of different types of E-Motors (SRL, SPMS, ASM) with computationally fast models, including thermal, electrical, ambient and mechanical effects was accurately described.
4. Total system (E-driveline) models integration and validation on test bench
To transfer drive cycle data from the field testing (S/T objective 1) to the duty cycle on the test benches emphasis was put on the development of procedures to more adequately characterize the real world performance of the (aged) components on the test bench.
To integrate models of the components into an overall simulation environment it was important to define interfaces and components in order to exchange and co-simulate component models between different tool-environments (AMESIM, CRUISE, PERFECTS). Models (e.g. Battery model for Battery Simulator) were integrated into the test bench environment and tests on test benches were performed in order to evaluate the quality of the simulation models as well as the application of real-life testing scenarios in simulation environments and the verification of possible efficiency and performance improvements by means of modifying/ exchanging components in the e-driveline.
The whole design and testing process for EVs and e-drivelines in ASTERICS was supported in such a way that the specific requirements and demands (efficiency, performance, life-time, stress-scenarios, etc...) can be fulfilled leading to high quality, reliable, more efficient, optimized electric vehicles in a 50% reduced development time.
Project Results:
The aim of ASTERICS was the development of a systematic and comprehensive approach for the design, development and test phase of next generation electrical vehicles that satisfy the requirements of increased performance, durability and safety. This was a very ambitious goal since it requires not only the development of accurate and real-time capable stand-alone models of all electric components but a complete vehicle simulation that allows for the co-simulation of different models from different vendors.
ASTERICS covered the development of realistic driving cycles and highly accurate simulation models for batteries, e-motors and inverter from the investigation of appropriate test procedures, equipment and tools through to a co-simulation platform that enabled project partners to integrate and co-simulate their models in their specific vehicle simulation environments.
Realistic driving conditions for EV’s
Within the last decade a lot of work was invested to collect data in order to derive real-world driving cycles that are supposed to representative the typical use of cars in European. Such a driving cycle can be considered as a standardized procedure to evaluate vehicle performance in a reproducible way under laboratory conditions. These can include simulation environment, power–adsorbing chassis dynamometer and/or testbeds. It is therefore clear that a single “driving cycle” cannot represent all the possible conditions under which the vehicle could be used during its entire lifecycle which is why some compromises in designing such cycles have to be made. A test procedure has to include a time–vehicle speed signal and a large set of boundary conditions.
At this point it has to be mentioned that EVs introduce new parameters for the evaluation of their performance and are affected by specific, new criticalities in comparison with conventional vehicles. The possibility of energy recovery during breaking for example could cause drivers to modify their way of driving in order to optimize the energy consumption. The limited range can cause drivers to display a particularly smooth, benign driving style. Some specific conditions (e.g. occasionally high daily distance driven, unavailability of charging points) are in particularly determining the so–called “range anxiety” phenomena and a different drivability leads further to a different perception of vehicle performances.
The way things are at the moment there are about 150 driving cycles available for vehicle type approval and available for tests in scientific literature, but only a few have been specifically built up for EVs. The above mentioned considerations were the driver for the development of a new set of reference cycles within ASTERICS. Data was obtained within the frame of three case studies in three different cities using different vehicles:
• in the city of Turin the data acquisition was performed using a Light-Delivery electric Vehicle (Iveco Daily EV, below 3.5t) running over a predetermined route consisting of urban, extra urban and mixed urban roads. The vehicle was tested by varying the load mass and the driving style in order to explore different conditions
• in the city of Lyon including the surrounding province data acquisition was performed using a Heavy Vehicle (Volvo Premium Hybrid, equipped with an innovative hybrid powertrain, above 10t – GVW 19.2t). The vehicle was driven under real conditions, there were no restrictive conditions about the daily path or about its distance. The vehicle was equipped with a data logging system for the powertrain and with a GPS system for geo-referencing, thus also data related to altitude variation were included in the analysis
• in the city of Florence data was acquired during normal use condition of the vehicles under study. The main advantage of such an approach is that data is representative for the very particular use to which the vehicle is subjected. The city of Florence is affected by intense traffic within its historical center. Data acquisition was performed on a small fleet of light vehicles including:
o electric quadricycles (Renault Twizy), mainly used within the area of the city for light freights delivery; different drivers were using the vehicles
o electric vans (Renault Kangoo ZE, below 3.5t class), mainly used within the area of the city for light freights delivery
o electric passenger cars (Peugeot iOn), used by various private and commercial users for general purpose trips.
The acquired data was analysed and synthetized into representative driving cycles adapting the major indications of common synthesis procedures, in order to cover various use conditions for electric/ hybrid vehicles and components. The methodology includes a subdivision of each “trip” into “kinematic sequences” (e.g. from stop to stop), analysis and grouping of “kinematic sequences” using common indicators (e.g. mean positive acceleration, average speed, number of stop per km).
According to different characteristics of the vehicles, the synthesis procedure was applied on filtered datasets in order to generate different cycles for categories such as, representative “average” electric vehicle, small class passenger vehicle, light delivery vehicle, low powered vehicles and most “unsteady” sequences. For each category, two different distances – “long” (95th percentile trip distance) and “mean” (mean distance excluding extra-urban phases) – were calculated. This analysis forms ten ASTERICS driving cycles. The “unsteady long” driving cycle was for example calculated and further used for the creation of current profiles for battery testing in this project.
Driving data management
Beside the driving cycle analysis a tool for data analysis and cycle synthesis was developed which has the same methodology implemented as briefly mentioned above. The tool is delivered as a Matlab-language product with a Graphical User Interface (GUI) and it includes two main functionalities.
The first one is named “builder” which is an interpreter of data that can be used to “produce” new cycles and to verify their similarity with original data. The user can set a few parameters like the target distance, the vehicle data to be used, the data “clusters” to be included and the acceptance thresholds. With these parameters set, a large number of cycles can be generated. If any of the created cycles fits the original dataset, the tool plots the generated “representative” cycle and saves it on a spreadsheet. Saved data include the signal, the acceptance results (number of similar parameters, Performance Values – PV – indicator, Sum Square Distance – SSD - of speed-acceleration density matrix – SAPD) and the general describing parameters.
The second function named “binder” is a cycle explorer, editor and creator. It can be used to “explore”, plot and save each cycle from a collection of data. The sources are on the one side the ASTERICS cycles and recent relevant driving cycles, e.g. ADAC highway cycles or WLTP (Worldwide harmonized Light vehicles Test Procedure) and on the other side the database created for the ARTEMIS project which includes legislative and research cycles.
As an editor, the function can be used to “trim” existing cycles and to “paste them together as required in order to reproduce some testing procedures as described for example in literature (e.g. “EPA-5” test, ADAC “ECO test”). The function then calculates the energy needed to run the cycle (in kWh) using the pre-calculated results coming from a simplified model. The latter is based on an M1, A-class vehicle, curb mass being about 1100kg; the results should be considered as a “rough” estimation of energy needed for a comparable vehicle, but are not validated for any specific case.
As a creator, the tool can be used to build new cycles using ASTERICS data assembled thorough the same methodology implemented in the “builder” function, but, in this case, any cycle is accepted: the product of the randomization has to be considered as an “extraction” from the whole data, rather than a synthetic cycle.
Data created with this tool can be used during testing (e.g. or SIL/MIL/HIL environment) or for preparation of vehicle energy management strategies, usually verified over a large number of use cases. The proposed tool that allows for the recombination of measured driving data is particularly suitable for batch simulation activities, offering multiple inputs aimed to represent real driving style variability.
Model based development methodology
The MBSE (Model Based Systems Engineering) methodology applied to vehicle design aims at finding the right vehicle component types, sizing and architecture, fulfilling design goals and allowing the realization of tradeoff between design goals, formally analyzing the impact of these choices on the vehicle performances and behavior. Relying on physical component models, the MBSE design process brings better confidence on preliminary design, as it limits the risk of designing physically incoherent systems, leading to wrong specification, and later development issues.
In the automotive domain, the vehicle must ideally fulfill several design goals. These design goals are driven by regulatory constraints, common automotive goals and carmaker specific goals. The resulting design is usually a tradeoff between all these goals (except regulation), that’s why the MBSE process must allow a fast and efficient way to balance design to find the right tradeoff between all the design goals. These vehicle-level design goals, for most of them, are easily identifiable. Some more specific ones related to EVs are not shared among the automotive community.
The examination of literature, of consolidated know-how and of selected case studies suggest that in order to maximally exploit the potential of model engineering it is needed to use a software language that can guarantee:
• the storage of different kind of data
• the communication between different models
• the communication between different environments.
To reach these goals, it is necessary to overcome the obstacle related to different tools and the different signals that can be necessary at each design stage; some critical aspects are:
• the data exchange/storing protocol that is initially defined
• over-dimensioning can be useful to let dataset expansion if needed (later engineering steps could require more detailed information than expected in early design stage)
• data exchange/conversion interfaces should be prepared
• the difficulty related to the access of a large number of people to the data.
Finally, the choice of solver characteristics and of main time step has to be carefully evaluated, since inappropriate calculation efforts should hinder the use of models that are formally correct; to avoid such risk, models should be easily scalable, while the possibility to establish a dialogue between solver and model subsystem is an option to increase the rate of “automation” in sub-models assembly.
Models and test procedures for battery, e-motor and inverter
The efficient design of the e-powertrain of hybrid or full electric vehicles requires the availability of e-component models for the battery, inverter and e-motor. However, in this regard it is not sufficient to have an exact model of the steady state behavior, it is rather indispensable to utilize models that model the dynamic behavior and lifetime degradation effects of the component. Within the runtime of ASTERICS such sophisticated models covering different variants (battery, PMSM, SRM, IM, etc.) that are capable to model electric, thermal and degradation behavior over the component life-time were developed and validated.
Battery Models
In a modern electric vehicle the battery plays a crucial role in two different ways. On the one hand does the power and operating distance depend from the battery but on the other hand is the battery also one of the biggest cost drivers of the e-vehicle. Thermal conditioning of a battery is in turn essential for the battery and therefore for the vehicle performance. Beside that major influence, impacts on safety, durability and life time are driven by battery temperatures. Hence, models which cover the thermal behavior of a battery are as essential as deriving cell data from measurements for calibration and validation of the models itself.
The purpose of developing the ASTERICS electrical battery model was to provide a smart battery simulation model with easily adjustable parametrization for the use in dynamic load cycle simulations, taking ageing of the battery cell into account. The model is built on empirical data from accelerated ageing testing of a Li-ion battery cell (LG Chem 41 Ah, NMC blend). To achieve this goal two different approaches - an equivalent circuit model approach, and a semi-empirical electro-thermal model approach based on the theoretical background of Shepherd - were investigated.
The user-defined inputs and model outputs are current, temperature, SOC (State of Charge), SOH (State of Health) and pack configuration, and voltage, power loss, SOC and SOH as output. There is also the possibility to tune some of the physical parameters of the model related to ageing, energy throughput, internal resistance, and SOC window which enables the simulation of different battery types and driving cycles.
The battery model is based on a dynamic battery cell model, reflecting the cell performance at BOL (begin of life). In order to take ageing into account, a separate simulation block sets the ageing conditions of the original battery cell characteristics in relation to the number of cycles and SOH. The trends of both simulated voltage curves coincide well with the experimental data over the entire analyzed interval.
During operation the battery pack generates heat. It is well known that besides electrical parameters, cell temperature decisively influences battery performance and aging. Therefore a cooling system that efficiently dissipates generated heat is of importance for ensuring high performance of the battery pack and in particular its longevity. Combining the electrical model with a thermal model is therefore essential.
Battery Testing
The purpose with the development of testing procedures for durability testing is to efficiently identify stress factors that influence the battery aging and performance under realistic operating conditions. To reduce the time needed for the evaluation of aging effects, the measurements are commonly accelerated by applying a driving cycle of high stress and with peak currents of fast charging, and/or operated at elevated temperatures. The testing in ASTERICS was performed within 3 months or until the battery cell capacity is degraded to 80% of initial capacity, a level often used as end-of-life criteria in vehicle applications, respectively. At strategic points during the cycle life test an extraction of parameters representing the electro-chemical cell properties was made at various SOC (state-of-charge) levels, temperatures and frequencies in order to relate the degradation and SOH (state-of-health) to different stress factors. A special focus was on the impedance spectroscopy test method which is used for the diagnosis of battery cells.
Within ASTERICS a specification of test procedures for battery cells to be used for sampling of empirical data to parameterize ageing models of battery cells, modules or systems for residual capacity determination at variable working conditions was written. The test procedures were developed for single cells, mainly fresh cells delivered directly from the cell suppliers. However, the parameter tests can also be applied to cells already aged.
All procedures are based on established test standards and experience from previous projects such as EUCAR, EUROLION, SUPERLIB, HELIOS and AMELIE as well as internal projects. However, the procedures were customized to specifically address the performance measures especially important for the applications within ASTERICS.
Adaptive test procedures
Since the purpose of the ageing test is to produce experimental data for fitting of two type of ageing models, AVL shepherd model and Volvo ECC model the design of the ageing tests has two different approaches, applied on the same battery cell type. The AVL approach is based on cycling at fixed c-rates, in the full SOC (State of Charge) -window of the cell, while the Volvo approach is based on a constant load cycle derived from vehicle drive cycles, in a delimited SOC window. For both methods of stressing the cell, the cycles are counted and the capacity degradation of the cell is the indicator of physical ageing, at a limit for EOL (End of life - 80% of initial capacity or acc. to manufacturer’s definition) set to 80% remaining capacity of the initial value. The test procedures defined in ASTERICS sets the outline for the reference tests that have been used in common for two different principles of ageing tests. This means that the results of the cell characterization can be compared on the same basis.
Several cells were tested for ageing under different load conditions to furnish a broad base for the modelling. The focus has been to cover various stress parameters, and less effort was made to produce replicate tests, this priority was decided since the scope of modelling was high while the test budget and time was restrained.
Cell Testing is the basis for understanding cell behavior. Beside cell data sheets and safety specification sheets, the standardized test procedures that were described in ASTERICS allow for a benchmark of different cell chemistries on the one hand, and allow the fitting of simulation models on the other hand.
In ageing tests the adaption of certain test parameters may be needed in order to respect the changes in the physical properties of the battery and to detect ageing trends that depend on single stress factors. Not adapting the test parameters to respect the ageing of the cell may result in ageing effects that depends on multiple and combined stress factors. Since the purpose with testing is to enhance the knowledge about the cell that can be applied in prediction and modelling, there is a clear benefit if at least the stress factor is kept under control.
The basic concept of battery cell testing is based on the definition of the c-rate which is the ratio between the current in the battery cell and the cell capacity. This ratio guarantees that the cell is stressed with very similar respectively comparable conditions, since the maximum current allowed for a cell depends directly on the capacity. Higher currents have the potential to damage the cell irreversible, i.e. cause high degradation. The cell is now measured and stressed via defined cycles with a specific or varied c-rate (dependent on the used model) and continuously adapted to the new cell capacity, which changes over time. This allows the adaptation of the c-rates to the changed health condition of the cell and avoid unwanted conditions, like higher temperatures. The test is interrupted time wise to identify the new cell capacity with special pulse cycle tests. When the new cell capacity is available, the test continuous with the adapted c-rate of the cell. This procedure allows the un-interrupted execution of cell tests over a very long time and therefore avoid any stand-still time of the test bed and reduce cost of the tests.
Inverter Modelling
The development of inverters that are used for the automotive environment requires adaptions to the conventional inverter design process. Constraints to operate close to the specification limits of individual inverter components in order to achieve a higher utilization factor results in more stress on the components and a greater wearout which in turn reduces the durability of the inverter. The involvement of high accurate simulation models with a higher level of detail can help to enhance the inverter development throughout all stages in the design process. Methodologies for inverter modeling considering multi-physics, including thermal and ageing effects were investigated and developed in the ASTERICS project.
State of the art component models operate at very low temporal resolution as they have to resolve the switching pattern of the inverter. They reach high accuracy in power loss assessment which is the basis for the efficiency analysis of this component. Nevertheless, due to their computational effort these models cannot be used in system analysis.
System simulation tools commonly use data map based inverter models that are sufficiently accurate for a vehicle efficiency and cooling layout simulation but lack the level of detail which is needed for investigations concerning ageing. For aging, thermal stress of the power electronic elements is a key factor, hence more detailed models are necessary that resolve the individual power electronic elements of the inverter and provide localized temperatures. Further, the simulation effort must be kept at an acceptable level with respect to a driving cycle analysis of several minutes in real time or more.
Two approaches for the evaluation of power losses were identified to be the most promising. Both rely on input data that is commonly made public by the manufacturers and are thus accessible to the system integrator. The averaged model considers the fundamental loss contribution that covers the vast majority of losses in common operation conditions. The switched model additionally considers the impact of current ripple on losses. This is relevant for customized torque controllers or turndown operation at high speeds. The power loss information serves as input for a thermal inverter model that is coupled back to the temperature dependent parameters of the electric inverter model. Accuracy of the models were assessed by means of simulation as well as physical inverter testing.
Inverter ageing depends on mechanical stress which is commonly treated adequately by the inverter base structure, but it also strongly depends on thermal stress. And thermal stress has two main drivers: active stress cycles from operation and passive stress cycles from changes in the ambient temperature over day and year. It is concluded from literature study that a combination of a driving cycle with a condensed test cycle that reflects environmental changes can capture the thermal stress impact.
Inverter models can be integrated in other simulation environments within the project using specific interfaces or the Functional Mock-up Interface (FMI) standard. As presented above, the more relevant choice for inverter models in vehicle environment is the averaged model, which is a good compromise between CPU time and accuracy. A Functional Mock-up Unit (FMU) for co-simulation of a three-three-phase inverter including static aging has been created, using the interfaces defined in the ASTERICS project.
Inverter Testing
Inverter testing in the ASTERICS project focused on the generation of test data to validate the inverter models. The inverter was tested together with the real e-motor on the e-motor testbed. The behavior of the battery was emulated through a battery emulator that used the generated battery models. The measurement procedure was carried out in following sequence:
• In the basic characterization of the system, all relevant sensors were calibrated, idle runs were executed, electrical and mechanical angles were synchronized, and the AVL xEV box (fast data acquisition system) was calibrated.
• For vibration analysis, surface velocity measurements were carried out, using ten 1D acceleration sensors. Velocity levels were measured at 4 stationary operation points with different speed and load configurations in thermal equilibrium, with parallel measurement of electrical quantities.
• For power flow analysis, dynamic electric measurements for power loss assessment at different speed and load configurations at constant DC voltage were executed in stationary motoric operation. Additional measurements were executed at reduced DC-link voltage for inverter behavior assessment without entering field weakening.
• For thermal analysis, thermal equilibrium response at different speed and load configurations was measured in stationary operation plus cooling down into cold in transient resolution. Measurement with high dynamic resolution at thermally stationary condition was executed, followed by a measurement of the electro-motive force with inverter shut down time resolved measurements quantities for voltage, current, speed and torque were taken for model validation.
E-Motor modelling
E-motors have been used in industrial applications (e.g. energy, machining) and transportation applications (e.g. rail, ship) for over a hundred of years. Whereas electric motors in industry and heavy transportation (rail, ships) are mainly running at a constant speed and experience only slow transients, electric motors in automotive applications experience high-speed transients. Nevertheless, they must comply in the same way as standard vehicle do. Accurate electric motor dynamic models are required to improve the technical performance and fasten the industrial development process of new electric powertrain. Depending on the system simulation needs, high frequency (short timescale and high accuracy) simulation models up to high level simulation models (long timescale, typically for full driving cycles) are needed. Beyond sizing, primary efficiency and NVH analysis, higher fidelity models also support advanced controls development, allowing to evaluate strategies up to automate optimal control generations without any physical prototyping.
The ASTERICS E-motor models were developed for each electric machine technology, namely PMSM, SRM and IM (Induction Machine). All the newly developed dynamic models integrate combined electric, magnetic, mechanic and thermal domains, considering also each domain influence on stress factors and – subsequently – loss of life. With the use of designed experiments and data pattern analysis, a substantial reduction of testing effort is possible by concentration of the most salient features of the load cycles. The main ageing mechanism investigated were demagnetization (PMSM), windings insulation loss of life (function of temperature) and mechanical stress (torque ripples). Several levels of models have been developed to address different simulation needs, from high frequency models up to fast reduced models. Induction machine models were developed following an equivalent circuit approach, integrating a thermal loss model and considering windings loss of life function of thermal stress. Permanent Magnet Synchronous Machine models include an iron loss model and consider magnet demagnetization function of thermal stress. Thermal model of the Permanent Magnet Synchronous Machine was also coupled with the Permanent Magnet Synchronous Machine (PMSM) model. The switched Reluctance Machine (SRM) models integrate an equivalent circuit with magnetic circuit to achieve higher prediction accuracy in terms of physical behaviour (mechanic, electric), losses (thermal, magnetic...) and model-based optimization for torque ripple reduction. Even if these drive-technologies present specificities and specific challenges and are addressed through different simulation tools and methodologies, the studies all follow the same scheme: electro-magneto-mechanical modelling, magnetic losses modelling, loss of lifetime or degradation or ageing (when applicable) evaluation, parameterization methodology and validation.
For PMSM, a new methodology was developed to calculate iron losses and induced eddy current losses in permanent magnets using a combination of analytical approach and magneto-static finite element calculation. Calculated losses in the case study of two types of PM motor are prepared for different operational conditions. Demagnetization in PMSM was also evaluated. The results of this case study demonstrate the torque performance drop due to partial demagnetization and the change of induced no-load voltage shape. This phenomenon can be used for testing and diagnostics of permanent magnet machine performance changes due to increased temperature. Completed study of partial demagnetization and its effects gives useful guidelines to overpass the demagnetization problem in permanent magnet machines.
A new IM model describing electric, magnetic, mechanic, thermic and ageing behavior was developed. It covers both symmetrically built and asymmetrically supplied machine. The thermal model is based on lumped circuit theory. The temperature distribution as output of thermal model feeds back electro-mechanic model to calculate proper value of temperature dependent stator and rotor winding resistances. The same output is used as an input for ageing model. The ageing model shows loss of life-time depending on winding temperature distribution Based on results, the proper cooling system and thermal management system can be established. During the development, a new method to calculate the induction machine parameters covering the nominal shaft power from 5 kW to 100 kW was established. The model was tested with induction machines from different European suppliers. The comparison is very good and shows a little difference between modelled results and measured quantities.
Switched Reluctance motor was studied using two complementary approaches. Reluctance network, exploiting the analogy of magnetic and electric domains, has proved to be a good compromise between the Finite Element method and analytical calculation regarding simulation speed and accuracy of the results. This method was validated on a first design both on static and dynamic operation. Improvement of SRM model accuracy was also addressed using a co-energy based approach accounting for coupling between the phases. A full numerical methodology was developed for motor accurate parameterization integrating accurate loss estimation, as well as optimal control generation to minimize losses and torque ripples. A critical progress was to achieve this full process without involving any physical prototyping, from CAD to drive virtual testing.
Beyond the electric motor models themselves, the integration capabilities of these models in full vehicle environment is critical. Despite the different simulation tools and electric machine addressed, all these models were developed following the same framework, allowing interoperable and substitutable models trough standardized ports and interfaces.
E-Motor Testing
As already stated before electric motors have been used in industrial applications (energy, machining) and transportation applications (rail, ship) for over a hundred of years. However, it is not possible to pick an industrial motor off the shelf and install it in a production car. The reason is that automotive applications differ in many ways from the applications in which E-motors have been used so far.
The main identified automotive electric motor specificities are:
- Environmental conditions: while industrial plants sustain specific and controlled environmental conditions, and while rail vehicles are adapted to specific exploitation conditions, automotive vehicles must be able to sustain a wide range of conditions (temperatures range, vibrations, moisture, saltiness...) without any specific adaptation, keeping the same durability goals (>10 years, >100.000 km).
- Integration: the electric motor use out of automotive applications usually didn’t have to deal with the strong automotive integration constraints. This level of integration requires to consider all component interactions (functional, electrical, mechanical, thermal...)
- Efficiency and power density: There is a strong push from automotive OEMs for increasing the power density and efficiency of E-motors even further. The increase in power density is mainly driven by the limited installation space inside a vehicle. A higher power density opens the possibility for new motor topologies such as e.g. in-wheel motors. The increase of efficiency is driven by the resulting reduction of energy consumption which has a direct influence of the range of the electric vehicle.
- Weight, costs and supply chain: addressing a highly competitive mass market, electric vehicles must be competitive, in terms of costs and performances, implying a specific stress on the electric motor costs, supply chain dependencies and weight. And while industrial and heavy transportation address a Total Cost of Ownership (TCO) at decades levels, the automotive domain mainly addresses acquisition costs. These constraints lead to tradeoffs in the electric motor technologies, manufacturing process, quality and materials, requiring specific investigations to ensure that these tradeoffs don’t impact the vehicle lifetime and performances significantly.
- Production process: The same type of electric motor needs to be installed in tens of thousands of vehicles. Due to the high competition in the automotive sector, electric motors need to be produced in a cost effective manner. This requires a highly automated production process.
- Fast transient use: whereas electric motors in industry and heavy transportation (rail, ships) are mainly running at a constant speed and experience only slow transients, electric motors in automotive applications experience high-speed transients.
- Regenerative use: most of dual mode (traction & regeneration) were used in the rail domain, more often for dissipation braking than real regenerative or rejecting braking. And even in these domains, the transient dynamics are far lower than the automotive ones.
Motivation for e-motor testing
Due to the specificities mentioned above, the E-motor test procedures developed for industrial applications and heavy transportation (rail, ships) are not suitable for the use in automotive. Specific test procedures are required which ensure the OEMs and carmakers that their E-motor will be suitable for their future electric vehicles. Whereas procedures for testing internal combustion engines against the specificities of the automotive environment are well established, testing procedures for automotive E-motors are relatively new. This motivates a further development of E-motor test procedures tailored to the automotive environment.
Accelerated testing
The term “accelerated testing” is used in two different contexts:
• In the context of reducing the time (and hence the cost) needed to perform a test.
• In the context of performing the test earlier in the design cycle of the electric vehicle and thereby reducing significantly the cost needed to solve issues.
New methodologies for accelerated testing are developed for the purpose of ageing of winding insulation, to combine simulation and measurements, such as e.g. parameter estimation and for operational NVH and efficiency testing
In ASTERICS several e-motor testing methods were investigated, defined and executed. As an example a vibro-accoustic test of an SRM (switched reluctance machine) was performed, since SRM’s have the tendency to vibrate and produce noise. The test setup consisted of accelerometers and microphones in the near of the e-machine. In total, sixty tri-axial accelerometers were mounted on the jacket and the side panels. Furthermore, four microphones were placed: two in the near-field, two in far-field. An acoustic insulation box was placed around the SR driveline to separate the acoustic motor noise from environmental noise. Localization of these sensors is a very important issue. The output of this study was a waterfall 3-D diagram of the frequency spectra over the speed of the SRM-rotor, which can be used further to analyze the main frequencies on tonal components of the produced vibration respectively noise. Such an analysis allow the acoustic engineer to identify potential improvements for the acoustic behavior and to adapt the NVH characteristic of the e-machine accordingly.
Further testing methods for IM-machine and PMSM (e.g. common and differential-mode impedances were measured with an impedance analyzer in the frequency range of interest) were developed and validated on a real test bed to ensure their applicability. The measurement data was also used to parameterize the e-machine models or at least to validate their goodness of fit.
Virtual prototyping
The virtual prototyping consisted of two phase studies: first the parameterization itself and then the exploitation of the models predictions. As parameterization of such advanced models are not straightforward, as requiring data that are usually not available and not directly measurable at industrial costs, the approach was to develop parameterization methodologies allowing to parameterize these models.
The main approaches followed were measurement or 2D/3D simulation-based. When based on measurement, specific tests were performed on the real device to identify parameters. When 2D/3D simulation-based, finite element models (magnetic, thermal...) simulations were used to identify parameters. These last approaches have the main advantage to not necessarily require a physical prototype.
The parameterized models were used to perform several types of virtual tests and experiments. Performance simulations, integrating all the multi-physics presented above, were the first tests done, in order to validate the methods, tools and models, first in terms of functionality, then in terms of accuracy. At this stage, several corrections were applied in order to ensure prediction accuracy and robustness. Various performance simulations for several drives, sizing and suppliers were performed and these results were validated versus matching drive data.
Thermal and ageing models of electric powertrain components were implemented into behavioral vehicle simulations. With that method, the ageing effects of different driving scenarios can directly be studied via the resulting warming of the E-Drive components, for instance. Such information can be used also for the acceleration of tests, e.g. via designed experiments (DoE). Such an approach was implemented and tested for a sub-cycle of the NEDC where most damaging scenarios for E-Drive components were extracted.
The investigated methods and the gained insight into E-Drive components’ ageing behavior helped to significantly reduce the test durations which can be reached by focusing on the most damaging load combinations. As load-dependent warming and ageing models have been developed for all electric motor components (E-motor, invertor, HV battery) and implemented in a vehicle simulation program. Therefore, the influence of vehicle load cycles can be traced directly to equivalent component stress. The effect of E-motor layout on its behavior in the vehicle and the associated protection and de-rating strategy can be directly estimated thus avoiding problems with drivability at a very early stage in a project. Testing is a very elaborate and costly process. Employing new data condensing methods, e.g. via Genetic Algorithms and Neural Networks, the concentration on the most preeminent features of the driving cycle is possible thus reducing testing effort and time.
Further, the SRM model (electrical, mechanical, magnetic) was validated against existing SRM measurements at different levels of the model. This procedure validated the parameterization method as well as the model itself. The resulting accurate models (especially regarding electro-mechanical behavior) make it possible to set up a control law optimization process allowing the SRM to reach higher efficiency (for a real drive cycle) while drastically reducing the typical torque ripple issues affecting SRM and their transmission durability (NVH impact). These new capabilities for the three machine technologies demonstrated the capacity to address electric drives durability issues following model-based approach, in the thermal (windings), magnetic (demagnetization) and mechanic (torque ripples) domains.
System design, integration and evaluation
A link between the simulation environments (Amesim, AVL-Cruise, GoFAST, GSP) was created by means of two different kinds of standard integration interfaces:
• Functional Mock-up interface (FMI)
• S-function interface
In particular, the potential of the Functional Mock-up Interface, in its Model Exchange and Co-simulation configuration, was investigated and the capability of each tool to support this standard was described. Before integrating physical models of different subsystems in the target vehicle applications, the capability of the different tools to export and import within each other was verified. Starting from simple mathematical models (gain, integrator, integrator + gain), the efficiency of each standard integration and the accuracy of the results was checked. After these preliminary tests, simple physical subsystem models of e-motor and battery were imported and successfully simulated in the full vehicle model of each partner. The Functional Mockup Interface has proven to be an efficient and more flexible tool because of the possibility to get integrated with simulators other than Simulink.
Different configurations of the vehicle in each environment by means of several cross sub-system model exchanges based on the FMI/FMU Standard were created. As a result, seamless coupling between Amesim, AVL-Cruise, Volvo and GoFAST Vehicle Simulation platforms was possible thanks to the close cooperation between all partners. For the integration, a common definition (standard) in terms of input/output signals of the sub system models was used. For electric machine DC voltage [V], Requested torque [Nm] and electric machine speed [rpm] was used as input and DC current [A], Torque [NM], Efficiency (EM + inverter) [/] and Overall losses (EM + inverter) [W] were used as output signals. For the battery Battery request current [A] was used as input and Battery voltage [V], Battery state of charge [%) and Battery power losses [W] were used as output signal. Moreover some parameters of the imported sub system models were modified according to the specification of the native corresponding models used in the vehicle. Each rescaling was carried out following the procedure suggested by the partner that provided the model to be imported.
This made the performances of the simulated vehicles very similar to the real ones. The above procedure was made possible through the custom interface of the FMUs.
The ASTERICS project delivered solutions to significantly enhance the development process of EV’s in short-term, so that faster development cycles with 50% less effort than before are feasible. Further, these methods and tools can be used to improve the performance, reliability and safety of EV’s and also hybrids, as similar modeling and testing capabilities are needed in all sorts of EV-variants.
Overall the ASTERICS project made a huge step forward in enabling the development of EV’s which are more competitive in the transportation sector and compelling for end users, which finally helps to reduce global CO2 emissions.
Potential Impact:
The primary goal of ASTERICS was to develop within 3 years from its start a “full next generation” vehicle model (from the virtual point of view) taking an existing vehicle as reference in order to achieve a decrease in development time by 50% and an improved vehicle efficiency by 20%.
Environmental impact of ASTERICS
Electric vehicles have a huge potential to support the requested CO2 reduction and achieving the climate protection goals of the EU. Although the real contribution to the reduction depends on the energy mix of the electricity, more or less most studies show a great benefit for the electric car. Especially when it comes to countries with very high percentage of renewable electrical energy, like Austria, Finland, Sweden or Latvia. The goal of the EU is a renewable energy rate of 20% by 2020 up to 27% in 2030 and an improvement for up to 50% in 2050. This target mix will enable much more CO2 reduction with electrical cars in future as it is possible nowadays. The current overall energy mix in the EU includes about 75% created via fossil fuels.
Europe is therefore committed to the Green Cars Initiative (GCI) which responds on the one side to the economic need to strengthen the automotive industry in Europe and on the other side to the need for a reduction of greenhouse gases along with a reduction of noxious emission caused by the transport industry. Initiatives in this direction shall lead to a reduction or avoidance of global warming, an improvement of public health and primary energy savings, thus moving towards more sustainable transport. To support these higher level goals road transport is required to move from conventional internal combustion engines (ICE) vehicles to pure electric vehicles (PEV) or hybrid electric vehicles (HEV).
Currently the performance - in terms of range, speed, reliability, durability and efficiency - of pure electric vehicles leaves much to be desired. To facilitate the transition from conventional ICE vehicles towards PEVs the performance must be doubled in order to have a realistic market chance whereas the cost of technology and development time has to be cut in half.
The technology of internal combustion engine is by large the most used for road transport vehicles. The proposal of different technologies for vehicles propulsion implies a number of modifications in the known industrial and social established habits. Changes could be needed in terms of production technologies, primary energy supply, material choice, while at the same moment also the support infrastructure has to be modified, in order to satisfy users’ expectations or to influence their acceptance. Before a deep change in the vehicle supply chain is started, a number of evaluations have to be performed to respond to question such as:
• what is the effective modification in terms of primary energy consumption for the deployment of a similar service using different vehicles technologies?
• what kind of performances are expected considering current state-of-the-art technologies?
• which technology is expected to have larger margins for improvement?
• what about the economical sustainability of the technology?
• what about the environmental sustainability of the technology?
• are there any possible future “bottlenecks” (e.g. resources having limited availability on national/continental/world scale) for the technology?
It is evident that in this kind of analysis the focus is not only on the vehicle itself but, in many case, on the general context where a fleet is going to operate; as a consequence, the expected result is related to general indicators such as primary energy consumption (e.g. MJ, or tons petroleum equivalent), resource consumption (renewable and not renewable) or to environmental impact indicators (e.g.: greenhouse gas emission; waste production; abiotic resource depletion potential; photochemical ozone creation potential; sometimes other indicators or a combination of them is also calculated). The build of so-called “Life Cycle Assessment” (LCA) is in many cases a suitable solution to offer a complete output for the analysis; the methodology is defined by the technical standard ISO 14040, which is aimed to reduce the arbitrariness and the uncertainties of the evaluation. LCA approach is able to summarize in a small number of indicators the impact related to the use of primary resources, of energy vectors, of the use of the fleet, of the technology applied; in general, of all the material and energy flows related to a so called “function unit”, that is the service or the product under examination. In case of a car, an example of functional unit definition is: a European, compact-sized, five-door gasoline vehicle for 5 passengers including a luggage compartment, and all functions of the defined reference scenario with a mileage of 150,000 km over 12 years, complying with the same emission standards.
Since full LCA studies have to take into account all the variables determining the impact of the vehicle on the environment, a very large amount of data is needed: material extraction, transport, production technology, component manufacturing, consumption during use phase, maintenance and end of life management are main examples. All of them are contributing to air/soil/water emission, to raw materials depletion and to any other relevant environmental impact that has to be taken into account. Similar general approach can be built also using economic indicators, thus defining the total cost of ownership through “Life Cycle Costing” (LCC) analysis. Considering that for vehicles (and, in general, for all durable goods) the use phase is by far the one producing the most relevant impact, the analysis can be sometimes focused only on it. The analysis of the use phase (sometimes called “fuel cycle” or “energy cycle”) is therefore a very important step to evaluate the environmental impact of the vehicle and it can be focused on different sub-analysis, main being:
1. well to tank: it includes the analysis related to fuel production/extraction and delivery
2. tank to vehicle: it includes on-board energy conversion and is mainly related on vehicle powertrain characteristics
3. vehicle to miles: it includes vehicle energy consumption and is mainly related to the driving/duty cycle selected.
4. an analysis including point two and three of the above list is also said “tank to wheel”; a full use phase analysis (including point 1, 2 and 3) is said “well to wheel”.
Tank to wheel analysis is therefore very related to the knowledge of vehicle technology and of its performances: at least, a reference mission and a fuel/energy consumption model are needed. Since in many cases the value of the analysis relies on the comparison of different alternatives, the model should be able to calculate the variability of the output (e.g. efficiency, energy consumption) on given impact factors (e.g. mass variation) over different scenarios (e.g. different driving cycle). A short literature review of preliminary analysis that was performed within ASTERICS showed that the tools used for similar calculation can be really different.
An electric vehicle doesn’t directly emit pollutants per se, but that doesn’t mean it doesn’t have an indirect environmental impact directly dependent to its energy consumption.
The pollutants emissions related to a full electric vehicle depend on the electricity energy mix and its global powertrain efficiency. It also depends on the type and time of battery charging, as the electric grid management can resort to different type of energy source depending on the consumption. In this respect the proportion of the slow, fast and ultra-fast charge is an important data to assess the indirect vehicle CO2 footprint. Even if there is a lack of global regulation regarding EV emission (like the EURO standards), public authorities in several countries (Germany, Japan, United States...) pay attention to the CO2 footprint of vehicles, meaning that, even if not shared and standardized, the CO2 indirect footprint should be taken in account as a vehicle design goal.
ASTERICS addressed the related issues by introducing a high-fidelity virtual modelling environment for e-components, realistic e-driving cycles and advanced testing concepts and tools for the entire e-drivetrain.
Within the lifespan of the project it was successfully demonstrated that current e-drivetrain concepts can be improved significantly by applying virtual development methods. It was further shown that testing efforts for e-components can be reduced while the quality of results can be increased. On that note ASTERICS is an important stepping stone towards the development and early market readiness of an “optimal” Full Electric Vehicle in terms of performance, reduced development and testing times and of course the related costs. Hence, a setup that is closer to the real use of the vehicle can be realised by using the ASTERICS tools already in the prototyping phase.
Strategic and economic impact of ASTERICS
The strategic shift of the energy mix towards renewables together with the shift of the private transport towards electrical cars will create huge environmental impact, but also huge economic impact. According to the report of the “International economic development council” from 2013 ( http://www.iedconline.org/clientuploads/Downloads/edrp/IEDC_Electric_Vehicle_Industry.pdf ), the introduction of electrical vehicles will create additional jobs in the automotive industry, battery development and production and in the research and development of electrical vehicles. It is also obvious that this trend will lead to some job losses in the oil industry however. Electric vehicles will lead to a decrease in overall energy consumption and support a better balance between day and night, because electrical vehicles will typically be charged at night, where electricity is currently cheapest.
Electric Vehicles (EVs) are the most promising technology to drastically reduce the environmental burden of road transport, i.e. emissions of CO2, air pollutants and noise of particularly passenger cars and light commercial vehicles. Within this context a study of CONCAWE-EUCAR JEC-WTW showed that PEV remains the most efficient solution in terms of well-to-wheel efficiency of the different power-trains using different types of primary energy sources. Carmakers are responsible for delivering the reductions. The company target is an average for all cars sold, not a fixed limit that no car may exceed. In this scenario, automotive manufacturers must seek and provide improved and innovative vehicle solutions to the open market, which are not only compliant with the energy and environmental conditions and with the new legal requirements, but they must also be competitive and increasingly appealing. In other words all the main global OEMs, including those based in Europe, are essentially being forced to find new technological solutions to respect the social and environmental needs (environmental friendly vehicles) on one hand and on the other hand meet the demands of the customer (like higher levels of driveability and comfort, better safety and styling) at acceptable costs (in terms of purchase and operation costs as the price of fuel continues to increase). There are currently several concepts for PEV (pure electric vehicle) and HEV (hybrid electric vehicle) that support this “clean mobility” demand. However, the development and improvement of the different concepts require a huge effort in analysis, design, implementation and testing, not to forget feeding back experience, results and knowledge to new generations of such cars.
Advanced modelling tools and testing procedures going from one-dimension to three dimensional approaches play a fundamental role in the optimization process during the earliest project phases for the energy dimensioning of PEV & HEV as well as their “energy management strategies”, while reducing the project’s development lead-time as well as to build up requirements for subsystems and their related control units. The ASTERICS’ simulation tool suit and testing procedures is designed to deliver these functionalities.
From the validation point of view, if a ‘new generation’ vehicle is available, during the prototyping phase, the only way to test it (from the energetic efficiency point of view) is to apply the usual test methods, on the roller bench test. Sometimes, the applied driving cycles (and, sometimes, the driver’s driving style during the test) shadows the real benefit of the next generation archetype. This would results in big efforts to repeat the experimental tests and, sometimes, loss of time to tune the simulation models without great results. Moreover, if the project is in the prototyping phase, a ‘try and fix’ loop could start, with further efforts and losses.
In the frame of ASTERICS two major software programs for fuel economy and energy management (AVL-CRUISE and LMS-AMEsim) were linked with a standard OEM simulation tool (Fiat Group-PerFECTS). This integration showcase created a set of shared tools that speed up the simulation time using the best capabilities of each simulation environment.
From the experimental perspective, within ASTERICS a set of roller bench test cycles were created, with the intent to fit the ‘next generation’ vehicles attitudes, taking into account climate conditions, driving style and, of course, the traction archetype in study. These aspects resulted in a more confident testing in the future, being the future test sessions based on the experience and tools developed within ASTERICS.
ASTERICS helped to make an important step ahead in the concept phase for non-conventional vehicles, taking into account the principal aspects related to both the e-power-train peculiarities and the real use of the vehicle (also in terms of drive style). Hence, in the prototyping phase, a set-up closer to the real use of the vehicle can be realized using the developed tools. As a result, the whole loop from concept to prototype will be more focused and faster leading to an estimated 20% energy efficiency increase and a 50% reduction of the development & testing time of the e-driveline of future generation PEVs.
The ASTERICS consortium investigated and demonstrated the way towards designing and building efficient, cost optimized vehicles with advanced testing and modelling approaches. This will further pave the way towards electrification of European’s private transport and support the overall targets of green mobility and de-carbonization, while maintaining and creating additional economic potential for the industry and academics in Europe.
Dissemination activities
The main objectives of ASTERICS dissemination activities are to:
• raise awareness of the community on ASTERICS activities, results and ASTERICS partners
• inform and educate the community about the research area concerned with advanced simulation and tests for system and components of battery electric vehicles (BEV)
• engage the community to receive input
• promote and exploit the output of the project inside the community
For this purpose several tools were set up and used. In order to represent ASTERICS in a standardized way project templates for reports, documents, meetings and presentations were prepared. The project logo was designed right at the start of the project and was used in all presentations, reports, documents, etc. that are related to the ASTERICS project in order to assure a uniform appearance.
Website – www.asterics-project.eu
The ASTERICS website provides information on the project and its activities. A download area was established for additional information like the project leaflet, project poster, newsletters and public deliverables. The website was updated on a regular basis with actual information in the news section and project activities performed. The website will be a central communication element. The website is accessible via www.asterics-project.eu and features many aspects of the project and the continuous work performed. Visitors of the site have the possibility to download public deliverables, newsletters and to contact the coordinator for further information on the project. The website will be accessible for five years after the project is closed.
Flyer and Newsletter
A general Flyer was created at the beginning of the project and distributed to all relevant stakeholders, using a confidential dissemination database containing over a hundred contact details. The flyer was also published on the website as a news item.
During the runtime of this project three Newsletter were prepared which are available for download on the ASTERICS website. In the first Newsletter the objectives of the project and systematic and comprehensive approach with its four building blocks were explained. The second Newsletter focused on the achievements within each single work package and the last Newsletter shows how each work package contributed to the achievement of the ASTERICS objectives.
Publications
The consortium set great value on a continuous re-presentation of the ASTERICS project to show the progress to a wide audience. Therefore ASTERICS was represented at several conferences to get into a conversation with e.g. relevant stakeholders and standardization bodies. Due to the presentations that were held at several occasions the consortium had many opportunities to promote debates to accelerate the implementation of the research results. Further, one book chapter and five technical papers were presented at conferences and are listed below.
Book chapter
Horst Pfluegl, Claudio Ricci, Laura Borgarello, Pacôme Magnin, Frank Sellier, Lorenzo Berzi, Marco Pierini, Carolien Mazal, Hellal Benzaoui. A framework for electric vehicle development: from modelling to engineering through real-world data analysis. Series Lecture Notes in Mobility: Electric Vehicle Systems Architecture and Standardization Needs, February 2015, pp 55-73
Technical presentations
Mathieu Sarrazin, Steven Gillijns, Jan Anthonis, Karl Janssens, Herman van der Auweraer, Kevin Verhaeghe. NVH analysis of a 3 phase 12/8 SR motor drive for HEV applications. EVS27, Barcelona, Spain, November 17-20, 2013, pp 1-10
Tomaž Munih, Damijan Miljavec. A Method for Accelerated Ageing of Electrical Machine Insulation. Mechatronika (ME), 2014 16th International Conference on Mechatronics, Brno, Czech Republic, 3-5 Dec. 2014, pp. 65-70
Ylva OLOFSSON, Jens GROOT, Tomaž KATRAŠNIK, Gregor TAVČAR. Impedance spectroscopy characterisation of automotive NMC/graphite Li-ion cells aged with realistic PHEV load profile: quantification of cell properties vs. temperature at different stages of ageing. V: 2014 IEEE International Electric Vehicle Conference, Florence, Italy – December 17-19, 2014. IEVC2014 : conference proceedings.: Institute of Electrical and Electronics Engineers – IEEE, 2014, pp 1-6
L. Berzi, M. Delogu, M. Pierini, G. Zonfrillo. Analisi del degrade di sistemi si accumulo energia per veicoli elettrici mediante simulazione estesa di scenari di uso. AIAS – ASSOCIAZIONE ITALIANA PER L’ANALISI DELLE SOLLECITAZIONI 44° CONVEGNO NAZIONALE, 2-5 SETTEMBRE 2015, – UNIVERSITÀ DI MESSINA, pp 1-10.
Chen C., Diwoky F., Pavlovic Z., and Wurzenberger J. An Application of the Linear and Time-Invariant Method for the System-Level Thermal Simulation of an EV Battery. SAE Technical Paper 2015-01-1197, 2015, pp 1-8
Exploitation of results
According to the ERTRAC road map, the future introduction of Electric Vehicles in the market will very strongly depend on the availability of functioning and cost efficient products from the suppliers. In their view, the first step of the implementation of electric vehicles was done by the adaptation of existing vehicles by 2012. In the intermediate step efficiency gains by all consumers and advanced system integration alongside high performance energy storage systems will become available by 2016. As the last milestone, powertrain systems providing unlimited range should be a fact by 2018-2020.
The European Green Car Initiatives’ multi-annual roadmap and long term strategy is in line with these milestones and states that the industries involved have agreed that eventually after 10 years the goal of an accumulated 5 million pure EVs and PHEVs on Europe’s roads may be achieved. The ASTERICS execution ran in parallel with these schedules, herewith facilitating the transition from milestone 1 (introduction) to milestone 2 (intermediate).
Exploitation by OEM partners (Volvo and CRF)
The commercial vehicles industry in Europe need to find transport solutions that are both efficient on the technological level and on the business level. The Project partners Volvo and FIAT built knowledge on e-driveline components and complete vehicles, defined and modelled drive cycle systems, integrated test results in use for model development for simulation and explored new concepts.
Furthermore, CRF is developing a global simulation platform in order to have analysis of fuel economy and transport efficiency on complete vehicle level, however, there is a need to further populate this simulation platform with alternative drivelines, e.g. hybrid and pure electric vehicles for city distribution.
Volvo developed a battery model and corresponding testing procedures for a large scale measurement and evaluation study, which results in a model for further use in design and concept of the next electric vehicle generation.
The main exploitable project results for OEMs are: vehicle simulation on e-driveline in comparison to a conventional vehicle. This system simulation is good for demonstration and optimisation of new e-vehicle based transport solution in a city environment. This creates the possibility to find vehicle concepts for internal product planning and product development.
Exploitation by Supplier partners
Siemens, AVL, GK, THIEN progressed on the modeling and simulation of electric motors and inverters as well as on testing methodologies for battery, inverter and electric motor testing.
Ultimately, the project results contribute to the development of innovative testing, measure and diagnostics services for electric drivetrains.
The impact of ASTERICS in the simulation and testing sector is an improvement of the competitiveness of the solutions, providing more accurate results and reduced system engineering time and efforts for the customers (OEMs and Tier1). The innovative aspect of developing new simulation models and test methodologies allows OEMs and Tier1 suppliers and the whole automotive sector to implement the results providing more energy efficient vehicles in less time with less efforts.
According to the supplier partners the main exploitable results are:
• New services (testing, methodology)
• New simulation model libraries (after the industrialization phase) for the use in office simulations and on the test bed
• New 4 channel battery testers (battery life cycle tests, fast charging, any charging power, impulse load, different current ripple factors)
• New product requirements and specifications for testing of e-drivetrains
These results will now be brought to a higher level allowing their commercialization; this action will reinforce and expand the commercial position of the participating suppliers.
Exploitation by academic partners (FH-J, UNIFI, UL)
The Universities improved their knowledge regarding real-world driving cycles, vehicle simulation frameworks, energy consumption assessment, battery-, inverter- and e-motor- ageing phenomena and the related environmental impact. The participation in the ASTERICS project was also aimed at a knowledge transfer from research to industry; to offer engineering students and young researchers a State-of-the-Art training on an emerging engineering field, which is very important in the next decades.
Some of the innovative results of ASTERICS among others are: better studies on system level inverter models, e-driveline system simulation, more detailed information on the behavior of e-driveline components (especially thermal and aging behavior); which will – via (inter)national lectures, meeting and conferences – spread knowledge among young students and researcher in the field.
In addition, the development of realistic driving cycles for PEV and the proposal of a management model for batteries and their integration provides to future engineers the knowledge to further help European industry to develop new products with reduced environmental impact and reduce production costs and time-to-market.
With this in mind, the target audience for the implemented results are not only automotive OEMs and Tier1 suppliers but also MSc and PhD students. Young people will be more aware of sustainable and electric mobility.
List of Websites:
ASTERICS website: www.asterics-project.eu
Contact details: DI Horst Pflügl
horst.pfluegl@avl.com
Office: +43 316 787 1587
Mobile: +43 664 8379 426
In September 2009, both the European Union (EU) and G8 leaders agreed that CO2 emissions must be cut by 80% by 2050 if atmospheric CO2 is to stabilize at 450 parts per million – and global warming stay below the safe level of 2°C. But 80% decarbonisation overall by 2050 may require 95% decarbonisation of the road transport sector. With the number of passenger cars set to rise to 273 million in Europe and 2.5 billion worldwide by 2050, this is not achievable by improvements of traditional internal combustion engines or alternative fuels: the traditional combustion engine is expected to improve by 30% in comparison to 2009, so achieving full decarbonisation is not possible through efficiency alone. It is also uncertain if large amounts of (sustainably produced) biofuels - i.e. more than 50% of the demand - will be available for passenger cars, given the potential demand for biofuels from other sectors, such as goods vehicles, aviation, marine, power and heavy industry. Combined with the increasing scarcity and cost of energy resources, it is therefore vital to develop a range of alternative technologies that will ensure the long-term sustainability of mobility in Europe. Electric Vehicles (EVs) are the most promising technology for a drastic reduction of the environmental burden of road transport, i.e. emissions of CO2, air pollutants and noise of particularly passenger cars and light commercial vehicles. In December 2008, a new regulation was adopted which nominally strives to reduce the average CO2 emissions from new cars to 130 g/km by 2015, which nowadays can be said is satisfied with huge efforts on the combustion engine, weight optimization and first mild hybrid concepts. Significantly the law added a 95 g/km target for 2020. There were several concepts for PEV (pure electric vehicle) and HEV (hybrid electric vehicle) at project start available that support this “clean mobility” demand. However, the development and improvement of the different concepts required and still require a huge effort in analysis, design, implementation and testing, not to forget feeding back experience, results and knowledge to design new generations of such cars. Advanced modeling tools and testing procedures going from one-dimension to three dimensional approaches played a fundamental role in the optimization process during the earliest project phases for the energy dimensioning of PEV & HEV as well as their “energy management strategies”, while reducing the project’s development lead-time as well as to build up requirements for subsystems and their related control units. Deep and thorough investigation of the ASTERICS’ simulation tool suite and testing procedures were done to deliver these functionalities. ASTERICS hence contributes to a better performance and thus customer acceptance of PEV’s by improving modeling and testing tools needed for the development of future PEVs throughout Europe. Primary goal was to develop a “full next generation” virtual vehicle model taking different existing vehicles and their use by customers as a reference. All subsystems and components of the EV are the base for other ‘next generation’ vehicle archetypes, also different from the one studied within ASTERICS. In this way, ASTERICS supports the competitiveness of the automotive sector in all its aspects: basic components, integrated components, sub-systems, algorithms, systems and OEMs applications.
Project Context and Objectives:
The concept of the ASTERICS project comprises a systematic and comprehensive approach for the design, development and testing phases of E-drivelines in Pure Electric Vehicles, which is based upon four major building blocks. The four building blocks, their features and the major innovations are as follows:
1. Real world environment and conditions based drive cycles
Real world environment, conditions and fleet data from other projects set the baseline by defining drive cycles, usage profiles, the real world use of the EVs and customer demands. This information was used to identify and specify representative operating conditions, performance requirements and constraints for EV components, as well as stress scenarios, use cases and assessment criteria.
2. Advanced testing methodologies and models for E-driveline components
Enhanced test methodologies using Software, Model and Hardware in the Loop (SiL, HiL, MiL) environments and advanced models capable of simulating the real world behavior with the required accuracy and calculation speed together with sophisticated test approaches (like DoE or online-adaptions) enable standardized, accurate and time-efficient testing of E-driveline components in different environments (Office, test-bench, chassis-roller, ...) and different development phases.
3. Descriptive/ predictive models for battery subsystem, power electronics and electric motor
Standardized procedures for battery ageing and models describing ageing effects and energetic approaches enable the comparison of batteries of the same type and models, LCA (Life Cycle Analysis) , LCC (Life Cycle Costing) as well as the identification of second life applications. Tests are done on test benches rather than in the field in order to determine models that describe ageing and other effects regarding mechanical, thermal and electrical behavior.
Standardized test procedures for power electronics and enhanced inverter models (e.g. based on MOS-FET) were developed which include thermal simulation of the environment and different materials of inverter components accurately that describe a wide range of power classes and efficiency of the components (Si, SiCa, GaN ). Advanced testing approaches for e-motor ageing (e.g. windings) and models describing mechanical, thermal and electric behavior in combination with environmental models were developed - 1D models for fast real-time simulation and 3D for accurate drive cycle in-depth timescale studies as well as linkage between construction parameters and model parameters was created.
4. Total system (e-driveline and EV)
Standards for the connection/integration of high-fidelity subsystem models by means of components and interfaces, which take into account existing standards (like Modelisar-FMI), enable accurate simulation, calibration and optimization of the entire drive train in terms of vehicle performance and energy efficiency, and the usage/ verification together with real world drive cycles in early stages of the development process were adapted.
The proposed approach a seamless model based development process from design to validation, has the largest potential to arrive at optimal Electric Vehicles, in terms of performance (range, speed, reliability, durability and efficiency), reduces development and testing times and related costs.
The overall objectives of the ASTERICS project were:
• to develop a systematic and comprehensive approach for the design, development and testing phases of E-drivelines in Fully Electric Vehicles;
• to reduce the overall development time and testing efforts for EV and EV components by 50% compared to the current time and efforts;
• to enable improvement and optimization of the overall efficiency and performance of electric vehicles by at least 20% compared to existing/known concepts.
Scientific and technological objectives of ASTERICS were:
1. Real world environment and condition based drive cycles
To define a set of real-world drive cycle usage profiles, the real world use of EVs and customer demands for EVs based on analyses of existing fleet data of electric vehicle prototypes. Since existing driving cycles are conceived to assess the characteristics of ICE/HEV vehicles, their capability to assess the performances (e.g. energy consumption, durability) of EV was investigated. Such analyses take into account specific EV issues, such as regenerative braking capabilities and the relationship with driver skills in order to analyze component stress (e.g. effects of peak currents on batteries) and evaluate the overall performance (e.g. range, energy consumption). Analyses further identified clues for advanced vehicle design (e.g. advanced power systems including flywheel/ super capacitors/ range extender devices) which in turn helped to build up a complete “use case” scenario, including information about charging strategy during vehicle life. Such assessments are needed for the evaluation of typical battery Depth-of-Discharge (DoD) and the related battery State-Of-Health (SoH) expected during vehicle life.
To describe real-world contexts, different driving cycles were provided, including urban/extra-urban/motorway cycles or specific applications, such as freight delivery and passenger public services.
2. Advanced Testing methodologies
Test procedures that automatically populate the component simulation models for Battery, Inverter and E-Motor were developed, which can be used in all phases of the development process in MIL (Model in the Loop), SIL (Software in the Loop), HIL (Hardware in the Loop) and other test environments.
Focus in the development of these test procedures was:
a) on autonomous running testing procedures, which take into account limitations of battery and/or other involved units without interruption.
b) on automatically adaptive procedures for batteries with respect to the need of the model building algorithms.
c) on finding appropriate combinations of test conditions and stress factors influencing the battery’s characteristic (including tolerances between individual cells, self-discharge rate, open-circuit voltage, isolation resistance, mechanical deformation/expansion) and their changes over life-time – especially on cell-level.
d) on the identification of stress factors that influence the ageing behavior of inverters (thermal, electric factors,..) and set up tests with sophisticated methods (e.g. DoE).
e) on the identification of components that have an influence on ageing effects of Inverters (e.g. MOS-FET, DC-link capacitor) for different topologies and compositions.
f) on the definition of influencing factors, needed measurement accuracy and dynamic resolutions (e.g. phase current, torque, efficiency, fluxes...) for different E-Motor types and models (SRM – switched reluctance machine, PMSM – permanent magnet synchronous machine, IM – induction machine).
g) to develop procedures for accelerated ageing of Battery, Inverter and E-Motor based on the knowledge of S/T objective 1, in order to shorten the testing time for this kind of tests.
Main aspects of these test procedures lay in the identification of stress factors and their relevant contribution to the accelerated ageing (plus degradation and abuse) of components as well as in the definition of standard accelerated ageing tests for EV components based on the knowledge about ageing/stress influencing factors. The development and set-up of a test system that is able to perform parallel (multi-channel) battery cell tests, in order to run different test procedures with different test excitation sequences on the same type of cells with higher accuracy and less energy consumption as known today is another aspect here.
3. Electric Sub-system models developments
Several accurate high fidelity model for Batteries, Inverters and E-Motors considering all relevant dependences from inputs/parameters were developed, valid for the whole life-cycle of the E-driveline components. Main aspects that were covered in the course of the battery model development were
a) to identify the influence of parameters like capacity, impedance, tolerances between the individual cells, self-discharge rate, temperature, open-circuit voltage, isolation resistance, mechanical deformation/expansion, etc...
b) to describe the dependence of battery performance, electrical range, internal resistance, power acceptance from “age” of the battery.
c) to establish a standard definition of “battery age” by means of an appropriate set of parameters to enable the comparison to other batteries of the same type.
d) to integrate the battery model into a thermal sub-system describing the cooling/heating transfer-rate with the sub-system.
e) to design a real-time capable representation of the battery model needed for real-time purposes like usage in battery simulator or entire vehicle models.
Concerning Inverter models emphasis was put on the development of a computationally fast model, covering different inverter topologies and compositions and that they can be used in higher power ranges (150kW) and higher speed drives (22000 rpm) as usual. The evaluation and identification of thermal and ageing effects as well as influences of component tolerances to the life-time behavior of Inverters was another important aspect that was covered.
A major aspect in the development of E-Motor models was to identify/ parameterize E-Motor models by transferring construction parameters automatically into the model description. Further the influence of combined stress factors and the subsequent identification of expected loss-of-life damaging models was determined and the dynamic behavior of different types of E-Motors (SRL, SPMS, ASM) with computationally fast models, including thermal, electrical, ambient and mechanical effects was accurately described.
4. Total system (E-driveline) models integration and validation on test bench
To transfer drive cycle data from the field testing (S/T objective 1) to the duty cycle on the test benches emphasis was put on the development of procedures to more adequately characterize the real world performance of the (aged) components on the test bench.
To integrate models of the components into an overall simulation environment it was important to define interfaces and components in order to exchange and co-simulate component models between different tool-environments (AMESIM, CRUISE, PERFECTS). Models (e.g. Battery model for Battery Simulator) were integrated into the test bench environment and tests on test benches were performed in order to evaluate the quality of the simulation models as well as the application of real-life testing scenarios in simulation environments and the verification of possible efficiency and performance improvements by means of modifying/ exchanging components in the e-driveline.
The whole design and testing process for EVs and e-drivelines in ASTERICS was supported in such a way that the specific requirements and demands (efficiency, performance, life-time, stress-scenarios, etc...) can be fulfilled leading to high quality, reliable, more efficient, optimized electric vehicles in a 50% reduced development time.
Project Results:
The aim of ASTERICS was the development of a systematic and comprehensive approach for the design, development and test phase of next generation electrical vehicles that satisfy the requirements of increased performance, durability and safety. This was a very ambitious goal since it requires not only the development of accurate and real-time capable stand-alone models of all electric components but a complete vehicle simulation that allows for the co-simulation of different models from different vendors.
ASTERICS covered the development of realistic driving cycles and highly accurate simulation models for batteries, e-motors and inverter from the investigation of appropriate test procedures, equipment and tools through to a co-simulation platform that enabled project partners to integrate and co-simulate their models in their specific vehicle simulation environments.
Realistic driving conditions for EV’s
Within the last decade a lot of work was invested to collect data in order to derive real-world driving cycles that are supposed to representative the typical use of cars in European. Such a driving cycle can be considered as a standardized procedure to evaluate vehicle performance in a reproducible way under laboratory conditions. These can include simulation environment, power–adsorbing chassis dynamometer and/or testbeds. It is therefore clear that a single “driving cycle” cannot represent all the possible conditions under which the vehicle could be used during its entire lifecycle which is why some compromises in designing such cycles have to be made. A test procedure has to include a time–vehicle speed signal and a large set of boundary conditions.
At this point it has to be mentioned that EVs introduce new parameters for the evaluation of their performance and are affected by specific, new criticalities in comparison with conventional vehicles. The possibility of energy recovery during breaking for example could cause drivers to modify their way of driving in order to optimize the energy consumption. The limited range can cause drivers to display a particularly smooth, benign driving style. Some specific conditions (e.g. occasionally high daily distance driven, unavailability of charging points) are in particularly determining the so–called “range anxiety” phenomena and a different drivability leads further to a different perception of vehicle performances.
The way things are at the moment there are about 150 driving cycles available for vehicle type approval and available for tests in scientific literature, but only a few have been specifically built up for EVs. The above mentioned considerations were the driver for the development of a new set of reference cycles within ASTERICS. Data was obtained within the frame of three case studies in three different cities using different vehicles:
• in the city of Turin the data acquisition was performed using a Light-Delivery electric Vehicle (Iveco Daily EV, below 3.5t) running over a predetermined route consisting of urban, extra urban and mixed urban roads. The vehicle was tested by varying the load mass and the driving style in order to explore different conditions
• in the city of Lyon including the surrounding province data acquisition was performed using a Heavy Vehicle (Volvo Premium Hybrid, equipped with an innovative hybrid powertrain, above 10t – GVW 19.2t). The vehicle was driven under real conditions, there were no restrictive conditions about the daily path or about its distance. The vehicle was equipped with a data logging system for the powertrain and with a GPS system for geo-referencing, thus also data related to altitude variation were included in the analysis
• in the city of Florence data was acquired during normal use condition of the vehicles under study. The main advantage of such an approach is that data is representative for the very particular use to which the vehicle is subjected. The city of Florence is affected by intense traffic within its historical center. Data acquisition was performed on a small fleet of light vehicles including:
o electric quadricycles (Renault Twizy), mainly used within the area of the city for light freights delivery; different drivers were using the vehicles
o electric vans (Renault Kangoo ZE, below 3.5t class), mainly used within the area of the city for light freights delivery
o electric passenger cars (Peugeot iOn), used by various private and commercial users for general purpose trips.
The acquired data was analysed and synthetized into representative driving cycles adapting the major indications of common synthesis procedures, in order to cover various use conditions for electric/ hybrid vehicles and components. The methodology includes a subdivision of each “trip” into “kinematic sequences” (e.g. from stop to stop), analysis and grouping of “kinematic sequences” using common indicators (e.g. mean positive acceleration, average speed, number of stop per km).
According to different characteristics of the vehicles, the synthesis procedure was applied on filtered datasets in order to generate different cycles for categories such as, representative “average” electric vehicle, small class passenger vehicle, light delivery vehicle, low powered vehicles and most “unsteady” sequences. For each category, two different distances – “long” (95th percentile trip distance) and “mean” (mean distance excluding extra-urban phases) – were calculated. This analysis forms ten ASTERICS driving cycles. The “unsteady long” driving cycle was for example calculated and further used for the creation of current profiles for battery testing in this project.
Driving data management
Beside the driving cycle analysis a tool for data analysis and cycle synthesis was developed which has the same methodology implemented as briefly mentioned above. The tool is delivered as a Matlab-language product with a Graphical User Interface (GUI) and it includes two main functionalities.
The first one is named “builder” which is an interpreter of data that can be used to “produce” new cycles and to verify their similarity with original data. The user can set a few parameters like the target distance, the vehicle data to be used, the data “clusters” to be included and the acceptance thresholds. With these parameters set, a large number of cycles can be generated. If any of the created cycles fits the original dataset, the tool plots the generated “representative” cycle and saves it on a spreadsheet. Saved data include the signal, the acceptance results (number of similar parameters, Performance Values – PV – indicator, Sum Square Distance – SSD - of speed-acceleration density matrix – SAPD) and the general describing parameters.
The second function named “binder” is a cycle explorer, editor and creator. It can be used to “explore”, plot and save each cycle from a collection of data. The sources are on the one side the ASTERICS cycles and recent relevant driving cycles, e.g. ADAC highway cycles or WLTP (Worldwide harmonized Light vehicles Test Procedure) and on the other side the database created for the ARTEMIS project which includes legislative and research cycles.
As an editor, the function can be used to “trim” existing cycles and to “paste them together as required in order to reproduce some testing procedures as described for example in literature (e.g. “EPA-5” test, ADAC “ECO test”). The function then calculates the energy needed to run the cycle (in kWh) using the pre-calculated results coming from a simplified model. The latter is based on an M1, A-class vehicle, curb mass being about 1100kg; the results should be considered as a “rough” estimation of energy needed for a comparable vehicle, but are not validated for any specific case.
As a creator, the tool can be used to build new cycles using ASTERICS data assembled thorough the same methodology implemented in the “builder” function, but, in this case, any cycle is accepted: the product of the randomization has to be considered as an “extraction” from the whole data, rather than a synthetic cycle.
Data created with this tool can be used during testing (e.g. or SIL/MIL/HIL environment) or for preparation of vehicle energy management strategies, usually verified over a large number of use cases. The proposed tool that allows for the recombination of measured driving data is particularly suitable for batch simulation activities, offering multiple inputs aimed to represent real driving style variability.
Model based development methodology
The MBSE (Model Based Systems Engineering) methodology applied to vehicle design aims at finding the right vehicle component types, sizing and architecture, fulfilling design goals and allowing the realization of tradeoff between design goals, formally analyzing the impact of these choices on the vehicle performances and behavior. Relying on physical component models, the MBSE design process brings better confidence on preliminary design, as it limits the risk of designing physically incoherent systems, leading to wrong specification, and later development issues.
In the automotive domain, the vehicle must ideally fulfill several design goals. These design goals are driven by regulatory constraints, common automotive goals and carmaker specific goals. The resulting design is usually a tradeoff between all these goals (except regulation), that’s why the MBSE process must allow a fast and efficient way to balance design to find the right tradeoff between all the design goals. These vehicle-level design goals, for most of them, are easily identifiable. Some more specific ones related to EVs are not shared among the automotive community.
The examination of literature, of consolidated know-how and of selected case studies suggest that in order to maximally exploit the potential of model engineering it is needed to use a software language that can guarantee:
• the storage of different kind of data
• the communication between different models
• the communication between different environments.
To reach these goals, it is necessary to overcome the obstacle related to different tools and the different signals that can be necessary at each design stage; some critical aspects are:
• the data exchange/storing protocol that is initially defined
• over-dimensioning can be useful to let dataset expansion if needed (later engineering steps could require more detailed information than expected in early design stage)
• data exchange/conversion interfaces should be prepared
• the difficulty related to the access of a large number of people to the data.
Finally, the choice of solver characteristics and of main time step has to be carefully evaluated, since inappropriate calculation efforts should hinder the use of models that are formally correct; to avoid such risk, models should be easily scalable, while the possibility to establish a dialogue between solver and model subsystem is an option to increase the rate of “automation” in sub-models assembly.
Models and test procedures for battery, e-motor and inverter
The efficient design of the e-powertrain of hybrid or full electric vehicles requires the availability of e-component models for the battery, inverter and e-motor. However, in this regard it is not sufficient to have an exact model of the steady state behavior, it is rather indispensable to utilize models that model the dynamic behavior and lifetime degradation effects of the component. Within the runtime of ASTERICS such sophisticated models covering different variants (battery, PMSM, SRM, IM, etc.) that are capable to model electric, thermal and degradation behavior over the component life-time were developed and validated.
Battery Models
In a modern electric vehicle the battery plays a crucial role in two different ways. On the one hand does the power and operating distance depend from the battery but on the other hand is the battery also one of the biggest cost drivers of the e-vehicle. Thermal conditioning of a battery is in turn essential for the battery and therefore for the vehicle performance. Beside that major influence, impacts on safety, durability and life time are driven by battery temperatures. Hence, models which cover the thermal behavior of a battery are as essential as deriving cell data from measurements for calibration and validation of the models itself.
The purpose of developing the ASTERICS electrical battery model was to provide a smart battery simulation model with easily adjustable parametrization for the use in dynamic load cycle simulations, taking ageing of the battery cell into account. The model is built on empirical data from accelerated ageing testing of a Li-ion battery cell (LG Chem 41 Ah, NMC blend). To achieve this goal two different approaches - an equivalent circuit model approach, and a semi-empirical electro-thermal model approach based on the theoretical background of Shepherd - were investigated.
The user-defined inputs and model outputs are current, temperature, SOC (State of Charge), SOH (State of Health) and pack configuration, and voltage, power loss, SOC and SOH as output. There is also the possibility to tune some of the physical parameters of the model related to ageing, energy throughput, internal resistance, and SOC window which enables the simulation of different battery types and driving cycles.
The battery model is based on a dynamic battery cell model, reflecting the cell performance at BOL (begin of life). In order to take ageing into account, a separate simulation block sets the ageing conditions of the original battery cell characteristics in relation to the number of cycles and SOH. The trends of both simulated voltage curves coincide well with the experimental data over the entire analyzed interval.
During operation the battery pack generates heat. It is well known that besides electrical parameters, cell temperature decisively influences battery performance and aging. Therefore a cooling system that efficiently dissipates generated heat is of importance for ensuring high performance of the battery pack and in particular its longevity. Combining the electrical model with a thermal model is therefore essential.
Battery Testing
The purpose with the development of testing procedures for durability testing is to efficiently identify stress factors that influence the battery aging and performance under realistic operating conditions. To reduce the time needed for the evaluation of aging effects, the measurements are commonly accelerated by applying a driving cycle of high stress and with peak currents of fast charging, and/or operated at elevated temperatures. The testing in ASTERICS was performed within 3 months or until the battery cell capacity is degraded to 80% of initial capacity, a level often used as end-of-life criteria in vehicle applications, respectively. At strategic points during the cycle life test an extraction of parameters representing the electro-chemical cell properties was made at various SOC (state-of-charge) levels, temperatures and frequencies in order to relate the degradation and SOH (state-of-health) to different stress factors. A special focus was on the impedance spectroscopy test method which is used for the diagnosis of battery cells.
Within ASTERICS a specification of test procedures for battery cells to be used for sampling of empirical data to parameterize ageing models of battery cells, modules or systems for residual capacity determination at variable working conditions was written. The test procedures were developed for single cells, mainly fresh cells delivered directly from the cell suppliers. However, the parameter tests can also be applied to cells already aged.
All procedures are based on established test standards and experience from previous projects such as EUCAR, EUROLION, SUPERLIB, HELIOS and AMELIE as well as internal projects. However, the procedures were customized to specifically address the performance measures especially important for the applications within ASTERICS.
Adaptive test procedures
Since the purpose of the ageing test is to produce experimental data for fitting of two type of ageing models, AVL shepherd model and Volvo ECC model the design of the ageing tests has two different approaches, applied on the same battery cell type. The AVL approach is based on cycling at fixed c-rates, in the full SOC (State of Charge) -window of the cell, while the Volvo approach is based on a constant load cycle derived from vehicle drive cycles, in a delimited SOC window. For both methods of stressing the cell, the cycles are counted and the capacity degradation of the cell is the indicator of physical ageing, at a limit for EOL (End of life - 80% of initial capacity or acc. to manufacturer’s definition) set to 80% remaining capacity of the initial value. The test procedures defined in ASTERICS sets the outline for the reference tests that have been used in common for two different principles of ageing tests. This means that the results of the cell characterization can be compared on the same basis.
Several cells were tested for ageing under different load conditions to furnish a broad base for the modelling. The focus has been to cover various stress parameters, and less effort was made to produce replicate tests, this priority was decided since the scope of modelling was high while the test budget and time was restrained.
Cell Testing is the basis for understanding cell behavior. Beside cell data sheets and safety specification sheets, the standardized test procedures that were described in ASTERICS allow for a benchmark of different cell chemistries on the one hand, and allow the fitting of simulation models on the other hand.
In ageing tests the adaption of certain test parameters may be needed in order to respect the changes in the physical properties of the battery and to detect ageing trends that depend on single stress factors. Not adapting the test parameters to respect the ageing of the cell may result in ageing effects that depends on multiple and combined stress factors. Since the purpose with testing is to enhance the knowledge about the cell that can be applied in prediction and modelling, there is a clear benefit if at least the stress factor is kept under control.
The basic concept of battery cell testing is based on the definition of the c-rate which is the ratio between the current in the battery cell and the cell capacity. This ratio guarantees that the cell is stressed with very similar respectively comparable conditions, since the maximum current allowed for a cell depends directly on the capacity. Higher currents have the potential to damage the cell irreversible, i.e. cause high degradation. The cell is now measured and stressed via defined cycles with a specific or varied c-rate (dependent on the used model) and continuously adapted to the new cell capacity, which changes over time. This allows the adaptation of the c-rates to the changed health condition of the cell and avoid unwanted conditions, like higher temperatures. The test is interrupted time wise to identify the new cell capacity with special pulse cycle tests. When the new cell capacity is available, the test continuous with the adapted c-rate of the cell. This procedure allows the un-interrupted execution of cell tests over a very long time and therefore avoid any stand-still time of the test bed and reduce cost of the tests.
Inverter Modelling
The development of inverters that are used for the automotive environment requires adaptions to the conventional inverter design process. Constraints to operate close to the specification limits of individual inverter components in order to achieve a higher utilization factor results in more stress on the components and a greater wearout which in turn reduces the durability of the inverter. The involvement of high accurate simulation models with a higher level of detail can help to enhance the inverter development throughout all stages in the design process. Methodologies for inverter modeling considering multi-physics, including thermal and ageing effects were investigated and developed in the ASTERICS project.
State of the art component models operate at very low temporal resolution as they have to resolve the switching pattern of the inverter. They reach high accuracy in power loss assessment which is the basis for the efficiency analysis of this component. Nevertheless, due to their computational effort these models cannot be used in system analysis.
System simulation tools commonly use data map based inverter models that are sufficiently accurate for a vehicle efficiency and cooling layout simulation but lack the level of detail which is needed for investigations concerning ageing. For aging, thermal stress of the power electronic elements is a key factor, hence more detailed models are necessary that resolve the individual power electronic elements of the inverter and provide localized temperatures. Further, the simulation effort must be kept at an acceptable level with respect to a driving cycle analysis of several minutes in real time or more.
Two approaches for the evaluation of power losses were identified to be the most promising. Both rely on input data that is commonly made public by the manufacturers and are thus accessible to the system integrator. The averaged model considers the fundamental loss contribution that covers the vast majority of losses in common operation conditions. The switched model additionally considers the impact of current ripple on losses. This is relevant for customized torque controllers or turndown operation at high speeds. The power loss information serves as input for a thermal inverter model that is coupled back to the temperature dependent parameters of the electric inverter model. Accuracy of the models were assessed by means of simulation as well as physical inverter testing.
Inverter ageing depends on mechanical stress which is commonly treated adequately by the inverter base structure, but it also strongly depends on thermal stress. And thermal stress has two main drivers: active stress cycles from operation and passive stress cycles from changes in the ambient temperature over day and year. It is concluded from literature study that a combination of a driving cycle with a condensed test cycle that reflects environmental changes can capture the thermal stress impact.
Inverter models can be integrated in other simulation environments within the project using specific interfaces or the Functional Mock-up Interface (FMI) standard. As presented above, the more relevant choice for inverter models in vehicle environment is the averaged model, which is a good compromise between CPU time and accuracy. A Functional Mock-up Unit (FMU) for co-simulation of a three-three-phase inverter including static aging has been created, using the interfaces defined in the ASTERICS project.
Inverter Testing
Inverter testing in the ASTERICS project focused on the generation of test data to validate the inverter models. The inverter was tested together with the real e-motor on the e-motor testbed. The behavior of the battery was emulated through a battery emulator that used the generated battery models. The measurement procedure was carried out in following sequence:
• In the basic characterization of the system, all relevant sensors were calibrated, idle runs were executed, electrical and mechanical angles were synchronized, and the AVL xEV box (fast data acquisition system) was calibrated.
• For vibration analysis, surface velocity measurements were carried out, using ten 1D acceleration sensors. Velocity levels were measured at 4 stationary operation points with different speed and load configurations in thermal equilibrium, with parallel measurement of electrical quantities.
• For power flow analysis, dynamic electric measurements for power loss assessment at different speed and load configurations at constant DC voltage were executed in stationary motoric operation. Additional measurements were executed at reduced DC-link voltage for inverter behavior assessment without entering field weakening.
• For thermal analysis, thermal equilibrium response at different speed and load configurations was measured in stationary operation plus cooling down into cold in transient resolution. Measurement with high dynamic resolution at thermally stationary condition was executed, followed by a measurement of the electro-motive force with inverter shut down time resolved measurements quantities for voltage, current, speed and torque were taken for model validation.
E-Motor modelling
E-motors have been used in industrial applications (e.g. energy, machining) and transportation applications (e.g. rail, ship) for over a hundred of years. Whereas electric motors in industry and heavy transportation (rail, ships) are mainly running at a constant speed and experience only slow transients, electric motors in automotive applications experience high-speed transients. Nevertheless, they must comply in the same way as standard vehicle do. Accurate electric motor dynamic models are required to improve the technical performance and fasten the industrial development process of new electric powertrain. Depending on the system simulation needs, high frequency (short timescale and high accuracy) simulation models up to high level simulation models (long timescale, typically for full driving cycles) are needed. Beyond sizing, primary efficiency and NVH analysis, higher fidelity models also support advanced controls development, allowing to evaluate strategies up to automate optimal control generations without any physical prototyping.
The ASTERICS E-motor models were developed for each electric machine technology, namely PMSM, SRM and IM (Induction Machine). All the newly developed dynamic models integrate combined electric, magnetic, mechanic and thermal domains, considering also each domain influence on stress factors and – subsequently – loss of life. With the use of designed experiments and data pattern analysis, a substantial reduction of testing effort is possible by concentration of the most salient features of the load cycles. The main ageing mechanism investigated were demagnetization (PMSM), windings insulation loss of life (function of temperature) and mechanical stress (torque ripples). Several levels of models have been developed to address different simulation needs, from high frequency models up to fast reduced models. Induction machine models were developed following an equivalent circuit approach, integrating a thermal loss model and considering windings loss of life function of thermal stress. Permanent Magnet Synchronous Machine models include an iron loss model and consider magnet demagnetization function of thermal stress. Thermal model of the Permanent Magnet Synchronous Machine was also coupled with the Permanent Magnet Synchronous Machine (PMSM) model. The switched Reluctance Machine (SRM) models integrate an equivalent circuit with magnetic circuit to achieve higher prediction accuracy in terms of physical behaviour (mechanic, electric), losses (thermal, magnetic...) and model-based optimization for torque ripple reduction. Even if these drive-technologies present specificities and specific challenges and are addressed through different simulation tools and methodologies, the studies all follow the same scheme: electro-magneto-mechanical modelling, magnetic losses modelling, loss of lifetime or degradation or ageing (when applicable) evaluation, parameterization methodology and validation.
For PMSM, a new methodology was developed to calculate iron losses and induced eddy current losses in permanent magnets using a combination of analytical approach and magneto-static finite element calculation. Calculated losses in the case study of two types of PM motor are prepared for different operational conditions. Demagnetization in PMSM was also evaluated. The results of this case study demonstrate the torque performance drop due to partial demagnetization and the change of induced no-load voltage shape. This phenomenon can be used for testing and diagnostics of permanent magnet machine performance changes due to increased temperature. Completed study of partial demagnetization and its effects gives useful guidelines to overpass the demagnetization problem in permanent magnet machines.
A new IM model describing electric, magnetic, mechanic, thermic and ageing behavior was developed. It covers both symmetrically built and asymmetrically supplied machine. The thermal model is based on lumped circuit theory. The temperature distribution as output of thermal model feeds back electro-mechanic model to calculate proper value of temperature dependent stator and rotor winding resistances. The same output is used as an input for ageing model. The ageing model shows loss of life-time depending on winding temperature distribution Based on results, the proper cooling system and thermal management system can be established. During the development, a new method to calculate the induction machine parameters covering the nominal shaft power from 5 kW to 100 kW was established. The model was tested with induction machines from different European suppliers. The comparison is very good and shows a little difference between modelled results and measured quantities.
Switched Reluctance motor was studied using two complementary approaches. Reluctance network, exploiting the analogy of magnetic and electric domains, has proved to be a good compromise between the Finite Element method and analytical calculation regarding simulation speed and accuracy of the results. This method was validated on a first design both on static and dynamic operation. Improvement of SRM model accuracy was also addressed using a co-energy based approach accounting for coupling between the phases. A full numerical methodology was developed for motor accurate parameterization integrating accurate loss estimation, as well as optimal control generation to minimize losses and torque ripples. A critical progress was to achieve this full process without involving any physical prototyping, from CAD to drive virtual testing.
Beyond the electric motor models themselves, the integration capabilities of these models in full vehicle environment is critical. Despite the different simulation tools and electric machine addressed, all these models were developed following the same framework, allowing interoperable and substitutable models trough standardized ports and interfaces.
E-Motor Testing
As already stated before electric motors have been used in industrial applications (energy, machining) and transportation applications (rail, ship) for over a hundred of years. However, it is not possible to pick an industrial motor off the shelf and install it in a production car. The reason is that automotive applications differ in many ways from the applications in which E-motors have been used so far.
The main identified automotive electric motor specificities are:
- Environmental conditions: while industrial plants sustain specific and controlled environmental conditions, and while rail vehicles are adapted to specific exploitation conditions, automotive vehicles must be able to sustain a wide range of conditions (temperatures range, vibrations, moisture, saltiness...) without any specific adaptation, keeping the same durability goals (>10 years, >100.000 km).
- Integration: the electric motor use out of automotive applications usually didn’t have to deal with the strong automotive integration constraints. This level of integration requires to consider all component interactions (functional, electrical, mechanical, thermal...)
- Efficiency and power density: There is a strong push from automotive OEMs for increasing the power density and efficiency of E-motors even further. The increase in power density is mainly driven by the limited installation space inside a vehicle. A higher power density opens the possibility for new motor topologies such as e.g. in-wheel motors. The increase of efficiency is driven by the resulting reduction of energy consumption which has a direct influence of the range of the electric vehicle.
- Weight, costs and supply chain: addressing a highly competitive mass market, electric vehicles must be competitive, in terms of costs and performances, implying a specific stress on the electric motor costs, supply chain dependencies and weight. And while industrial and heavy transportation address a Total Cost of Ownership (TCO) at decades levels, the automotive domain mainly addresses acquisition costs. These constraints lead to tradeoffs in the electric motor technologies, manufacturing process, quality and materials, requiring specific investigations to ensure that these tradeoffs don’t impact the vehicle lifetime and performances significantly.
- Production process: The same type of electric motor needs to be installed in tens of thousands of vehicles. Due to the high competition in the automotive sector, electric motors need to be produced in a cost effective manner. This requires a highly automated production process.
- Fast transient use: whereas electric motors in industry and heavy transportation (rail, ships) are mainly running at a constant speed and experience only slow transients, electric motors in automotive applications experience high-speed transients.
- Regenerative use: most of dual mode (traction & regeneration) were used in the rail domain, more often for dissipation braking than real regenerative or rejecting braking. And even in these domains, the transient dynamics are far lower than the automotive ones.
Motivation for e-motor testing
Due to the specificities mentioned above, the E-motor test procedures developed for industrial applications and heavy transportation (rail, ships) are not suitable for the use in automotive. Specific test procedures are required which ensure the OEMs and carmakers that their E-motor will be suitable for their future electric vehicles. Whereas procedures for testing internal combustion engines against the specificities of the automotive environment are well established, testing procedures for automotive E-motors are relatively new. This motivates a further development of E-motor test procedures tailored to the automotive environment.
Accelerated testing
The term “accelerated testing” is used in two different contexts:
• In the context of reducing the time (and hence the cost) needed to perform a test.
• In the context of performing the test earlier in the design cycle of the electric vehicle and thereby reducing significantly the cost needed to solve issues.
New methodologies for accelerated testing are developed for the purpose of ageing of winding insulation, to combine simulation and measurements, such as e.g. parameter estimation and for operational NVH and efficiency testing
In ASTERICS several e-motor testing methods were investigated, defined and executed. As an example a vibro-accoustic test of an SRM (switched reluctance machine) was performed, since SRM’s have the tendency to vibrate and produce noise. The test setup consisted of accelerometers and microphones in the near of the e-machine. In total, sixty tri-axial accelerometers were mounted on the jacket and the side panels. Furthermore, four microphones were placed: two in the near-field, two in far-field. An acoustic insulation box was placed around the SR driveline to separate the acoustic motor noise from environmental noise. Localization of these sensors is a very important issue. The output of this study was a waterfall 3-D diagram of the frequency spectra over the speed of the SRM-rotor, which can be used further to analyze the main frequencies on tonal components of the produced vibration respectively noise. Such an analysis allow the acoustic engineer to identify potential improvements for the acoustic behavior and to adapt the NVH characteristic of the e-machine accordingly.
Further testing methods for IM-machine and PMSM (e.g. common and differential-mode impedances were measured with an impedance analyzer in the frequency range of interest) were developed and validated on a real test bed to ensure their applicability. The measurement data was also used to parameterize the e-machine models or at least to validate their goodness of fit.
Virtual prototyping
The virtual prototyping consisted of two phase studies: first the parameterization itself and then the exploitation of the models predictions. As parameterization of such advanced models are not straightforward, as requiring data that are usually not available and not directly measurable at industrial costs, the approach was to develop parameterization methodologies allowing to parameterize these models.
The main approaches followed were measurement or 2D/3D simulation-based. When based on measurement, specific tests were performed on the real device to identify parameters. When 2D/3D simulation-based, finite element models (magnetic, thermal...) simulations were used to identify parameters. These last approaches have the main advantage to not necessarily require a physical prototype.
The parameterized models were used to perform several types of virtual tests and experiments. Performance simulations, integrating all the multi-physics presented above, were the first tests done, in order to validate the methods, tools and models, first in terms of functionality, then in terms of accuracy. At this stage, several corrections were applied in order to ensure prediction accuracy and robustness. Various performance simulations for several drives, sizing and suppliers were performed and these results were validated versus matching drive data.
Thermal and ageing models of electric powertrain components were implemented into behavioral vehicle simulations. With that method, the ageing effects of different driving scenarios can directly be studied via the resulting warming of the E-Drive components, for instance. Such information can be used also for the acceleration of tests, e.g. via designed experiments (DoE). Such an approach was implemented and tested for a sub-cycle of the NEDC where most damaging scenarios for E-Drive components were extracted.
The investigated methods and the gained insight into E-Drive components’ ageing behavior helped to significantly reduce the test durations which can be reached by focusing on the most damaging load combinations. As load-dependent warming and ageing models have been developed for all electric motor components (E-motor, invertor, HV battery) and implemented in a vehicle simulation program. Therefore, the influence of vehicle load cycles can be traced directly to equivalent component stress. The effect of E-motor layout on its behavior in the vehicle and the associated protection and de-rating strategy can be directly estimated thus avoiding problems with drivability at a very early stage in a project. Testing is a very elaborate and costly process. Employing new data condensing methods, e.g. via Genetic Algorithms and Neural Networks, the concentration on the most preeminent features of the driving cycle is possible thus reducing testing effort and time.
Further, the SRM model (electrical, mechanical, magnetic) was validated against existing SRM measurements at different levels of the model. This procedure validated the parameterization method as well as the model itself. The resulting accurate models (especially regarding electro-mechanical behavior) make it possible to set up a control law optimization process allowing the SRM to reach higher efficiency (for a real drive cycle) while drastically reducing the typical torque ripple issues affecting SRM and their transmission durability (NVH impact). These new capabilities for the three machine technologies demonstrated the capacity to address electric drives durability issues following model-based approach, in the thermal (windings), magnetic (demagnetization) and mechanic (torque ripples) domains.
System design, integration and evaluation
A link between the simulation environments (Amesim, AVL-Cruise, GoFAST, GSP) was created by means of two different kinds of standard integration interfaces:
• Functional Mock-up interface (FMI)
• S-function interface
In particular, the potential of the Functional Mock-up Interface, in its Model Exchange and Co-simulation configuration, was investigated and the capability of each tool to support this standard was described. Before integrating physical models of different subsystems in the target vehicle applications, the capability of the different tools to export and import within each other was verified. Starting from simple mathematical models (gain, integrator, integrator + gain), the efficiency of each standard integration and the accuracy of the results was checked. After these preliminary tests, simple physical subsystem models of e-motor and battery were imported and successfully simulated in the full vehicle model of each partner. The Functional Mockup Interface has proven to be an efficient and more flexible tool because of the possibility to get integrated with simulators other than Simulink.
Different configurations of the vehicle in each environment by means of several cross sub-system model exchanges based on the FMI/FMU Standard were created. As a result, seamless coupling between Amesim, AVL-Cruise, Volvo and GoFAST Vehicle Simulation platforms was possible thanks to the close cooperation between all partners. For the integration, a common definition (standard) in terms of input/output signals of the sub system models was used. For electric machine DC voltage [V], Requested torque [Nm] and electric machine speed [rpm] was used as input and DC current [A], Torque [NM], Efficiency (EM + inverter) [/] and Overall losses (EM + inverter) [W] were used as output signals. For the battery Battery request current [A] was used as input and Battery voltage [V], Battery state of charge [%) and Battery power losses [W] were used as output signal. Moreover some parameters of the imported sub system models were modified according to the specification of the native corresponding models used in the vehicle. Each rescaling was carried out following the procedure suggested by the partner that provided the model to be imported.
This made the performances of the simulated vehicles very similar to the real ones. The above procedure was made possible through the custom interface of the FMUs.
The ASTERICS project delivered solutions to significantly enhance the development process of EV’s in short-term, so that faster development cycles with 50% less effort than before are feasible. Further, these methods and tools can be used to improve the performance, reliability and safety of EV’s and also hybrids, as similar modeling and testing capabilities are needed in all sorts of EV-variants.
Overall the ASTERICS project made a huge step forward in enabling the development of EV’s which are more competitive in the transportation sector and compelling for end users, which finally helps to reduce global CO2 emissions.
Potential Impact:
The primary goal of ASTERICS was to develop within 3 years from its start a “full next generation” vehicle model (from the virtual point of view) taking an existing vehicle as reference in order to achieve a decrease in development time by 50% and an improved vehicle efficiency by 20%.
Environmental impact of ASTERICS
Electric vehicles have a huge potential to support the requested CO2 reduction and achieving the climate protection goals of the EU. Although the real contribution to the reduction depends on the energy mix of the electricity, more or less most studies show a great benefit for the electric car. Especially when it comes to countries with very high percentage of renewable electrical energy, like Austria, Finland, Sweden or Latvia. The goal of the EU is a renewable energy rate of 20% by 2020 up to 27% in 2030 and an improvement for up to 50% in 2050. This target mix will enable much more CO2 reduction with electrical cars in future as it is possible nowadays. The current overall energy mix in the EU includes about 75% created via fossil fuels.
Europe is therefore committed to the Green Cars Initiative (GCI) which responds on the one side to the economic need to strengthen the automotive industry in Europe and on the other side to the need for a reduction of greenhouse gases along with a reduction of noxious emission caused by the transport industry. Initiatives in this direction shall lead to a reduction or avoidance of global warming, an improvement of public health and primary energy savings, thus moving towards more sustainable transport. To support these higher level goals road transport is required to move from conventional internal combustion engines (ICE) vehicles to pure electric vehicles (PEV) or hybrid electric vehicles (HEV).
Currently the performance - in terms of range, speed, reliability, durability and efficiency - of pure electric vehicles leaves much to be desired. To facilitate the transition from conventional ICE vehicles towards PEVs the performance must be doubled in order to have a realistic market chance whereas the cost of technology and development time has to be cut in half.
The technology of internal combustion engine is by large the most used for road transport vehicles. The proposal of different technologies for vehicles propulsion implies a number of modifications in the known industrial and social established habits. Changes could be needed in terms of production technologies, primary energy supply, material choice, while at the same moment also the support infrastructure has to be modified, in order to satisfy users’ expectations or to influence their acceptance. Before a deep change in the vehicle supply chain is started, a number of evaluations have to be performed to respond to question such as:
• what is the effective modification in terms of primary energy consumption for the deployment of a similar service using different vehicles technologies?
• what kind of performances are expected considering current state-of-the-art technologies?
• which technology is expected to have larger margins for improvement?
• what about the economical sustainability of the technology?
• what about the environmental sustainability of the technology?
• are there any possible future “bottlenecks” (e.g. resources having limited availability on national/continental/world scale) for the technology?
It is evident that in this kind of analysis the focus is not only on the vehicle itself but, in many case, on the general context where a fleet is going to operate; as a consequence, the expected result is related to general indicators such as primary energy consumption (e.g. MJ, or tons petroleum equivalent), resource consumption (renewable and not renewable) or to environmental impact indicators (e.g.: greenhouse gas emission; waste production; abiotic resource depletion potential; photochemical ozone creation potential; sometimes other indicators or a combination of them is also calculated). The build of so-called “Life Cycle Assessment” (LCA) is in many cases a suitable solution to offer a complete output for the analysis; the methodology is defined by the technical standard ISO 14040, which is aimed to reduce the arbitrariness and the uncertainties of the evaluation. LCA approach is able to summarize in a small number of indicators the impact related to the use of primary resources, of energy vectors, of the use of the fleet, of the technology applied; in general, of all the material and energy flows related to a so called “function unit”, that is the service or the product under examination. In case of a car, an example of functional unit definition is: a European, compact-sized, five-door gasoline vehicle for 5 passengers including a luggage compartment, and all functions of the defined reference scenario with a mileage of 150,000 km over 12 years, complying with the same emission standards.
Since full LCA studies have to take into account all the variables determining the impact of the vehicle on the environment, a very large amount of data is needed: material extraction, transport, production technology, component manufacturing, consumption during use phase, maintenance and end of life management are main examples. All of them are contributing to air/soil/water emission, to raw materials depletion and to any other relevant environmental impact that has to be taken into account. Similar general approach can be built also using economic indicators, thus defining the total cost of ownership through “Life Cycle Costing” (LCC) analysis. Considering that for vehicles (and, in general, for all durable goods) the use phase is by far the one producing the most relevant impact, the analysis can be sometimes focused only on it. The analysis of the use phase (sometimes called “fuel cycle” or “energy cycle”) is therefore a very important step to evaluate the environmental impact of the vehicle and it can be focused on different sub-analysis, main being:
1. well to tank: it includes the analysis related to fuel production/extraction and delivery
2. tank to vehicle: it includes on-board energy conversion and is mainly related on vehicle powertrain characteristics
3. vehicle to miles: it includes vehicle energy consumption and is mainly related to the driving/duty cycle selected.
4. an analysis including point two and three of the above list is also said “tank to wheel”; a full use phase analysis (including point 1, 2 and 3) is said “well to wheel”.
Tank to wheel analysis is therefore very related to the knowledge of vehicle technology and of its performances: at least, a reference mission and a fuel/energy consumption model are needed. Since in many cases the value of the analysis relies on the comparison of different alternatives, the model should be able to calculate the variability of the output (e.g. efficiency, energy consumption) on given impact factors (e.g. mass variation) over different scenarios (e.g. different driving cycle). A short literature review of preliminary analysis that was performed within ASTERICS showed that the tools used for similar calculation can be really different.
An electric vehicle doesn’t directly emit pollutants per se, but that doesn’t mean it doesn’t have an indirect environmental impact directly dependent to its energy consumption.
The pollutants emissions related to a full electric vehicle depend on the electricity energy mix and its global powertrain efficiency. It also depends on the type and time of battery charging, as the electric grid management can resort to different type of energy source depending on the consumption. In this respect the proportion of the slow, fast and ultra-fast charge is an important data to assess the indirect vehicle CO2 footprint. Even if there is a lack of global regulation regarding EV emission (like the EURO standards), public authorities in several countries (Germany, Japan, United States...) pay attention to the CO2 footprint of vehicles, meaning that, even if not shared and standardized, the CO2 indirect footprint should be taken in account as a vehicle design goal.
ASTERICS addressed the related issues by introducing a high-fidelity virtual modelling environment for e-components, realistic e-driving cycles and advanced testing concepts and tools for the entire e-drivetrain.
Within the lifespan of the project it was successfully demonstrated that current e-drivetrain concepts can be improved significantly by applying virtual development methods. It was further shown that testing efforts for e-components can be reduced while the quality of results can be increased. On that note ASTERICS is an important stepping stone towards the development and early market readiness of an “optimal” Full Electric Vehicle in terms of performance, reduced development and testing times and of course the related costs. Hence, a setup that is closer to the real use of the vehicle can be realised by using the ASTERICS tools already in the prototyping phase.
Strategic and economic impact of ASTERICS
The strategic shift of the energy mix towards renewables together with the shift of the private transport towards electrical cars will create huge environmental impact, but also huge economic impact. According to the report of the “International economic development council” from 2013 ( http://www.iedconline.org/clientuploads/Downloads/edrp/IEDC_Electric_Vehicle_Industry.pdf ), the introduction of electrical vehicles will create additional jobs in the automotive industry, battery development and production and in the research and development of electrical vehicles. It is also obvious that this trend will lead to some job losses in the oil industry however. Electric vehicles will lead to a decrease in overall energy consumption and support a better balance between day and night, because electrical vehicles will typically be charged at night, where electricity is currently cheapest.
Electric Vehicles (EVs) are the most promising technology to drastically reduce the environmental burden of road transport, i.e. emissions of CO2, air pollutants and noise of particularly passenger cars and light commercial vehicles. Within this context a study of CONCAWE-EUCAR JEC-WTW showed that PEV remains the most efficient solution in terms of well-to-wheel efficiency of the different power-trains using different types of primary energy sources. Carmakers are responsible for delivering the reductions. The company target is an average for all cars sold, not a fixed limit that no car may exceed. In this scenario, automotive manufacturers must seek and provide improved and innovative vehicle solutions to the open market, which are not only compliant with the energy and environmental conditions and with the new legal requirements, but they must also be competitive and increasingly appealing. In other words all the main global OEMs, including those based in Europe, are essentially being forced to find new technological solutions to respect the social and environmental needs (environmental friendly vehicles) on one hand and on the other hand meet the demands of the customer (like higher levels of driveability and comfort, better safety and styling) at acceptable costs (in terms of purchase and operation costs as the price of fuel continues to increase). There are currently several concepts for PEV (pure electric vehicle) and HEV (hybrid electric vehicle) that support this “clean mobility” demand. However, the development and improvement of the different concepts require a huge effort in analysis, design, implementation and testing, not to forget feeding back experience, results and knowledge to new generations of such cars.
Advanced modelling tools and testing procedures going from one-dimension to three dimensional approaches play a fundamental role in the optimization process during the earliest project phases for the energy dimensioning of PEV & HEV as well as their “energy management strategies”, while reducing the project’s development lead-time as well as to build up requirements for subsystems and their related control units. The ASTERICS’ simulation tool suit and testing procedures is designed to deliver these functionalities.
From the validation point of view, if a ‘new generation’ vehicle is available, during the prototyping phase, the only way to test it (from the energetic efficiency point of view) is to apply the usual test methods, on the roller bench test. Sometimes, the applied driving cycles (and, sometimes, the driver’s driving style during the test) shadows the real benefit of the next generation archetype. This would results in big efforts to repeat the experimental tests and, sometimes, loss of time to tune the simulation models without great results. Moreover, if the project is in the prototyping phase, a ‘try and fix’ loop could start, with further efforts and losses.
In the frame of ASTERICS two major software programs for fuel economy and energy management (AVL-CRUISE and LMS-AMEsim) were linked with a standard OEM simulation tool (Fiat Group-PerFECTS). This integration showcase created a set of shared tools that speed up the simulation time using the best capabilities of each simulation environment.
From the experimental perspective, within ASTERICS a set of roller bench test cycles were created, with the intent to fit the ‘next generation’ vehicles attitudes, taking into account climate conditions, driving style and, of course, the traction archetype in study. These aspects resulted in a more confident testing in the future, being the future test sessions based on the experience and tools developed within ASTERICS.
ASTERICS helped to make an important step ahead in the concept phase for non-conventional vehicles, taking into account the principal aspects related to both the e-power-train peculiarities and the real use of the vehicle (also in terms of drive style). Hence, in the prototyping phase, a set-up closer to the real use of the vehicle can be realized using the developed tools. As a result, the whole loop from concept to prototype will be more focused and faster leading to an estimated 20% energy efficiency increase and a 50% reduction of the development & testing time of the e-driveline of future generation PEVs.
The ASTERICS consortium investigated and demonstrated the way towards designing and building efficient, cost optimized vehicles with advanced testing and modelling approaches. This will further pave the way towards electrification of European’s private transport and support the overall targets of green mobility and de-carbonization, while maintaining and creating additional economic potential for the industry and academics in Europe.
Dissemination activities
The main objectives of ASTERICS dissemination activities are to:
• raise awareness of the community on ASTERICS activities, results and ASTERICS partners
• inform and educate the community about the research area concerned with advanced simulation and tests for system and components of battery electric vehicles (BEV)
• engage the community to receive input
• promote and exploit the output of the project inside the community
For this purpose several tools were set up and used. In order to represent ASTERICS in a standardized way project templates for reports, documents, meetings and presentations were prepared. The project logo was designed right at the start of the project and was used in all presentations, reports, documents, etc. that are related to the ASTERICS project in order to assure a uniform appearance.
Website – www.asterics-project.eu
The ASTERICS website provides information on the project and its activities. A download area was established for additional information like the project leaflet, project poster, newsletters and public deliverables. The website was updated on a regular basis with actual information in the news section and project activities performed. The website will be a central communication element. The website is accessible via www.asterics-project.eu and features many aspects of the project and the continuous work performed. Visitors of the site have the possibility to download public deliverables, newsletters and to contact the coordinator for further information on the project. The website will be accessible for five years after the project is closed.
Flyer and Newsletter
A general Flyer was created at the beginning of the project and distributed to all relevant stakeholders, using a confidential dissemination database containing over a hundred contact details. The flyer was also published on the website as a news item.
During the runtime of this project three Newsletter were prepared which are available for download on the ASTERICS website. In the first Newsletter the objectives of the project and systematic and comprehensive approach with its four building blocks were explained. The second Newsletter focused on the achievements within each single work package and the last Newsletter shows how each work package contributed to the achievement of the ASTERICS objectives.
Publications
The consortium set great value on a continuous re-presentation of the ASTERICS project to show the progress to a wide audience. Therefore ASTERICS was represented at several conferences to get into a conversation with e.g. relevant stakeholders and standardization bodies. Due to the presentations that were held at several occasions the consortium had many opportunities to promote debates to accelerate the implementation of the research results. Further, one book chapter and five technical papers were presented at conferences and are listed below.
Book chapter
Horst Pfluegl, Claudio Ricci, Laura Borgarello, Pacôme Magnin, Frank Sellier, Lorenzo Berzi, Marco Pierini, Carolien Mazal, Hellal Benzaoui. A framework for electric vehicle development: from modelling to engineering through real-world data analysis. Series Lecture Notes in Mobility: Electric Vehicle Systems Architecture and Standardization Needs, February 2015, pp 55-73
Technical presentations
Mathieu Sarrazin, Steven Gillijns, Jan Anthonis, Karl Janssens, Herman van der Auweraer, Kevin Verhaeghe. NVH analysis of a 3 phase 12/8 SR motor drive for HEV applications. EVS27, Barcelona, Spain, November 17-20, 2013, pp 1-10
Tomaž Munih, Damijan Miljavec. A Method for Accelerated Ageing of Electrical Machine Insulation. Mechatronika (ME), 2014 16th International Conference on Mechatronics, Brno, Czech Republic, 3-5 Dec. 2014, pp. 65-70
Ylva OLOFSSON, Jens GROOT, Tomaž KATRAŠNIK, Gregor TAVČAR. Impedance spectroscopy characterisation of automotive NMC/graphite Li-ion cells aged with realistic PHEV load profile: quantification of cell properties vs. temperature at different stages of ageing. V: 2014 IEEE International Electric Vehicle Conference, Florence, Italy – December 17-19, 2014. IEVC2014 : conference proceedings.: Institute of Electrical and Electronics Engineers – IEEE, 2014, pp 1-6
L. Berzi, M. Delogu, M. Pierini, G. Zonfrillo. Analisi del degrade di sistemi si accumulo energia per veicoli elettrici mediante simulazione estesa di scenari di uso. AIAS – ASSOCIAZIONE ITALIANA PER L’ANALISI DELLE SOLLECITAZIONI 44° CONVEGNO NAZIONALE, 2-5 SETTEMBRE 2015, – UNIVERSITÀ DI MESSINA, pp 1-10.
Chen C., Diwoky F., Pavlovic Z., and Wurzenberger J. An Application of the Linear and Time-Invariant Method for the System-Level Thermal Simulation of an EV Battery. SAE Technical Paper 2015-01-1197, 2015, pp 1-8
Exploitation of results
According to the ERTRAC road map, the future introduction of Electric Vehicles in the market will very strongly depend on the availability of functioning and cost efficient products from the suppliers. In their view, the first step of the implementation of electric vehicles was done by the adaptation of existing vehicles by 2012. In the intermediate step efficiency gains by all consumers and advanced system integration alongside high performance energy storage systems will become available by 2016. As the last milestone, powertrain systems providing unlimited range should be a fact by 2018-2020.
The European Green Car Initiatives’ multi-annual roadmap and long term strategy is in line with these milestones and states that the industries involved have agreed that eventually after 10 years the goal of an accumulated 5 million pure EVs and PHEVs on Europe’s roads may be achieved. The ASTERICS execution ran in parallel with these schedules, herewith facilitating the transition from milestone 1 (introduction) to milestone 2 (intermediate).
Exploitation by OEM partners (Volvo and CRF)
The commercial vehicles industry in Europe need to find transport solutions that are both efficient on the technological level and on the business level. The Project partners Volvo and FIAT built knowledge on e-driveline components and complete vehicles, defined and modelled drive cycle systems, integrated test results in use for model development for simulation and explored new concepts.
Furthermore, CRF is developing a global simulation platform in order to have analysis of fuel economy and transport efficiency on complete vehicle level, however, there is a need to further populate this simulation platform with alternative drivelines, e.g. hybrid and pure electric vehicles for city distribution.
Volvo developed a battery model and corresponding testing procedures for a large scale measurement and evaluation study, which results in a model for further use in design and concept of the next electric vehicle generation.
The main exploitable project results for OEMs are: vehicle simulation on e-driveline in comparison to a conventional vehicle. This system simulation is good for demonstration and optimisation of new e-vehicle based transport solution in a city environment. This creates the possibility to find vehicle concepts for internal product planning and product development.
Exploitation by Supplier partners
Siemens, AVL, GK, THIEN progressed on the modeling and simulation of electric motors and inverters as well as on testing methodologies for battery, inverter and electric motor testing.
Ultimately, the project results contribute to the development of innovative testing, measure and diagnostics services for electric drivetrains.
The impact of ASTERICS in the simulation and testing sector is an improvement of the competitiveness of the solutions, providing more accurate results and reduced system engineering time and efforts for the customers (OEMs and Tier1). The innovative aspect of developing new simulation models and test methodologies allows OEMs and Tier1 suppliers and the whole automotive sector to implement the results providing more energy efficient vehicles in less time with less efforts.
According to the supplier partners the main exploitable results are:
• New services (testing, methodology)
• New simulation model libraries (after the industrialization phase) for the use in office simulations and on the test bed
• New 4 channel battery testers (battery life cycle tests, fast charging, any charging power, impulse load, different current ripple factors)
• New product requirements and specifications for testing of e-drivetrains
These results will now be brought to a higher level allowing their commercialization; this action will reinforce and expand the commercial position of the participating suppliers.
Exploitation by academic partners (FH-J, UNIFI, UL)
The Universities improved their knowledge regarding real-world driving cycles, vehicle simulation frameworks, energy consumption assessment, battery-, inverter- and e-motor- ageing phenomena and the related environmental impact. The participation in the ASTERICS project was also aimed at a knowledge transfer from research to industry; to offer engineering students and young researchers a State-of-the-Art training on an emerging engineering field, which is very important in the next decades.
Some of the innovative results of ASTERICS among others are: better studies on system level inverter models, e-driveline system simulation, more detailed information on the behavior of e-driveline components (especially thermal and aging behavior); which will – via (inter)national lectures, meeting and conferences – spread knowledge among young students and researcher in the field.
In addition, the development of realistic driving cycles for PEV and the proposal of a management model for batteries and their integration provides to future engineers the knowledge to further help European industry to develop new products with reduced environmental impact and reduce production costs and time-to-market.
With this in mind, the target audience for the implemented results are not only automotive OEMs and Tier1 suppliers but also MSc and PhD students. Young people will be more aware of sustainable and electric mobility.
List of Websites:
ASTERICS website: www.asterics-project.eu
Contact details: DI Horst Pflügl
horst.pfluegl@avl.com
Office: +43 316 787 1587
Mobile: +43 664 8379 426