Energy and Thermal Management, Air-Conditioning, and Waste Heat Utilization: 2nd ETA Conference, November 22-23, 2018, Berlin, Germany

The volumes includes selected and reviewed papers from the 2nd ETA Conference on Energy and Thermal Management, Air Conditioning and Waste Heat Recovery in Berlin, November 22-23, 2018. Experts from university, public authorities and industry discuss the latest technological developments and applications for energy efficiency. Main focus is on automotive industry, rail and aerospace.

• Language: English
• Publisher: Springer
• Release date: Nov 4, 2018
• ISBN: 9783030008192

Book preview

Energy and Thermal Management

© Springer Nature Switzerland AG 2019

Christine Junior and Oliver Dingel (eds.)Energy and Thermal Management, Air-Conditioning, and Waste Heat Utilizationhttps://doi.org/10.1007/978-3-030-00819-2_1

Choice of Energetically Optimal Operating Points in Thermal Management of Electric Drivetrain Components

Carsten Wulff¹  , Patrick Manns², David Hemkemeyer², Daniel Perak², Klaus Wolff² and Stefan Pischinger¹

(1)

RWTH Aachen University, Institute for Combustion Engines, Forckenbeckstr. 4, 52074 Aachen, Germany

(2)

FEV Europe GmbH, Neuenhofstr. 181, 52078 Aachen, Germany

Carsten Wulff

Email: [e-mail]

Abstract

Increasing the efficiency of electric vehicles is a development focus in the automotive industry in order to reach the range targets set by customer requirements. Thermal management can have a positive effect on the system efficiency of electric vehicles. In this contribution, a simulation model of the drivetrain and cooling system of an electric vehicle has been build up. The aim is to investigate the influence of the cooling system control and resulting component temperatures on the drivetrain efficiency. Thus, energetically optimal target temperatures for inverter and motor can be identified and implemented in the cooling system control.

This approach goes beyond the state of the art control strategy of keeping the temperatures under the component protection threshold. Related research suggests that the component efficiency of inverter and motor can be increased by reducing their operation temperature. The simulation results in this article show that choosing target temperatures for inverter and motor below the components’ safety limit can have a small, positive impact on the system efficiency of the electric vehicle.

As the model is yet to be validated, these results implicate that the optimal component target temperatures for inverter and motor regarding system efficiency are below the protective limit. As a next step, the model will be validated with comprehensive component and vehicle measurement data in order to give a quantitative statement on the possible benefits of optimized thermal management control.

Keywords

Electric vehiclesThermal managementOptimal control

1 Introduction

Vehicle range shows to be a major contributor to the consumer acceptance of battery electric vehicles. As the battery capacity installed into a vehicle is limited by cost- as well as weight-considerations, one development focus for electric vehicles lies in the improvement of the system efficiency. [1] Thermal management is seen as a considerable factor in the system efficiency of battery electric vehicles [2].

This paper aims to investigate the effects of the cooling of electric drivetrain components on the system efficiency of a battery electric vehicle. To this end, a simulation model is developed which simulates the energy flows within the electric drivetrain of an A-Segment BEV.

The model includes map-based models for an inverter as well as motor and transmission, which simulate the effects of component temperatures onto their efficiency. The simulation model features comprehensive models for the cooling system as well as the vehicle longitudinal dynamics in order to simulate the system energy consumption. This model is used to determine the energy consumption of the drivetrain as well as the cooling circuit components under various ambient and operating conditions. Finally, an analysis of these results is conducted to find energetically optimal operating points and control strategies for the cooling system of battery electric vehicles.

2 Simulation Model

The simulation model is composed of three main parts:

1.

The drivetrain model, which consists of a simplified longitudinal dynamics model for the calculation of the loads for the drivetrain, and map-based models for the transmission, electric motor and inverter.

2.

The cooling circuit model, which consists of physical models for the coolant tubes as well as degas-bottle and map-based models for the coolant pump and radiator.

3.

The map-based underhood-model, which thermally links the other submodels by calculating the relative air speeds and ambient temperatures for all other components.

The model has been implemented in Matlab Simulink. The following sections provide a detailed description of these submodels.

2.1 Drivetrain Model

The drivetrain is modeled as an inverse model in which the desired vehicle speed from the drive pattern acts as an input to a signal path. Along this path the required power demand in order to follow the drive pattern is calculated (see Fig. 1).

../images/464817_1_En_1_Chapter/464817_1_En_1_Fig1_HTML.png

Fig. 1.

Signal flow in the inverse model of the drivetrain

Within the drivetrain model, the model control provides the desired speed and gradient to the vehicle model. In the vehicle model, the drive resistance resulting from the given drive pattern is being calculated with a simple longitudinal dynamics mode [3, 4]. The resulting wheel torque and speed are propagated to the transmission model. The transmission model calculates the resulting motor speed with the final drive ratio and uses an efficiency map to calculate the required motor torque. This efficiency map uses wheel torque and transmission oil temperature as inputs. The consecutive motor and inverter model also use efficiency maps to calculate the resulting power demand for the given drive pattern. These efficiency maps use the component temperatures as an additional dimension.

2.2 Cooling Circuit and Underhood Model

The cooling circuit model consists of physical models for the coolant tubes as well as the degas bottle. The models for the radiator and the coolant pump are map-based (see Fig. 2). For a given pump speed the pump model calculates the volume flow in the cooling circuit based in the resulting pressure drop of the cooling circuit. Motor as well as inverter are part of the cooling circuit model, with physical hydraulic models for the calculation of the pressure drop [5].

../images/464817_1_En_1_Chapter/464817_1_En_1_Fig2_HTML.png

Fig. 2.

Integration of drivetrain and cooling circuit into underhood model

The underhood model consists of a single air volume, which represents the thermal mass of the engine compartment air within the vehicle. The inlet air flow is calculated with a map depending on vehicle and fan speed. This airflow is zero when the radiator shutter is closed. The radiator shutter also changes the drag resistance coefficient within the vehicle model depending on its state.

2.3 Energy Flows Within the Model

As this model is designed to simulate the influence of the cooling system on the electric drivetrain components, in addition to the electric and mechanical energy flows all relevant thermal energy flows are modeled. This enables a more precise prediction of the temperatures of the electric drivetrain components.

The thermal energy flows that have been included comprise all heat transfer mechanisms. The radiation losses towards the engine compartment are modeled physically based on the components’ temperature, surface area and emissivity. Conductive heat transfer is considered between the transmission and motor, as those are physically joined in the reference vehicle. Conductive heat transfer is also considered within the thermal networks that model each components’ thermal behavior. The amount of conductive heat transfer is determined by the temperature difference between the thermal masses and fixed thermal resistances.

Three thermal masses are considered for the motor, the component housing, coolant within the component and abstract inner thermal mass to simulate the relevant temperatures for the component efficiency maps. For the motor, the temperature of the inner mass represents the stator temperature. For the inverter, the thermal masses of the housing is combined with the inner thermal masses. The resulting temperature of the inverter’s thermal mass aims to simulate the temperature of the power electronics. A separate thermal mass for the coolant is also part of the inverter model. For the gearbox, only two thermal masses are considered. These are the combination of the gears and housing and the oil.

The convective heat transfers considered are those between the components and the engine compartment air as well as the heat transfer to the coolant circulating between the drivetrain components. For these physical models, the heat transfer classes as described in [6] are used. Also, the heat losses from the coolant tube surfaces to the engine compartment are considered. For the heat transfer via the radiator, a map-based approach is used, while the pump and degas-bottle are considered as adiabatic.

2.4 Model Parametrization

As the main aim of this article is to investigate the effects of the cooling circuit on the system efficiency, the parametrization of the drivetrain components’ efficiency maps is crucial. The efficiency maps for the components within this model are not only dependent on speed and torque, but also on the component temperature. This enables a simulation of the temperature-dependent behavior of the drivetrain. As the efficiency maps that are provided by the manufacturers do not reflect the component temperature, several assumptions have to be made in order to model the behavior of the drivetrain components. The process of generating these temperature-dependent efficiency maps shall be explained in the following chapters.

Inverter Efficiency Map.

The inverter efficiency map provided by the manufacturer has been measured at a constant coolant temperature. Information concerning the actual temperature of the different inverter components at the time of measurement is not available. Therefore, the temperature of the IGBTs and diodes have to be estimated in order to separate the influences of the load and the device temperature within the efficiency map. It is assumed, that the efficiency map has been measured in stationary conditions and that all power losses within the inverter are dissipated by the coolant. Furthermore, it is assumed that the temperature of the IGBTs and diodes TInverter is equal to the derating temperature of the device TDerate when it is operated at maximum continuous load. In peak load conditions, the junction and diode temperatures are assumed as being equal to the derating temperature. For operating points below the maximum continuous inverter load, the junction and diode temperatures are assumed to be proportional to the inverter power loss in that point. At no load, these temperatures are assumed to be equal to the coolant temperature TCoolant. The inverter temperatures are calculated using (1)

$$ T_{\text{Inverter}} { = }T_{\text{Coolant}} + \left( {T_{\text{Derate}} { - }T_{\text{Coolant}} } \right)\frac{{\frac{{P_{{{\text{act}}.}} }}{{\eta_{{{\text{Mot}},{\text{act}}}} }} (\frac{1}{{\eta_{{{\text{Inv}},{\text{act}}}} }}{ - 1)}}}{{\frac{{P_{{{\text{Max, cont}}.}} }}{{\eta_{{{\text{Mot}},c{\text{ont}}}} }} (\frac{1}{{\eta_{{{\text{Inv}},{\text{cont}}}} }}{ - 1)}}} $$

(1)

with PAct as the actual motor outlet power of a given point within the efficiency map of the inverter, PMax,cont as the maximum continuous motor outlet power in the efficiency map of the inverter and ηMot,act, ηInv,act, ηMot,cont, and ηInv,cont, as the respective efficiencies in these points.

For the determination of the temperature-dependant losses of the inverter, the approach developed by Feix et al. [7] is used. According to this approach, the switching losses as well as the conduction losses can be calculated by using correlations. For the conduction losses, Feix et al. [7] provide Eq. (2)

$$ P_{\text{on}} \left( T \right) = \left( {\left( {\frac{{f_{V} { - 1}}}{{ 1 0 0 \,^{ \circ } {\text{C}}}}T + \frac{{ 5 { - }f_{V} }}{4}} \right)V_{{{\text{T0,25}}\;^{ \circ } {\text{C}}}} + \left( {\frac{{f_{R} { - 1}}}{{ 1 0 0 \,^{ \circ } {\text{C}}}}T + \frac{{ 5 { - }f_{R} }}{4}} \right)R_{{{\text{on,25}}\;^{ \circ } {\text{C}}}} I_{\text{on}} } \right)I_{\text{on}} $$

(2)

with $$ T $$ as the inverter temperature, $$ f_{V} $$ and $$ f_{R} $$ as material-specific factors and $$ R_{{{\text{on,25}}\,\,^{ \circ } {\text{C}}}} $$ and $$ V_{{{\text{T0,25}}\,\,^{ \circ } {\text{C}}}} $$ as device-specific parameters. The conduction losses $$ P_{\text{on}} $$ then result depending on temperature and the current $$ I_{\text{on}} $$ , which is assumed to be equal to the DC-current of the inverter. The DC-current can be calculated from the Inverter input power and the Voltage of the DC source.

For the switching losses Feix et al. [7] provide another Eq. (3)

$$ E_{\text{SW}} \left( T \right){ = }\left( {\frac{{f_{T} { - 1}}}{{ 1 0 0 \,^{ \circ } {\text{C}}}}T + \frac{{ 5 { - }f_{T} }}{4}} \right) \frac{1}{{f_{T} }} E_{{ 1 2 5\,\;^{ \circ } C}} $$

(3)

with $$ f_{T} $$ and $$ E_{{ 1 2 5\,\,^{ \circ } C}} $$ as device specific parameters and $$ E_{\text{SW}} $$ as the resulting switching energy. Multiplied with the switching frequency, this results in the switching losses.

For the determination of the temperature-dependent maps, the device-specific parameters need to be known. To this end, a regression analysis is done with the known temperatures, currents and losses from the given efficiency map for the prior calculated temperatures in the given map (4)

$$ P_{\text{Loss}} { = }a E_{\text{SW}} \left( T \right) f_{\text{SW}} { + }b P_{\text{on}} \left( T \right) $$

(4)

where $$ P_{\text{Loss}} $$ is the power loss in a given point of the efficiency map, a and b as weighing factors and the constant switching frequency $$ f_{\text{SW}} $$ . As a result of the regression analysis, a set of parameters is created which can be used for the generation of the temperature-dependent efficiency map for the inverter.

Motor Efficiency Map.

For the motor efficiency map, assumptions have to be made as well due to limited information at hand. It is assumed, that the temperature-dependency of the motor losses are mainly linked to the copper losses. Therefore, the temperature-dependency of friction losses and iron losses within the motor is neglected [8]. The copper losses of the motor are assumed to be solely linked to the known phase resistance of the motor (5)

$$ P_{\text{Loss, copper}} \,{ = }\, R\left( T \right) (I_{q}^{2} { + }I_{d}^{2} ) $$

(5)

where the copper losses of the Motor

$$ P_{\text{Loss, copper}} $$

are a result of the linearly temperature-dependent phase resistance $$ R\left( T \right) $$ and the two components of the phase current $$ I_{q} $$ and $$ I_{d} $$ [8].

In order to calculate a temperature-dependent efficiency map, the losses in the known motor efficiency map need to be linked to respective temperatures. The approach applied here is analog to the one applied to the inverter. With the known temperatures and currents for the efficiency map of the motor, the copper losses can be calculated within the given efficiency map. When deduced from the total losses, a map of constant residue losses remains, which is not assumed to be temperature-dependent. The full efficiency map for the motor is then calculated by adding the temperature-dependent copper losses according to (5) to the map of residue losses for different temperatures, thus adding the third dimension to the efficiency map.

Further Parametrization.

The further parametrization of the model is being done by using maps and parameters as provided by the manufacturers of the components within the reference vehicle. For a comprehensive overview of the vehicle specifications, please refer to the Annex.

3 Simulation Approach

As this contribution aims to evaluate the influence of the cooling system on the drivetrain efficiency of an electric vehicle, the choice of the control strategy for the cooling system is crucial for this investigation. Also, the choice of boundary conditions for the simulation strongly influences the results. The control strategy as well as the choice of boundary conditions are subject of the following chapters.

3.1 Cooling System Control Strategy

The control strategy for the cooling system applied within the model aims to control the temperatures of the inner thermal mass of the electric motor and the inverter. The temperatures of the inner thermal mass determine the efficiency of the component together with the load point in the efficiency map. Therefore, the aim of the cooling circuit controller is to keep the component temperatures below a desired target temperature at minimum power consumption.

The actuators, which need to be controlled, are the coolant pump, the vehicle fan and the radiator shutter. While all component temperatures are well below the desired target temperature, the radiator shutter is closed and coolant pump and fan switched off. When the component temperature reaches within 5 °C of the set target, the radiator shutter is opened, enabling an airflow through the engine compartment. This affects the drag resistance in the vehicle model. When the component temperature reaches the desired target temperature, the coolant pump is switched on. The pump speed is controlled by a PI-controller depending on the deviation of the component temperature from the set target. If the deviation increases even if the pump has reached full speed, the vehicle fan is engaged and also controlled with a PI-controller depending on the component temperature. This control strategy is engaged when either of the components reaches its target temperature, with the maximum of both temperature deviations being the input for the controllers.

3.2 Boundary Conditions

The boundary conditions for the simulations carried out in this investigation refer to the choice of driving cycle, ambient temperatures, start temperatures of the components as well as the target temperatures set for the control of the cooling system. The WLTP Class 3 is an industry standard in the evaluation of the power consumption of both conventional and electric vehicles [3]. This representative driving cycle is chosen for the evaluation of the drivetrain power consumption, as effects are evaluated on a system level. Ambient temperatures of 20 °C and 40 °C are chosen to be evaluated in order to compare normal and higher load conditions of the cooling system. Also, the starting temperatures of the components are varied in order to evaluate the effect of the components’ thermal mass on the load for the cooling system. Finally, the target temperatures for the inverter are varied in a range between 60 °C and 130 °C and for the motor in a range between 60 °C and 140 °C. Table 1 provides an overview over the different boundary conditions set for the simulations.

Table 1.

Boundary conditions for the simulations

With this set of boundary conditions, 360 simulations have been carried out in total. The following chapter gives an overview over the main findings that can be deduced from the simulation results.

4 Simulation Results

The simulations carried out in this contribution show that the impact of the cooling system on drivetrain efficiency is very low. Figure 3 gives an overview of the drivetrain efficiency for three different boundary conditions. For an ambient temperature of 20 °C, the component start temperatures are also set to 20 °C. In this case, the target temperatures of the components for the cooling system control are equal to their maximum allowed temperature. This means that the cooling system in this case only acts to prevent damage to the component. The same control strategy is applied in a case where the ambient temperature is set to 40 °C and the component start temperatures are set to 60 °C. This Scenario reflects an operational scenario for the cooling system with higher load. For the same ambient and starting temperature conditions, the drivetrain efficiency is also shown for a simulation with the target temperatures for the cooling system set to the energetically optimal temperatures of 90 °C for the inverter and 100 °C for the motor.

../images/464817_1_En_1_Chapter/464817_1_En_1_Fig3_HTML.png

Fig. 3.

Comparison of electric drivetrain efficiency in WLTP Class 3 for different ambient conditions and component target temperatures

The results show that the drivetrain efficiency actually increases for higher component temperatures, which is mainly due to the increased gearbox efficiency at higher oil temperatures. Furthermore, it can be seen that the drivetrain efficiency does not change significantly depending on the target component temperatures set for the cooling system control. With the optimum component temperatures, the energy required for the completion of the drive cycle can be reduced by merely 0,05%. This means that the energy required for the coolant pump as well as the compensation of the additional drag by opening the radiator shutter almost completely outweighs the inverter efficiency gains by reducing the component temperature.

The main contributor to this phenomenon seems to be the fact that the cooling system actually is not required to be active for the most part of the drive cycle. Figure 4 gives an overview of the component temperatures in 20 °C ambient and component starting temperature conditions. It can be seen that the maximum component temperatures are not reached for neither inverter nor motor.

../images/464817_1_En_1_Chapter/464817_1_En_1_Fig4_HTML.png

Fig. 4.

Component temperatures and system efficiency for WLTP class 3 at 20 °C ambient and component starting temperature with inactive cooling system

A similar behavior can be seen for higher ambient and component starting temperatures (see Fig. 5). The inverter reaches its maximum temperature only towards the end of the WLTP at about 1600 s, which means that the cooling system does not need to be activated before that.

../images/464817_1_En_1_Chapter/464817_1_En_1_Fig5_HTML.png

Fig. 5.

Component temperatures and system efficiency for WLTP class 3 at 40 °C ambient and 60 °C component starting temperature with component maximum temperatures as target for the cooling system

If the cooling system is controlled in such a way that the minimum energy is required to complete the drive cycle (see Fig. 6), it needs to be activated much earlier and therefore also requires energy for conditioning much earlier. This almost outweighs the efficiency gains for the inverter which thus can be achieved.

../images/464817_1_En_1_Chapter/464817_1_En_1_Fig6_HTML.png

Fig. 6.

Component temperatures and system efficiency for WLTP class 3 at 40 °C ambient and 60 °C component starting temperature with optimum target temperatures for the cooling system

In conclusion, it can be said that the control of the cooling system can have a positive effect on the drivetrain