Scholarly article on topic 'Development of Advanced Conventional and Hybrid Powertrains by Mechanistic System Level Simulations'

Development of Advanced Conventional and Hybrid Powertrains by Mechanistic System Level Simulations Academic research paper on "Mechanical engineering"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{"System Level Vehicle Model" / "Hybrid Electric Vehicle" / "Conventional Vehicle" / "Transient Drive Cycle" / "Powertrain Performance" / "Engine Out and Tailpipe Out Emissions^p"}

Abstract of research paper on Mechanical engineering, author of scientific article — Tomaž Katrašnik, Johann C. Wurzenberger

Abstract The paper presents a mechanistic system level simulation approach for modeling hybrid and conventional vehicles. The paper addresses the dynamic interaction between the different domains: internal combustion engine, exhaust aftertreatment devices, electric components, mechanical drive train, cooling circuit system and corresponding control units. Both vehicle topologies are powered by a spark ignition and compression ignition engine. Analyses concentrate on the transient phenomena caused by high interdependency of the sub–systems. Thereby the applicability of mechanistic system level models to adequately represent specific characteristics of the components is highlighted. To achieve high fidelity results of multi–domain simulations featuring high predictability and high computational speed it is necessary to develop adequate simulation tools considering all characteristic time scales of different domains and the nature of their interaction. Analyses are based on the verified models powertrain models. Simulation results of vehicles driven according to a legislative cycle provide the basis for comparative analyses of energy efficiency and exhaust gas emissions.

Academic research paper on topic "Development of Advanced Conventional and Hybrid Powertrains by Mechanistic System Level Simulations"

Available online at www.sciencedirect.com

SciVerse ScienceDirect

Procedia - Social and Behavioral Sciences 48 (2012) 3371 - 3388

Transport Research Arena- Europe 2012

Development of Advanced Conventional and Hybrid Powertrains by Mechanistic System Level Simulations

Tomaz Katrasnika'*9 Johann C. Wurzenbergerb

"University of Ljubljana, Faculty of Mechanical Engineering, Askerceva 6, 1000 Ljubljana, Slovenia bA VL List GmbH, Hans List Platz 1, 8020 Graz, Austria

Abstract

The paper presents a mechanistic system level simulation approach for modeling hybrid and conventional vehicles. The paper addresses the dynamic interaction between the different domains: internal combustion engine, exhaust aftertreatment devices, electric components, mechanical drive train, cooling circuit system and corresponding control units. Both vehicle topologies are powered by a spark ignition and compression ignition engine. Analyses concentrate on the transient phenomena caused by high interdependency of the sub-systems. Thereby the applicability of mechanistic system level models to adequately represent specific characteristics of the components is highlighted. To achieve high fidelity results of multi-domain simulations featuring high predictability and high computational speed it is necessary to develop adequate simulation tools considering all characteristic time scales of different domains and the nature of their interaction. Analyses are based on the verified models powertrain models. Simulation results of vehicles driven according to a legislative cycle provide the basis for comparative analyses of energy efficiency and exhaust gas emissions.

© 2012Published by Elsevier Ltd. Selection and/or peer review under responsibility of the Programme Committee of the T ransport Research Arena 2012

Keywords: System Level Vehicle Model; Hybrid Electric Vehicle; Conventional Vehicle; Transient Drive Cycle; Powertrain Performance; Engine-Out and Tailpipe-Out Emissions

1. Introduction

The development of future powertrains is strongly driven by steadily increasing demands on improvements in fuel consumption, emissions and performance. A promising approach is to combine the

* Corresponding author. Tel.: +386-(0)l-4771-305; fax: +386-(0)l-4771-310. E-mail address: tomaz.katrasnik@fs.uni-lj.si.

1877-0428 © 2012 Published by Elsevier Ltd. Selection and/or peer review under responsibility of the Programme Committee of the Transport Research Arena 2012

doi:10.1016/j.sbspro.2012.06.1302

conventional combustion engine with powertrain components like turbocharging devices, electrical propulsion components and dedicated exhaust aftertreatment devices. An efficient development process should thus address both the development and optimization of the individual components and also their integration on the system engineering level. Dynamic interactions among various components, their multidisciplinary nature and their different characteristic response times additionally increase the complexity of the system analyses.

Nomenclature

a Element acceleration vector (m/sA2)

B Capacity matrix (variable)

B Momentum-to-mass connection matrix (-)

c Constrain matrix for accelerations (-)

c Capacitance (C/V)

D Constrain matrix for momentums (-)

F Flux vector (variable)

f generic right-hand-side function

g constrain function

H Enthalpy flow (W)

I Current (A)

M Mass matrix (kg.mA2)

m Mass (kg)

m Element momentum (Nm)

P Power (W)

R Electrical resistance (Q)

S System matrix (variable)

t time (s)

u Voltage (V)

u Specific Internal Energy (J/kg)

w Mass fraction (kg/kg)

X State vector (variable)

X Momentum (Nm)

® State vector (variable)

9 Angular position (rad)

ra Angular velocity (rad/s)

Abbreviations

BMEP Brake Mean Effective Pressure

CI Compression Ignited Engine

CO Carbon Monoxide

DOC Diesel Oxidation Catalyst

DPF Diesel Particulate Filter

ECU Engine Control Unit

EG Electrical Generator

EM Electric Motor

HC Hydrocarbons

HEV Hybrid Electric Vehicle

ICE Internal Combustion Engine

NOx Nitric Oxides

PHEV Plug-In Hybrid Electric Vehicle

ROHR Rate-of-heat-Release

SI Spark Ignited Engine

SCR Selective Catalytic Reduction converter

TC Turbocharger

TWC Three Way Catalyst

UDC Urban Drive Cycle

The application of system level simulations is therefore indispensable (see Gschweitl 2007) especially during the early concept design phase to efficiently support analysis of a large number of system topologies and configurations. To efficiently support the development during early concept and design phases it is required that simulation models feature a high level of predictability at low computational expenses. Especially the predictability strongly influences the potential of system simulations to support the development of power trains in very early development stages where measurements are not yet (fully) available. Here, physically based models ensure a high reliability of results when optimizing power trains, since a mechanistic modeling approaches enable adequate response to changed parameters of individual components and adequate interaction between the components on the system level.

Modeling of hybrid powertrains on a system engineering level covers several different physical domains: the vehicle, drivetrain, electrical circuit, the engine including exhaust aftertreatment and the cooling system. Particular domains or combinations of domains were already intensively investigated in literature using a broad variety of models reaching from course map-based approaches to more dimensional, typically non-real-time simulations. A brief review on recent system level publications shows increasing interest in assessing not only performance and fuel consumption but also in giving

estimates on tail pipe emissions for HEVs and PHEVs (Plug-in Hybrid Electrical Vehicle), which imposes additional complexity compared to the system level analyses of conventional vehicles (see Lee and Filipi

(2010), Simpson et al. (2009), Lindenkamp et al. (2009), Smith et al. (2010) and Gao et al. (2011)). One common approach of all cited studies is that they do not rely on mechanistic engine models. This needs to be considered when assessing the model accuracy during transient driving conditions, the predictability when investigating different control strategies and particularly when scaling specific components.

The application of predictive approaches significantly reduces the time needed to set up the models, since potentially only a limited number of physically based parameters need to be adjusted. This reduces the workload compared to the required effort for developing non-mechanistic engine model approaches like maps (see Boland et al. 2010, Millo et al. 2010) or surrogate models (see e.g. Isermann and Müller 2001), which need to be re-populated or re-trained whenever changes in the hardware configurations are performed. It also needs to be considered that non-mechanistic, surrogate engine models are generally valid only within the trained data range. Moreover, sophisticated methodologies and sound basis of reference data are needed when generating surrogate models (see Deregnacourt and Wurzenberger 2011) capable of covering transient effects caused by turbo-lagging, exhaust gas recirculation and cold-start. Here hybrid-models are often applied combining surrogate cylinder approaches with a physical based mean value description of the air path. The cylinder models typically use Static Neural Networks (see Winsel et al. 2004), He and Lin 2007) or Support Vector Machines (see Canova et al. 2005) trained on experimental data or a high fidelity reference model.

The application of pure mechanistic engine and even overall system models is advantageous, since they inherently reflect the impact of changes in engine controls and engine configuration on the thermal behavior of the engine, on the variations in gas state and composition and turbocharger lagging if a turbocharger is included. Moreover, mechanistic engine/cylinder models are capable of evaluating the actual in-cylinder air-fuel ratio on a mechanistic basis, which is a major benefit compared to the pure map based engine outlet emission models. Thus, a mechanistic modeling approach enables a more adequate modeling of engine performance, fuel consumption, pollutant production and exhaust aftertreatment.

However, solving all domains with unique time stepping that is dictated by the domain with the shortest characteristic response time would certainly not comply with the fast computational times needed in system level simulations. Therefore, a novel approach relying on innovative time domain coupling of individual simulation domains is presented in the paper. The approach is based on tailored solver technologies taking into account different time scales of different domains. Such an approach is crucial to comply with the typically contradictory goals on numerical efficiency and high fidelity of the mechanistic based models. Since the characteristic time scales of all addressed domains differ significantly the overall system matrix is stiff as discussed by Pfau et al. (2011). Therefore, a selective and innovative coupling of individual simulation domains is needed to optimize the computational expenses while preserving high fidelity of the results.

In order to highlight the complex interaction between the internal combustion engine, exhaust aftertreatment, cooling, driveline, electrical network and controls the present work aims to analyze the performance and emissions of an HEV in comparison to a conventional powertrain during transient Urban Drive Cycle (UDC). Both investigated powertrains are powered once by turbocharged SI (Spark Ignited) and once by turbocharged CI (Compression Ignited) direct injection engines. In this context, the cross-influences between control strategies, turbocharging, engine heating and catalyst conversion are investigated in the light of energy consumption and exhaust emissions. All analyses are performed by a real-time capable system engineering simulation package AVL BOOST RT (2011) and AVL CRUISE

(2011). Both tools are based on one single software framework and solver technology allowing simulating system engineering models in a most coupled manner without performing any kind of co-simulation.

2. Model

The schematics of the conventional vehicle and HEV models analyzed in the present study are shown in Figure 1. The model of the turbocharged CI engine with indicated domains is shown in a separate Figure 2 to preserve readability of the figures. The turbocharged SI engine features a similar air path topology as the CI engine, whereas it does not feature the exhaust gas recirculation line. The exhaust aftertreatment systems of the lambda controlled SI comprises a standard TWC (Three Way Catalyst) and the one of the lean operated CI engine comprises a comprehensive assembly of a DOC (Diesel Oxidation Catalyst), DPF (Diesel Particulate Filter) and urea SCR (Selective Catalytic Reduction converter). All control models are first built in Simulink, are then exported via Real-Time Work Shop and are finally dynamically linked to the engine and the HEV drivetrain model, respectively.

Fig. 1. Schematics of the (a) HEV and (b) conventional vehicle models analyzed in the present study with indicated domains; yellow - mechanical system, blue - electrical system, grey - external Matlab/Simulink HEV controller and red - internal combustion engine shown in Figure 2

2.1. Driveline System

The vehicle and the driveline can be seen as a mechanical multi-body system represented mainly by the rotational degree of freedom. According to Pfau et al. (2011) such systems can be represented by

(M B ^ ( a ^ (

where M is the diagonal mass matrix and B represents the acting of momentums between the components and on the components. C describes constraints between the mechanical parts and D covers the relation between the momentums. The accelerations of the individual components are represented by a . m describes the momentums in the system, r comprises all momentums calculated in the individual components which acts on the mechanic system as well as losses.

2.2. ElectricalSystem

In hybrid electric powertrains, the electrical and mechanical systems are strongly coupled (see Figure 1). However, typically electrical system responds much faster compared to the mechanical system (Pfau

et al. 2011). Following Kirchoff's law and the bond-graph approach all currents (I) in the nodes sum up to zero. As discussed by Pfau et al. (2011) the applied electric system consists of resistors and capacities that typically depend on current, voltage but also on the mechanical state of attached mechanical components. On the system level, inductivities might be neglected (Pfau et al. 2011), which enables by introducing the voltage as an additional variable to reformulate the governing equations of the electric system to yield an index of one (Pfau et al. 2011).

The characteristics of the generator and e-machine are modeled by the efficiency or the power loss and maximum torque which typically depend on different aspects of the state of the machine. The efficiencies or power losses are given in maps as a function of rotational speed and torque. The input data are taken from specification sheets of the electric machines. The battery is modeled as an ideal voltage source depending on the actual SOC (State Of Charge), possible additional dependency and an equivalent circuit of resistances and capacities. The equivalent circuit models the different electrical losses of the battery. In the analyzed case the NiMH battery module features a nominal voltage of 7.2V and provides a maximum charge of 6.5Ah. The battery pack is assembled of 28 individual modules connected in series configuration.

Fig. 2. Topology of the CI engine model with the exhaust aftertreatment with indicated domains and different transfer paths; cyan - gas path of the engine, dark blue - fuelling, grey - mechanical part, red crossed - heat transfer functionality, green - liquid cooling circuits, light blue - exhaust aftertreatment, red - vehicle model shown in Figure 1

2.3. Engine

The engine model used in the present study describes the transport of mass, energy and species through typical engine components such as air cleaner, compressor, inter-cooler, throttle valve (SI engine), cylinder, turbine and catalyst. Internal combustion engines (Figure 2) typically feature three different characteristic time scales. The shortest time scale is associated with the in-cylinder phenomena, wave dynamics in the engine manifold and torque oscillations at the engine shaft. A larger time scale is associated with filling and emptying the engine manifolds during transient operation of turbocharged engines or during changed throttle position. Third, the largest time scale is given by the thermal response of the engine.

Modeling of the complete engine considering the shortest time scale would certainly lead to the highest accuracy of the results. However, despite the convincing physical depth of the ID gas dynamic models these simulations (see Regner et al. 2000) typically suffer from rather long computational times.

In the present study, the complete hybrid vehicle model considering all domains targets to run close to or even faster than real-time. To achieve the required computational performance target the engine model was subjected to several optimization measures such as code optimization and tailored model depth reduction. The engine model therefore consists of a crank angle resolved cylinder model embedded in a mean value based gas path description (Figure 2). For engines that do not exhibit significant enhancements of volumetric efficiency due to wave dynamics it is acceptable to neglecting these effects (see Hrauda et al. 2010). The applicability of the chosen approach for the analyzed engines is also confirmed in the results of the present work. As discernable from Figure 2, the mean value model covers the interaction of the gas path with the turbocharger dynamics and also describes EGR on a mechanistic basis. This is crucial for simulating transient engine operation (see Wurzenberger et al. 2010).

To ensure a high level of predictability and accuracy during transient engine simulations it is necessary to model the cylinder with mechanistic models on a crank angle basis (see Wurzenberger et al. 2009, 2010). Thereby, in-cylinder heat and mass transfer as well as combustion phenomena are modeled on a physical basis, which is beneficial for capturing transient results with an appropriate accuracy. The cylinder is described by state-of-art single and two zone models (see Wurzenberger et al. 2011a). The CI engine in-cylinder combustion is modeled by ROHR tables given at 35 different engine operating condition (see Wurzenberger et al. 2010). The ROHR tables are derived from indicating measurements using the online Gas Exchange and Combustion Analysis tool GCA (see Leifert et al. 2008). The SI engine in-cylinder combustion is modeled by the Vibe model where the variation of Vibe parameters at different engine operation conditions is again derived from the indicating measurements.

One of the aims in this analysis is also to appropriately describe and compare engine outlet and tailpipe emissions of the vehicle during transient legislation cycles. In terms of modeling engine outlet emissions, the evaluation of the actual in-cylinder air-fuel ratios on the mechanistic basis is one of the major benefits of the crank angle cylinder approach compared to purely operating map based emission models. In this study two approaches are pursued to analyze and validate different emission modeling approaches. In the CI engine model the two zone approach enables evaluation of a burned zone temperature, which is together with species concentrations and the pressure used to model kinetically driven NOx, CO and soot formation, whereas in this study special emphasis is put to the NOx emissions. Due to the fact that a large number of experimental data is available for the emissions of the SI engine it is intended to develop a model taking advantage of the extensive experimental data pool and simultaneously using the benefits of the crank angle cylinder model. Therefore, the experimental data are used to establish correlations between emissions and other thermodynamic and combustion parameters including an adequate response to air-fuel ratio variations. The considered emissions are carbon monoxide, hydrocarbons and nitric oxides which are of special importance in stoichiometrically operated engines.

Mean value gas path models typically link so-called storage and transfer components in an alternating manner. This approach follows the a-causal (not signal oriented) bond-graph theory as presented by Wurzenberger et al. (2009, 2010, 2011a, 2011b, 2012).

2.4. Catalytic Converters

Transient ID models can be seen as a reasonable compromise to describe with sufficient accuracy single catalysts or also entire exhaust lines (see Figure 2). The present study applies such a generic transient ID approach where the chosen conversion and storage reaction mechanism are selected to represent either a TWC used together with the stoichiometrically operated SI engine or to model a DOC-DPF-SCR system together with the lean operated CI engine.

Within the bond-graph concept the catalyst can be seen as a hybrid component combining characteristics of both, transfer and storage components. The flow through the catalyst can be categorized

as transfer characteristics. The axial variation of species composition due to conversion reactions is considered with the help of a ID finite volume discretization. The thermal inertia of the catalyst substrate and the surface reactions with possible storage reaction mechanism can be categorized as storage characteristics. The model equations (described in detail in Wurzenberger et al. (2011a, 2011b, 2012) cover the base effects of convection in the gas phase, heat conduction in the substrate and heat mass transfer between the two phases.

2.5. Cooling Circuit

In powertrain configurations heat is exchanged within the solid structure via heat conduction and between solid walls and any fluid via heat transfer (Figure 2). Following the a-causal bond-graph modeling approach, heat transfer networks can be described by lumped solid mass elements using temperature as state variable and heat transfer elements returning heat flows.

Characteristic times of the temperature changes of the lumped solid mass elements are much larger compared to the integration time steps of the mechanical network. The advantage of this difference can further be utilized when optimizing overall computational performance by decoupling the equations of the different domains by physical and therefore well-defined considerations.

2.6. Coupled Overall System

The different domains described above are modeled by equations representing different characteristics. For an efficient and coupled solution they can be transformed into a similar form. Thus, the second order mechanical system is transformed into a first order system of doubled size. The gas path equations already form the system of the first order ODEs. The mechanical system typically features additional linear constraints which can be substituted into the system. Additional remaining constrains are given by the electrical circuits.

All different states can be merged into a common state vector of unknowns x. Its time derivate multiplied with the system matrix S equals a right-hand-side function depending on x and time t. This coupled system of ODEs, as it is combined with algebraic constrain functions g(x) , leads to an overall semi-implicit Differential Algebraic Equation (DAE) system,

0 = g (x)

where the system matrix S may additionally depend on the state x. An efficient solution of the overall system is made possible by the novel approach relying on innovative time domain coupling of individual simulation domains. The approach is based on tailored solver technologies as discussed in the following.

Due to the special nature of the electrical system the index of the overall DAE system is one. To optimize the solution procedure of such model it is thus beneficial to solve first for the mechanical system and then for the electrical system assuming a frozen mechanical state (Pfau et al. 2011). This time integration step can be seen as a mixed discretization where the mechanical variables are discretized explicitly and the electrical variables are discretized implicitly.

The engine model consists of a crank angle resolved cylinder model embedded in a mean value based gas path description as discernible from Figure 2. Such an approach enables an adequate interaction of the cylinder with the intake and exhaust manifold, since the exchange of mass and enthalpy flows is based on

the physical models. Similarly, the heat transfer due to wall heat losses and the mass transfer due to fuel injection are also modeled on the crank angle basis. This is beneficial for the accuracy of transient results, since variations in temperature of the solid structure and arbitrary injection strategies are adequately captured. Due to the fact that cylinders are calculated in a different time domain than the other engine air path components, cycle averaged fluxes are exchange with the surrounding. These fluxes need to be evaluated in the components directly attached to the cylinder (marked by the black rectangle in Figure 2) based on the crank angle resolved data. Thereby a consistent exchange of flows is ensured in different time domains. System level simulations typically aim to optimize vehicle performance and emissions. Thus, such system analyses do not cover vibrational modes of the drive train enabling the exchange of data between the engine and the mechanical system either with the frequency of the integration of the mechanical system or even with the frequency of engine cycles.

Chemical reactions taking place in catalysts typically feature an Arrhenius type exponential temperature dependency. Therefore adaptive step size solver techniques are often mandatory to efficiently solve the balance equations of catalytic converters. In the present approach, the gas path and catalytic surface reaction equations are not solved in a fully coupled manner with the goal to optimize the overall computational performance. Instead, an independent time domain of the catalytic solver allows, depending on the problem, either smaller or even bigger integration time steps compared to the gas path solver.

The characteristic time scale related to the thermal inertia of the solid structure is generally much larger compared to the integration time steps of other domains. Therefore the elements of the solid structure represents numerically stable interacting elements to couple heat fluxes from different time domains to these elements.

The novel numerical coupling approach enables very time efficient mechanistic based system level simulation of the vehicles, since it inherently considers the characteristic response times of different domains. The overall real-time factor measured for an offline UDC simulation on a standard PC is in the range of 1.6 for the conventional and 1.7 for the hybrid power train. Real-time factors decrease below 0.9 if chemical reactions in the catalysts are not considered. Moreover, the present approach also enables a high level of accuracy and predictability, since it allows the incorporation of multiple mechanistic component models in a common system level model. The capability of the overall vehicle and powertrain model to adequately simulate transient vehicle operation was presented in Wurzenberger et al. (2010), comparing chassis-dynamometer measurements to simulated results of a conventional vehicle powered by an internal combustion and equipped with a 6-speed manual transmission.

3. Results

3.1. 'Validation of the Engine Model

In the present study two engine models are used. First, a four cylinders 1.4 liter high speed direct injection turbocharged diesel engine with EGR (see Figure 2) (denoted the CI engine) is investigated. Second, a four cylinders 1.6 liter turbocharged direct injection gasoline engine (denoted the SI engine) is simulated. The CI engine is controlled by the cyclic fuel delivery, vane position of the variable geometry turbine and the EGR rate. The SI engine operation is controlled by the position of the throttle in the intake, by the position of the turbine waste-gate, by the cyclic fuel delivery and the injection profile. These actuator values are controlled by the ECU (Engine Control Unit) which uses the input data that are based on the experimental values measured for a wide range of parameter variations. The CI engine is equipped by a DOC featuring 500 CPSI and 1.2 liter, a DPF featuring 300 CPSI and 2 liter and a urea

dosing unit and a SCR featuring 400 CPSI and 2.2 liter. The simulations were started with an empty DPF. The SI engine is equipped by a TWC featuring 400 CPSI and 1.625 liter.

-sim 1000 a exp1000

-sim 4000 □ exp 4000

-sim 6400 exp 6400

0.2 0.4 0.6 o.a

Load Signal (-)

Fig. 3. Validation of the CI engine model

Fig. 4. Validation of the SI engine model

A comprehensive validation of the CI engine model is presented in (Wurzenberger et al. (2010, 2011a 2011b) therefore only the validation of basic engine parameters is presented here. Figure 3 shows the capability of the model to accurately predict engine performance and efficiency or equivalently specific fuel consumption at full load (Figure 3 a) and b)). For the purpose of this study it is also very important that the engine model consisting of the crank angle cylinder model embedded in the mean value gas path model is capable to predict volumetric efficiency with appropriately high accuracy (Figure 3 c)). This is advantageous in terms of the transient engine performance modeling and even more importantly in the context of steady-state and transient engine out emission modeling. The validation of the engine out NOx emissions and of the performance of the corresponding aftertreatment devices is summarized in Wurzenberger et al. (2011a, 2011b).

The comprehensive validation of the SI engine model is presented in the Wurzenberger et al. (2012) therefore only validation of basic engine parameters is presented in this study. Figure 4 shows the comparison of measured and simulated values of BMEP (Brake Mean Effective Pressure), boost pressure and exhaust temperature. The validation of the SI engine model engine was performed at eight engine speeds from 700rpm to 6400rpm with 5 load signal variations, whereas Figure 4 shows load variation for three selected engine speeds. Very good agreement of measured and simulated BMEP values (Figure 4 a)) at identical fuel injection indicates adequate values of the effective efficiency of the engine. It is also very important that measured and simulated boost pressure values agree well (Figure 4 b)), since this indicates good agreement of measured and simulated volumetric efficiency. This is crucial for an adequate simulation of transient engine performance. Good agreement of the boost pressure is also

influenced by adequate modeling of the compressor, turbine and the thermodynamic state in the exhaust manifold. In addition, good agreement in exhaust temperature (Figure 4 c)) also indicates adequacy of the cylinder model and of the model of heat losses. The validation of the TWC model is also given in Wurzenbergeret al. (2012).

50 100 150

Time (s)

100 150

Time (s)

Fig. 5. Vehicle and engine parameters for the UDC 3.2. Drive Cycle Powertrain Performance Results

The overall system engineering model is used to compare the two powertrains topologies both powered by two different engines during a transient drive cycle. This analysis does not target to fully optimize performance and emissions of the considered vehicle configurations but to show the interaction of the domains as they are described by the mechanistic system engineering models. Thus, the controller of the engine and of the complete hybrid vehicle is simplified. However, they still ensure adequate vehicle behavior and performance and adequate operating regimes of powertrain components.

The simulations of the hybrid electric vehicle are based on a 2004 Toyota Prius model (Figure 1 a)). The model uses realistic data of the vehicle, tires, of the mechanical driveline components and of the electric components. For the purpose of the presented analysis the original gasoline engine is replaced by the turbocharged CI and SI engines presented in the previous section. This is done to expose the complex

interaction between dynamically interacting domains, which highlights the applicability and versatility of system level simulations featuring the mechanistic modeling depth.

Results of the hybrid electric vehicle were compared against results of conventional vehicle configurations powered by the same engines. To enable a valid comparison between the hybrid and the conventional vehicle, a vehicle model featuring the same geometric characteristics and tires was used for all simulations. However, the mass of the conventional vehicle is 140 kg less than the mass of the hybrid vehicle to account for the mass of electric components. The conventional vehicle is equipped with a 6-speed manual transmission in the case of the CI engine and with a 5-speed manual transmission to adequately consider engine speed ranges.

The results are shown for the UDC. Due to the fact that for a cold start of the first UDC the catalysts do not feature significantly high conversion efficiency for all vehicles, since light off temperature is not yet reached, all vehicles were run for four consecutive UDC cycles. The results are shown only for the last UDC to preserve readability of the figures. The hybrid vehicle is operated with the neutral SOC of the batteries (difference in SOC at the end and at the beginning of the drive cycle less than 0.1% SOC) and thus its fuel consumption adequately reflects its energy consumption.

In this and in the following section the following notation is adopted conv-SI: conventional vehicle with the SI engine, HEV -SI: hybrid vehicle with the SI engine, conv-CI: conventional vehicle with the CI engine, HEV -CI: hybrid vehicle with the CI engine.

Figure 5 shows vehicle performance of all analyzed configurations during the UDC. Figure 5 a) confirms that all configurations adequately follow the UDC velocity trace. From Figure 5 b) it is discernable that in both hybrid vehicles the engine is turned on mainly during vehicle accelerations and thus for a much shorter period of time compared to the conventional vehicle. Moreover, in the hybrid vehicles, the engine is coupled to the wheels over a planetary gear box that allows decoupling of the engine and the vehicle speed. Due to this and due to availability of electric machines it is possible to operate the engine in the HEV at higher BMEP values (Figure 5 c)) and thus at higher efficiencies (Figure 5 d)). As a result the HEVs consume less fuel than the conventional vehicles with the same engine type (Figure 5 e)). It can be observed that the engine operation in both conventional vehicles is characterized by low engine loads and thus low BMEP during steady-state cruising at relatively low vehicle speed. This leads to low effective efficiency of the engine being the main reason for the higher fuel consumption of the particular engine in the conventional vehicle compared to HEV.

A useful complementary information is given by the effective work (Figure 5 f)) produced by the engine over the entire cycle while powering different vehicles. The values in Figure 5 f) are obtained by integrating engine power over the entire cycle, whereas only non-negative values were considered. It can be observed that the effective work of the engine in the HEV is larger compared to the effective work of the same engine in the conventional vehicle. This is mainly due to two facts. First, the hybrid vehicle has larger mass and thus more work is needed for its propulsion. Second, the transformation between mechanical and electrical energy as well as transport and storage of electric energy is associated with higher losses than the transport of pure mechanical energy. If regenerative braking is not capable of compensating for all these losses, additional energy produced by the engine is needed to cover for these losses if the hybrid vehicle is operated with neutral SOC (more details on the energy conversion phenomena in hybrid electric vehicle can be found in Katrasnik (2009) and Banjac et al. (2009).

Due to mechanistic basis of the model it is possible to analyze the fuel consumption and the effective work results in more detail. First, results of both conventional vehicles are compared. At first sight it is surprising that the effective works of CI and SI engines differ (Figure 5 f)); note: a backward facing model that evaluates power trace of the engine out of the given vehicle velocity profile would predict identical or very similar values if rotational speed difference of the driveline components notably influence their efficiencies. The difference in the effective works is caused by the interaction of the driver

model, the ECU and the engine topology as well as by the mechanistic model basis that is capable of responding to these parameters. At engine loads characteristic for the UDC the SI engine is mainly controlled by the position of the throttle in the intake and by the cyclic fuel delivery. Since it is intended to keep the air fuel ratio close to the stoichiometric value at these loads the engine power is to a large extend determined by the air mass flow and thus by the pressure in the intake manifold. The intake manifold features a finite capacity and therefore a certain response time elapses between the driver load signal and thus also the throttle signal variation and the required power increase of the engine. It was already addressed above that simplified controller models were applied in this analysis and also the driver's responses might be better in the real world emission test drive. However, exactly the non-perfect behavior of controls reveals the potentials of the mechanistic models to adequately respond to these changes. This is furthermore the basis for a reasonable application of the model when developing the control functionalities in the virtual environment. Due to the above mentioned fact the SI engine responds slight slower when starting the vehicle from stand still and thus the drive increases the load in this period. As a result sharp increases can be observed in engine torque (corresponds to the BMEP Figure 5 c)) and engine speed (Figure 5 b)) when starting the vehicle from stand still. Additionally, it can be seen that due to the same reason the BMEP of the SI engine is also higher during certain deceleration periods and during certain transitions from deceleration to idling, which results in higher effective work of the SI engine.

3.3. Drive CycleEmissions Results

In this section engine-out emissions and emission conversion in exhaust aftretreatment devices is compared for the particular engine, i.e. CI or SI engine, while powering conventional and HEV powertrain. The emission results correspond to the performance results shown in the previous section.

The operation of the engine at high loads is clearly favored if the fuel economy is the main objective. However, operating the engine constantly at high loads as is the case for the engine in HEV- CI (Figure 5 c)) results in very low EGR rates (Figure 6 a)), since sufficient amount of oxygen should enter the cylinders to allow for complete combustion, which is required to keep sufficiently low soot emissions. Figure 6 a) gives the concentration of the combustion products in the intake manifold, where for the lean operating engines it is also necessary to consider the concentration of the burned fuel (Figure 6 b)), since it represents the available oxygen in the exhaust gases. Due to the operation of the HEV- CI engine at high loads, i.e. high cyclic fuel deliveries, and at low EGR rates, the temperature of the burned zone is significantly larger compared to CI engine in the conventional vehicle. This leads to significantly higher engine out NO and N02 emissions (Figure 6 f)) of the HEV- CI engine compared to the conv-CI engine. Due to the operation at higher loads, the engine in the hybrid vehicle also feature higher exhaust gas temperatures (Figure 6 c)).

The DOC is the first aftertreatment device and it thus features high temperature allowing significant oxidation of NO to N02 (Figure 6 d) and e)). Conversion from NO to N02 is more pronounced for the hybrid case due to higher exhaust gas temperatures, however mole concentration of the N02 at the DOC outlet exceeds the mole concentration of the NO for both vehicles. This might occur in the specific temperature window as presented by Wurzenberger and Wanker (2005). The DOC out emissions is very similar to SCR inlet emissions and thus these results are not shown to preserve readability of the figures. The SCR is modeled by the calibrated model featuring the simplified reaction scheme consisting of NO oxidation and "fast SCR reaction" as well as of ammonia adsorption and desorption reactions as given in Wurzenberger et al. (2008). In might be noted that for the case where N02 concentration significantly exceeds the NO concentration this simplified scheme might become inaccurate. This might be the case specific operating points of the HEV -CI engine. It can be observed that NO is very efficiently removed

in the SCR of the both vehicles due to a sufficiently high SCR temperature, high N02 and sufficient NH3 concentrations. Unlike, it can be seen that the HEV -CI vehicle emits a very high amount of N02, which is related to its high SCR inlet concentration and partial NO oxidation in the SCR. Therefore overall N0X emissions are only slightly reduced by the exhaust aftertreatment line of the HEV. Due to much smaller difference between N02 and NO concentration at the SCR inlet the conventional vehicle features higher N02 conversion of the SCR and thus the exhaust aftertreatment line notably reduces overall N0X emissions. These results clearly indicate the need to optimize performance of exhaust aftertreatment devices in a coupled approach to the desired engine operation by additionally considering adequate positions of the devices. The latter namely notably influence the temperature levels.

NO DOC In - N02 DOC Out

N02 DOC In - NO SCR Out

NO DOC Out - N02 SCR Out

0 50 1 00 1 54) 200 0 50 100 150 200

TilHi; (S : Tillli: ¡.S:

Fig. 6. Comparison of a) combustion products and b) fuel burned concentrations together with the c) exhaust gas temperatures and f) engine-out NO emissions of the CI engine in the conventional vehicle and in the HEV. Comparison of accumulated NO and N02 emissions at the inlet and outlet of catalytic converters for the CI engine in d) conventional vehicle and e) HEV.

In the case of the SI engine the in-cylinder air-fuel ratio even stronger influences the engine-out emissions compared to the CI engine. Thus, modeling of the actual in-cylinder air-fuel ratio provides a good basis for determining engine-out emissions of the SI engine. As shown Figure 7 c) it is challenging to control stoichiometric air-fuel ratios especially during transient operation. As a result the air-fuel ratio fluctuates around stoichiometric conditions especially during transient engine operation (Figure 7 c)). The

excess air ratio of the HEV engine is infinite during the engine stop phase (after engine is turned off it is decelerated to zero engine speed), since the engine is not fired and only air is pumped through the gas path. During the UDC the engine is not operated at high engine speeds and loads in the conventional vehicle, whereas this is also the case for the HEV with the exception of the short peak in during the first non-zero vehicle velocity period. At low engine speeds and/or loads the engine targets to be is operated at stoichiometric conditions. Due to the capability of the model to predict actual air-fuel ratios during the transients and due to the fact that the SI engine-out emission model captures the dependency of engine-out emissions on varying air-iuel ratios actual transient engine-out emissions (integrated over the cycle) are higher compared to steady-state emissions at the equivalent engine speed and load points.

mnv In - HEV In

convOut - HEV Out

Time (s) Time (s)

Fig. 7. Comparison of a) TWC inlet temperature b) exhaust gas mass flow c) in-cylinder excess air ratio of the SI engine in the conventional vehicle and in the HEV. Comparison of accumulated d) NO, e) HC and f) NOx emissions at the TWC inlet and outlet for the conventional vehicle and the HEV.

One of the characteristics of gasoline engines is that the mass flow through the engine is mainly determined by the throttle position and engine speed (Figure 7 b)). Moreover, the exhaust gas temperature is strongly influenced by the engine load. Thus, the HEV engine - as it is operated at higher load points (see Figure 5 c)) - features higher exhaust gas and consequently higher catalyst inlet temperatures (Figure 7 a)) compared to the conventional vehicle configuration. Nevertheless the HEV is characterized also by pronounced cooling effects taking place immediately after the engine operation is turned off. After the

engine in the HEV is not fired any longer it decelerated and continues pumping air. This favors the oxygen storage in the Ce02.

The accumulated engine outlet and tailpipe outlet emissions of carbon monoxide, hydrocarbons and nitric oxides are shown in Figure 7 d). The step-like profile of the engine outlet emissions of the HEV powertrain indicates an engine shut-off phase where no further engine emissions are accumulated. The accumulated engine outlet emissions of the conventional engine do not show completely horizontal profiles indicating a low level of emissions during motored engine operation. Figure 7 f) confirms an expected trend that higher NOx emissions are characteristic for engine operation at higher loads and thus the HEV engine configuration characteristically features higher engine out NOx emissions compared to the conventional vehicle configuration. Furthermore it can be seen from Figure 7 e) that the HEV engine generates lower hydrocarbon engine out emissions, which is mainly the consequence of not operating the engine at low loads, where more THC emissions are produced. More explanation might be needed on higher CO engine out emissions (see Figure 7 d)) of the hybrid powertrain. This is the consequence of very unsteady engine operation at high load, which is caused by a not optimized hybrid controller model. Due to this the HEV- SI engine is more frequently operated at high loads and rich air-fuel ratios compared to the conv-SI engine. This results in higher CO engine out emissions of the HEV- SI engine.

The performance of the TWC in converting all three pollutants can be derived by the difference between the engine-out, i.e. TWC Inlet - In, and tailpipe-out, TWC Outlet - Out, lines in Figures 7 d) -f). The latter curves also feature a characteristic step-like profile. Flat phases indicate full or very high conversion, whereas the individual steps represent pollutant breakthroughs mainly caused by lower catalyst brick temperatures, elevated exhaust mass flows (Figure 7 b)) and rich or lean combustion product composition (Figure 7 c)). The comparison of the overall conversion performance of the HEV and conventional powertrains shows higher conversion rates for the TWC of the HEV. This is caused by higher exhaust gas temperature (Figure 7 a)) and thus also higher catalyst brick temperatures especially during the last 60 seconds of the drive cycle. Moreover, more frequent operation of the HEV- SI engine at high loads and rich air-fuel ratios favors good conversion of N0X in the TWC, whereas well operating oxygen storage functionality in the TWC enables conversion of CO and HC also during these rich conditions.

4. Conclusions

A comprehensive mechanistic model of a hybrid and conventional vehicle is presented in the paper. The model comprises all relevant domains, which are necessary to adequately model performance and exhaust emissions of the vehicle. It was highlighted that mechanistic modeling approach provides a good basis for development and optimization of hybrid and conventional powertrains. This isjustified by a high level of predictability and adequacy in describing the interaction of all different domains during transient operation. An appropriate coupling of the domains in a common multi-physics solver is additionally crucial to achieve computational speed requirements. Unlike many steady-state surrogate models, the applied mechanistic approach appropriately covers transient operating regimes that may significantly deviate from the steady-state regimes. The physical based real-time approach is also capable of reflecting changes of the hardware configuration. The drive cycle simulations of the four different powertrains emphasize not only the tight interaction of all domains but also the influence of the control system. Here the appropriate transient response of the plant model is mandatory to evaluate e.g. thermal management measures for avoiding pollutant breakthroughs in the catalyst. This highlights the applicability of physical based real-time approaches in the development process of modified control strategies. The direct impact of changed engine calibrations on engine outlet and tail pipe emissions allows optimizing tail pipe emissions in the light of engine performance and drivability. Moreover, in HEV configurations, the

complex dependency of emissions, fuel consumption, performance and drivability on all hardware and software domains requires appropriate system engineering models to support the entire development process.

References

Banjac, T., Trenc, F. & Katrasnik, T. (2009) Energy conversion efficiency of hybrid electric heavy-duty vehicles operating according to diverse drive cycles. Energy Convers. Manage. 50, pp. 2865-2878.

Boland, D., Berner, H.-J., & Bargende, M. (2010). Optimization of aCNG Driven SI Engine Within a Parallel Hybrid Power Train by Using EGR and an Overized Turbocharger with Active-WG Control. SAE

Technical Paper 2010-01-0820.

BOOST RT User Guide. (2011). AVL List GmbH.

Canova, M., Fiorani, P., Gambarotta, A., & Tonetti, M. (2005). A real-time model of a small turbocharged Multijet Diesel engine: application and validation. SAE Technical Paper 2005-24-065.

CRUISE Users Guide. (2011). AVL List GmbH.

Deregnaucourt, M. & Wurzenberger, J.C. (2011). Optimales dynamisches DoE mit AVL BOOST Ladungswechselsimulation. 2. Workshop„ModellbasierteKalibriermethoden", TU-Wien.

Gao, Z., Conklin, J.C., Daw, C.S., Chakravarthy, V.K. (2010). A proposed methodology for estimating transient engine-out temperature and emissions from steady-state maps. Int. J. Engine Res. 11(2), pp. 137151.

Gao, Z., Chakravarthy, V. K., & Daw C. S. (2011). Comparison of the simulated emissions and fuel efficiency of diesel and gasoline hybrid electric vehicles. Proc. IMechEPart D: J.Automobile Engineering, 225, pp. 944-959.

Gschweitl, K., Ellinger, R. & Loibner, E. (2007). Tools and Methods for the Hybrid Development Process. AVI Conference Engine & Environment. Graz.

He, Y., & Lin. C.-C. (2007). Development and Validation of a Mean Value Engine Model for Integrated Engine And Control System Simulation. SAE Technical Paper 2007-01-1304.

Hrauda, G., Strasser, R. & Aschaber, M. (2010). Gas Exchange Simulation from Concept to Start of Production - AVL's Tool Chain in the Engine Development Process. THIESEI 2010, Conference on Thermo- and Fluid Dynamic Processes in Diesel Engines.

Isermann, R. and Müller, N. (2001). Modeling and Adaptive Control of Combustion Engines with Fast Neural Networks", eunite, URL http://www.eunite.org. 1st International Workshop on Hybrid Methods for Adaptive Systems: Hybrid Systems.

Katrasnik, T. (2009) Analytical framework for analyzing the energy conversion efficiency of different hybrid electric vehicle topologies. Energy Convers. Manage. 50, pp. 1924-1938.

Lee, T., & Filipi, Z. (2010). Simulation Based Assessment of Plug-In Hybrid Electric Vehicle Behavior During Real-Word 24-Hour Missions. SAE Technical Paper 2010-01-082.

Leifert, T., Fairbrother, R. & Moreno Nevado, F. (2008) Combustion and Gas Exchange Analysis of Transient Engine Cycles at the Test Bed. 8th International Symposium on Internal Combustion Diagnostics, Baden-Baden.

Lindenkamp, N., Stöber-Schmidt, C.P., & Eilts P. (2009). Strategies for Reducing NOX- and Particulate Matter Emissions in Diesel Hybrid Electric Vehicles. SAE Technical Paper 2009-01-1305.

Millo, F., Badami, M., Ferraro, C.V., Lavarino, G. andRolando, L. (2010). A Comparisonbetween Different Hybrid Powertrain Solutions for a European Mid-Size Passenger Car. SAE Technical Paper 2010-01-0818.

Pfau, R.U. & Schaden, T. (2011). Real-Time Simulation of Extended Vehicle Drivetrain Dynamics.

Multibody Dynamics— Computational Methods in Applied Sciences. 23, pp. 195-214, Springer.

Regner, G., Loibner, E., Krammer, J., Walter, L. & Truemner, R. (2000). Analysis of Transient Drive Cycles using CRUISE-BOOST Co-Simulation Techniques. SAE Technical Paper 2000-01-0627.

Simpson, A., Fleck, R., Kee R., Douglas R., Steele D., Hanna A., & Maybin. B. (2009). Development of aheavy Duty Hybrid Vehicle Model. SAE Technical Paper 2009-01-2933.

Smith, D. E., Lohse-Busch, H., & Irick D. K. (2010). A Preliminary Investigation into the Mitigation of Plug-in Hybrid Electric Vehicle Tailpipe Emissions through Supervisory Control Methods. SAE

Technical Paper 2010-01-1266.

Winsel, T., Ayeb, M., Theuerkauf, H. J., Pischinger, S., Schernus, C., & Lütkenmeyer G. (2004). HiL-Calibration of SI Engine Cold Start and Warm-Up Using Neural Real-Time Model. SAE Technical Paper 2004-01-1362.

Wurzenberger, J.C., & Wanker, R. (2005). Multi-Scale SCR Modeling, ID Kinetic Analysis and 3D System Simulation. SAE Technical Paper 2005-01-0948.

Wurzenberger, J.C., Wanker, R., & Schussler, M. (2008). Simulation of Exhaust Gas Aftertreatment Systems - Thermal Behavior During Different Operating Conditions. SAE Technical Paper 2008-010865.

Wurzenberger, J.C., Heinzle, R., Schuemie, H.A. & Katrasnik, T. (2009). Crank-Angle ResolvedReal-Time Engine Simulation -Integrated Simulation Tool Chain from Office to Testbed. SAE Technical Paper 2009-01-0589.

Wurzenberger, J.C., Bartsch, P. & Katrasnik, T. (2010). Crank-Angle Resolved Real-Time Capable Engine and Vehicle Simulation - Fuel Consumption and Performance. SAE Technical Paper 2010-010748.

Wurzenberger, J.C., Bardubitzki, S., Bartsch, P. & Katrasnik, T. (2011a). Real Time Capable Pollutant Formation and Exhaust Aftertreatment Modeling-HSDI Diesel Engine Simulation. SAE Technical Paper 2011-01-1438.

Wurzenberger, J.C., Bardubitzki, S., Heinzle, R. & Katrasnik, T. (2011b) Physical Based Real Time System Modeling—Model and Application Workflow. 13th EAEC European Automotive Congress. Paper A32.

Wurzenberger, J.C., Prah, I., Tominc, P. & Katrasnik, T. (2012) Optimization of Hybrid Power Trains— Physical Based Modeling for Concept Design. SAE Technical Paper 2012-01-0359.