Effects of engine thermal transients on the energy management of series hybrid solar vehicles

Control Engineering Practice 18 (2010) 1231–1238

Contents lists available at ScienceDirect

Control Engineering Practice journal homepage: www.elsevier.com/locate/conengprac

Effects of engine thermal transients on the energy management of series hybrid solar vehicles Ivan Arsie , Gianfranco Rizzo, Marco Sorrentino Department of Mechanical Engineering-University of Salerno, 84084 Fisciano (SA), Italy

a r t i c l e in fo

abstract

Article history: Received 9 March 2009 Accepted 27 January 2010 Available online 16 February 2010

The paper focuses on investigating thermal-transients effects, associated to intermittent use of internal combustion engine (ICE), on fuel economy and hydrocarbon (HC) emissions of series hybrid solar vehicles (HSVs). An offline, non-linear constrained optimization is set-up to individuate the ICE power trajectory that simultaneously minimizes fuel consumption, suitably operates the battery and fully exploits daily solar contribution. The results highlight the importance of including thermal transients in HSV energy management. The combined effects of engine, generator and battery losses, along with cranking energy and thermal transients, produce non-trivial solutions for the engine/generator group, which should not necessarily operate at its maximum efficiency. & 2010 Elsevier Ltd. All rights reserved.

Keywords: Engine modeling Engine control Hydrocarbon emissions Cold-start Hybrid vehicles Solar vehicles

1. Introduction In the last years, increasing attention is being spent towards the integration of solar energy with either electric or hybrid cars. While solar cars do not represent a practical alternative to cars for normal use, the concept of a hybrid electric car assisted by solar panels appears more realistic (Letendre, Perez, & Herig, 2003; Saitoh, Hisada, Gomi, & Maeda, 1992; Sasaki, Yokota, Nagayoshi, & Kamisako, 1997; Seal & Campbell, 1995; Seal, 1995). In fact, thanks to some relevant research efforts, in the last decade hybrid electric vehicles (HEV) have evolved to industrial maturity, and represent now a realistic solution to important issues, such as the reduction of gaseous pollution in urban drive as well as the energy saving requirements. The use of solar energy on cars has been considered with certain skepticism by most users, including automotive engineers. This may be due to the simple observation that the net power achievable in a car with current photovoltaic panels is about two order of magnitude less than maximum power of most of today cars. But a deeper energetic analysis evidences that this perception may be misleading. In fact, there are a large number of drivers who use car everyday for short trips and with limited power demand. For instance, some recent studies conducted by ¨ INRETS (Andre´, Hammarstrom, & Reynaud, 1999) report that about 71% of European car users drive 61 minutes a day, with 12

 Corresponding author. Tel.: + 39 089 964080; fax: + 39 089 964037.

E-mail address: [email protected] (I. Arsie). 0967-0661/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.conengprac.2010.01.015

minutes average trip length and one passenger for most of the time (i.e. the driver). On the other hand, a solar panel, if properly located, oriented and controlled, can operate near its maximum power for many hours a day, and in those conditions the daily solar energy collected by solar panels on the car may represent a significant fraction of the energy required for traction (Arsie, Rizzo, & Sorrentino, 2007b). Despite their potential interest, solar hybrid cars have received relatively little attention in literature until a few years ago (Letendre et al., 2003). Some prototypes have been developed in Japan (Saitoh et al., 1992; Sasaki et al., 1997), at Western Washington University (Seal, 1995; Seal & Campbell, 1995) and at the Queensland University, while the first model of hybrid vehicle assisted by a solar panel, even if finalized to air conditioning, has been launched in 2009 by Toyota. Although these works demonstrate the general feasibility of such an idea, detailed presentation of results and performance, along with a systematic approach to hybrid solar vehicle design, were missing in literature until a few years ago. Therefore, appropriate methodologies are required to address both the rapid changes in the technological scenario and the increasing availability of innovative, more efficient components and solutions. The current study focuses on the extension of the methodologies presented in previous papers (Arsie, Graziosi, Pianese, Rizzo, & Sorrentino, 2005; Arsie, Di Martino, Rizzo, & Sorrentino, 2007a) for the energy management of an hybrid solar vehicle. Particularly the effects of engine thermal transients on fuel consumption and hydrocarbon (HC) exhaust emissions (Cheng & Santoso, 2002; Zavala, Sanketi, Wilcutts, Kaga, & Hedrick, 2007) are accounted for while

1232

I. Arsie et al. / Control Engineering Practice 18 (2010) 1231–1238

investigating the most suitable powertrain management in case of ICE intermittent operation. The paper is organized into 4 main sections. After a general description of vehicle model, the issues related to energy management in HSV are addressed in Section 3. Section 4 is devoted to the analysis of ICE thermal transients through detailed description of experiments and modeling approach. Finally, in Section 5 optimization results are presented and discussed.

PV

EG

EN

ICE

B

EM

2. The solar hybrid vehicle model Different architectures can be applied to HEVs: series, parallel, and parallel-series. The choice can depend on vehicle size, performance and targeted usage. In this case, as for other solar hybrid vehicles (Letendre et al., 2003), the series structure has been adopted. Although the latter exhibits lower global efficiency than parallel, a series hybrid powertrain presents some very interesting features, such as:

 It is simpler, with fewer constraints for vehicle layout.  There are no mechanical links between generator and wheels,   

therefore very effective vibration insulation can be achieved. It is possible to use in-wheel motors with advanced traction control techniques. Engines specifically optimized for steady operation can be used, with high peak efficiency and/or with favorable weight/ power ratio (i.e. micro gas turbines). Series architecture acts as a bridge towards the introduction of fuel cell powertrains.

Fig. 1. Scheme of the series hybrid solar vehicle.

vehicle wheels: I

d$ ¼ TEM TR dt

ð1Þ

where I is the vehicle inertia, TEM is the EM output torque and TR is the resistant torque due to aerodynamic and rolling resistance (Arsie, Flora, Pianese, Rizzo, & Serra, 2001). The required traction energy depends on vehicle weight and aerodynamics, which in turn depend on the size of propulsion system components, vehicle dimensions and solar panel area. Battery, electric motor and generator are simulated by the ADVISOR model (Burch et al., 2009). 2.2. Vehicle weight

Moreover, their main disadvantage (i.e. lower global efficiency, due to electric generator/converters losses and not optimized engines for the specific application) is expected to become less concerning in next future, thanks to a significant research effort in such fields. The growing interest paid by the automotive industry towards the series structure is confirmed by the announced launch of the Chevrolet Volt, a plug-in hybrid electric vehicle to be produced by GM. Unlike most current commercially available electric hybrids, the actual propulsion of the Volt is accomplished exclusively by the electric motor (Bullis, 2007). In the series structure, the photovoltaic panels (PV) assist the electric Generator (EG), powered by the internal combustion engine (ICE), in recharging the battery pack (B) in both parking mode and driving conditions, through the electric node (EN). The electric motor (EM) can either provide the mechanical power for propulsion or restore part of the braking power during regenerative braking (Fig. 1). In this structure, the thermal engine can work mostly at constant power, corresponding to its optimal efficiency, while the electric motor EM is designed to assure the attainment of the vehicle peak power.

2.1. Solar energy for vehicle propulsion The estimation of net solar energy captured by PV panels in real conditions (i.e. considering clouds, rain etc.) and available for propulsion is accomplished by a solar calculator developed at the US National Renewable Energy Lab (Arsie et al., 2007b). The maximum panel area can be estimated as function of car dimensions and shape by means of a simple geometrical model. The instantaneous power (P(t)) is estimated for assigned vehicle data and driving cycle by integrating a longitudinal vehicle model, expressed by the following Newton law reduced to

The parametric HSV weight model results, on one hand, from the addition of the hybridizing devices (PV panels, battery pack, ICE, generator, electric motor, inverter) onto the conventional vehicle (CV) equipped with ICE (WCV) and, on the other, by properly revaluating the contribution of the components resized or not present in the HSV (i.e. ICE, gearbox, clutch) (Arsie et al., 2007b). Considering the layout described in Fig. 1, the required nominal battery power is PB ¼ PEM PEG

ð2Þ

In Eq. (2) the power contribution from the PV array is not accounted for because of two reasons: the first is the unpredictability of sunshine availability (e.g. rainy days); the second is linked to the relatively small PV nominal power that can be installed on cars at the current technology stage. Therefore the number of battery modules is evaluated as NB ¼

PEM PEG PB;u

ð3Þ

where PB,u is the nominal power of a single battery module, here set to 1.67 kW for a 12 V, 25 Ah lead-acid module (Burch et al., 2009). The power of the electric machine (PEM) is computed imposing that the HSV power to weight ratio (PtWHSV), corresponds to a 1250 kg conventional vehicle (CV) powered by a 75 kW gasoline engine, as reported in Table 1 PtWHSV ¼

PICE;CV Wbody;CV

PEM ¼ PtWHSV WHSV

ð4Þ ð5Þ

Table 1 also shows that a battery pack consisting of 27 leadacid modules was yielded by Eq. (3). Such an energy reservoir

I. Arsie et al. / Control Engineering Practice 18 (2010) 1231–1238

Table 1 Vehicle technical data.

PICE (kW) Fuel PEG (kW) PEM (kW) NB [/] APVH (m2) Aerodynamic drag coefficient Cx [/] Rolling coefficient Cr [/] Frontal area (m2) W (kg)

CV

HSV

75 Gasoline 0 0 0 0 0.33 0.01 2.36 1250

46 Gasoline 43 90 27 3 0.33 0.01 2.36 1465

allows performing adequate regenerative braking, while achieving a 25 km pure electric range on the ECE-EUDC driving cycle. It is worth noting that further optimization of the regenerative braking mode can be accomplished through addition of supercapacitors to assist the battery pack during severe braking phases (Burke, Hardin, & Dowgiallo, 1990; Cundev & Mindl, 2008; Rodatz, Paganelli, Sciarretta, & Guzzella, 2005).

3. Energy flow management and control in a hybrid solar vehicle Hybrid solar vehicles have of course many similarities with hybrid electric vehicles, for which several studies on the optimal management and control of energy flows have been presented in last decade (Arsie et al., 2005; Pisu & Rizzoni, 2007; Powell, Bailey, & Cikanek, 1998; Sciarretta & Guzzella, 2007). Nevertheless, the presence of solar panels and the adoption of a series structure may require to study and develop specific solutions for optimal management and control of an HSV. In fact, in most electric hybrid vehicles a charge sustaining strategy is adopted: at the end of a driving path, the battery state of charge should remain unchanged. With a solar hybrid vehicle, a different strategy should be adopted as battery is charged during parking hours as well. In this case, a different goal can be pursued, namely restoring the initial state of charge within the end of the day rather than after a single driving path (Arsie et al., 2007b). Moreover, the series configuration suggests an efficient solution, namely to operate the engine in an intermittent way at almost constant operating conditions. In such case, the engine– generator system may be designed and optimized to maximize its efficiency, emissions and noise at design point, while in current automotive engines the maximum efficiency is usually sacrificed to the need of assuring stable operation and good performance in the whole operating range. The techniques developed for HEV, mostly adopting parallel or series/parallel structure, tend to treat the engine as a continuous system working in the whole range of operating conditions. This approach is also followed in some recent studies on HSV, based on the application of Dynamic Programming and Model Predictive Control (Preitl, Bauer, Kulcsar, Rizzo, & Bokor, 2007). In case of engine intermittent operation, the effects exerted on fuel consumption and emissions by the occurrence of thermal transients in engine and catalyst should be considered. These effects are neglected in most studies on HEV, where a steady-state approach is usually used to evaluate fuel consumption and emissions. A preliminary analysis of HSV energy management has been presented by considering a single period for ICE operation within the driving cycle module, at specified position (i.e. at the end of driving cycle) (Arsie et al., 2007b). This approach allows taking into account the key aspects related to control, in a framework where the main target was to estimate the effects of different

1233

vehicle and powertrain variables on energy flows. This procedure has been integrated in the vehicle dynamic model, also considering weight and costs, and used to study optimal vehicle design. In order to develop a supervisory control to be implemented on the vehicle, a more accurate analysis of the optimal ICE power distribution over an arbitrary driving cycle has to be performed. Specifically, in this paper an off-line procedure was developed to demonstrate the importance of accounting for ICE thermal transients if real-world effectiveness of supervisory control strategies must be guaranteed.

4. Effects of ice intermittent operation ICE intermittency causes the occurrence of thermal transients, both in engine and in catalyst, thus influencing fuel consumption and exhaust emissions. These effects should be analyzed and taken into account to properly control energy flows and enhance the development of suitable solutions for vehicle thermal management. In this paper, the aspects related to engine thermal transients and their impact on fuel consumption and HC emissions are considered. Actually HC emissions are of major concern when frequent engine start/stop maneuvers are performed, as in the case of the ICE intermittent use proposed for HSV operation. Therefore, estimation of HC emissions during engine thermal transient is useful to optimize energy management strategies also with respect to environmental impact. The optimal ICE power trajectory was estimated by solving the following constrained optimization problem: Z _ f ;HSV ðXÞdt minX m ð6Þ _ f ;HSV Þ minimization is subject to the where fuel flow rate ðm constraints:

DSOCday ðPEM ; PICE ðXÞ; PSUN Þ ¼ 0

ð7Þ

SOC 4SOCmin

ð8Þ

SOC oSOCmax

ð9Þ

The decision variables X include, for each ICE-on event, starting time, duration and ICE power level, while the number N of ICE-on phases has been assigned, in order to analyze its influence on the results. The constraint expressed by Eq. (7) allows restoring the initial state of charge (SOC) within the end of the day, also considering parking phases. It requires the integration of the vehicle dynamic model over the day. PEM is known from the assigned mission profile, while also net power from sun is considered known. PICE depends on the decision variables X. It is worth noting that such control strategy is based on the knowledge of the vehicle route, thus being unsuitable for realtime control. Nevertheless the proposed approach is consistent with the purpose of the paper, namely analyzing the effects of engine operation on fuel efficiency. This task will be helpful for the future development of supervisory HSV control, extending to HSV the approach based on provisional load estimate previously applied to HEV real-time control (Arsie et al., 2005). The minimum and maximum allowed values for SOC are imposed considering battery reliability, while the limit on maximum SOC during driving phases is due to the exigency to assure a battery capacity sufficient to store the expected solar energy during parking time. Further constraints are introduced to limit the ICE power within the operating range and to avoid ICE operation phases overlapping.

1234

I. Arsie et al. / Control Engineering Practice 18 (2010) 1231–1238

Vehicle speed [km/h]

150

100

50

0 0

200

400

600 Time [s]

800

1000

1200

and HC emissions can be accounted for through the engine thermal state as well, since AFR dynamics is driven by wall wetting phenomenon and injection time that are in turn dependent only on engine temperature, in case of fixed load and speed. Figs. 3–5 show the experimental trajectories of coolant temperature, specific fuel consumption (SFC) and concentration of HC tailpipe emissions along the warm-up transient. It is worth noting that Figs. 4 and 5 show similar trends of SFC and HC emissions, with a dramatic increase during the early stages of engine operation, due to fuel enrichment and misfiring. Particularly, this latter results in harsh torque delivery with negative impact on SFC. Afterwards, SFC and HC trends follow

Fig. 2. Module of the ECE-EUDC driving cycle.



energy management of hybrid vehicles, including previous paper of the authors (Arsie et al., 2005; Preitl et al., 2007) has not been proposed for two reasons: (i) the presence of thermal transients in the engine model would require to adopt a further state variable, besides SOC and engine power, with an increase in dimensionality and in computational burden in DP problem; (ii) the presence of an asymptotic behavior in the dependence of fuel consumption with engine temperature would represent a problem in the backward computation required by DP algorithm. One of the objectives in this paper is to study the influence of the number of engine-on phases on the results, for a given driving cycle. In this case, all the remaining decision variables are continuous, and the problem can be efficiently solved by classical algorithms. When N has to be optimized, the optimization problem becomes a mixed-integer programming, and different techniques, like as Genetic Algorithms, can be selected (Chipperfield, Fleming, Polheim, & Fonseca, 2009; Sakawa, Kato, Ushiro, & Inaoka, 2001).

Engine Temp. [°C]

 The use of Dynamic Programming, adopted in many papers on

100 Measured Predicted

80 60 40 20 0

100

300

400

1000 Measured Predicted

800 600 400 0

100

4.1. Experimental set-up Experimental data were collected at the test bench to analyze the effects of thermal transients on ICE performance and emissions. The experiments were carried out at University of Salerno on a Multi-Port injection engine, 4 cylinders, 1.2 l. The test bench was equipped with (i) an AVL gravimetric balance to measure fuel consumption, (ii) a dSPACE& MicroAutobox equipment, in place of the engine control system, to accomplish a full control of the engine tasks and (iii) an ABB Hartmann & Braun Multi-FID analyzer to detect HC tailpipe emissions. The experiments were accomplished by measuring the main engine variables (e.g. engine speed, brake torque, fuel consumption, coolant and oil temperature, air–fuel ratio, etc.) and tailpipe HC emissions during a cold-start transient. The whole warm-up maneuver was performed by imposing constant load (i.e. throttle opening) and engine speed. This way control variables (i.e injection time and spark advance) and dynamics of engine performance and emissions along the transient were influenced by anything else than engine and catalyst thermal state, respectively. It is worth noting that the AFR impact on performance (i.e. brake torque and specific fuel consumption)

200 Time [s]

Fig. 3. Measured and predicted engine temperature along a cold-start maneuver on the engine test bench.

SFC [g/KWh]

The driving scheduling is composed of 4 modules of the ECEEUDC driving cycle, as the one shown in Fig. 2. To solve the optimization problem (6–9), a classical second order Quasi-Newton algorithm for non-linear constrained optimization has been selected. The reasons for this choice are the following:

200 Time [s]

300

400

Fig. 4. Measured and predicted specific fuel consumption along a cold-start maneuver on the engine test bench.

HC [ppm] 1000 Measured Predicted

800 600 400 200 0 0

100

200 Time [s]

300

400

Fig. 5. Measured and predicted concentration of tail-pipe HC emissions along a cold-start maneuver on the engine test bench.

I. Arsie et al. / Control Engineering Practice 18 (2010) 1231–1238

1235

engine and catalyst thermal dynamics, respectively. Along the warm-up transient, SFC approaches the steady-state value after about 300 s while HC emissions reach almost negligible concentration after about 400 s. The experimental data were used to identify black-box models aimed at predicting the effects of thermal transients on engine performance and emissions, as it will be described in the next sections.

Once the warm-up transient is extinguished, degradation factor reaches the unit and specific fuel consumption is computed by a map over engine operating condition (i.e. load and speed) derived from experimental data collected on the test bench in steady-state conditions, by ranging engine load and speed in a wide operating domain. Fig. 4 shows the comparison between measured and predicted SFC along engine warm-up, evidencing satisfactory accuracy of the proposed modeling approach.

4.2. Modeling of engine thermal transients

4.3. Modeling of HC emissions

The effects of thermal transients are considered assuming that specific fuel consumption depends not only on ICE power but also on actual engine temperature. The coolant temperature T is assumed as engine reference temperature. Its time variation, during warm-up operation or following ICE swith-off, may be estimated assuming a first order process, as usually performed for simulating thermal systems dynamics:

A simple model for HC emissions was developed, based on the experimental data collected at the test bench. According to the observed results and to experimental evidences derived from literature analyses (Cheng & Santoso, 2002; Zavala et al., 2007), in the first 20 s HC emissions are modeled as a first order process, as follows:

dT 1 ¼ ðT TÞ dt tT ss

ð10Þ

The values of steady state temperature Tss and time constant tT can be identified from experimental observations. It is worth noting that, in case of ICE warm-up, the first order dynamic response is characterized by a relatively high steadystate temperature that corresponds to the natural thermal equilibrium condition. However, during in real operation, this steady-state temperature is never reached since as the coolant temperature approaches approximately 80 1C, the cooling system thermostat starts opening and the temperature stops increasing and starts fluctuating around a constant value, due to alternative thermostat opening/closing (see Fig. 3). Moreover, the experimental data shown in Fig. 3 evidence that, in the relatively narrow range of variation experienced (i.e. from 27 to 80 1C), the time history of coolant temperature is well approximated by a linear regression. From these considerations, in the present work the coolant temperature transient during ICE warm-up is predicted by a first order polynomial regression over time, identified from the experimental data. Comparison between experimental measurements and model prediction is shown in Fig. 3. Once the set-point of 80 1C is reached, the temperature is assumed constant until ICE is switched-off. Afterwards, coolant temperature decay is expressed by Eq. (10), with steady-state temperature Tss and time constant tT equal to 27 1C and 600 s, respectively. Fuel consumption during engine thermal transient is estimated by correcting the steady-state value, corresponding to thermal equilibrium condition, by a degradation factor dSFC expressed by a map over the ratio of actual and steady-state engine temperature. The map was derived by the experimental measurements of specific fuel consumption along the warm-up transient, shown in Fig. 4 SFCðtÞ ¼ SFCss =dSFC

ð11Þ

Being based on the data collected along one warm-up transient, the proposed model should guarantee good estimation accuracy only when similar maneuvers are accomplished (i.e. same load and speed time history). However, it is worth noting that in case of series HEV application, it is reasonable assuming that engine warm-up is performed following one fixed transient by imposing time history of load and speed. In fact, in a series hybrid, the actual ICE operating range is much more limited than in a conventional vehicle, and located near the region of maximum efficiency.

dHC 1 ¼ ðHCss HCÞ dt tHC

ð12Þ

The initial value is computed by correcting the experimental one (see Fig. 5), measured during the cold-start test, by a factor dHC to account for actual engine temperature. In this way the model can predict HC emissions even in case of engine warm start HCin ¼ HCin_cold dHC

ð13Þ

The correcting factor dHC corresponds to a fuel delivery efficiency that accounts for the wall wetting process and the effect of fuel enrichment on HC formation mechanism (Cheng & Santoso, 2002). As previously mentioned (Section 4.1), assuming fixed engine speed and load, the correcting factor mainly depends on the engine thermal state (Cheng & Santoso, 2002; Arsie, Pianese, & Rizzo, 1999). The steady-state value HCss in Eq. (12) is computed by a map expressing the dependence of HC concentration over engine temperature, derived from the experimental measurements (Fig. 5). Once the early transient has been extinguished (i.e. after the first 20 s), HC emissions are evaluated by a steady-state approach by the map just mentioned, without any explicit dependency on exhaust gas properties (i.e. composition and temperature) and catalyst thermal state. When the engine is warmed up and the catalyst light off has been reached, HC emissions are drastically reduced assuming almost a full oxidation in the TWC. This hypothesis is consistent with the fact that in case of series HEV application, the engine does not experience fast load transients and no AFR excursion from stoichiometry is expected. The comparison between measured and predicted HC tailpipe emissions along engine warm-up is shown in Fig. 5, evidencing satisfactory accuracy of the proposed modeling approach. The proposed assumptions are in accordance with the way the experiments were accomplished, as mentioned in Section 4.1. If fixed load and speed are assumed along engine start transient then exhaust gas properties and, consequently, catalyst thermal state, are not affected by anything else than engine temperature. Nevertheless it is worth noting that catalyst thermal dynamics differs significantly from engine. However, although a simplification is introduced by expressing HC emissions as function of engine temperature time history instead of catalyst one, the proposed approach can be considered acceptable for the purposes of this work, focused on the estimation of the influence of thermal transients on energy management in a series hybrid, and on checking the limits of accuracy of a steady-state approach. The possible use of a more detailed thermal model accounting for catalyst thermal dynamics deserves to be further investigated in future developments.

1236

I. Arsie et al. / Control Engineering Practice 18 (2010) 1231–1238

Fuel consumption [kg]

Engine Temp. [°C]

2.2 80

60 T =27 °C in

40

T =50 °C in

20 0

200

400 600 Time [s]

800

2 1.8 1.6 1.4 0

1000

2 4 6 Number of ICE-on events

8

Fig. 8. Effects of number of ICE-on events (N) on fuel consumption.

Fig. 6. Simulated engine temperature in case of cold (27 1C) and warm (50 1C) engine start maneuver.

1.4 HC emissions [kg]

0.8 T = 27 °C in

0.6

T = 50 °C

HC [g]

in

0.4 0.2

1.2 1 0.8 0.6 0.4

0 0

200

400 600 Time [s]

800

0

2 4 6 Number of ICE-on events

1000

8

Fig. 9. Effects of number of ICE-on events (N) on HC emissions.

The results of a simulated engine start maneuvers are shown in Figs. 6 and 7 in case of cold and warm operation. Particularly, Fig. 7 evidences that the mass of HC emitted at the tail-pipe differs in the two cases, due to the different temperature transients shown in Fig. 6 that in turn influence the catalyst light off time.

5. Simulation results The simulations were performed by solving the constrained optimization problem, as defined in Eqs. (6)–(9), for the HSV configuration described in Table 1. It is worth mentioning that such configuration corresponds to the one that maximizes fuel economy, as indicated by the authors themselves in a previous work (Arsie et al., 2007b). The ICE intermittent operation was analyzed for a number (N) of ICE-on phases ranging from 1 to 8, thus allowing to extensively assess the effect of several engine start/stop maneuvers on both fuel consumption and emissions. Figs. 8 and 9 show the dependence of fuel consumption and HC emissions on the number of ICE-on events (i.e. N), respectively. It emerges from the figure how fuel consumption initially decreases towards a minimum value, in correspondence of N =4. This is explained by the tendency, with higher number of ICE-on phases, to operate the engine close to the most efficient operating point (i.e. 22 kW, which is approximately half the maximum power as reported in Table 1). Then, due to the negative effect of cranking energy and of thermal transient on ICE efficiency, taken into account by Eq. (11), the advantage related to the higher number of ICE operation vanishes, thus causing fuel consumption to increase.

Engine temperature [°C]

Fig. 7. Simulated mass of HC emitted at the tail-pipe in case of cold (27 1C) and warm (50 1C) engine start maneuver.

80

N=3 N=4

60

40

20 0

1000

2000 3000 Time [s]

4000

5000

Fig. 10. Simulated engine temperature trajectory in case of three and four (N = 3 and 4) ICE-on events.

On the other hand, HC emissions show an undesirable increasing trend when more ICE-on events are considered. This happens because of the initial large amount of HC released when ICE is turned on, due to the higher impact exerted by some of the mechanisms responsible for HC formation, such as non-stoichiometric air-fuel mixture and flame quenching (Heywood, 1988). Thus, letting the engine turn-on once, at the power level required for ensuring charge-sustaining operation, would result in the lowest HC emission, as indicated by the case N = 1 in Fig. 9. Nevertheless, the low N =1 HC emissions are significantly countercompensated by the high fuel-consumption simulated in the corresponding case (see Fig. 8). On the other hand, the analysis of HC emissions associated with N>2 cases indicates that the impact of HC formation mechanisms tends to reduce, due to the relatively slow decay in engine temperature at the following engine starts (as evidenced in Fig. 7). Particularly, Fig. 10 shows that in case of N =3 the engine is turned on for the second and third time at

I. Arsie et al. / Control Engineering Practice 18 (2010) 1231–1238

HSV power [kW] - N = 4

40 20 0 -20 -40 0

drive gen 1000

2000 3000 Time [s]

4000

5000

State of charge [/] - N = 4

Fig. 11. Simulated power trajectory for the selected driving cycle in case of four ICE-on events.

0.8 0.75 0.7 0.65 0.6 0

1000

2000 3000 Time [s]

4000

5000

Fig. 12. Simulated state of charge trajectory for the selected driving cycle in case of four ICE-on events.

Table 2 Analysis of engine thermal transient impact on supervisory strategies effectiveness. Cases: (a) Decision variables (see Eq. (6)) optimized without thermal transients; (b) optimal decision variables of case (a) with thermal transients; (c) decision variables optimized with thermal transients. Case

Fuel consumption [Kg]

a b c

1.4559 1.5361 1.5185

HSV power [kW] - N = 4

temperature as low as 30/40 1C, whereas in case of N=4 ICE is always turned on at higher temperatures. Moreover, the last ICE-on event for N=4 is enabled at temperature around 60 1C, which results in a small HC release in this phase. Therefore, fuel delivery efficiency for N=4 is, on average, higher than N=3, thus explaining the local minimum of HC emissions occurring in correspondence of N=4. The above considerations on fuel consumption and HC emissions lead to select N= 4 engine starts as the most convenient strategy to be adopted for the selected driving route. Figs. 11 and 12 show the model outputs computed in case of 4 ICE-on events. Particularly, Fig. 11 shows the power contributions from the electric generator (dotted line) to meet the traction power demand, whereas battery and solar panels (the latter being constant and quite negligible during driving phases) power trajectories were omitted for sake of clarity. It is interesting to note how for N = 4 the constrained optimization analyses led to place the ICE-on events mostly in correspondence of the highest power demands. This result can be explained by considering that it allows reducing low-load, less efficient ICE operations. Fig. 12 shows that the final state of charge differs from the initial one by the value corresponding to the energy storable through the solar panels during parking hours with sunlight (here set to 8.7 h whereas driving time is 1.3 h per day). Finally, a specific numerical analysis was carried out to assess the impact of engine thermal transient on control strategies efficacy in real world applications. Table 2 lists simulated fuel consumption in three scenarios, all referring to the ECE-EUDC cycle and 4 ICE-on events: (a) decision variables (see Eq. (6)) optimized neglecting thermal transient effects; (b) simulation of decision variables found in case (a) but accounting for engine thermal transients; (c) decision variables optimized by taking into account thermal transient effects. As expected, case (a) and (b) correspond to minimum and maximum fuel consumption, respectively. On the

1237

40 Case a) Case c) Maximum ICE efficiency point

30 20 10 0 0

1000

2000 3000 Time [s]

4000

5000

Fig. 13. Simulated generator trajectories related to scenarios (a) and (c) described in Table 2.

other hand case (c) indicates that considering thermal transients within the optimization task allows appreciably decreasing fuel consumption with respect to case (b). Furthermore, the comparison of decision variables (i.e. ICE-on starting time, duration and power level), yielded by the optimization task (i.e. Eqs. (6)–(9)) and described in Fig. 13, indicates that the values of such variables vary significantly going from case (a) to case (c). The results evidence that the optimal energy management of a series hybrid is not as straightforward as it could appear at a first glance. In fact, it could be expected that there is no relationship between engine-on operation and traction power demand, since the engine is mechanically disconnected from the wheels, and that the ICE can be operated anyway in its maximum efficiency region regardless of power demand. Instead, the results show that the engine does not work necessarily at highest efficiency and that, when compatible with SOC constraints, ICE operation is concentrated at the highest power demand. Such observations can be explained considering that, besides the losses in engine/ generator group, the models also take into account energy losses associated to battery charge/discharge, as well as other nonnegligible effects, such as cranking energy and thermal transients induced by ICE intermittency. Particularly, the fact that the engine operation tends to be concentrated at the highest traction power is justified since in this case a larger amount of generator power is directly supplied to the electric motor to meet power demand, while the remaining smaller part, which is stored in the battery, is subject to energy losses for charge/discharge. Instead, if the energy generation occurs at lower power demand, the fraction of the energy to be charged into the battery (and at a later time discharged from it) would increase, so the associated energy losses. Moreover, to minimize the incidence of the losses associated to cranking energy, the frequency of engine starts should be reduced, while, to limit the fuel consumption increase due to engine temperature decay, engine-off phases should be shortened. Thus, the combined effect of these phenomena can determine the selection of engine power values different from the maximum efficiency point.

1238

I. Arsie et al. / Control Engineering Practice 18 (2010) 1231–1238

6. Conclusions

References

A study on the effects of engine thermal transients on the energy flow management in a hybrid solar vehicle with series structure has been presented. Although the study was conducted on a hybrid solar vehicle, most of the conclusions are valid for hybrid electric vehicles with series structure as well. This work represents a step toward the development of optimal strategies for the energy management, suitable for both off-line application and real-time implementation. The analysis considers the impact of thermal transients on engine power, fuel consumption and HC emissions during the engine start maneuver. The optimal ICE power trajectory was found by solving a non-linear constrained optimization accounting for fuel mileage and state of charge and considering solar contribution during parking mode. The results show that the engine thermal transients due to start–stop operation cause a non-negligible reduction of fuel economy compared to steady-state warmed-up operation. Moreover, the distribution of ICE operation modes differs from the steady-state case, indicating that such effects should be taken into account in HEV and HSV control where ICE start–stop events take place. Concerning HC emissions, simulation results evidence that the intermittent ICE operation usually results in higher HC, though a local minimum can be detected considering the tradeoff between number of engine start events and catalyst light off. The results also suggest that, for future developments, further refinement of engine and catalyst thermal models would be useful. While some results are, of course, specific to the driving cycle examined, the methodology can be applied to any driving cycle. Moreover, some general trends and conclusions appear quite general and could be extrapolated to other cycles. The tendency to locate engine operation in the region of high power demand. The presence of a trade-off for the choice of the number of engine start events since a larger value allows modulating the power demand but leads to higher energy losses for cranking. The tendency to avoid too long engine-off phases that effect engine temperature decay and cause an increase of fuel consumption and HC emissions and, finally, the prospect to operate the engine not necessarily at its maximum efficiency. In conclusion, it emerges that the energy management of a series hybrid is not as straightforward as it could appear at a glance, if the main phenomena involved are properly considered. Future work will include the development of a supervisor control for the HSV prototype under development at University of Salerno.

¨ Andre´, M., Hammarstrom, U., & Reynaud, I. (1999). Driving statistics for the assessment of air pollutant emissions from road transport. INRETS report (p. 111), LTE9906. Available at /http://www.inrets.fr/infos/cost319/index.htmlS, Bron, France. Arsie, I., Di Martino, R., Rizzo, G., & Sorrentino, M. (2007a). Toward a supervisory control of a hybrid solar vehicle. In 5th IFAC Symposium on Advances in Automotive Control, August 20–22. Seascape Resort, USA. Arsie, I., Flora, R., Pianese, C., Rizzo, G., & Serra, G. (2001). A computer code for S.I. engine control and powertrain simulation. SAE 2000 Transactions–Journal of Engines, 109(3), 935–949. SAE paper 2000-01-0938. Arsie, I., Graziosi, M., Pianese, C., Rizzo, G., & Sorrentino, M. (2005). Control strategy optimization for hybrid electric vehicles via provisional load estimate. JSAE Review of Automotive Engineering, 26(3), 341–348. Arsie, I., Pianese, C., & Rizzo, G. (1999). Models for the prediction of performance and emissions in a spark ignition engine—A sequentially structured approach. SAE 1998 Transactions-Journal of Engines, 106(3), 1065–1079. SAE paper 980779. Arsie, I., Rizzo, G., & Sorrentino, M. (2007b). Optimal design and dynamic simulation of a hybrid solar vehicle. SAE 2006 Transactions—Journal of Engines, 115(3), 805–811. SAE paper 2006-01-2997. Bullis, K. (2007). GM’s new electric vehicle. Available at /http://www.technolo gyreview.com/energy/18042/?a=fS. Burch, S., Cuddy, M., Johnson, V., Markel, T., Rausen, D., Sprik S., et al. (2009). ADVISOR: Advanced vehicle simulator. Available at /http://www.avl.com/ advisorS. Burke, A. F., Hardin, J. E., & Dowgiallo, E. J. (1990). Application of ultracapacitors in electric vehicle propulsion systems. In 34th International Power Sources Symposium (pp. 328–333). 90CH2863. Cheng, W. K., & Santoso, H. (2002). Mixture preparation and hydrocarbon emissions behaviors in the first cycle of SI engine cranking. SAE paper 200201-2805. Chipperfield, A. J., Fleming, P. J., Polheim, H., & Fonseca, C. M. (2009). Genetic algorithm toolbox—Matlab tutorial. Available at /http://www.shef.ac.uk/ acse/research/ecrg/getgat.htmlS. Cundev, D., & Mindl, P. (2008). European driving schedule of hybrid electric Vehicle with electric power splitter and supercapacitor as electric storage unit. In 18th International Conference on Electrical Machines (pp. 1–5). September 6– 9, Vilamoura—Algarve, Portugal. Heywood, J. B. (1988). Internal combustion engine fundamentals. New York, USA: MC Graw Hill. Letendre, S., Perez, R., & Herig, C. (2003). Vehicle integrated PV: A clean and secure fuel for hybrid electric vehicles. American Solar Energy Society Annual Conference. Boulder, CO. Pisu, P., & Rizzoni, G. (2007). A comparative study of supervisory control strategies for hybrid electric vehicles. IEEE Transactions on Control Systems Technology, 15(3), 506–518. Powell, B. K., Bailey, K. E., & Cikanek, S. R. (1998). Dynamic modeling and control of hybrid electric vehicle powertrain systems. IEEE Control Systems Magazine, 18(5), 17–33. Preitl, Z., Bauer, P., Kulcsar, B., Rizzo, G., & Bokor, J. (2007). Control solutions for hybrid solar vehicle fuel consumption minimization. In IEEE Intelligent Vehicles Symposium (pp. 767–772). June 13–15, Istanbul, Turkey. Rodatz, P., Paganelli, G., Sciarretta, A., & Guzzella, L. (2005). Optimal power management of an experimental fuel cell/supercapacitor-powered hybrid vehicle. Control Engineering Practice, 13(1), 41–53. Sakawa, M., Kato, K., Ushiro, S., & Inaoka, M. (2001). Operation planning of district heating and cooling plants using genetic algorithms for mixed integer programming. Applied Soft Computing, 1, 139–150. Saitoh, T., Hisada, T., Gomi, C., & Maeda, C. (1992). Improvement of urban air pollution via solar-assisted super energy efficient vehicle. In ASME JSES KSES International Solar Energy Conference (pp. 571–577). Maui, HI. Sasaki, K., Yokota, M., Nagayoshi, H., & Kamisako, K. (1997). Evaluation of an electric motor and gasoline engine hybrid car using solar cells. Solar Energy Material and Solar Cells, 47(1–4), 259–263. Sciarretta, A., & Guzzella, L. (2007). Control of hybrid electric vehicles. IEEE Control Systems Magazine, 27(2), 60–70. Seal, M.R., & Campbell, G. (1995). Ground-up hybrid vehicle program at the vehicle research institute. SAE paper 951904. Seal, M. R. (1995). Viking 23-zero emissions in the city, range and performance on the freeway. Northcon—Conference Record, IEEE, 1RC-108, 264–268. Zavala, J. C., Sanketi, P. R., Wilcutts, M., Kaga, T., & Hedrick, J. K. (2007). Simplified models of engine HC emissions, exhaust temperature and catalyst temperature for automotive cold start. In 5th IFAC Symposium on Advances in Automotive Control (August 20–22). Seascape Resort, USA.

Acknowledgements This research has been financed by University of Salerno, European Commission ‘‘Education and Culture’’ within the Leonardo Program ‘‘Energy Conversion Systems and Their Environmental Impact’’ (www.dimec.unisa.it/leonardo), Italian Ministry of University and Research MIUR within the PRIN project ‘‘Integration of Photo-Voltaic Systems in Conventional and Hybrid Vehicles’’.