Professional Documents
Culture Documents
Energy
journal homepage: www.elsevier.com/locate/energy
a r t i c l e i n f o a b s t r a c t
Article history: With the growing desire for “lucid waters and lush mountains”, energy savings and emission reductions
Received 14 June 2018 of automobiles have received unprecedented attention. As a result, how to achieve economical driving is
Received in revised form also a focus of current research. The engine dynamic fuel consumption model is the basis for research on
8 November 2018
automobile fuel economy. To overcome the deficiency of the original transient fuel consumption models
Accepted 11 December 2018
based on “steady-state prediction plus transient correction”, a new model called BIT-TFCM-3 was
Available online 11 December 2018
developed. The new model was verified using measured fuel consumption data from Argonne National
Laboratory. The results show that, compared with the original models, the new model not only has a
Keywords:
Energy savings
higher computing speed but also a significantly improved prediction accuracy.
Emission reductions © 2018 Elsevier Ltd. All rights reserved.
Economical driving
Fuel consumption model
Model optimization
https://doi.org/10.1016/j.energy.2018.12.067
0360-5442/© 2018 Elsevier Ltd. All rights reserved.
H. Guang, H. Jin / Energy 169 (2019) 508e514 509
Table 1
Fuel consumption models based on the engine power or torque.
where m_ f (mg/s) is the fuel consumption rate; gi is the model regression coefficient, i ¼ 0; 1;/;5; P(kW) is the instantaneous engine power; and ne (rpm) and Te (Nm) are the
engine speed and torque, respectively.
that under steady-state conditions [10e15]. This is the reason why The transient fuel consumption model developed in this paper
the transient model can estimate the fuel consumption of the consists of the steady-state module and transient module. The in-
vehicle more accurately under various operating conditions puts of the steady-state module are the engine speed and torque,
compared with the steady-state model. The modelling method and the inputs of the transient module are the vehicle speed and
includes transient-correction-based modelling and direct dynamic- acceleration. The steady-state module is similar to the steady-state
variables-based modelling. The steady-state module of the fuel model which can accurately predict the fuel consumption of the
consumption model established by the first method is a steady- vehicle under steady-state conditions. However, the prediction will
state fuel consumption map of the engine that uses the engine have a large deviation under unsteady conditions. Therefore, the
speed and torque as inputs. The inputs of the transient module are transient module is introduced to make the model accurately
the air-fuel ratio or the transients in the engine speed and torque; predict the fuel consumption under various working conditions.
such transients are difficult to measure. The second method is to fit Since the engine torque is different when the vehicle is running at
fuel consumption as a function of vehicle speed and acceleration. the same speed and acceleration on a level road and a sloped road,
The representative models of the two methods and transient var- the steady-state module is indispensable. So the two modules
iables are shown in Table 2. complement each other. It was verified that the developed model
The VT-Micro model developed by Ahn and Pakha predicts can overcome the deficiency of the existing models.
instantaneous fuel consumption based on the instantaneous speed
and acceleration [16]. The combination of speed and acceleration
with different powers is used to fit the vehicle's instantaneous fuel 2. Transient fuel consumption model
consumption. The model takes the positive and negative intervals
of the acceleration into consideration and uses a logarithm function Based on our previous research, the new transient fuel con-
to ensure that the output is positive. Wei Lei replaced the acceler- sumption model is developed based on BIT-TFCM-1 [20] and BIT-
ation in the VT-Micro model with a composite acceleration in 9 s to TFCM-2 [21], whose basic structure is “steady-state estimation
develop the MEF model [17]. To date, many people, such as Tang T Q plus transient correction”. The modelling data are derived from the
[18] and Llopis-Castello D [19], still use the VT-Micro model. Downloadable Dynamometer Database (D3) in the Advanced
Powertrain Research Facility (APRF) at Argonne National Laboratory
Table 2
Transient models and dynamic variables.
Transient-correction-based Chiara F et al. [13] Engine fuel Map þ start-up & fuel-to-air ratio limitation correction m_ air
Lindgren M. [14] Engine fuel Map þ engine speed and torque change correction DT; Du
Pelkmans L et al. [15] Engine fuel Map þ torque change correction DT
8
Direct Dynamic-variable-based Ahn K et al. [16] >
> X3 X 3 a
>
> ðLi;j vi aj Þ a0
>
<
i¼0 j¼0
lnðMOEe Þ ¼
>
> X3 X 3
>
> ðMi;j vi aj Þ a < 0
>
:
i¼0 j¼0
8
Lei W et al. [17] >
> X3 X 3 a
>
> ðLi;j vi aj Þ a0
>
<
i¼0 j¼0
lnðMOEe Þ ¼
>
> X3 X 3
>
> ðMi;j vi aj Þ a < 0
>
:
i¼0 j¼0
X
9
aðtÞ ¼ a,aðtÞ þ ð1 aÞ aðt iÞ=9
i¼1
where “Engine fuel Map” represents the traditional engine fuel map; m_ air (g/s) is the transient air mass flow rate; DT(Nm/s) and Du(rad/s2) are the transients in engine torque
and speed, respectively; MOEe is the instantaneous fuel consumption rate (mg/s); Li;j and Mi;j are the model regression coefficients; v(km/h) and a(m/s2) are the vehicle
instantaneous speed and acceleration, respectively; aðtÞ(m/s2) is the composite acceleration at time t(s); and a is the acceleration impact factor.
510 H. Guang, H. Jin / Energy 169 (2019) 508e514
(ANL) [22]. The data are collected under the steady-state cycle, the
UDDS cycle, the Highway cycle and the US06 cycle, as shown in where m_ c (cc/s) is the transient fuel consumption correction; gi is
Fig. 1. As depicted in Fig. 1, the vehicles mainly run at medium and the model regression coefficient, i ¼ 1; 2; /; 9; and a(m/s2) and
low speeds under UDDS cycle condition and at high speed under v(km/h) are the vehicle acceleration and speed, respectively.
US06 cycle, and both of the two cycles have rapid acceleration and
rapid deceleration conditions. Therefore, combining the data from
2.2. BIT-TFCM-2 transient fuel consumption model
UDDS driving cycle and US06 driving cycle can take the fuel con-
sumption characteristics at high and low speed into account. And it
The steady-state module still uses steady-state cycle data. A
is helpful to reduce the prediction error. Then, all data were divided
polynomial is used to fit steady-state fuel consumption with the
into two groups: one for model development (UDDS&US06 cycles
engine speed and torque as the inputs. The transient module
(D)) and the other for model validation (UDDS&US06 cycles (V)).
modelling data are from the UDDS&US06 cycles (D). The final
structure is given by equation (3):
2.1. BIT-TFCM-1 transient fuel consumption model 8
>
> ms ¼ g0 þ g1 Te þ g2 ne þ g3 Te 2 þ g4 ne Te þ g5 Te 3 þ g6 ne T 2 e
>
>
< X3 X 1
The steady-state module is a traditional engine map based on mc ¼ bi;j ai vj
fuel consumption data from the steady-state cycle. As shown in >
>
>
> i¼1 j¼0
: _
Fig. 1, the vehicle steady speed of the steady-state cycle ranged from mf ¼ eðms þmc Þ
15 km/h to 130 km/h. The inputs of the module are the engine
speed and torque, and the vehicle's fuel consumption under steady- (3)
state working condition can be calculated using the method of 2D
where ms (cc/s) is steady-state fuel consumption; mc is the multiple
interpolation.
ratio of logarithmic fuel consumption; m_ f (cc/s) is the instantaneous
The inputs of the transient correction module are the instanta-
fuel consumption rate of vehicles.
neous speed and acceleration of the vehicle. This module can
further correct the predictions given by the steady-state module
based on the Vehicle Specific Power (VSP) partition. VSP was pro- 2.3. Verifications and shortcomings of the models
posed by Jime nez-Palacios to reflect the relationship between
vehicle output power and fuel consumption [23].The corrected The two models were verified under a variety of working con-
structure is shown in equation (2): ditions; the verification results under the US06 Driving cycle (V) are
8
>
< g1 þ g2 v þ g3 v3 þ g4 þ g5 v þ g6 v3 a þ g7 þ g8 v þ g9 v3 a2 VSP 0
m_ c ¼ 0:3984 VSP < 0kW=t&v 50km=h (2)
>
:
0:0038 VSP < 0kW=t&v < 50km=h
H. Guang, H. Jin / Energy 169 (2019) 508e514 511
Table 3
Operation time under the US06 cycle and Highway cycle.
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
Xnv .
RMSE ¼ m _ p ðiÞ m_ m ðiÞ 2 nv (7)
i¼1
5. Conclusion
new vehicle control strategy, such as minimum-energy speed VSP Vehicle Specific Power
profile optimization for intelligent vehicles, etc. However, the fuel VT-Micro Virginia Tech Microscopic model
consumption of the vehicles is also affected by the actual road
conditions, traffic conditions and driving behaviors, the modelling
data is the typical driving cycle data from the chassis dynamom- References
eter. Therefore, to make sure that the new model can be applied to
complex and varied driving conditions, the real vehicle tests [1] World oil outlook. 2017. https://www.opec.org/opec_web/static_files_project/
media/downloads/publications/WOO_2017.pdf.
should be conducted to further verify the model accuracy in the
[2] Van Mierlo J, Maggetto G, Van de Burgwal E, et al. Driving style and traffic
future. measures - influence on vehicle emissions and fuel consumption. Proc Inst
Mech Eng - Part D J Automob Eng 2004;218(1):43e50.
Acknowledgements [3] Yue H. Mesoscopic fuel consumption and emission modeling. Virginia Poly-
technic Institute and State University; 2008.
[4] Post K, Kent JH, Tomlin J, et al. Fuel consumption and emission modelling by
This work was supported by the National Natural Science power demand and a comparison with other models. Transport Res Gen
Foundation of China (Grant No. 51875040). 1984;18(3):191e213.
[5] Chang DJ, Morlok EK. Vehicle speed profiles to minimize work and fuel con-
sumption. J Transport Eng 2005;131(3):173e82.
Nomenclature [6] Ahn K, Rakha HA, Moran K. A simple fuel consumption model based on
instantaneous vehicle power. In: Proceedings of the transportation research
board 89th annual meeting; 2010. p. 10e4.
a Vehicle instantaneous acceleration [m/s2] [7] Rakha HA, Ahn K, Moran K, et al. Virginia tech comprehensive power-based
aðtÞ Composite acceleration at time t [m/s2] fuel consumption model: model development and testing. Transport Res
I Interval of corresponding point set Transport Environ 2011;16(7):492e503.
[8] Saerens B, Rakha H, Ahn K, et al. Assessment of alternative polynomial fuel
l Interval length consumption models for use in intelligent transportation systems applica-
Li;j , Mi;j Model regression coefficients tions. J Intell Transport Syst 2013;17(4):294e303.
m_ air Transient air mass flow rate [g/s] [9] Passenberg B, Kock P, Stursberg O. Combined time and fuel optimal driving of
trucks based on a hybrid model. In: Proceedings of the 10th European control
m_ c Transient fuel consumption correction [cc/s]
conference, 2009, 4955e4960.
mc Multiple ratio of logarithmic fuel consumption [cc/s] [10] Lindgren M, Hansson PA. Effects of transient conditions on exhaust emissions
m_ f Fuel consumption rate [mg/s] from two non-road diesel engines. Biosyst Eng 2004;87(1):57e66.
[11] Callahan T, Ryan T, Dietzmann H, et al. The effects of discrete transient in
m_ fm Fuel consumption rate [ml/min] speed and load on diesel engine exhaust emissions. SAE Transactions Paper,
m_ m ðiÞ ith measured fuel consumption [cc/s] 1985.
[12] Hansson PA, Lindgren M, Nordin M, et al. A methodology for measuring the
m_ p ðiÞ ith predicted fuel consumption [cc/s]
effects of transient loads on the fuel efficiency of agricultural tractors. Appl
ms Steady-state fuel consumption [cc/s] Eng Agric 2003;19(3):251e7.
MOEe Instantaneous fuel consumption rate [mg/s] [13] Chiara F, Wang J, Patil CB, et al. Development and experimental validation of a
control-oriented Diesel engine model for fuel consumption and brake torque
ne Engine speed [rpm]
predictions. Math Comput Model Dyn Syst 2011;17(3):261e77.
nv Total number of samples used for model validation [14] Lindgren M. A transient fuel consumption model for non-road mobile ma-
P Instantaneous engine power [kW] chinery [J]. Biosyst Eng 2005;91(2):139e47.
q Overlap ratio [15] Pelkmans L, Debal P, Hood T, et al. Development of a simulation tool to
calculate fuel consumption and emissions of vehicles operating in dynamic
S Point set conditions. SAE Technical Paper, 2004.
Te Engine torque [Nm] [16] Ahn K, Rakha H, Trani A, et al. Estimating vehicle fuel consumption and
DT Engine torque transient [Nm/s] emissions based on instantaneous speed and acceleration levels. J Transport
Eng 2002;128(2):182e90.
v Vehicle instantaneous speed [km/h] [17] Lei W, Chen H, Lu L. Microscopic emission and fuel consumption modeling for
light-duty vehicles using portable emission measurement system data. World
Greek Acad Sci Eng Technol 2010;66:918e25.
[18] Tang TQ, Wang T, Chen L, et al. Analysis of the trip costs of a traffic corridor
a Acceleration impact factor with two entrances and one exit under car-following model. Phys A Stat Mech
b Parameter related to the vehicle [ml/(min$kW)] Appl 2017;486:720e9.
l Parameter related to the vehicle [ml/min] [19] Llopis-Castello D, Pe
rez-Zuriaga AM, Camacho-Torregrosa FJ, et al. Impact of
horizontal geometric design of two-lane rural roads on vehicle co 2 emissions.
g Model regression coefficient Transport Res Transport Environ 2018;59:46e57.
Du Engine speed transient [rad/s2] [20] Zhou M, Jin H. Development of a transient fuel consumption model. Transport
Res Transport Environ 2017;51:82e93.
[21] Jin H, Zhou M, Li SHJ. Transient fuel consumption modelling based on tran-
Abbreviations
sient correction. Trans Beijing Inst Technol 2017;37(05):473e477þ484.
BIT-TFCM Transient Fuel Consumption Model of Beijing Institute [22] Argonne National Laboratory. Transportation Technology R&D Center.
of Technology Downloadable Dynamometer Database. http://www.transportation.anl.gov/
MAPE Mean Absolute Percentage Error D3/.
[23] Jimenez-Palacios JL. Understanding and quantifying motor vehicle emissions
MEF Microscopic Emission and Fuel consumption model with vehicle specific power and TILDAS remote sensing. Massachusetts
RMSE Root Mean Square Error Institute of Technology; 1998.