Solar Cars do not represent realistic alternative to ânormalâ cars, due to: ⢠Limited ... Considering the daily energy spent for driving, it emerges that solar energy can give a ... 1:4.875. Electric Motor .... Scenario 1: Rule-based strategy;. ⢠Scenario ...
Experimental validation of an heuristic strategy to optimize on-board energy management of a hybrid solar vehicle G. Coraggio, C. Pisanti, G. Rizzo, M. Sorrentino DIIN,University of Salerno
RHEVE11, Rueil-Malmaison, France 6-7 December 2011 1
Summary •
Background:
•
Rule-Based Control Strategy for HSV management:
•
Importance of sustainable mobility; Solar energy; Solar cars; Hybrid electric vehicles; Is solar contribute significant? Hybrid solar vehicles; The prototype of HSV at the University of Salerno.
Control strategy for HSV; HSV modeling; Algorithm for the real time control; Structure and validation of RB algorithm; RB analysis applied at the prototype: sensitive analysis; Implementation in LabVIEW; Experimental results.
Conclusions and future developments. Extra Slides
2
Background
3
Importance of sustainable mobility 1/3 • Growing demand for mobility. The Chindia factor, 1/3 of world population. • 400% and 205% increase in cars for China and India from 1990 2000. • The oil production will reach the peak in the next feature in the west of Europe, while it could be already falling in several countries
4
Importance of sustainable mobility 2/3 • The oil price is going to increase and is subject to large and unpredictable oscillations. • The degree of electrification is expected to grow significantly in next years in terms of fleet distribution
5
Importance of sustainable mobility 3/3 • Strong correlation between CO2 and mean temperature. • Measurements in last 150 years evidence a significant increase of both CO2 and temperature.
•
CO2 emission for transport is increased in last 30 years both in relative and absolute values and there could bedrammatic consquences. 6
Possible Solutions? Kyoto Protocol: A possible solution to fossil fuel depletion and global warming is an increased recourse to Renewable Energy (RE).
• Application to cars: • Fuels/H2/Energy from RE • Solar Cars 7
Solar energy Nuclear fusion into the sun produces an enormous amount of energy, irradiated into the space.
Solar energy is a renawable, free and largely diffused source. •
A pictorial view of the potentialities of photovoltaic: the areas defined by the dark disks could provide more than the world's total primary energy demand (assuming a conversion efficiency of 8%). 8
Photovoltaic cells • PV cost is decreasing due to a fast growing diffusion of this technology; • Their production is increasing; • PV efficiency is increasing. 9
Solar Cars Various propotypes of solar cars have been developed since 70’s, mainly for racing and demonstrative purposes
Solar Cars do not represent realistic alternative to “normal” cars, due to: • Limited power and performance. • Limited range. • Discontinuous energy source. • High cost.
10
Hybrid Electric Vehicles Buick Skylark, 1974
Toyota Prius
F.Porsche, 1900 Honda Insight
GM Precept
Ford Escape
Mercedes S400 Hybrid-Diesel Peugeot 308 Hybrid-Diesel
11
Is solar contribute significant? 1/4 Power (kW)
Car
70
PV Panel
0,3
Ratio
0,004
PV panels power is about two order of magnitudes lower than engine power.
12
Is solar contribute significant? 2/4 Power (kW)
Average Power (kW)
Car
70
10
PV Panel
0,3
0,3
0,004
0,03
Ratio
Speed [km/h] 150
PV panels power is about two order of magnitudes lower than engine power. The average power in urban driving is about one order of magnitude less than maximum power. 100
50
0
0
200
400
600
800
1000
1200
800
1000
1200
Power [KW]
Speed [km/h]
60
150
40 100
20 0
50
-20
0
-40 0
200
400
600
800
1000
1200
0
200
400
600 Time [s]
13
Is solar contribute significant? 3/4 Power (kW)
Average Power (kW)
Time (h/day)
Car
70
10
1
PV Panel
0,3
0,3
10
0,004
0,03
10
Ratio
PV panels power is about two order of magnitudes lower than engine power. The average power in urban driving is about one order of magnitude less than maximum power. In many cases, cars are used for no more than one hour per day.
(Source: Labour Force Survey, http://www.statistics.gov.uk/CCI/nscl.asp?ID=8027) 14
Is solar contribute significant? 4/4 Power (kW)
Average Power (kW)
Time (h/day)
Energy (kWh/day)
Car
70
10
1
10
PV Panel
0,3
0,3
10
3
0,004
0,03
10
0,3
Ratio
PV panels power is about two order of magnitudes lower than engine power. The average power in urban driving is about one order of magnitude less than maximum power. In many cases, cars are used for no more than one hour per day. Considering the daily energy spent for driving, it emerges that solar energy can give a substantial contribute. 15
Recent HSV Toyota Solar Prius, with an aftermarket 215 W monocristalline solar module with peak power tracking and a 95% efficiency DC-DC Converter
Astrolab Venturi 16
HSV Prototype at UniSa Vehicle Length Width Height Drive ratio Electric Motor Continuous Power Peak Power Batteries
Piaggio Porter 3.370 m 1.395 m 1.870 m 1:4.875 BRUSA MV 200 – 84 V 9 KW 15 KW 16 6V Modules Pb-Gel
Mass Capacity Photovoltaic Panels Surface Weight Efficiency
520 Kg 180 Ah Polycrystalline 1.44 m2 60 kg 0.13
Electric Generator
Diesel Yanmar S 6000
Power COP/LTP Specific fuel cons. Weight
5.67/6.92 kVA 272 g/kWh 120 kg
Overall weight (with driver) Weight
1950 kg
A prototype of hybrid solar vehicle with series structure has been developed at the University of Salerno, within the EU Leonardo Program “Energy Conversion Systems and Their Environmental Impact” (www.dimec.unisa.it/leonardo) 17
RB Control Strategy for HSV management
18
Control strategy for HSV • Two different approaches have been followed to determine the best control strategy (i.e. the enginegenerator operation): An off-line optimization with Genetic Algorithms (GA), based on previous knowledge of the driving cycle (non casual solution); it is used as a benchmark. A Rule-Based (RB) strategy, implementable online, that does not require the previous complete knowledge of the driving cycle (casual).
G Rizzo, M Sorrentino, I Arsie (2010) Rule-Based Optimization of Intermittent ICE Scheduling on a Hybrid Solar Vehicle SAE International Journal of Engines 2. 521-529 March M Sorrentino, I Arsie, R Di Martino, G Rizzo (2010) On the Use of Genetic Algorithm to Optimize the Onboard Energy Management of a Hybrid Solar Vehicle Oil & Gas Science and Technology - Revue de l'IFP 65: 1. 133-143 Jan-Feb
19
RB control strategy for HSV
HSV modeling • A dynamic model for the simulation of a HSV and of the reference conventional vehicle over a driving cycle has been used: Vehicle longitudinal model; Simulation of engine thermal transient in start and stop operation, Specific fuel consumption computed from steady-state engine maps, taking into account the effects of engine temperature.
Reference vehicle speed [km/h] 140 120 100 80 60 40 20 0
0
200
400
600 Time [s]
800
1000
1200
20
Algorithm for the real-time control
SOC Low SOC: more energy can be accumulated during parking
P [kW]
SOC SOCup dSOC
An external loop, necessary to define the final state of charge (SOCf), depending on the solar energy of the next parking phase (estimated by real-time weather forecast). This step is specific to hybrid solar vehicle. An internal loop, to obtain the power of the electric generator and dSOC, in depending on the power of the vehicle. This step can be applied to every series hybrid vehicle.
High SOC: reduced energy losses for charging and discharging and longer battery life
SOCf dSOC
• A real-time Rule-Based (RB) control strategy has been developed. The strategy is composed by two loops.
SOClo
PEG Ptr ICE-ON
ICE-OFF Time [min]
Psun
21
Structure of the RB Algorithm • •
The external loop defines the final state of charge (SOCf), depending on the solar energy of the next parking phase (i.e. solar factor); The internal loop obtains the power of the electric generator and dSOC, depending on the power of the vehicle. SOClo=SOCf+dSOC
SOCf=f(Sf )
SOCup=SOCf-dSOC P [kW] dSOC
SOC SOCup
dSOC
SOCf SOClo
PEG Ptr
Solar factor:
Sf RB control strategy for HSV
Esun, day
ICE-ON
ICE-OFF
Psun
Time [min]
Esun, day year-based average
dSOC=f(PEM) PEG=f(PEM)
22
Results – Internal loop The optimal generator power PEG and dSOC have been determined by optimization analysis, at different vehicle power Ptr in steady-state conditions.
40 35 30 25
P [kW]
The optimal generator power PEG is not equal to minimum fuel consumption conditions (red line), due to electric and battery losses and to thermal transient effects.
20 15 P
10 5 0
rule
EG
PEG,opt=21.5 kW average Ptr 0
5
10 15 20 average P [kW]
25
30
tr
dSOC rule [/]
0.055
The optimal SOC variation is also dependent on vehicle power.
0.05 0.045 0.04 0.035 0.03 0.025
0
5
10 15 20 average P [kW] tr
25
30
23
Validation of RB strategy A comparison between RB strategy and GA optimitation has been made, for various time intervals for estimating vehicle power.
% FE
FEGA FE RB 100 FEGA
% FE between RB and GA benchmark 5
FE evaluated with Backward strategy is about 3% greater than reference case (GA).
4
[%]
3 Backward best case
2 1 0 10
Backward Forward
Forward best case
15
20 th [min]
25
For the Forward strategy, in the best case fuel economy is less than 1% of the one computed with GA (22.5 km/l). RB control strategy for HSV
30
24
RB strategy applied at the prototype: main data. DRIVING CYCLE
HSV specifications Nominal ICE-EG power [kW]
gasoline
EM peak power [kW]
15
Number of Lead-acid battery modules
16
Battery capacity [kWh]
17
PV horizontal surface [m2]
1.44
PV efficiency
0.10
Weight [kg]
1900
Esun,day=4.31 kWh/m2
Driving cycle Electric motorspeed power
40 15 35
PEM [kW] [km/h] Speed
Fuel
6
30 10 25 20 5 15 10 50 0 -5 -5 00
200 200
400 600 400 600 Time Time [s] [s]
800 800
1000 1000
25
RB control strategy for HSV
Sensitivity analysis Fuel economies - Scenario 1
Fuel economies - Scenario 2
18.5 18
20
backward forward
0.25 0.3125 0.375 0.4375 0.5 0.5625 0.625 0.6875 0.75
18 [km/l]
[km/l]
17.5 17 16.5
14 12
16 15.5
16
0.8
1 1.2 Sun factor [\]
1.4
10 0.7
0.9
1.1 Sun factor[\]
1.3
1.5
Fuel economies - Scenario 3 19 0.01 0.0113 0.0125 0.0138 0.015 0.0163 0.0175 0.0188 0.02
[km/l]
18
17
16
15 0.7
0.9
1.1 Sun factor[\]
1.3
1.5
• • •
Scenario 1: Rule-based strategy; Scenario 2: Parametric analysis with dSOC=f(PEM) and PEG=[0.25÷0.75]; Scenario 3: Parametric analysis with dSOC=[0.01÷0.02] and PEG=f(PEM).
26
Best cases of the scenarios • Best cases for the three scenarios 18.5 18
[km/l]
17.5
Scenario 1 Best case Scenario 2 Best case Scenario 3
•
17
16.5
•
16 15.5 0.7
0.9
1.1 Sun factor[\]
1.3
In this figure the best results of the parametric analysis and results obtained with RB strategy are compared; The proposed RB heuristic strategy performs near to the optimal conditions, without requiring previous knowledge of the driving cycle; RB was then implemented on-board into a National Instrument compact Rio (NI cRio ) platform.
1.5
27
RB control strategy for HSV
Rule BasedCompactRIO Vehicle data
MATLAB model
Fuel economy Performance
Vehicle experimental data
LabVIEW model
Fuel economy Performance
Real Time Vehicle data
LabVIEW cRIO
Fuel economy Performance
28
Implemetatio on cRIO platform • In this flow diagram the implementation of RB strategy, realized with our model (developed in LabVIEW) , is shown.
29
Experimental results 1/2 SOC [/] [/] SOC
0.712 0.712
• RB strategy has been applied at the prototype trough an algorithm developed in Labview; • These are the results for the previous cycle.
SOC SOC SOCmin SOCmin SOCmax SOCmax
0.711 0.711 0.71 0.71
5 5
1 1 0.709 0.709 2 2 0.708 0.708 0.707 0.707
2 2
6 6
7 7
4 4
8 8
9 11 9 11
6 6
7 7
10 10
0 0 3 3
Electric Electric generator generator power power [kW] [kW]
1.5 1.5
15 15
5 5
10 10
3 3
Vehicle Vehicle speed speed [km/h] [km/h]
20 20
2 2 1 1
8 8
4 4
11 11 9 9
Power Power at at wheels wheels [kW] [kW]
2 2 1 1
1 1
4 4
11 11
5 5
0 0 0.5 0.5 0 0 4 4
-1 -1
10 10
3 3
-2 -2 Average power of electric motor [kW] Average power of electric motor [kW] 2.5 2.5
Electric Electric motor motor power power [kW] [kW]
3 3
2 2
2 2
1.5 1.5
1 1
1 1
0 0
0.5 0.5
-1 -1
0 0
-2 -20 0
100 100
200 300 200 300 Time Time [s] [s]
400 400
500 500
-0.5 -0.50 0
100 100
200 300 200 300 Time Time [s] [s]
400 400
30 500 500
Experimental results 2/2 SOC [/]
Vehicle speed [km/h]
0.712 SOC SOCmin SOCmax
0.711 0.71 0.709
5
6
7
1
0.708 0.707
8 2
9 11
20 15
4 5
10
3
0
Electric generator power [kW]
7
10
2
8
4
11 9
1 Power at wheels [kW]
3
2
6
2
1.5
1
4
11
5
1
0 0.5
0
-1
10
3
-2
Electric motor power [kW]
Average power of electric motor [kW] 2.5
4 3
2
2
1.5
1
1
0
0.5
-1
0
-2 0
100
200 300 Time [s]
400
500
-0.5 0
100
200 300 Time [s]
400
500
1. In this section HSV’s speed and power traction are zero, so there is no variation of SOC. 2. There is a strong SOC decrease due to a rapid acceleration from 0 to 10 km/h. 3. The value of SOC decreases until the value of SOCmin (point 3), as imposed by the RB strategy. 4. Despite the ICE is on, there are two little battery discharges due to two strong accelerations. 5. There is a linear increase of SOC, as a consequence of the almost constant value of traction power. 6. The value of SOC grows till the SOCmax (point 6) value imposed by the RB strategy. 7. There is a rapid SOC decrease due to the high acceleration. 8. The ICE is off but there is a brief recharge. This happens because there is a strong braking (regenerative braking) as it is possible to see in the speed plot. 9. The value of SOC decreases again with a strong gradient due to high acceleration. 10. The value of SOC decreases until another value of SOCmin (point 10) that is imposed by the RB strategy. 11. The speed is almost low and constant, so the recharge of the battery is slow and linear. 12. It is clear, from the trend of SOC and its thresholds, that after a few seconds before the end of the cycle, the ICE shuts down. In fact after a few seconds the value of SOC
31
Conclusions And Future Developments
32
Conclusions •
•
• • •
•
Due to the growing mobility, researchers are studying an alternative to traditional cars. Here are presented the application of the RB methodology on a prototype of series hybrid solar vehicle developed at the University of Salerno. The strategy has been first successfully validated via simulation analysis against GA optimization. Through a sensitive analysis RB strategy has been validated. The RB strategy has been then finalized to the prototype through the development of a LabView algorithm, which was then implemented onto a NI cRio platform. Thanks to preliminary experimental tests that highlight the high potential of the developed experimental test-bench, now it is possible to improve this strategy. 33
Future developments • In the next future we are going to extend the numerical analyses to other driving cycles and/or HSV architecture. • We are going also to test the correspondence of real fuel consumption measured on the vehicle and the ones obtained by a simulator developed in LabVIEW.
34
Extra Slides
35
Longitudinal model • Longitudinal model of the HSV protoype: Pw
PEM
M HSV g v C r cos
sin
Pw
dv v dt
0.5 C x A v 3
M eff
if
Pw
0
if
Pw
0
if
Pw
0
if
Pw
0
tr
PEM PEM
PB
PEG EM Pw
PEM
tr
AC / DC
PB
PPV
EM
PEG
AC / DC
PPV
= experimentally characterized
Power at wheels
EM Power
Battery recharge power 36
ICE thermal transients
T t
Tss
Tin Tss e
t K
Specific Fuel Consumption and power are related to the ratio between actual temperature and its steady state value, starting from experimental data for a SI engine SFC t
SFCss P T f eng Tss
Engine temperature 90 80 70
T [°C]
Engine temperature dynamics is estimated by a first order dynamic model
60 50 40 30 20
N=1 N=4 0
1000
2000 3000 Time [s]
4000
5000
Teng
f
Pt
Pss f
1
2
e
Tss
3
Teng Tss
37
Off-line approach: the choice of optimization method If most cases, Dynamic Programming (DP) is adopted for off-line optimal energy management in hybrid vehicles, with steady state hypothesis for thermal engine. • The use of DP has not been considered for this problem since: Due to the presence of thermal transients, the dimensionality of the DP problem would be increased. The asymptotic behavior with engine temperature would represent a problem for backward computation involved in DP.
As both integer and real variables are involved (Mixed Integer Programming), Genetic Algorithms (GA) have been used to perform offline optimization of ICE-EG scheduling. 38
Why is series HEV appealing? • PV Panels
VMU ICE
EG
EM
Battery
•
PROS ICE may be designed and optimized for steady conditions Less constraints for vehicle layout No mechanical links Acts as a bridge towards the introduction of fuel cell powertrains. More suitable for stationary domestic use (cogeneration)
CONS – More inefficiencies on the energy line – Weight
But the features of electrical components improved significantly in the last years! 39